Structural basis of the signal transduction via transmembrane domain of the human growth hormone receptor

BBA - General Subjects 1862 (2018) 1410–1420

Contents lists available at ScienceDirect

BBA - General Subjects journal homepage: www.elsevier.com/locate/bbagen

Structural basis of the signal transduction via transmembrane domain of the human growth hormone receptor

T

⁎

Eduard V. Bocharova, , Dmitry M. Lesovoya, Olga V. Bocharovaa, Anatoly S. Urbana, Konstantin V. Pavlovb, Pavel E. Volynskya, Roman G. Efremova,c,d, Alexander S. Arsenieva a

Department of Structural Biology, Shemyakin–Ovchinnikov Institute of Bioorganic Chemistry RAS, str. Miklukho-Maklaya 16/10, Moscow 117997, Russian Federation Federal Clinical Center of Physical-Chemical Medicine of FMBA, str. Malaya Pirogovskaya 1a, Moscow 119435, Russian Federation c Higher School of Economics, 20 Myasnitskaya, Moscow 101000, Russian Federation d Moscow Institute of Physics and Technology (State University), 9 Institutskiy per., Dolgoprudny, Moscow Region, 141700, Russian Federation b

A R T I C LE I N FO

A B S T R A C T

Keywords: Growth hormone receptor Transmembrane domain Alternative dimerization Spatial structure Protein-lipid interactions NMR

Background: Prior studies of the human growth hormone receptor (GHR) revealed a distinct role of spatial rearrangements of its dimeric transmembrane domain in signal transduction across membrane. Detailed structural information obtained in the present study allowed elucidating the bases of such rearrangement and provided novel insights into receptor functioning. Methods: We investigated the dimerization of recombinant TMD fragment GHR254–294 by means of high-resolution NMR in DPC micelles and molecular dynamics in explicit POPC membrane. Results: We resolved two distinct dimeric structures of GHR TMD coexisting in membrane-mimicking micellar environment and providing left- and right-handed helix-helix association via different dimerization motifs. Based on the available mutagenesis data, the conformations correspond to the dormant and active receptor states and are distinguished by cis-trans isomerization of Phe-Pro266 bond in the transmembrane helix entry. Molecular dynamic relaxations of the structures in lipid bilayer revealed the role of the proline residue in functionally significant rearrangements of the adjacent juxtamembrane region supporting alternation between protein-protein and protein-lipid interactions of this region that can be triggered by ligand binding. Also, the importance of juxtamembrane SeS bonding for signal persistency, and somewhat unusual aspects of transmembrane region interaction with water molecules were demonstrated. Conclusions: Two alternative dimeric structures of GHR TMD attributed to dormant and active receptor states interchange via allosteric rearrangements of transmembrane helices and extracellular juxtamembrane regions that support coordination between protein-protein and protein-lipid interactions. General significance: This study provides a holistic vision of GHR signal transduction across the membrane emphasizing the role of protein-lipid interactions.

1. Introduction The human growth hormone receptor (GHR), as a type I cytokine receptor utilizing Janus kinase 2 (JAK2) for signaling, plays a critical role in a wide range of cellular growth and differentiation pathways [1,2]. It is closely related to receptors for erythropoietin, prolactin, interleukin, thrombopoietin, leptin, and ciliary neurotrophic factor among others [2]. Dysregulated signaling from these receptors has been

shown to play significant roles in promotion of a number of human diseases [3–5]. Hormone binding to the dormant GHR, residing in the cell membrane as pre-formed inactive dimer, results in formation of active asymmetric 1:2 hormone-receptor ternary complex that leads to conformational changes in the receptor and the associated dimeric JAK2 kinase, including realignment of their domains [6–8]. The kinases cross-phosphorylate themselves and phosphorylate the receptor to initiate complex networks of intracellular signaling pathways [1,2].

Abbreviations: GHR, growth hormone receptor; GHRtm, recombinant fragment GHR254–294; JAK2, Janus kinase 2; RTK, receptor tyrosine kinase; TMD, transmembrane domain; ECD, extracellular domain; ICD, intracellular domain; JM, juxtamembrane; ecJM, extracellular juxtamembrane region; LID, lipid interaction domain; NMR, nuclear magnetic resonance; TROSY, transverse relaxation optimized spectroscopy; HSQC, heteronuclear single quantum coherence spectroscopy; HSQC-CT, constant-time version of HSQC experiment; NOE, nuclear Overhauser effect; τR, effective rotation correlation time; MD, molecular dynamics; MHP, molecular hydrophobicity potential; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine; DPC, n-dodecylphosphocholine; D/P, detergent/peptide molar ratio ⁎ Corresponding author at: Dept. of Structural Biology, IBCh RAS, Str. Mikluho-Maklaya, 16/10, Moscow 117997, Russian Federation. E-mail address: [email protected] (E.V. Bocharov). https://doi.org/10.1016/j.bbagen.2018.03.022 Received 24 September 2017; Received in revised form 13 March 2018; Accepted 19 March 2018 Available online 21 March 2018 0304-4165/ © 2018 Elsevier B.V. All rights reserved.

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

pH 7.3 and within one week acquired a series of 1He15N TROSY spectra. The 6 mM tris(2-carboxyethyl)phosphine was used as a reducing agent to break the inter-molecular SeS bridge between the Nterminal C259 residues (inhering to the ecJM sequence). The SeS bridge formation or distortion was monitored by an appearance of the characteristic signals of the cysteine CβH2 group in the 1H/13C-HSQC spectra. In order to achieve homogenization of the sample, several freeze–thaw cycles were carried out, followed by sonication. The sample was placed in Shigemi NMR tubes with glass plunger. The 15 N/13С-labeled sample of 0.5 mM P266A-GHRtm was prepared at the detergent/peptide molar ratio (D/P) of 80 and 190. For the dimeric GHRtm structure determination two NMR samples of GHRtm were prepared at D/P of ~70: 15N/13С-labeled sample (0.7 mM) and a 2:1 mixture of unlabeled and 15N/13С-labeled peptide (1.9 mM in the sum) referred to as “isotopic-heterodimer” sample [14]. The self-association of GHRtm was studied under variation of D/P within the range of 70 to 6000. NMR spectra were acquired at 313 K on 600 and 800 MHz AVANCE III spectrometers (Bruker BioSpin, Germany) equipped with pulsed-field gradient triple-resonance cryoprobes. The 1H, 13C, and 15N chemical shifts of GHRtm were assigned with the CARA software [15] using twoand three-dimensional heteronuclear experiments [16]: 1H/15N-TROSY (transverse relaxation optimized spectroscopy), 1H/15N-HSQC (heteronuclear single quantum coherence spectroscopy), 1H/13C-HSQC, carbon detected 2D 1H/13C-HETCOR, 1H/15N-HNHA, 1H/15N-HNHB, 1 1 1 H/13C/15N-HNCA, H/13C/15N-HN(CO)CA, H/13C/15N-HNCO, 1 13 15 1 13 13 H/ C/ N-HN(CA)CO, H/ С-HСCH-TOCSY, C- and 15N-edited NOESY-HSQC/TROSY. The backbone resonances were assigned using the BEST-TROSY version of triple resonance experiments [17]. Spectra were recorded with non-uniform sampling of indirect dimensions and processed using the qMDD software [18]. Constant-time versions of 1 H/13C-HSQC-CT, 1H/13C-HCCH-TOCSY-CT and 1H/13C-NOESY-HSQCCT with evolution period of 28.6 ms were utilized [19] to assign the sidechain resonances. The intra- and inter-monomeric NOE (nuclear Overhauser effect) distance restraints for dimeric GHRtm were derived through the analysis of three-dimensional 13C- and 15N-edited NOESYHSQC/TROSY and 13C-F1-filtered/F3-edited-NOESY spectra (mixing times were 60 ms) [20] acquired at D/P of 70. The backbone (φ and ψ) and sidechain (χ1) torsion angle restraints were estimated from the chemical shift values using the web-based programs TALOS-N [21] and SHIFTOR [22], respectively. In order to characterize the intra-molecular dynamics, the effective rotation correlation times (τR) were estimated for individual amide groups of GHRtm based on 15N CSA/dipolar cross-correlated transverse relaxation experiment [23]. The spatial structure of the GHRtm dimer was calculated with the CYANA-3.97 software [24] based on proton-proton NOE connectivities, torsion angle restraints and hydrogen bond restrains for stable (based on high τR values) helix region, as described in [25]. Prior to that, after preliminary structure calculations of TMD subunits, the hydrogen bond restraints were imposed on the α-helical region covering residues 267–288 (amide NH groups 271–288 (donors) with appropriate i-4 CO acceptors) based on the collected structure-dynamic information. Standard rotameric constraints from CYANA software and Ramachandran constraints were used for flexible juxtamembrane regions. The two alternative dimer structures were solved from a single experimental data set, and discrimination between the two sets of intermolecular NOEs related to the two dominant dimeric states relied on the following observations: (i) > 92% of the protein was in dimeric state; (ii) overall intermolecular contacts pattern covered the opposing sides of the TMD helix, which would have been impossible assuming only one dimerization mode; and (iii) some signals were clearly from major cross-peak components of the 1H/13C-HSQC-CT spectrum, whereas a few were attributable exclusively to minor components (e.g., the key signal from Cγ2H3 T279 residue) – importantly, the signals attributed to the major and minor components were from the residues on the opposite sides of the TMD helix. The chemical shift assignments, atomic coordinates and

The GHR is composed of an extracellular domain (ECD) and an intracellular domain (ICD) connected by a single-pass helical transmembrane domain (TMD) via flexible juxtamembrane (JM) regions. As reported in [2], the ECD (residues 19–262, the numbering is according to UniProt ID P10912) consists of two subdomains joined through a hinge region: the N-terminal part of the ECD is involved in ligand binding, whereas the membrane-proximal part containing the dimerization motif has a structural support function. The flexible extracellular JM region (ecJM, residues 251–262) has a C259 cysteine residue (not conserved among type I cytokine receptors), which can reportedly form an inter-monomeric disulfide bond upon hormone binding [8,9]. The following GHR TMD (residues 263–288) is critical for ligand-independent receptor association and is presumably capable of switching between two dimeric configurations corresponding to the inactive basal and the ligand-induced receptor states [1,2]. The ICD (residues 289–638) is rather flexible and has multiple functional conformational states with easy transitions between them, as typical for intrinsically disordered regions [10–12]. The GHR ICD contains a number of lipid interaction domains (LIDs), which can associate with and/or submerge into the cytoplasmic membrane leaflet [11,12]. The LID1 adjacent to TMD is transiently folded into α-helical structure (populated between 10% and 30%) [11,12] and includes functionally important proline-containing Box1 motif (residues 298–304), which, together with the less-conserved distal Box2 sequence of aromatic and acidic residues, serves to bind the dimeric JAK2 kinase in close proximity to the inner side of the cell membrane [2,11,12]. An overall mechanistic model of signal transduction by GHR was recently proposed by Brooks et al. [8] based on integration of crystallographic, fluorescence resonance energy transfer, mutagenesis and modeling data. The proposed mechanism suggests a distinct role of TMD rearrangements in signal transduction; however, no detailed structural information was available to date to elucidate the bases of such rearrangement. The GHR TMD structure obtained by us in the membrane-mimicking micellar environment was indicative of the presence of two distinct TMD dimer configurations associated with different arrangements of extracellular JM (ecJM) regions and possibly transient interactions of TMD with water molecules. This information augmented by computational modeling generally supports the proposed receptor activation mechanism, providing additional insights into the nature of the atomistic drivers of the processes occurring inside the membrane. 2. Materials and methods 2.1. Nuclear magnetic resonance (NMR) spectroscopy of GHR TMD in micellar environment The 15N- and 15N/13С-labeled and unlabeled samples of a recombinant 43-residue peptide GHRtm (GSM254SQFTCEEDFYFPWL LIIIFGIFGLTVMLFVFLFSKQQRIK294, including fragment GHR254–294) and its mutant P266A form (P266A-GHRtm) were produced in continuous exchange cell-free system using Escherichia coli extract as described earlier [13] and solubilized in an aqueous suspension of d38dodecylphosphocholine (d38-DPC, 98%, CIL) micelles. (The sequence numbering corresponds to the Swiss-Prot annotation of the human GHR, P10912.) The peptide powder was first dissolved in 2:1 (v/v) trifluoroethanol–water mixture with the addition of DPC, placed for several minutes in an ultrasound bath and then lyophilized overnight. After that, the dried GHRtm were dissolved at pH 5.0 in 300 μl of buffer solution containing 10 mM citric acid, 20 mM Na2HPO4, 0.3 mM sodium azide, and 5% D2O (v/v). We used samples at pH 5.0 for structuredynamic studies by NMR methods, since in the absence of histidine residues there were no ionogenic groups with pKa in the range of ~5 to ~8, whereas the exchange of solvent-exposed amide groups with water decelerated notably at such a low pH value compared to neutral conditions. However, for measuring the exchange rates between water and HN protons we reconstituted lyophilized GHRtm sample in D2O at 1411

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

(caption on next page)

1412

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

Fig. 1. 1H/15N NMR spectra of GHRtm in a membrane-mimicking environment of DPC micelles. (A) 1H/15N-TROSY NMR spectrum acquired at 313 K is presented for dimeric GHRtm embedded into the DPC micelles at D/P of 70, pH 5.0. The fraction of the monomeric GHRtm is ~8% (see also Supplementary Fig. S1). The 1He15N resonance assignments of the GHRtm amino groups are shown. Distinct cross-peaks (non-overlapped) identified for the minor dimeric conformation of GHRtm are indicated by blue arrows. Two inner boxes surround the cross-peak regions of the sidechain indole group of W267 (solid line) and backbone amide group of G274 and G277 (dashed line). Similar regions are shown at the bottom with overlaid NMR spectra of (B) GHRtm at D/P of 70 and 310 (in gray and red, respectively), (C) unlinked and SeS bonded GHRtm at D/P of 70 (in gray and orange, respectively) and (D) mutant P266A form of GHRtm at D/P of 80 and 190 (in magenta and blue, respectively). In the boxes, the cross-peaks corresponding to cis-trans isomerization of the FeP266 peptide bond and monomer-dimer transition of GHRtm are marked by ‘c’, ‘t’ and ‘M’, ‘D’, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

barostat with 0.5 and 10 ps relaxation parameters, respectively, and a compressibility of 4.5 × 10−5 bar−1 for the barostat. Temperature of protein, lipids and water molecules were coupled separately. Analysis of conformational dynamics of a protein and its van-der-Walls contacts with lipid and water molecules was done using utilities from the GROMACS package. In order to map the protein-protein, protein-lipid and protein-water contacts, the number of direct van-der-Waals contacts between atoms with 4 Å distance cut-off were estimated.

structure factors of the left- and right-handed configurations of the GHRtm dimer have been deposited in the Protein Data Bank (http:// www.rcsb.org/) under accession codes PDB ID: 5OEK and 5OHD. The resultant GHRtm dimer structure was analyzed and visualized with MOLMOL [26] and PYMOL (Schrödinger, LLC). Surface MHP distributions for TMD helical segments were calculated as described in [27] using the PREDDIMER software [28]. 2.2. Molecular dynamics (MD) of GHR TMD in explicit lipid bilayer

3. Results In order to the estimate dependence of the protein behavior on its starting structure, NMR models obtained in the micellar environment were further subjected to MD-relaxation in the explicit POPC bilayer. This was done for the monomeric, left- and right-handed dimeric conformations of the GHR TMD models (GHR254–294 fragment) using GROMACS 5.1.4 package [29] and Amber99SB-ILDN force-field [30] with TIP3P water model [31] and lipid parameters as described elsewhere [32]. Such a combination of force fields proved to be effective for simulation of different proteins in various lipid bilayers [33–35]. Leftand right-handed variants of the GHR TMD dimer models having cis- or trans- configurations of the FeP266 peptide bond as well as those with and without the SeS bridge between the ecJM C259 residues were tested. MD-relaxations of GHRtm monomeric forms derived from the NMR structures of the dimers were also performed. Charged forms of the residues Glu, Asp, Lys, and Arg were assumed, with the electric charge totaling at zero. Therefore, no buffer ions were modeled, which can be expected to result in the electrostatic interactions being somewhat overestimated. The initial configurations of the simulated systems were obtained by inserting the NMR-derived dimer models into a preequilibrated lipid bilayer comprised of 200 1-palmitoyl-2-oleoyl-snglycero-3-phosphocholine (POPC) molecules. Then the systems were solvated, and water molecules from the bilayer interior were removed; the system was energy minimized and subjected to 4 consequent steps of MD equilibration with restrained positions of protein heavy atoms. In the first two MD runs, Berendsen barostat was used. The length of these runs was 100,000 steps with a time step of 1 fs in the first run and 2 fs in the second one. In the next two runs, the Parrinello-Rahman barostat was used. Their lengths were 500,000 steps and the time steps were 1 fs and 2 fs. This relaxation allows removing cavities in the starting systems, which appeared after their generation, and excluding water molecules from the hydrophobic membrane interior. Finally, MD production runs of 100 ns were carried out for all the systems using a 2 fs time step. The simulations time is relatively small to describe long-time processes, which occur in the system, such as lipid redistribution [36–38] – for the latter, at least microsecond timescale is needed. On the other hand, because the goal of MD simulations was to estimate the dependence of protein behavior on the starting structure, the computational protocol seems to be adequate. The spherical cut-off function (15 Å) was used for truncation of van-der-Waals interactions, while electrostatic interactions were treated using the particle-mesh Ewald summation (real space cut-off 15 Å and 1.2 Å grid with fourth-order spline interpolation). MD simulations were carried out with imposed 3D periodic boundary conditions in the isothermal-isobaric (NPT) ensemble with the semi-isotropic pressure of 1.013 bar scaled independently along the bilayer normal and in the bilayer plane at a constant temperature of 310 K. The pressure and temperature were controlled using Nose-Hoover thermostat and Parrinello-Rahman

3.1. Diverse dimerization of the GHR TMD in the micellar environment The recombinant fragment GHR254–294 (hereafter GHRtm), which included TMD (residues F263 – S288), the full-length extracellular JM region (residues M254–D262 between the ECD and TMD) and initial JM part of the flexible ICD (residues K289–K294) was prepared in the aqueous suspension of DPC micelles [13]. The shapes of cross-peaks in the NMR spectra proved rather complex, implying that GHR TMD participates in several slow (micro-millisecond time scale) conformation exchange processes. These included a slow monomer-dimer transition in the micellar environment detected as signal broadening and doubling for the GHRtm amide (Fig. 1A, B) and methyl (Supplementary Figs. S1 and S4) groups varying within the detergent/peptide molar ratio (D/P) range used in the NMR experiments. An additional signal doubling independent of the D/P ratio was attributable to cis-trans transition of FeP266 peptide bond, as was directly confirmed by disappearance of the doubling in the P266A mutant spectra (Fig. 1D). The conformational exchange was maintained in the entire investigated pH range of 3.3 to 7.3 (Supplementary Fig. S5), i.e. independently of the state of protonation of the ionogenic sidechain groups in the juxtamembrane region. The pattern of intra-monomeric nuclear Overhauser effect (NOE) connectivities along with the distribution of the positive secondary 13 Cα chemical shifts indicate distinct TM helical structure from W267 to S288 (Supplementary Fig. S6A, B). Based on the analysis of the NMR structure and its MD-relaxation in explicit lipid bilayer, the GHR TMD helix can be defined as W267 – K289. In addition, the backbone flexibility derived from 15N-relaxation and 1H,13C,15N-chemical shift data (Supplementary Fig. S6C–E) revealed a somewhat restricted backbone mobility in the N-terminus of the GHR TMD (residues F263–P266) and adjacent ecJM region (residues S255–D262) as well as in the C-terminal region (residues Q290–R292) flanking the TMD helix. In particular, transitory elements of secondary structure occur in the ecJM part, as seen both in the MD traces (Supplementary Fig. S7) and in the NMR data (e.g., T258–E261 region). The observed inter-molecular NOE contacts could not be adequately described by a single pattern (Supplementary Fig. S2), indicating coexistence of alternative dimeric configurations employing different dimerization interfaces. (The patterns of chemical shift difference, e.g. Fig. S6F, are generally not informative in terms of dimerization mode discrimination [39].) Comparative structural analysis revealed parallel left- and right-handed dimers formed via I270xxFGxFGxxVM281 and W267LxxIIxxIFxxTVxxFVxxF287 interfaces (distinguished by mutual rotation of TMDs around helix axes by ~160°) with the distance d between the helix axes of ~9.5 Å and the average helix-helix crossing angles θ of 58° and −44°, respectively (Figs. 2 and 3A, B; Supplementary Table S1 and Fig. S6). Based on the molecular hydrophobicity 1413

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

roughly equally spaced. The former dimerization interface was primarily stabilized by polar and stacking interactions of the glycine Cα-H groups and aromatic rings of the (FG274/FG277)2 motif, whereas the latter interface having larger contact surface area relied on the van der Waals interactions of bulk sidechains of leucine and isoleucine residues as well as stacking interactions between opposite aromatic rings of W267/F283/F287 and polar contacts of T279 with F286. The dimeric component of G277 amide group signal contained a distinct NOE cross peak with water hydroxyl group (at the frequency shifted relative to the bulk water signal, as is typical for OH group in a nonpolar environment) (Supplementary Fig. S3). Water penetration into the TMD center was confirmed by deuterium exchange experiments. The measured values of the exchange half-times are given in Supplementary Table S2. In particular, the amide group of G277 is in ~1.5 times faster exchange compared to G274. Overall, this experiment supports the interpretation of MD implying water penetration deep into the hydrophobic core. Moreover, the pattern of the exchange time distribution suggests that the C-terminal side of the helix is more water-accessible, as is to be expected based on MD interpretation and MHP distribution. In MD relaxations, water molecules were seen to form transient contacts with the TMD surface and be temporarily arrested in a polar cavity formed by FG274/FG277 residues (Figs. 3C and 4A, B; Supplementary Fig. S7A, B). In the minor dimeric conformation (as well as in the monomeric state) the cavity is exposed to lipid tails and water lifetime in the cavity is limited to several nanoseconds, whereas in the major conformation the cavity is almost fully enclosed and water molecules in the cavity participate in dimerization forming a network of hydrogen bonds and polar contacts (Fig. 4A, B; Supplementary Fig. S8C). Furthermore, presence of water molecule in the cavity within the dimerization interface should affect the dimer geometry and hence be reflected in the observed complex cross-peak shapes (Fig. 1; Supplementary Figs. S1–S3). Indeed, the NMR structure calculations for the left-handed dimer yielded two slightly different alternative configurations with the crossing angles differing by about 10° (Fig. 2B), primarily due to difference in the relative position of the sidechains of M281 and F276. According to MD simulations, the smaller crossing angle occurred in the presence of water in the dimeric interface when the polarity of the cavity was not attenuated by M281 methyl groups (Figs. 4B, C; Supplementary Fig. S7B, C). 3.2. Right- and left-handed conformations of the GHR TMD dimer are distinguished by cis-trans isomerization of the FeP266 peptide bond It is important to keep in mind that two processes occur in parallel – cis-trans isomerization of FeP266 peptide bond in the N-terminus of each subunit (“monomer”), and alterations between different dimerization modes. Only a few isolated (non-overlapping) cross-peaks in the NMR spectra could be credibly assigned to the “minor” dimerization mode. Doubling of the Cγ2H3 T279 signal in the conditions of nearly complete (~92%) dimerization (Supplementary Fig. S1) allowed estimating relative occupancies of the minor and major conformations of the GHRtm dimer as 1:5, respectively. The fact that this ratio remains unchanged in the course of D/P titration supports the assumption that both the minor and the major configurations are dimeric [40]. For the major conformation, D/P titration experiment revealed the dimerization energy [40] determined by the unique polar FG274/FG277 dimerization interface at about −4.0 kcal/mol (Supplementary Fig. S4). From this value and the occupancy ratio, the minor conformation dimerization energy can be estimated at −3.0 kcal/mol. Taking into account the relative occupancy of cis- and trans-configurations of the FeP266 peptide bond of 2:5 estimated based on the cross-peak intensity ratio and the 1:5 ratio between the minor and major dimeric states, it can be provisionally inferred that the minor conformation of the GHRtm dimer exists only in the cis/cis configuration. Disappearance of the minor component of T279 signal (exclusively assigned to the minor conformation) in case of the P266A mutant is

Fig. 2. Representative ensembles of the NMR-derived spatial structures of the alternative GHRtm dimers. (A) Inter-monomeric NOE connectivities used for calculation of left- and right-handed GHRtm dimer spatial structures are shown with red and blue lines, respectively. (B) and (C) The sets of 20 NMR structures calculated with trans/trans left-handed and cis/cis right-handed configurations of the GHRtm dimer (PDB ID: 5OEK and 5OHD), respectively, are superimposed on backbone atoms of the TMD helix regions (W267–K289)2. Backbone heavy atom bonds are shown for the region (F263–Q291)2 in black, flexible N- and Cterminal JM regions are in gray. The left-handed dimeric GHRtm structures (B) are split on two subsets (sidechain heavy bonds in magenta and orange) having slightly different TMD helix-helix crossing angles. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

potential (MHP) analysis [28], the former interface is polar, and the latter is hydrophobic (Fig. 3B, C). In the left-handed configuration, the distance between the N-termini of the TMD helix was smaller than between the C-termini, whereas in the right-handed dimer they were

1414

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

Fig. 3. Comparison of the alternative dimerization modes of GHR TMD identified in the DPC micelles. (A) Ribbon diagrams of the NMR-derived structures of the right- and left- handed GHRtm dimers formed via hydrophobic 267WLxxIIxxIFxxTVxxFVxxF287 and relatively polar 270IxxFGxFGxxVM281 motifs are presented at the top and the bottom, respectively (see also Fig. 2A and Supplementary Fig. S6H). Based on the available mutagenesis data, the conformations correspond to dormant and active receptor states and are distinguished by cis-trans isomerization of FeP266 peptide bond in the TMD helix entry. Residues C259, P266 (cis-configuration is designated as ‘cP’) and K289 situated in the central part of ecJM region (folded into transient structures), N- and C-termini of the TMD helix, respectively, are marked. Heavy atom bonds are depicted. Approximate position of the membrane boundaries is highlighted by yellow strips. (B) and (C) Hydrophobic and hydrophilic (polar) surfaces of the GHRtm dimer subunit are shown in blue and orange, respectively, according to the MHP value. The MHP map (B) is plotted in cylindrical coordinates associated with the TMD helix. Percentages of the inter-monomeric contact surfaces occupied by the residues composing the alternative TMD helix packing interfaces (encircled by the red ovals) are shown on the left side of the maps. Polar cavity formed by FG274/FG277 residues is indicated by red arrow.

FYFPW267 (Figs. 4B, C), explaining the restricted backbone mobility in this part of ecJM (Supplementary Fig. S6C–E). Such interaction of ecJM region facilitates inter-molecular SeS bridge formation via native C259 situated within the flexible JM region in the left-handed dimeric conformation (Fig. 4C). This is supported by convergence of the opposite C259 residues (up to 3 Å between Cα atoms) along the MD traces in the left-handed conformation, which was not observed for the right-handed dimer (Supplementary Fig. S7A, B). Somewhat counterintuitively, the inter-monomeric SeS bridging further increases multiplicity and complexity of cross-peaks in the NMR spectra of GHRtm, with apparently minimal influence on the peaks corresponding to trans-configuration and possible appearance of additional minor peaks related to cis-configuration (Fig. 1C). Together with MD results, this can be interpreted as cis-configuration being less compatible with the SeS bridge. In the trans-configuration, on the other hand, the SeS bridge provides obvious stabilizing effect on the major conformation. More specifically, in the MD simulations in POPC bilayer, the major conformation in the absence of the SeS bridge tends to evolve with time into a somewhat different conformation (probably due to multiple degrees of freedom introduced by diverse interactions of flexible ecJM regions), as can be seen from time dependence of protein-protein interaction patterns (Supplementary Fig. S7B). In the presence of the SeS bridge, the MD traces become stationary, with all the key parameters including angles, distances and interaction patterns preserved over the simulation time (Supplementary Fig. S7C). Besides stabilizing the left-handed conformation, the SeS

consistent with this conclusion (Supplementary Fig. S1). More generally, the NMR spectra of P266A-GHRtm were much simpler than those of the wild type protein, without the additional minor cross-peaks (conformational states), implying that the GHRtm spectra complexity observed for the wild type protein was indeed associated with the conformational diversity (alternative dimerization modes and cis-trans isomerization within each subunit of the dimer). As suggested by Pro mutant measurements and the observed occupancy of the minor conformation in the wild type, there was a positive correlation (not necessarily 100%) between the cis/cis configuration and the minor dimerization mode. So, trans/trans, trans/cis, cis/trans, and perhaps (though not likely based on the occupancy values) even cis/cis states could be present in the ensemble corresponding to the “major” dimer, the trans form apparently having been predominant (Supplementary Figs. S1 and S2). Therefore, the GHRtm dimer structures calculated from the NMR data pool that were finally deposited in the Protein Data Bank (PDB ID: 5OHD and 5OEK) correspond to cis/cis right-handed and trans/trans left-handed configurations. Cis- and trans-configurations of the FeP266 peptide bond correspond to different orientations of FYF265 residues (flanking the TMD helix entry), lying under the plane of lipid heads or traversing this plane respectively, as confirmed by MD relaxations (Fig. 4; Supplementary Fig. S7). In the major (left-handed) dimeric conformation, this allows the ecJM regions immediately adjacent to TMD to form transient intraand inter-molecular stacking interactions [41] of aromatic sidechains of 1415

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

Fig. 4. MD relaxation of the alternative dimeric spatial structures of GHR TMD in explicit POPC lipid bilayer. In the top, ribbon structures of the (A) right- and (B) left-handed conformations of the GHRtm dimer (residues F257–K294) with cis-configuration of both FeP266 peptide bonds and a water molecule bound to the polar cavity formed by FG274/FG277 residues (indicated by red arrows). Hydrogen bonds and polar contacts of the water molecule with the protein, along with hydrogen bonding of the side-chain hydroxyl ОγН group of T279 with the backbone carbonyl group of I275 are depicted by dashed yellow lines. To improve visibility, only phosphorous atoms of lipids are shown by orange balls. (C) Ribbon structure of the left-handed conformation of the GHRtm dimer with trans-configuration of both FeP266 peptide bonds and inter-monomeric SeS bridge between the C259 residues situated in the flexible ecJM regions. The representative structures correspond to the MD snapshots of POPC bilayer presented in Supplementary Fig. S7. The ribbon structures are rotated by c.a. -20o relative to the NMR structures presented in Fig. 3 on the left. In the center, the central part of the structures is zoomed and rotated for better visibility. The residues participating in the dimer formation and/or binding water molecules are indicated for one subunit of the GHRtm dimer. In the bottom, extracellular N-terminal part of the structures is zoomed. Residues of one subunit of the GHRtm dimer are labeled.

4. Discussion

bridge also appears to expel water from the polar cavity on the TMD surface and generally prevents water from interaction with the TMD surface. Thus, in the MD simulation with water molecule initially placed into the cavity the molecule rapidly left the dimer interface, and spontaneous occurrence of water contacts with the TMD residues observed in cis conformation became considerably less frequent. By contrast, in the minor (right-handed) conformation, the TMD helix-helix interface is stable without the SeS bridge, with only minor fluctuations of structural parameters (Supplementary Fig. S7A).

4.1. Right- and left-handed dimerization modes of GHR TMD presumably correspond to the basal and active receptor states and can be complemented by different protein-lipid interactions The obtained dimerization interfaces and overall GHR TMD dimer configurations are fully consistent with those predicted in [8] based on biochemical and modeling data, the left-handed conformation corresponding to the active state of the receptor, and the right-handed one – 1416

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

properties of lipid bilayers and membrane proteins. Water penetration into the TMD dimeric interface buried in the membrane hydrophobic core was also observed earlier, both in NMR experiments [46,47] and MD simulations (with the duration of 100 ns [48]). The fact that the GHR TMD coexist in both conformations in the micellar environment might be seen as contradicting the lipid-mediated signal transduction hypothesis. However, the apparent contradiction is easily resolved since compared to lipid bilayer the micelles are capable of more flexibly adapting to the demands of alternative conformations taking into account certain degree of conformational freedom of the TMD helix entry associated with cis-trans isomerization of the FeP266 peptide bond.

to the basal inactive state. The only notable difference is a considerably larger crossing angle in the inactive conformation in our structures. Nevertheless, the ~5 Å change in the distance between the TMD Ctermini (between the positively charged K289) associated with the transition between the two conformations derived from the NMR structures and MD simulations (Figs. 2, 3 and 4) is consistent with the decrease in fluorescence resonance energy transfer observed in [8] upon receptor activation. In order to accommodate the active dimer configuration with the larger C-termini distance and helix-crossing angle into the same membrane thickness, the TMD entry sequence needs to move deeper into the membrane (Fig. 4C; Supplementary Fig. S7C). Apparently, the SeS bridge formation (overcoming repulsion of negatively charged EED262 residues of ecJM) and stacking interactions between the residues forming the TMD entry facilitated by it enable such a rearrangement by stabilizing the entry sequence in the orientation perpendicular to the membrane surface plane. In the inactive conformation, the smaller helix-crossing angle is compatible with cisconfiguration and orientation of the TMD entry sequences parallel to the membrane surface. Consistently with the cross-linking experimental data [8], the ecJM regions in this case retain certain degree of flexibility. Such an orientation implies stronger interaction of the protein with membrane lipid headgroups, and hence stronger perturbation of the lipid ordering state that can propagate deeper into the hydrophobic core of the membrane. This can make the inner parts of TMD more accessible for water molecules, including the polar cavity formed by FG274/FG277 residues, which is exposed to lipid tails in the inactive dimeric conformation. The dimerization interfaces employed by the TMD dimer in the active and inactive states differ in two important aspects – polarity of the interacting residues and the interface length (Fig. 3B, C). The inactive interface spans the entire depth of the membrane, with a couple of interacting residues per every loop of the TMD helix and a polar spot exposed to the hydrophobic core of the membrane in relative proximity to its extracellular side. In the active configuration, the dimerization interface condenses around this polar spot, and is considerably shorter at the expense of the C-terminal part of the helix. This suggests that polar interactions within the TMD dimerization interface have higher weight in the active conformation of the receptor, implying higher degree of local hydrophobicity around the TMD dimer. The implication is consistent with the observed rearrangement of the ecJM regions, introducing less perturbation into the lipid bilayer structure in the active dimeric conformation due to transition from protein-lipid to protein-protein interaction. Thus, the geometry of the TMD dimer closely correlates with local properties of lipid bilayer around it, and thus a change in these local properties triggered by ecJM interactions can propagate across the membrane inducing a change of interaction of the intracellular JM regions (e.g. LID1 including Box1) with the membrane and in their position relative to each other. This is similar to a lipidmediated receptor activation model we recently described for receptor tyrosine kinases from the HER/ErbB family, in case of which a transition from the active conformation employing a more polar interface to the passive conformation (with the less polar interface) was directly demonstrated to be caused by a change of the properties of lipid environment [25,33,42–44]. In case of GHR, interactions with water molecules capable of penetrating deep into the dimerization interface or along the TMD dimer surface also appear to play a notable role in the conformational transitions associated with the receptor activation. Based on the transitory nature of water molecule interactions with TMD residues, the interactions can be responsible for forming intermediate meta-states lowering the overall free energy barrier for switching between the active and inactive state, making water act as a “lubricant” for the protein machinery in the lipid environment. Clearly, water penetration into the membrane also depends on the local state of lipid bilayer. As follows from modern atomistic MD simulations, water dynamics near and inside lipid bilayers is nontrivial ([45] and references therein), water molecules being active players defining the local

4.2. Signal transduction by GHR can be mediated by coupled proteinprotein and protein-lipid interactions The proline residue at the entry to the extracellular side of the membrane appears to be important for allowing the adjacent ecJM region to alternate between protein-protein and protein-lipid interaction through cis-trans isomerization of the FeP266 peptide bond. Circumstantial evidence of that is relatively high conservatism of this residue among the type I receptors [49]. In the absence of proline in this position, reversible folding of the N-cap or first turn of TMD helices (e.g., observed in [34]) can serve to enable similar ligand-induced rearrangement of the ecJM regions. On the cytoplasmic side, it is Box1 region that was shown to be important for GHR activation and to associate (as part of LID1) with the membrane [11,12] in certain conditions, similarly to the JM-A region of the HER/ErbB receptor tyrosine kinases [35,44,50–55]. The concerted changes of the TMD configuration and of the properties of local lipid environment can cause partial release from the membrane and rearrangement (including separation) of this fragment, thus making it free to expose appropriate interaction sites for dimeric JAK2 kinase. Fig. 5 illustrates the sequences of key events leading to receptor activation and signal transduction without and with explicit consideration of the role of lipid-protein interaction. Note that the behavior of the cytoplasmic JM regions is inferred from our experiments and published data, rather than supported by direct observations. To reflect this fact, this part of the protein is shaded with a semi-transparent gray oval. The simplified representation provided by Fig. 5A essentially coincides with the model described in [8], where the functionally crucial consequence of ligand binding by GHR is divergence of the TMD C-termini translated into (a somewhat larger) divergence of Box1 domains destabilizing inhibitory dimerization mode of JAK2 and causing its activation. Though attractive in its pithiness, the model apparently fails to provide a robust mechanism by which a mere increase of geometric distance between the TMD ends connected to the JAK2 via extended intrinsically disordered ICDs (possibly with transient elements of secondary structure [11,12]) could reliably induce transition from the inhibitory to the active dimerization mode of the JAK2. Explicit consideration of the local changes of lipid ordering properties and hence of the lipid-protein interactions in addition to the mere geometric rearrangement of the protein itself provides a more holistic picture with a deterministic outcome of the ligand binding. The fact that ICDs interact specifically with hallmark lipids of the inner plasma membrane leaflet through conserved basic clusters and hydrophobic motifs [11,12] implies that protein-lipid interactions are indeed functionally significant, and therefore should be taken into consideration in the receptor signaling model, as shown in Fig. 5B. In the initial dormant state, GHR exists as a symmetric dimer, its extracellular juxtamembrane domains flexibly interacting with the membrane surface. Based on the TMD structural information, cis-configuration of the FeP266 peptide bond in which TMD dimer exists in the inactive conformation facilitates “snorkeling” of the ecJM regions under the lipid headgroups, while repulsion between the charged EED262 residues supports the cis-configuration of the bonds in both monomers. In line with the hypothesis of [8], asymmetric ligand 1417

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

Fig. 5. Key stages of the proposed mechanism of signal transduction by GHR inferred from structural information and supported by the available biochemical/ biophysical data. (A) Simplified sequence (left to right) of structural rearrangements of GHR disregarding the accompanying changes of lipid properties. Black arrows indicate the directions of displacement of structural elements of the receptor (ECD, ecJM, TMD, and LID1). The LID1/Box1 regions are shown by a dotted rectangle to designate their potential ability to form transient elements of α-helical structure [11,12]; the black squares at their ends illustrate the presumably functionally important C-termini of Box1 (increase of distance between which is associated with the receptor activation [8]). Negatively charged EED262 and aromatic ring FYF265 patches situated in the ecJM region and adjacent TMD entry are colored in magenta and black, respectively. SeS bonded and unlinked C262 residues are schematically presented as solid and open orange ovals. Cis-configuration of the FeP266 peptide bond is designated as ‘cP’. The polar and hydrophobic dimerization motifs of GHR TMD are shown in cyan and orange, respectively. Since we obtained no direct experimental evidence related to the behavior of the cytoplasmic JM regions, on the figure we covered the hypothetic events (see text for details) occurring on the cytoplasmic side with a semi-transparent gray oval. (B) Holistic representation of the GHR activation process taking credit for cross-talk between the membrane and the receptor in the course of ligand-induced receptor activation. The lipid molecules highlighted in green correspond to less ordered (more hydrophilic) lipid environment. Consideration of possible consequences of the local rearrangements of lipid environment consistent with the observed changes of polarity of the exposed TMD surface suggests a somewhat different sequence of events on the intracellular side of the membrane. Such an assumption is corroborated by the examples of other class 1 receptors, for which functional importance of protein-lipid interactions has been reported based on direct evidence [25,33].

of the Box1 C-termini in the membrane plane with appropriate realignment and modifying their interaction with the membrane (possibly with partial release, as observed with intracellular JM regions of other receptor tyrosine kinase proteins). There is no direct information about the exact nature of rearrangements of the intracellular JM regions. However, analogy with other HER/ErbB family receptors [35,51,56] and overall resemblance of the conformational transitions occurring in the TMD upon activation and capable of conveying and amplifying the signal transduced to the JAK2 kinase allow making certain plausible assumptions about them. More specifically, the localization of a perturbed lipid spot (possibly preferentially recruiting certain lipid

binding to the ECD dimer induces rearrangement of the ecJM C-termini accompanied by cis-trans switching of the FeP266 peptide bond and subsequent retraction of at least one of the ecJM regions from the membrane, which allows overcoming the EED262 repulsion and leads to enhancement of protein-protein interaction between the ecJM regions forming the TMD entry. This results in decrease of the ecJM region interaction with lipids and hence of local lipid bilayer perturbation, inducing transition to dimerization via the more polar and shorter interface allowing maximal separation of the TMD C-termini. The resultant changes of the TMD geometry and of local lipid properties affect the intracellular JM regions, simultaneously allowing further separation 1418

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

Appendix A. Supplementary data

species) in the cone between the TMD C-termini in the active conformation suggests that the ICD LID1/Box1 regions switch between parallel and antiparallel interaction modes rather than shift linearly in relation to each other. This would ensure deterministic separation of the Box1 C-termini by a fixed distance, unlike simple divergence of the TMD C-termini, which can result in a broad range of relative positions of the LID1 regions due to flexible nature of the ICD. Moreover, in such a case, the observed fluctuations of the distance between the C-termini would not bring the ICD out of “productive” mutual position and orientation.

Description of some additional results obtained for GHR TMD in membrane mimetics is presented in Supplementary Figs. S1–S8 and Tables S1, S2. Supplementary material to this article can be found online at https://doi.org/10.1016/j.bbagen.2018.03.022. References [1] A.J. Brooks, M.J. Waters, The growth hormone receptor: mechanism of activation and clinical implications, Nat. Rev. Endocrinol. 6 (2010) 515–525, http://dx.doi. org/10.1038/nrendo.2010.123. [2] M.J. Waters, A.J. Brooks, JAK2 activation by growth hormone and other cytokines, Biochem. J. 466 (2015) 1–11, http://dx.doi.org/10.1042/BJ20141293. [3] R. Roskoski Jr., Janus kinase (JAK) inhibitors in the treatment of inflammatory and neoplastic diseases, Pharmacol. Res. 111 (2016) 784–803, http://dx.doi.org/10. 1016/j.phrs.2016.07.038. [4] Y. Chhabra, M.J. Waters, A.J. Brooks, Role of the growth hormone-IGF-1 axis in cancer, Expert. Rev. Endocrinol. Metab. 6 (2011) 71–84, http://dx.doi.org/10. 1586/eem.10.73. [5] J. Guevara-Aguirre, A.L. Rosenbloom, Obesity, diabetes and cancer: insight into the relationship from a cohort with growth hormone receptor deficiency, Diabetologia 58 (2015) 37–42, http://dx.doi.org/10.1007/s00125-014-3397-3. [6] R.J. Brown, J.J. Adams, R.A. Pelekanos, Y. Wan, W.J. McKinstry, K. Palethorpe, R.M. Seeber, T.A. Monks, K.A. Eidne, M.W. Parker, M.J. Waters, Model for growth hormone receptor activation based on subunit rotation within a receptor dimer, Nat. Struct. Mol. Biol. 12 (2005) 814–821, http://dx.doi.org/10.1038/nsmb977. [7] D. Poger, A.E. Mark, Turning the growth hormone receptor on: evidence that hormone binding induces subunit rotation, Proteins 78 (2010) 1163–1174, http:// dx.doi.org/10.1002/prot.22636. [8] A.J. Brooks, W. Dai, M.L. O'Mara, D. Abankwa, Y. Chhabra, R.A. Pelekanos, O. Gardon, K.A. Tunny, K.M. Blucher, C.J. Morton, M.W. Parker, E. Sierecki, Y. Gambin, G.A. Gomez, K. Alexandrov, I.A. Wilson, M. Doxastakis, A.E. Mark, M.J. Waters, Mechanism of activation of protein kinase JAK2 by the growth hormone receptor, Science 344 (2014) 1249783, , http://dx.doi.org/10.1126/science. 1249783. [9] Y. Zhang, J. Jiang, J.J. Kopchick, S.J. Frank, Disulfide linkage of growth hormone (GH) receptors (GHR) reflects GH-induced GHR dimerization. Association of JAK2 with the GHR is enhanced by receptor dimerization, J. Biol. Chem. 274 (1999) 33072–33084, http://dx.doi.org/10.1016/j.phrs.2016.07.038. [10] V.N. Uversky, The multifaceted roles of intrinsic disorder in protein complexes, FEBS Lett. 589 (2015) 2498–2506, http://dx.doi.org/10.1016/j.febslet.2015.06. 004. [11] G.W. Haxholm, L.F. Nikolajsen, J.G. Olsen, J. Fredsted, F.H. Larsen, V. Goffin, S.F. Pedersen, A.J. Brooks, M.J. Waters, B.B. Kragelund, Intrinsically disordered cytoplasmic domains of two cytokine receptors mediate conserved interactions with membranes, Biochem. J. 468 (2015) 495–506, http://dx.doi.org/10.1042/ BJ20141243. [12] K. Bugge, E. Papaleo, G.W. Haxholm, J.T. Hopper, C.V. Robinson, J.G. Olsen, K. Lindorff-Larsen, B.B. Kragelund, A combined computational and structural model of the full-length human prolactin receptor, Nat. Commun. 7 (2016) 11578, , http://dx.doi.org/10.1038/ncomms11578. [13] O.V. Bocharova, P.K. Kuzmichev, A.S. Urban, S.A. Goncharuk, E.V. Bocharov, A.S. Arsenyev, Preparation of transmembrane fragments growth hormone receptor GHR in a cell-free expression system for structural studies, Bioorg. Khim. 41 (2015) 631–637, http://dx.doi.org/10.1134/S1068162015060047. [14] E.V. Bocharov, P.E. Volynsky, K.V. Pavlov, R.G. Efremov, A.S. Arseniev, Structure elucidation of dimeric transmembrane domains of bitopic proteins, Cell Adhes. Migr. 4 (2010) 284–298, http://dx.doi.org/10.4161/cam.4.2.11930. [15] R.L.J. Keller, The Computer Aided Resonance Assignment Tutorial, CANTINA Verlag, Goldau, Switzerland, 2004 ISBN: 3-85600-112-3 http://cara.nmr-software. org/downloads/3-85600-112-3.pdf. [16] J. Cavanagh, W.J. Fairbrother, A.G. Palmer, N.J. Skelton, Protein NMR Spectroscopy: Principles and Practice, 2nd ed, Academic Press, San Diego, CA, USA, 2006 ISBN:9780121644918 http://www.elsevier.com/books/protein-nmrspectroscopy/cavanagh/978-0-12-164491-8. [17] A. Favier, B. Brutscher, Recovering lost magnetization: polarization enhancement in biomolecular NMR, J. Biomol. NMR 49 (2011) 9–15, http://dx.doi.org/10.1007/ s10858-010-9461-5. [18] K. Kazimierczuk, V.Y. Orekhov, A comparison of convex and non-convex compressed sensing applied to multidimensional NMR, J. Magn. Reson. 223 (2012) 1–10, http://dx.doi.org/10.1016/j.jmr.2012.08.001. [19] D. Uhrı́n, S. Uhrı́nová, C. Leadbeater, J. Nairn, N.C. Price, P.N. Barlow, 3D HCCH3TOCSY for resonance assignment of methyl-containing side chains in 13C-labeled proteins, J. Magn. Reson. 142 (2000) 288–293, http://dx.doi.org/10.1006/jmre. 1999.1951. [20] C. Zwahlen, P. Legault, S.J.F. Vincent, J. Greenblatt, R. Konrat, L.E. Kay, Methods for measurement of inter-molecular NOEs by multinuclear NMR spectroscopy: application to a bacteriophage λ N-peptide/boxB RNA complex, J. Am. Chem. Soc. 119 (1997) 6711–6721, http://dx.doi.org/10.1021/ja970224q. [21] Y. Shen, A. Bax, Protein backbone and sidechain torsion angles predicted from NMR chemical shifts using artificial neural networks, J. Biomol. NMR 56 (2013) 227–241, http://dx.doi.org/10.1007/s10858-013-9741-y.

5. Conclusions In summary, detailed spatial structures of TMD dimers in the presumed active and inactive conformation were obtained for the first time for GHR in membrane-mimicking environment. As can be inferred from the structures, the conformational transitions underlying signal transduction by GHR are mediated by changes of protein- protein and protein-lipid interactions of JM regions on both sides of the membrane coupled with geometric rearrangement of the TMD dimer. Apparently, formation of C259 SeS bridge stabilizing the active conformation is an important event in signaling process enabling the signal persistence, and transitory interactions of TMD residues with water molecules might play a considerable role in optimizing kinetics of signal transduction. The proposed model of the transmembrane signal transduction pathway builds upon the model suggested by Brooks et al. [8], integrating the traditionally considered protein-protein interactions with protein-lipid and protein-water interactions, which sometimes tend to be neglected. However, our understanding of the processes occurring on the intracellular side of the GHR/JAK2 complex remains limited, especially as far as it concerns the behavior of cytoplasmic intrinsically disordered regions. Transparency document The http://dx.doi.org/10.1016/j.bbagen.2018.03.022 associated with this article can be found in online version. Acknowledgments The authors express their sincere thanks to Dr. K.A. Beyrit for helpful discussions. Experiments were partially carried out using the equipment provided by the IBCH core facility (CKP IBCH, supported by Russian Ministry of Education and Science, grant RFMEFI62117X0018). Access to computational facilities of the Supercomputer Center “Polytechnical” at the St. Petersburg Polytechnic University and Joint Supercomputer Center RAS (Moscow) is greatly appreciated. Funding sources This work was supported by Russian Science Foundation (project #14-50-00131) in part of structural analysis of GHR TMD by NMR spectroscopy. MD simulation work was supported by the Russian Foundation for Basic Research (project #16-04-00578), by the RAS Program “Molecular and Cellular Biology”, and by the Russian Academic Excellence Project “5–100”. Accession numbers The chemical shift assignments, NMR-derived constraints and atomic coordinates of the left- and right-handed configurations of the GHRtm dimer have been deposited in the Biological Magnetic Resonance Bank (http://www.bmrb.wisc.edu) and Protein Data Bank (http://www.rcsb.org/) under accession codes BMRB ID: 34160 and 34164; PDB ID: 5OEK and 5OHD. 1419

BBA - General Subjects 1862 (2018) 1410–1420

E.V. Bocharov et al.

[22] S. Neal, M. Berjanskii, H. Zhang, D.S. Wishart, Accurate prediction of protein torsion angles using chemical shifts and sequence homology, Magn. Reson. Chem. 44 (2006) Spec No:S158–167 https://doi.org/10.1002/mrc.1832. [23] J.H. Chill, J.M. Louis, J.L. Baber, A. Bax, Measurement of 15N relaxation in the detergent-solubilized tetrameric KcsA potassium channel, J. Biomol. NMR 36 (2006) 123–136, http://dx.doi.org/10.1007/s10858-006-9071-4. [24] P. Güntert, Automated NMR protein structure calculation, Prog. Nucl. Magn. Reson. Spectrosc. 43 (2003) 105–125, http://dx.doi.org/10.1016/S0079-6565(03) 00021-9. [25] E.V. Bocharov, D.M. Lesovoy, K.V. Pavlov, Y.E. Pustovalova, O.V. Bocharova, A.S. Arseniev, Alternative packing of EGFR transmembrane domain suggests that protein-lipid interactions underlie signal conduction across membrane, Biochim. Biophys. Acta 1858 (2016) 1254–1261, http://dx.doi.org/10.1016/j.bbamem. 2016.02.023. [26] R. Koradi, M. Billeter, K. Wüthrich, A program for display and analysis of macromolecular structures, J. Mol. Graph. 14 (1996) 51–55, http://dx.doi.org/10.1016/ 0263-7855(96)00009-4. [27] R.G. Efremov, A.J. Alix, Environmental characteristics of residues in proteins: threedimensional molecular hydrophobicity potential approach, J. Biomol. Struct. Dyn. 11 (1993) 483–507, http://dx.doi.org/10.1080/07391102.1993.10508011. [28] A.A. Polyansky, A.O. Chugunov, P.E. Volynsky, N.A. Krylov, D.E. Nolde, R.G. Efremov, PREDDIMER: a web server for prediction of transmembrane helical dimers, Bioinformatics 30 (2014) 889–890, http://dx.doi.org/10.1093/ bioinformatics/btt645. [29] M.J. Abraham, T. Murtola, R. Schulz, S. Páll, J.C. Smith, B. Hess, E. Lindahl, GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers, SoftwareX, 1–2 2015, pp. 19–25, , http://dx. doi.org/10.1016/j.softx.2015.06.001. [30] K. Lindorff-Larsen, S. Piana, K. Palmo, P. Maragakis, J.L. Klepeis, R.O. Dror, D.E. Shaw, Improved side-chain torsion potentials for the Amber ff99SB protein force field, Proteins 78 (2010) 1950–1958, http://dx.doi.org/10.1002/prot.22711. [31] W. Jorgensen, J. Chandrasekhar, J. Madura, R. Impey, M. Klein, Comparison of simple potential functions for simulating liquid water, J. Chem. Phys. 79 (1983) 926–935, http://dx.doi.org/10.1063/1.445869. [32] P.M. Joakim, J.P. Jambeck, A.P. Lyubartsev, An extension and further validation of an all-atomistic force field for biological membranes, J. Chem. Theory Comput. 8 (2012) 2938–2948, http://dx.doi.org/10.1021/ct300342n. [33] E.V. Bocharov, P.E. Bragin, K.V. Pavlov, O.V. Bocharova, K.S. Mineev, A.A. Polyansky, P.E. Volynsky, R.G. Efremov, A.S. Arseniev, The conformation of the epidermal growth factor receptor transmembrane domain dimer dynamically adapts to the local membrane environment, Biochemistry 56 (2017) 1697–1705, http://dx.doi.org/10.1021/acs.biochem.6b01085. [34] E.V. Bocharov, D.M. Lesovoy, S.A. Goncharuk, M.V. Goncharuk, K. Hristova, A.S. Arseniev, Structure of FGFR3 transmembrane domain dimer: implications for signaling and human pathologies, Structure 21 (2013) 2087–2093, http://dx.doi. org/10.1016/j.str.2013.08.026. [35] P.E. Bragin, K.S. Mineev, O.V. Bocharova, P.E. Volynsky, E.V. Bocharov, A.S. Arseniev, HER2 transmembrane domain dimerization coupled with self-association of membrane-embedded cytoplasmic juxtamembrane regions, J. Mol. Biol. 428 (2016) 52–61, http://dx.doi.org/10.1016/j.jmb.2015.11.007. [36] A.H. Beaven, A.M. Maer, A.J. Sodt, H. Rui, R.W. Pastor, O.S. Andersen, W. Im, Gramicidin A channel formation induces local lipid redistribution I: experiment and simulation, Biophys. J. 112 (6) (2017 Mar 28) 1185–1197, http://dx.doi.org/10. 1016/j.bpj.2017.01.028. [37] L. Zhang, A. Polyansky, M. Buck, Modeling transmembrane domain dimers/trimers of plexin receptors: implications for mechanisms of signal transmission across the membrane, PLoS One 10 (4) (2015 Apr 2) e0121513, , http://dx.doi.org/10.1371/ journal.pone.0121513. [38] L. Zhang, A.J. Sodt, R.M. Venable, R.W. Pastor, M. Buck, Prediction, refinement, and persistency of transmembrane helix dimers in lipid bilayers using implicit and explicit solvent/lipid representations: microsecond molecular dynamics simulations of ErbB1/B2 and EphA1, Proteins 81 (3) (2013 Mar) 365–376, http://dx.doi.org/ 10.1002/prot.24192. [39] D.M. Lesovoy, K.S. Mineev, P.E. Bragin, O.V. Bocharova, E.V. Bocharov, A.S. Arseniev, NMR relaxation parameters of methyl groups as a tool to map the

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

[54]

[55]

[56]

1420

interfaces of helix-helix interactions in membrane proteins, J. Biomol. NMR 69 (3) (2017 Nov) 165–179, http://dx.doi.org/10.1007/s10858-017-0146-1. K.S. Mineev, D.M. Lesovoy, D.R. Usmanova, S.A. Goncharuk, M.A. Shulepko, E.N. Lyukmanova, M.P. Kirpichnikov, E.V. Bocharov, A.S. Arseniev, NMR-based approach to measure the free energy of transmembrane helix-helix interactions, Biochim. Biophys. Acta 1838 (2014) 164–172, http://dx.doi.org/10.1016/j. bbamem.2013.08.021. G.B. McGaughey, M. Gagné, A.K. Rappé, pi-Stacking interactions. Alive and well in proteins, J. Biol. Chem. 273 (1998) 15458–15463, http://dx.doi.org/10.1074/jbc. 273.25.15458. A.A. Polyansky, P.E. Volynsky, R.G. Efremov, Multistate organization of transmembrane helical protein dimers governed by the host membrane, J. Am. Chem. Soc. 134 (2012) 14390–14400, http://dx.doi.org/10.1021/ja303483k. E.V. Bocharov, G.V. Sharonov, O.V. Bocharova, K.V. Pavlov, Conformational transitions and interactions underlying the function of membrane embedded receptor protein kinases, Biochim. Biophys. Acta 1859 (2017) 1417–1429, http://dx.doi. org/10.1016/j.bbamem.2017.01.025. E.V. Bocharov, K.S. Mineev, K.V. Pavlov, S.A. Akimov, A.S. Kuznetsov, R.G. Efremov, A.S. Arseniev, Helix-helix interactions in membrane domains of bitopic proteins: specificity and role of lipid environment, Biochim. Biophys. Acta 1859 (2017 Apr) 561–576, http://dx.doi.org/10.1016/j.bbamem.2016.10.024. N.A. Krylov, V.M. Pentkovsky, R.G. Efremov, Nontrivial behavior of water in the vicinity and inside lipid bilayers as probed by molecular dynamics simulations, ACS Nano 7 (2013) 9428–9442, http://dx.doi.org/10.1021/nn4042392. E.V. Bocharov, Y.E. Pustovalova, K.V. Pavlov, P.E. Volynsky, M.V. Goncharuk, Y.S. Ermolyuk, D.V. Karpunin, A.A. Schulga, M.P. Kirpichnikov, R.G. Efremov, I.V. Maslennikov, A.S. Arseniev, Unique dimeric structure of BNip3 transmembrane domain suggests membrane permeabilization as a cell death trigger, J. Biol. Chem. 282 (2007) 16256–16266, http://dx.doi.org/10.1074/jbc.M701745200. E.V. Bocharov, M.L. Mayzel, P.E. Volynsky, K.S. Mineev, E.N. Tkach, Y.S. Ermolyuk, A.A. Schulga, R.G. Efremov, A.S. Arseniev, Left-handed dimer of EphA2 transmembrane domain: helix packing diversity among receptor tyrosine kinases, Biophys. J. 98 (2010) 881–889, http://dx.doi.org/10.1016/j.bpj.2009.11.008. C.L. Ried, C. Scharnagl, D. Langosch, Entrapment of water at the transmembrane helix-helix interface of Quiescin sulfhydryl oxidase 2, Biochemistry 55 (2016) 1287–1290, http://dx.doi.org/10.1021/acs.biochem.5b01239. T. Schmidt, F. Ye, A.J. Situ, W. An, M.H. Ginsberg, T.S. Ulmer, A conserved ectodomain-transmembrane domain linker motif tunes the allosteric regulation of cell surface receptors, J. Biol. Chem. 291 (2016) 17536–17546, http://dx.doi.org/10. 1074/jbc.M116.733683. N. Jura, N.F. Endres, K. Engel, S. Deindl, R. Das, M.H. Lamers, D.E. Wemmer, X. Zhang, J. Kuriyan, Mechanism for activation of the EGF receptor catalytic domain by the juxtamembrane segment, Cell 137 (2009) 1293–1307, http://dx.doi. org/10.1016/j.cell.2009.04.025. N.F. Endres, R. Das, A.W. Smith, A. Arkhipov, E. Kovacs, Y. Huang, J.G. Pelton, Y. Shan, D.E. Shaw, D.E. Wemmer, J.T. Groves, J. Kuriyan, Conformational coupling across the plasma membrane in activation of the EGF receptor, Cell 152 (2013) 543–556, http://dx.doi.org/10.1016/j.cell.2012.12.032. K.S. Mineev, S.V. Panova, O.V. Bocharova, E.V. Bocharov, A.S. Arseniev, The membrane mimetic affects the spatial structure and mobility of EGFR transmembrane and juxtamembrane domains, Biochemistry 54 (2015) 6295–6298, http://dx. doi.org/10.1021/acs.biochem.5b00851. G. Hedger, M.S. Sansom, H. Koldsø, The juxtamembrane regions of human receptor tyrosine kinases exhibit conserved interaction sites with anionic lipids, Sci. Rep. 5 (2015) 9198, , http://dx.doi.org/10.1038/srep09198. A. Doerner, R. Scheck, A. Schepartz, Growth factor identity is encoded by discrete coiled-coil rotamers in the EGFR juxtamembrane region, Chem. Biol. 22 (2015) 776–784, http://dx.doi.org/10.1016/j.chembiol.2015.05.008. C. Matsushita, H. Tamagaki, Y. Miyazawa, S. Aimoto, S.O. Smith, T. Sato, Transmembrane helix orientation influences membrane binding of the intracellular juxtamembrane domain in Neu receptor peptides, Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 1646–1651, http://dx.doi.org/10.1073/pnas.1215207110. R. Trenker, M.J. Call, M.E. Call, Progress and prospects for structural studies of transmembrane interactions in single-spanning receptors, Curr. Opin. Struct. Biol. 39 (2016) 115–123, http://dx.doi.org/10.1016/j.sbi.2016.07.001.