Citrate stabilized gold nanoparticles interfere with amyloid fibril formation: D76N and ΔN6 \b{eta}2-microglobulin variants
Giorgia Brancolini, Maria Celeste Maschio, Cristina Cantarutti, Alessandra Corazza, Federico Fogolari, Vittorio Bellotti, Stefano Corni, Gennaro Esposito
NNanoscale
PAPER
Cite this:
Nanoscale , 2018, , 4793Received 12th September 2017,Accepted 13th February 2018DOI: 10.1039/c7nr06808e rsc.li/nanoscale Citrate stabilized gold nanoparticles interfere withamyloid fi bril formation: D76N and Δ N6 β † Giorgia Brancolini, * a Maria Celeste Maschio, a Cristina Cantarutti, b Alessandra Corazza, b,c
Federico Fogolari, b,c
Vittorio Bellotti, d,c,e
Stefano Corni f and Gennaro Esposito * a,c,g Protein aggregation including the formation of dimers and multimers in solution, underlies an array ofhuman diseases such as systemic amyloidosis which is a fatal disease caused by misfolding of native glob-ular proteins damaging the structure and function of a ff ected organs. Di ff erent kind of interactors caninterfere with the formation of protein dimers and multimers in solution. A very special class of interactorsare nanoparticles thanks to the extremely e ffi cient extension of their interaction surface. In particularcitrate-coated gold nanoparticles (cit-AuNPs) were recently investigated with amyloidogenic protein β β m). Here we present the computational studies on two challenging models known fortheir enhanced amyloidogenic propensity, namely Δ N6 and D76N β m naturally occurring variants, anddisclose the role of cit-AuNPs on their fi brillogenesis. The proposed interaction mechanism lies in theinterference of the cit-AuNPs with the protein dimers at the early stages of aggregation, that inducesdimer disassembling. As a consequence, natural fi bril formation can be inhibited. Relying on the compari-son between atomistic simulations at multiple levels (enhanced sampling molecular dynamics andBrownian dynamics) and protein structural characterisation by NMR, we demonstrate that the cit-AuNPsinteractors are able to inhibit protein dimer assembling. As a consequence, the natural fi bril formation isalso inhibited, as found in experiment. Introduction
The interest in the interaction of nanoparticles (NPs) withamyloidogenic proteins is continuously growing due to thehuge number of possible applications in nanomedicine andnanotechnology. – In particular, the interaction between goldnanoparticles (AuNPs) and the biological systems has received great attention due to the development of novel therapeuticand diagnostic tools, and due to concerns regarding theirsafety in vivo . It is widely accepted that the contact betweenthe surface of NPs and proteins triggers a competition betweendi ff erent biological molecules to adsorb on the surface of theNPs either transiently or permanently, in the so-called soft orhard corona layer. As a consequence, the protein structureand/or function may be perturbed to di ff erent extent orremain conserved.Understanding protein – inorganic nanoparticle interactionsis central to the rational design of new tools in biomaterialsciences, nanobiotechnology and nanomedicine. Theoreticalmodelling and simulations provide complementaryapproaches for experimental studies.We have recently studied the interaction of citrate-cappedgold nanoparticles (cit-AuNPs) with β β m), the light chain component of class I major histocompatibilitycomplex (MHCI), see Fig. 1. In long-term hemodialysedpatients, this protein precipitates into amyloid deposits andaccumulates in the collagen-rich tissues of the joints, originat-ing a pathology referred to as dialysis related amyloidosis(DRA). Contrary to expectations, based on previous studies of † Electronic supplementary information (ESI) available. See DOI: 10.1039/C7NR06808E a Center S3, CNR Institute Nanoscience, Via Campi 213/A, 41125 Modena, Italy.E-mail: [email protected] b Dipartimento di Scienza Mediche e Biologiche (DSMB), University of Udine,Piazzale Kolbe 3, 33100 Udine, Italy c Istituto Nazionale Biostrutture e Biosistemi, Viale medaglie d ’ Oro,305 - 00136 Roma, Italy d Dipartimento di Medicina Molecolare, Universita ’ di Pavia, Via Taramelli 3,27100 Pavia, Italy e Division of Medicine, University College of London, London NW3 2PF, UK f Department of Chemical Science, University of Padova, via VIII Febbraio 2,35122 Padova and Center S3, CNR Institute Nanoscience, Via Campi 213/A,41125 Modena, Italy g Science and Math Division, New York University at Abu Dhabi, Abu Dhabi,United Arab Emirates. E-mail: [email protected]
This journal is © The Royal Society of Chemistry 2018
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online
View Journal | View Issue m with various nanoparticle systems of larger sizes, noclear aggregation promotion and/or inhibition was detected, inthe presence of citrate-coated AuNPs with diameter of 5 nm. Here, we progress further the investigation of nanoparticlee ff ects on more challenging amyloidogenic β m proteinspecies, namely D76N and Δ N6 that can undergo fibrillogen-esis under mild conditions at neutral pH. D76N is a naturallyoccurring variant of β m bearing an asparagine residue at posi-tion 76 instead of an aspartate. This single point mutantASP76ASN (D76N) is associated with the late onset of a fatalhereditary systemic amyloidosis characterised by extensivevisceral amyloid deposits. This variant readily forms fibrils byagitation at neutral pH exhibiting the highest amyloidogenicability amongst all known β m variants. The Δ N6 is a truncated form of β m, lacking the first sixN-terminal residues. This cleaved variant is the major com-ponent of ex vivo amyloid plaques ( ∼ ff ected by DRA. While there is a broad agreement regard-ing the ability of Δ N6 to prime the fibrillar conversion ofWild-Type β m in vitro under physiological conditions, themechanism by which this occurs is not consensual.Notwithstanding that a prion-like mechanism of Δ N6 hasbeen proposed to drive the fibrillogenesis of the β m nativeform, Bellotti and coworkers has challenged the prion-likehypothesis by reporting that the Wild-Type β m does notfibrillate with monomeric Δ N6 but rather with preassembledfibrils of Δ N6. The major goal of the present work is to address the inter-action mechanism between cit-AuNPs and D76N and Δ N6adducts via enhanced molecular dynamics simulations andNMR experiments. The focus is placed on the interference ofthe cit-AuNPs with the protein at the early stages of aggrega-tion, namely monomeric and dimeric adducts.By using molecular simulations at multiple levels(enhanced sampling molecular dynamics and Browniandynamics) we provide a map of the preferential interactionsites between monomeric and dimeric protein aggregates and the cit-AuNP. The achieved results on D76N demonstrate thatsimulations and NMR data provide a picture in which theinteraction with cit-AuNP occurs via protein dimers,suggesting the presence of preassembled D76N dimers in solu-tion at neutral pH. For Δ N6 variant, preferential interactionsare mostly occurring through the amino-terminal region inboth the monomeric and dimeric species. At physiological pH, Δ N6 variant may be present as monomers and/or dimers insolution and the interaction with the cit-AuNPs may occurwith both, indistinctly.In all cases, this binding of nanoparticles is able to blockthe active sites of protein domains used for the binding withanother protein, thus leading to an inhibition of the fibrilla-tion activity as found in experiments.
Results and discussion
The nature of the binding of D76N and Δ N6 variants on cit-AuNP, is characterized by a comprehensive multiple level mod-eling investigation, spanning from rigid-body protein – surfacedocking to enhanced molecular dynamics simulations. In thissection we describe the employed computational approachand the results obtained for the monomeric and dimericadducts. In the Experimental part we will report the experi-mental NMR and UV-vis data and the comparison withsimulations.Brownian dynamics (BD) simulations are initially per-formed to generate protein – surface monomeric and/or dimericencounter complexes, by keeping the internal structure of theproteins and the surface rigid during the docking. Morespecifically, the adsorption free energies of the encounter com-plexes are computed for the structures resulting from thedocking and the BD simulation trajectory are clustered toidentify di ff erent orientations. For each of the most populatedcomplexes, which are ranked by size, a representative structureis selected for each system and refined by enhanced MD.The BD interaction energy of the protein with the cit-AuNPsurface is described by four main terms: van der Waalsenergy described by site – site Lennard-Jones interactions, E LJ ,adsorbate – metal electrostatic interaction energy, U EP and thedesolvation energy of the protein, U pds , and of the metalsurface, U mds . For the monomeric assemblies, the simulations are startedfrom the NMR structure (PDB:1JNJ) upon inclusion of vari-ations, whereas for the dimeric assemblies a preliminaryprotein – protein dockings is also performed to obtain theinitial more favourable association complexes (results arereported in section “ Protein – protein docking: the dimericinterface ” ).After performing the initial docking simulations, the stabi-lity of the docked encounter complexes is assessed by runningReplica-Exchange simulations in solvent and on cit-AuNPinvolving multiple simulations at di ff erent temperatures(T-REMD). The adopted simulation protocol includes 20 (or30) ns of replica exchange molecular dynamics (REMD) at Fig. 1
The native structure of wild-type human β -2 microglobulin ( β m)(top) and its secondary structure content together with that of Δ N6 andD76N β m variants (bottom). Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
The native structure of wild-type human β -2 microglobulin ( β m)(top) and its secondary structure content together with that of Δ N6 andD76N β m variants (bottom). Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online i ff erent temperatures (T-REMD), yielding an aggregated simu-lation time of 640 (or 960) ns. Monomers on cit-AuNPDocking of monomers on negative gold.
In this section, wefocus on the docking of D76N and Δ N6 monomers on cit-AuNP. The density of negative charge of the gold surface atomsis chosen according to an atomistic model which is able tomimic the electrochemical potential of the cit-AuNPs surfaceunder aqueous conditions and at physiological pH. – After this docking procedure was applied to D76N netchg =+3.00 e ( X netchg = total net charge of X species) and Δ N6 netchg = − e monomers on negative charged gold surface atoms(Au netchg = − e ), a hierarchical clustering algorithm (based ona minimum distance linkage function) was applied to thedi ff usional encounter complexes. Two main orientations arefound namely A and B, accounting for 71% and 29% of thetotal encounter complexes, respectively. In the case of Δ N6 monomer, docking provided a single orientation i.e. complex I, accounting for the 96% of the total encounter com-plexes. Protonation state of the proteins is determined asexplained in the Methodology.The representative structures of the resulting complexes areshown in Fig. 2. The complexes stability and the protein resi-dues contacting the surface are listed in Table 1.From Table 1, the binding of D76N on cit-AuNP in com-plexes A and B is driven mostly by the electrostatic terms. Thebinding in complex A and B is stabilised mostly by the electro-static terms. The preferred orientation involves the residues atthe N-terminal (ILE1 GLN2 ARG3) tail and DE-loop (LYS58).The strong and highly populated binding seems to be associ-ated with the total charge of the gold surface atoms and theamount of charged residues (ARG3, LYS58) contacting thesurface and this is due to the fact that in presence of negativelycharged gold the protein is able to use simultaneously morethan one charged contact in order to optimise the binding. Onthe contrary, binding of Δ N6 on cit-AuNP in complex I is
Fig. 2
Most populated encounter complexes of D76N netchg = +3.00 e and Δ N6 netchg = − e on negatively charged gold (Au negchg = − e ) obtainedby BD simulation. Complexes A, B are the representative structures ofthe two clusters obtained for D76N (including 71 and 29% of the orien-tations, respectively). Complex I is the representative complex of themost populated cluster identi fi ed for Δ N6 (including the 96% of theprotein – surface reciprocal orientations). The protein backbone is shownin cartoon representation (with the yellow colour for D76N and graycolor for Δ N6). The residues contacting the gold surface are shown instick representation.
Table 1
Encounter complex from rigid-body BD docking of D76N netchg = +3.00 e and Δ N6 netchg = − e (obtained from PDB:1JNJ and modi fi cated,truncated manually) to an Au (111) surface D76N MonomerLabel RelPop % a U Repr b E LJ c E LJ + U pds + U mds d U EP e Contact residues f A 71 − − − − − Δ N6 MonomerLabel RelPop % a U Repr b E LJ c E LJ + U pds + U mds d U EP e Contact residues f I 96 − − − a Relative population of this cluster. b U Repr : total interaction energy of the representative of the given cluster in kT with T = 300 K. c E LJ : Lennard-Jones energy term for the representative complex, U pds : non-polar (hydrophobic) desolvation energy of the representative complex, in kT. d U mds :surface desolvation energy of the representative complex, in kT. e U EP : total electrostatic energy of the representative complex, in kT. f Residueswith atoms contacting gold at distances ≤ Nanoscale Paper
This journal is © The Royal Society of Chemistry 2018
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online riven by E LJ interactions but electrostatic is also relevant dueto the charge contact of the negative surface with a positivelycharged LYS58.From the present docking results, we may conclude that thee ff ect of two variations located in significantly di ff erentprotein domains i.e. N-TER for Δ N6 and EF-loop for D76N,does not significantly a ff ect the global orientation of proteinbound complexes to cit-AuNP respect to the native β mprotein. The most populated A and I complexes of the two var-iants are contacting cit-AuNP through DE-loop (LYS58). Enhanced sampling of monomers on cit-AuNP.
To assessthe stability of the monomeric docked encounter complexesand to include the e ff ect of structural relaxation, Replica-Exchange simulations in solvent and on cit-AuNP involvingmultiple simulations at di ff erent temperatures (T-REMD) areperformed starting with the most representatives and popu-lated monomeric complexes obtained from rigid-body BDdocking.Simulation results of Complex A for D76N monomer andComplex I for Δ N6 interacting with cit-AuNP are summarised
Fig. 3
Top panel. On the left: Time evolution of contacting residues for the monomeric D76N with respect to the surface of the nanoparticle inter-face ( i.e. protein residues within 3 A from the surface), extracted from the total 20 ns T-REMD and central and right panels report the two mostrepresentative structures of the D76N monomer during T-REMD. Bottom panel. Δ N6 binding to citrate-auNP is conserved during the entire 20 nslength of T-REMD since the protein remains anchored through the DE-loop residues (LYS58, TRP60) and BC-loop residue (HIS31). In addition, the Δ N6 monomeric protein exhibited few contacts with C-TER (ARG97, MET99) residues in the very last part of the 20 ns simulation. The capability ofthe Δ N6 protein to remain anchored to the citrate surface during T-REMD is in line with the intensity reduction which were observed experimentallyfor Δ N6 on citAuNP.
Table 2
Most populated encounter complex for D76N and Δ N6 protein – protein complexes by BD simulation. The structure of a single complex isrepresentative for the 97% of the total encounter complexes for D76N dimers, whereas the Δ N6 dimers is representative for the 70% of the totalcomplexes. The protein backbone is shown in cartoon representation. For nomenclature, see Fig. 1
Label RelPop % a U Repr b U pds c U EP d Spread e Representative of cluster f D76NA 97 − − − Δ N6I 70 − − − − − a Relative population of this cluster. b U Repr : total interaction energy of the representative of the given cluster in kT with T = 300 K. c U pds : non-polar(hydrophobic) desolvation energy of the representative complex, in kT. d U EP : total electrostatic energy of the representative complex, in kT. e RMSD of the structures within the cluster with respect to the representative complex. f Representative of a given cluster.
Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online n Fig. 3. Panels (a) report time evolution of D76N contactingresidues on cit-AuNP, panels (b) and (c) report the final repre-sentative structures of the two most recurrent orientationsfound on the D76N and Δ N6 variants interacting with cit-AuNP. For the Δ N6 variant a longer simulation time of 30 nswas required in order to perform a satisfactory sampling of thecontacting patches with the negatively charged citrate layer.Orientations are obtained following the replica at the lowesttemperature during the 20 ns and 30 ns of T-REMD,respectively.For D76N mutant, (see upper panels of Fig. 3), a stableinteraction between the N-terminal, BC and DE loop and theNP surface is confirmed by T-REMD, but also AB loop exhibi-ted systematic contacts with the surface. This AB-loop isshown to be only loosely bound during the simulation see Fig. 3(c) and it can detach itself from the surface. On the con-trary, in Fig. 3(a) the contact patch through N-TER and DEloop is well conserved during the entire 20 ns length ofT-REMD and the protein remains anchored through theN-terminal residues (ILE1, GLN2, ARG3) and DE-loop residues(LYS58, TRP60).T-REMD results revealed that D76N mutant, if compared tothe wild-type β m, is characterised by a greater flexibility alongthe AB loop (res 12 –
20) and EF loop (res 71 – “ Protein – protein docking: pre-diction of dimeric interface ” ), looking for a better comparisonwith available experimental data.For Δ N6 variant, in the lower panel of Fig. 3, the contactpatch identified by docking is confirmed to be well conserved,since the protein remains anchored through the DE-loop resi-dues (LYS58, TRP60) and BC-loop residue (HIS31) during theentire 30 ns lenght of T-REMD. An additional contact is foundthrough C-TER only in the last 20 ns of simulation. The capa-bility of the Δ N6 protein to remain anchored to the citratesurface during T-REMD and the partial involvement of C-TERregion is in line with the behavior of native protein, and ingood agreement with the experimental data, as it will be dis-cussed in the “ Experimental part ” . Protein – protein docking: the dimeric interface In order to provide a complementary approach to the interpret-ation of available experimental data, a preliminary docking to
Fig. 4
Reciprocal orientations of two identical D76N and Δ N6 proteinswithin dimers obtained starting from the docking after 400 ns MDre fi nement at 300 K. A series of four simulations were performed on theinitial complex obtained from docking, each with di ff erent initial vel-ocities. (a) Initial orientation from rigid docking, (b) and (c) most stablezipped and unzipped fi nal orientation after MD re fi nement of D76N(green and cyan) and Δ N6 (orange and mangenta), respectively.
Table 3
Resultant cit-AuNP-dimer encounter complexes from rigid-body BD docking of D76N netchg = +6.00 e and Δ N6 netchg = − e to a negativeAu(111) surface. A hierarchical clustering algorithm (based on a minimum distance linkage function) was applied to the di ff usional encounter com-plexes after docking to a bare negative gold (Au negchg = − e ) surface. The reported complexes represent for (D76N) -AuNP the 99.9% of theencounter complexes obtained by BD simulation, respectively and for ( Δ N6) -AuNP the 99% Label RelPop % a U Repr b E LJ c E LJ + U pds + U mds d U EP e Contact residues f D76N1-Ad1 99.9 − − − − − − − − Δ N61-Id1 99 − − − − − − a Relative population of this cluster. b U Repr : total interaction energy of the representative of the given cluster in kT with T = 300 K. c E LJ : Lennard-Jones energy term for the representative complex, U pds : non-polar (hydrophobic) desolvation energy of the representative complex, in kT. d U mds :surface desolvation energy of the representative complex, in kT. e U EP : total electrostatic energy of the representative complex, in kT. f Residueswith atoms contacting gold at distances ≤ Nanoscale Paper
This journal is © The Royal Society of Chemistry 2018
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online uild D76N dimers is performed. For the sake of complete-ness, Δ N6 dimer are also considered. The association of wild-type β m or variants into dimers and, to reduced extents, larger oligomers – in solution has been frequently observed.Here, rigid-body docking method implemented in SDA 7.2 areapplied to predict the dimeric interfaces of the modifiedencounter complexes, which are supposed to exist pre-assembled in solution before the addition of cit-AuNP.We wish to remark that the protein – protein BrownianDynamics is performed as an initial sampling stage ofprotein – protein di ff usional association in the presence ofimplicit solvent. The docked configurations obtained at thisstage are then grouped with a hierarchical clustering algorithminto ensembles that represent potential protein – proteinencounter complexes. Flexible refinement of selected represen-tative structures is thus done by molecular dynamics (MD)simulations in explicit solvent. The advantage of usingBrownian Dynamics is that it mimics e ffi ciently the physicalprocess of di ff usional association of the unbound proteinswhereas the atomistic refinement is accounting for the proteinconformational flexibility upon association.The protein – protein docking reported in this section andthe MD refinements reported in the next section, represents apreliminary step towards the docking of D76N and Δ N6dimeric adducts on cit-AuNP.The adsorption free energies of the protein – protein encoun-ter complexes of D76N – D76N and Δ N6 – Δ N6 are reported inTable 2 along with the clustered trajectories.
D76N dimers.
Docking to build proteins dimers startingfrom two identical D76N monomers was applied and it pro-vided one main orientation accounting for more than 97% ofall the protein – protein encounter complexes, as reported inTable 2. The representative structure of the most relevantcomplex is shown in the last column of the same Table 2. Thecontact residues are di ff erent for the two monomers ( i.e. sub-units) forming the dimer. Fig. 5
Most populated encounter complexes of dimeric D76N netchg =+6.00 e and Δ N6 netchg = − e on negatively charged gold nanocluster(Au negchg = − e ) obtained by BD simulation. The protein backbone isshown in cartoon representation. The residues contacting the goldsurface are shown in stick representation. Fig. 6
Top panel: (on the left) Time evolutions of D76N dimers contacting residues respect to the surface of the cit-AuNPs along the entire TREMDdynamics ( i.e. protein residues within 3 A from the surface). The binding patches established by each sub-unit with the cit-AuNP are di ff erentiated bycolor (green for sub-unit 1 and cyan for sub-unit 2) (on the right). Most stable orientations of the 1-Ad4 dimer of d76N interacting with cit-AuNP.Direct contacts of the sub-unit 1 (green) and sub-unit 2 (cyan), are highlighted with balls on the α carbon atoms. Bottom panel: (left and right) Thesame representation is reported for second dimeric complex of D76N, namely 1-Ad1, interacting with cit-AuNP. Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Top panel: (on the left) Time evolutions of D76N dimers contacting residues respect to the surface of the cit-AuNPs along the entire TREMDdynamics ( i.e. protein residues within 3 A from the surface). The binding patches established by each sub-unit with the cit-AuNP are di ff erentiated bycolor (green for sub-unit 1 and cyan for sub-unit 2) (on the right). Most stable orientations of the 1-Ad4 dimer of d76N interacting with cit-AuNP.Direct contacts of the sub-unit 1 (green) and sub-unit 2 (cyan), are highlighted with balls on the α carbon atoms. Bottom panel: (left and right) Thesame representation is reported for second dimeric complex of D76N, namely 1-Ad1, interacting with cit-AuNP. Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online ub-unit 1, depicted in cyan, involves in the dimeric inter-face the binding N-terminal (THR4 PRO5), BC loop (HIS31)and FG loop (THR86 SER88). Sub-unit 2, depicted in green,shows B strand (PHE22) CD loop, D strand (ILE46, GLU47,LYS48, GLU50, ASP53) and E strand (TYR67, GLU69) as inter-acting residues.In the case of D76N – D76N dimers, the binding is drivenboth by Lennard-Jones and electrostatic interactions. Δ N6 dimers.
The docking provided two di ff erent orien-tations accounting for 70 and 30 per cent of the encountercomplexes, respectively. The most stable and populatedcomplex has a residue interface for sub-unit 1 (orange) involv-ing SER33, TRP60, PHE62, LEU54 and sub-unit 2 (magenta)involving HIS31, PRO32, ASP34, THR86.The most representative and most populated complexes foreach systems, are shown in Table 2.The results show that D76N variant has a more favourableattraction between monomers that facilitates aggregation withrespect to Δ N6 (and native protein), at pH around neutrality.This can be interpreted as a consequence of the asparaginesubstitution for aspartate which has a substantial impact inthe variant protein, despite the survived interaction betweenresidues 42 and 76.
For the sake of completeness, the stability of the protein – protein dimers has been examined using 400 ns of standard Fig. 7
Top panel: (on the left) Time evolutions of Δ N6 dimers contact-ing residues respect to the surface of the cit-AuNPs along the entireTREMD dynamics ( i.e. protein residues within 3 A from the surface). Thebinding patches established by each sub-unit with the cit-AuNP aredi ff erentiated by color (magenta for sub-unit 1 and orange for sub-unit2) (on the right). Most stable orientations of the 1-Id1 dimer of Δ N6interacting with cit-AuNP. Direct contacts are occurring only throughthe subunit 1 (magenta) and they are highlighted with balls on the α carbon atoms. Bottom panel: (left and right) The same representation isreported for second dimeric complex 1-Id3 of Δ N6 interacting with cit-AuNP.
Fig. 8
D76N β m NMR results. (a) Overlay of N – H HSQC spectra of N-labelled D76N β m 18 microM in the free form in blue and in the pres-ence of 90 nM cit-AuNP. (b) Bar plot of relative intensity calculated from the comparison between the spectra reported in (a). (c) D76N β m cartoonhighlighting the residue locations that proved most a ff ected by cit-AuNPs. i.e. displaced one standard deviation at least with respect to the averagerelative intensity. Nanoscale Paper
This journal is © The Royal Society of Chemistry 2018
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online
D simulations in solution, see Fig. 4, starting with the mostrepresentative dimeric complexes obtained from rigid-body BDdocking. Simulations were repeated four times using adi ff erent seed for the initial velocity distribution (d1, d2, d3,d4) for each system to improve the statistics of the search ofthe energy minima on the potential energy surface. Only thefinal most stable dimeric complexes are reported (more detailsare reported in the ESI † ). Dimers on cit-AuNPDocking of dimers to negative gold.
The docking procedureis thus applied to the dimers on negative gold surface. FromMD simulations in Fig. 4, two di ff erent dimers are obtainedfor each variant. More specifically, complexes A-d1, A-d4 pertain to D76N dimers and complexes I-d1 and I-d3 to Δ N6dimers.Docking results in Table 3 indicate that the surface chargehas a crucial influence on the binding of the dimeric com-plexes on the negative AuNP. The electrostatic interactions playan important role in changing the relative stability of the mostpopulated and stable complexes.From Fig. 5, complex 1-Ad1 of D76N is stabilised via
LYS58residue and the interacting patch is characterised by the pres-ence of both NTER and CTER (MET99) close to the Ausurface. Complex 1-Ad4 of D76N is also stabilised via
LYS58residue and it is shifting the NTER towards the negativesurface. Both D76N dimers ( Δ N6 netchg = +6.00 e ) benefits from afavourable electrostatic interactions with the negativelycharged surface. Fig. 9 Δ N6 β m NMR results. (a) Overlay of N – H HSQC spectra of N-labelled Δ N6 β m 17 μ M in the free form in blue and in the presence of 30nM cit-AuNP. In the inserts an example of two chemical shift variations (F30 and S55) and the doubling of D96 signal are presented. (b) and (c) Barplots of chemical shift variations and relative intensity, respectively, calculated from the comparison between the spectra reported in (a). (d) Δ N6 β m cartoon highlighting the residue locations that proved most a ff ected by cit-AuNPs. i.e. displaced one standard deviation at least with respect tothe average chemical shift variation. Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
LYS58residue and it is shifting the NTER towards the negativesurface. Both D76N dimers ( Δ N6 netchg = +6.00 e ) benefits from afavourable electrostatic interactions with the negativelycharged surface. Fig. 9 Δ N6 β m NMR results. (a) Overlay of N – H HSQC spectra of N-labelled Δ N6 β m 17 μ M in the free form in blue and in the presence of 30nM cit-AuNP. In the inserts an example of two chemical shift variations (F30 and S55) and the doubling of D96 signal are presented. (b) and (c) Barplots of chemical shift variations and relative intensity, respectively, calculated from the comparison between the spectra reported in (a). (d) Δ N6 β m cartoon highlighting the residue locations that proved most a ff ected by cit-AuNPs. i.e. displaced one standard deviation at least with respect tothe average chemical shift variation. Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online n the contrary, Δ N6 dimers are both accompanied by anunfavourable electrostatic contribution due to the total nega-tive charge of dimer ( Δ N6 netchg = − e ). Resulting complexes1-Id1 and 1-Id3 of Δ N6 interact textitvia the LYS58 and TRP60and LYS58, ASP59, TRP60, respectively. Complex 1-Id3 isaccompanied by a more unfavourable electrostatic energywhich is partially compensated by the LJ interaction.
Enhanced sampling of dimers on cit-AuNP.
As a final step ofthe computational strategy, T-REMD simulations are appliedto refine the interactions between the dimers of D76N (com-plexes 1-Ad4 and 1-Ad1) and of Δ N6 (complexes 1-Id1 and1-Id3) on cit-AuNP.Our simulations illustrate that the dimers of the two var-iants display a distinct behaviour towards the negativelycharged surface of the cit-AuNP, due to their di ff erent totalcharges. The initial contact of both dimers onto the surface ofthe citrate layer is facilitated by Coulomb interactions betweenthe positively charged residues at N-TER and/or and loop DE(LYS58) and the oxygen anions of the citrate molecules.However, once protein flexibility is introduced, this molecularpicture changes as the competition between protein – proteinand protein – cit-AuNP interaction depends on electrostatics.We found that the interaction with D76N dimers with cit-AuNP leads to complete dissociation of the dimeric adducts, whereas for Δ N6 dimers the dissociation cannot be seen at thetime length of the simulation. The gold – dimer interface of Δ N6 is found to be labile with respect to its gold – monomerinterface and also to the gold – dimer interface of D76N. Theelectric field created by the cit-AuNP, in fact, is not strong enough to prevent the formation of stable complexes with Δ N6 monomers ( Δ N6 netchg = − e ) but it weakens the inter-action with dimers carrying a larger negative charge ( Δ N6 netchg = − e ) due to an enhanced protonation state after dimerisa-tion as explained in Methodology, see Fig. 7. D76N 1-Ad4.
Results reported in Fig. 6, account for theprotein approaching the cit-AuNPs at the N-terminal tail andat the DE loop of the single sub-unit 2 (green) within D76Ndimer. The interaction between the protein and the AuNPsurface exhibits an initial state where just the sub-unit 2(green) is involved in the vicinity of the surface. However, afterrunning T-REMD, the final state displays a configurationwhere the dimer is essentially disassembled (see Fig. 6, withsub-unit-1 and sub-unit 2 interacting with the citrate layerthrough the N-terminal fragment or the DE loop).The crucial points of this result is the significance of theinteraction with cit-AuNPs that essentially leads to the com-plete dissociation of the dimer, i.e. disruption of the very firststep of aggregation. D76N 1-Ad1.
For the sake of completeness, docking with the “ zipped ” dimer is reported, showing a direct contact with cit-AuNP involving the unique sub-unit 2, see Fig. 6. Sub-unit 2touches the cit-AuNP surface through N-TER, AB loop (res 1, 3,12, 13, 19) while DE loop (LYS58) and CTER (93, 97, 99) resi-dues of the same sub-unit contact the surface only upon struc-tural relaxation at the interface. Results are reported in Fig. 6.Interestingly, the binding patch of sub-unit 2 with cit-AuNPis identical to the binding patch of sub-unit 2 with sub-unit 1(see Fig. 4), suggesting that the interaction of this dimer with Table 4
Direct comparison between experimental chemical shift deviations and the computed contacting residues of D76N (top) and Δ N6(bottom) monomers and dimers at the protein – NP interface from T-REMD re fi nement Structure region NMR attenuations Comp. monomer Comp. dimer Comp. dimerD76N D76N-A 1-Ad1 (zipped) 1-Ad4 (unzipped)N-ter, A strand 2sc, 6, 7, 8sc 1, 3, 6 1, 3 4, 5AB loop 13, 16, 17 11,12,19 12, 13, 19B strand 21sc, 24sc, 28 26 22BC loop 30, 33, 34 31CC ′ , C ′ D loops 42sc, 43 40 46, 47, 48D strand 53DE loop 58, 59 58, 60 58E strand 64, 65, 66, 70 75 67, 69F strand 83scFG loop 86, 88G strand, C-ter 91, 93, 95sc, 97 93, 97, 99Structure region NMR chemical Comp. monomer Comp. dimer Comp. dimer Δ N6 shifts Δ N6-I 1-Id1 (zipped) 1-Id3 (unzipped)N-ter, A strandAB loop 8, 10, 11,12,19 19, 20B strand 25, 26, 28BC loop 30, 31, 34, 38 30, 31 41CC ′ , C ′ D loops 47, 48D strand 52, 54, 55DE loop 58, 60 58, 59, 60E strandF strand 77 75FG loop 85 88G strand, C-ter 98, 99 97, 98, 99
Nanoscale Paper
This journal is © The Royal Society of Chemistry 2018
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online he cit-AuNPs is potentially able to block active sites of onemonomer for the binding to another protein, inhibiting thegrowth of further protein – protein interactions. Given the moreextended protein/protein dimeric interface at the zippeddimer, the detachment of sub-unit 1 from sub-unit 2 for thiscomplex, is not seen at the time length of the simulation butan overall weakening of the protein – protein interface isobserved as a consequence of salt-bridges breaking (see ESI † ). Δ N6 1-Id1, 1-Id3.
Results illustrate that in the presence of Δ N6 dimers, the protein – cit-AuNPs interface is more labile ata physiologically relevant pH, as a consequence, 1-Id1 (zipped)dimer can attach and detach from the surface during TREMDbut several contacts are observed. We wish to remark that thedimeric interface of (zipped) dimer with cit-AuNP is indistin-guishable from the monomeric interface (res 58, 59, 60). Infact, the direct interaction with the surface is occurring via asingle sub-unit, as shown in Fig. 7. This suggests that for thisspecies the dominant interaction may occur both with themonomers and with dimer ( i.e. the monomer within thedimer), providing the same binding interface. Also in thiscase, the complete detachment of sub-unit1 from sub-unit 2,is not seen at the time length of the simulation but a changein the protein – protein interface leading to hydrogen bondbreaking is seen during the simulations.In the case of 1-Id3 (unzipped) it is not possible to identifya really stable binding patch for Δ N6 dimers at the time lengthof our simulations but only some unstable contact with posi-tively charged residues ( e.g.
LYS48 and LYS75).
NMR experimental evidence
To map experimentally the interaction of the two β m variantstested in the simulations, N – H HSQC spectra of the two pro-teins without and with cit-AuNPs were collected using protein/NP ratios of 213 and 567 for D76N and Δ N6, respectively. Inspite of the rather conspicuous ratio di ff erence, the examinedsolutions had approximately equal protein concentrations (theNP concentration was around 90 or 30 nM), which rules outartifacts due to the actual behaviour of the proteins reflectingessentially the amount of free species. The e ff ect of the nano-particle presence on the intensity and position of nitrogen – hydrogen correlation peaks was assessed. As we reportedbefore, for D76N variant we observed no chemical shift vari-ation but an intensity decrease with an average intensity ratiobetween the signals in the two spectra of 0.78 ± 0.04. The rela-tive intensity profile shows a di ff erential pattern indicatingthat there are specific residues preferentially a ff ected by thepresence of cit-AuNP. Concerning Δ N6 variant, in addition tointensity decrease, resulting in an average value of 0.78 ± 0.12,some peaks undergo also slight change in their chemical shift(see Fig. 8 and 9). Other two features can be noticed in Δ N6HSQC spectrum recorded in presence of cit-AuNP: the dou-bling of D96 peak and the intensity increase of two peaks,namely V85 and D34. Similar e ff ects were observed also when Δ N6 was monitored at lower concentration (4 μ M, not shown),with the same cit-AuNP preparation. In general, intensity andchemical shift changes may report either the protein inter- action surface with cit-AuNP and/or the protein – protein inter-action evolution in the presence of cit-AuNP, as observed withD76N. Besides the direct contact e ff ects, any such inter-actions may prove capable of altering local conformations,which also may lead to intensity and/or chemical shift devi-ations. This must then be the case also with the changes in Δ N6 spectra. The clustering pattern of the chemical shift andintensity deviations depicted in Fig. 8 and 9, along with thepreviously reported results for wild-type and D76N, suggest Fig. 10
Final orientations resulted from T-REMD re fi nement of D76Nand Δ N6 on cit-AuNP. Dimeric complexes 1-Ad4 and 1-Ad1 are obtainedfor D76N. Monomeric complex and 1-Id1, 1-Id3 dimeric complexes areobtained for Δ N6.
Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online hat a typical protein/cit-AuNP contact interface encompassesthe N-terminal surrounding apical region (BC, DE and FGloops along with the N-terminal stretch where present). Theadditional involvements should reflect changes due to theshift of the protein/protein association equilibria that are eli-cited by competing protein/NP interactions and may concerneither the association interface or allosteric structural e ff ectsthat propagate to buried regions. All of these e ff ects werealready recognised in D76N/cit-AuNP systems and seem tooccur also with Δ N6, with appreciable e ff ects also on chemicalshifts. In particular, in Table 4 and Fig. 10 we report a directcomparison between experimental chemical shift deviationsand the contacting residues of D76N and Δ N6 variants at theprotein – NP interface. Our simulations both reproduce andexplain the experimentally observed data. The deviationsobserved at B-strand, F-strand and C-terminal fragment of Δ N6 may feature a decrease of association, with consequentlocal rearrangements at strand B, in the presence of cit-AuNP.The cross-peak splitting distinctly observed for D96 amidesignal is likely to arise from a decreased inter-conversion rateof two limiting local conformers following the associationpattern change. The simulation results support the establish-ment of di ff erent conformation of the C-terminal region uponinteraction with the cit-AuNP, as shown in Fig. 11. The plot ofthe distance between D96 residue and the neighbouring V9show the presence of three distinct peaks associated todi ff erent conformers, namely conformer α in solvent and con-formers β and γ with cit-AuNP, in which the C-terminal tail isclose/distant to strand A (V9). The cross-peak splittingobserved experimentally could then reflect either the inter-con-version between the most populated peak α in solvent andpeak γ on cit-AuNP or between β and γ . Since those conformers β and γ are not observed in solvent, their onset can be ascribedto the interaction with the cit-AuNP surface. On the otherhand, the intensity increase coupled to resonance shift of V85and D34 amide cross-peaks could reflect local decrease ofdipolar or/and exchange broadening arising from cit-AuNPcontact, associated to a conformational change at BC and DEloop as resulted from simulations (Fig. 12). The conformation- al changes of those loops are known to play an important rolein fibrillation process. Di ff usion coe ffi cient determinations NMR 2D DOSY spectra were collected to measure thetranslational di ff usion coe ffi cients of the β m variants inthe absence and presence of cit AuNPs. The results clearlyshow that the di ff usion coe ffi cients of the proteins increasewhen cit AuNPs are present in solution Fig. 13. This is con-sistent with an e ff ect of cit AuNP on the protein associationequilibria that prove all shifted towards the monomericspecies. Fig. 13
DOSY map overlays of D76N and Δ N6 β m without (black) andwith (red) cit-AuNP. Proteins were always 4 μ M and cit-AuNP was 20 – ff usion coe ffi -cient ( D in m s − ) is observed in the presence of cit-AuNP. The over-whelming signals from Hepes bu ff er and citrate, in addition to theresidual solvent peak, restrict the region where isolated protein peakscan be observed. Fig. 11 Δ N6 variant: the plot report the distance between D96 residueand the neighbouring V9, in solvent (red plot) and for the protein inter-acting with cit-AuNP (black plot). Results are obtained analysing themost populated structures after clustering. The fi gure show the pres-ence of two distinct peaks associated to di ff erent conformers in whichthe C-terminal tail is bound/unbound to strand A. Fig. 12 Δ N6 variant: residues V85 and D34 (yellow dots) undergo adi ff erent conformation of BC loop and DE loop going from solvent tocit-AuNP. Nanoscale Paper
This journal is © The Royal Society of Chemistry 2018
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online onclusions
In conclusion, we have proposed a comprehensive and consist-ent mechanism for how citrate covered gold NP influenceprotein aggregation (dimerization) and thus fibril formationfor the highly amyloidogenic variants D76N and Δ N6. Thecharacterisation of the interaction between cit-AuNPs andD76N, Δ N6 β m naturally occurring variants, by atomisticsimulations has shed light on the microscopic mechanism ofthe process. By NMR we have mapped preferential interactionsites on the protein surface that have been properly repro-duced by modelling calculations. We propose that the promi-nent dimer population of D76N at neutral pH undergoese ffi cient reshaping and eventually splitting in the presence of acit-AuNP surface. Conversely, the similar distribution of associ-ation adducts of Δ N6 protein interacts less e ffi ciently, i.e. withreduced turnover frequency, with cit-AuNPs. The reducede ffi ciency accompanies the electrostatic repulsion contri-butions that are estimated more unfavorauble for the dimerwith respect to monomers. While in simulations this e ff ectdetermines indistinct extent of interaction of monomers anddimers with cit-AuNPs, the experimental NMR pattern of Δ N6shows more pronounced consequences on the chemical shifts,with respect to the corresponding pattern observed with D76N.For both proteins, however, the NMR evidence also demon-strates consistently a reduction of the association extent inpresence of nanoparticles as inferred from the increase of thetranslational di ff usion coe ffi cient. In some of the simulatedsystems, we found that the interaction of D76N dimers withcit-AuNP leads to complete dissociation of the dimericadducts, whereas for Δ N6 dimers the dissociation could notbe seen at the time length of the simulation. However, theinteraction with cit-AuNPs is always seen to interfere at thesites of protein – protein interaction and to lead, conceivably, toan inhibition of the fibrillation events. MethodologyBrownian dynamics simulations.
Rigid-body docking simu-lations were carried out using Brownian dynamics (BD) tech-niques with the ProMetCS continuum solvent model forprotein – gold surface interactions. The calculations wereperformed using the SDA version 7 software. The β m struc-ture was taken from the NMR solution structure (PDB id: 1JNJ)and the mutation at residue 76 was introduced manually, asthe truncation of the first six residues of Δ N6.Titratable protein side chains, were assigned at pH 7.2 withH++. As in ref. 9, in addition to HIS51 and HIS84 even HIS31is protonated, given the presence of the negative citrate adlayerwhich may stabilise the protonated regime. For dimers the proto-nation state was assigned after dimerisation in explicit solvent.We wish to remark that the Δ N6 monomers in solvent has aninitial Δ N6 netchg = − e but after dimerisation the protonationstate of each monomeric sub-unit becomes Δ N6 netchg = − e .5000 BD trajectories were computed starting with the pro-teins positioned randomly with its center at a distance of 70 Åfrom the surface where the protein – surface interaction energy is negligible. The specified number of docked complexes wasextracted directly from the runs and clustered with a clusteringalgorithm. The relative translational di ff usion coe ffi cient was0.0123 Å ps − and the rotational di ff usion coe ffi cient for theprotein was 1.36 × 10 − in radian ps − . The simulation timestep was set to 0.50 ps. Parameters for the calculation of hydro-phobic desolvation energy/forces was set to − − Å − and for the electrostatic desolvation energy/forces to 1.67according to ref. 19. BD trajectories were generated in a rec-tangular box (ibox = 1); the dimensions of the ( x , y ) plane,describing the symmetry of the simulation volume as well asthe surface size, were given as input parameters. At each BDstep, the protein – surface interaction energy and forces actingon the protein were computed using the implicit-solventProMetCS forcefield, developed and parametrised forprotein – gold surface interactions. The energy terms includedin ProMetCS have been described in the main text.We applied a single-linkage clustering method (based onCA atoms, with RMSD = 3.0 Å) algorithm and parameters pro-viding the smallest number of physically distinct orientationsof β m on cit-AuNP, for all the results given in the manuscript. Molecular dynamics simulations.
We used our own forcefield parameters for the citrate anions based on ab initio calcu-lations. The same protein and gold structures as for the BDsimulations were used for the initial coordinates for the MDsimulations. A rectangular simulation box of dimensions(101.5 Å × 99.6 Å × 101.5 Å) including SPC/E water molecules,the protein monomers and dimers and the gold surface wasbuilt. The protein was placed at the positions of the represen-tatives of the docked clusters obtained from the BD dockingsimulations. Before the addition of the water molecules, thecenter of mass of the protein was placed at 47 Å from thesurface, retaining the original docked orientation with respectto the surface. The choice of this distance was motivated byvarious tests that we performed showing that if the simu-lations were started with the protein in direct contact to thesurface (or at smaller distances), it was in a kinetically trappedstate where only minor relaxation could take place on the time-scale of tens of ns. During equilibration dynamics, all systemscontacted the surface within the first 1 ns of MD without re-orienting respect to the surface.All simulations were performed with the Gromacs 5.2.1package. GolP and OPLS/AA parameters were used for thesurface and the protein and the SPC/E water model wasapplied. The lengths of bonds were constrained with theLINCS algorithm. Surface gold atoms and bulk gold atomswere frozen during all simulations but gold dipole chargeswere left free. Classical MD simulations were performed atconstant volume and temperature ( T = 300 K). Periodic bound-ary conditions and the Particle-Mesh-Ewald algorithm wereused. A 2 fs integration time step was used. For the citrateanions we have implemented new force field parameters basedon ab initio calculations (that take into account the quantumnature of such small chemical species) in a consistent andcompatible way with the existing GolP force field for theprotein – AuNP surface interactions.
Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online e worked out an enhanced MD sampling performed withT-REMD (Temperature Replica Exchange) simulations in expli-cit water on the most relevant encounter complex found alongthe docking. The sampling was enhanced by introducingtemperature swapping moves between states with similardensity at di ff erent temperatures. We employ a total of 32 repli-cas, covering the temperature range between 290 and 320 K.Principal component analysis, clustering analysis, hydrogenbond and salt bridges analysis were also performed usingGROMACS. NMR. N-labelled D76N and Δ N6 β m solutions in 25 mMphosphate bu ff er and 50 mM HEPES, respectively, at pH 7were analysed with and without cit-AuNPs by recording 2D N – H HSQC. Spectra were collected at 14.0 T, on the BrukerAvance III NMR facility of the Core Technology Platform atNew York University Abu Dhabi. The spectrometer, equippedwith cryoprobe and z -axis gradient unit, operated at 600.13and 60.85 MHz to observe H and N, respectively. Spectralwidths of 40 ppm ( N, t1) and 15 ppm ( H, t2) were used. Foreach t1 dimension point, 128 or 64 scans were accumulatedand quadrature in the same dimension was accomplished bygradient-assisted coherence selection (echo-antiecho). Processing with t1 linear prediction, apodization and zero-filling prior to Fourier transformation led to 2K1K real spectra.Water suppression was achieved by using a flip-back pulse inthe HSQC experiments. All measurements were performed at25 C. Spectra were processed with Topspin 2.1 and analysedwith Sparky. Chemical shift deviations were calculated as Δ δ (ppm) = [( Δ δ H)2 + ( Δ δ N/6.5)2]1/2 where Δ δ H and Δ δ N are thechemical shift variations for H and N, respectively, whereas the relative intensity is the ratio of the peak intensityin the presence of cit-AuNP and in the absence.Di ff usion coe ffi cients were determined by means of 2D HDSTEBPP (Double STimulated Echo BiPolar Pulse) experi-ments. Protein concentration was 4 μ M in 50 mM Hepes, pH= 7 in 95/5 H O/D O, either in absence and in presence of cit-AuNP. Sodium citrate (1.5 mM) was present in the absence ofNP. The z -axis gradient strength was varied linearly from 10 to90% of its maximum value ( ∼
60 G cm − ) and matrices of 2048by 40 points were collected by accumulating 512 scans per gra-dient increment. Water suppression was carefully adjusted byappending to the DSTEBPP sequence a pair of WATERGATE elements in the excitation-sculpting mode. Careful settingwas the acquired data were processed using the Bruker soft-ware Dynamics Center to extract the di ff usion coe ffi cients. Con fl icts of interest There are no conflicts to declare.
Acknowledgements
Funding from MIUR through PRIN 2012A7LMS3_003 isgratefully acknowledged. S. C. acknowledges funding from ERC under the grant ERC-CoG-681285 TAME-Plasmons. TheISCRA sta ff at CINECA (Bologna, Italy) is acknowledged forcomputational facilities and technical support. Oak RidgeNational Laboratory by the Scientific User FacilitiesDivision, O ffi ce of Basic Energy Sciences, U.S. Departmentof Energy is acknowledged for the supercomputing projectCNMS2013-064. Facilities of the National Energy ResearchScientific Computing Center (NERSC), which is supportedby the O ffi ce of Science of the U.S. Department of Energyunder Contract No. DE-AC02-05CH11231, are alsoacknowledged. References
Nat.Nanotechnol. , 2010, , 671.2 A. E. Nel, L. Mädler, D. Velegol, T. Xia, E. M. V. Hoek,P. Somasundaran, F. Klaessig, V. Castranova andM. Thompson, Nat. Mater. , 2009, , 543.3 D. F. Moyano and V. M. Rotello, Langmuir , 2011, ,10376.4 A. M. Gobin, E. M. Watkins, E. Quevedo, V. L. Colvin andJ. L. West, Small , 2010, , 745.5 P. Wang, X. Wang, L. Wang, X. Hou, W. Liu and C. Chen, Sci. Technol. Adv. Mater. , 2015, , 1.6 N. Gilbert, Nature , 2009, , 937.7 J. A. Kim, A. Salvati, C. Aberg and K. A. Dawson,
Nanoscale ,2014, , 14180.8 I. Lynch and K. A. Dawson, Nano Today , 2008, , 40.9 G. Brancolini, A. Corazza, M. Vuano, F. Fogolari,M. C. Mimmi, V. Bellotti, M. Stoppini, S. Corni andG. Esposito, ACS Nano , 2015, , 2600.10 F. Gejyo, T. Yamada, S. Odani, Y. Nakagawa, M. Arakawa,T. Kunitomo, H. Kataoka, M. Suzuki, Y. Hirasawa,T. Shirahama, et al. , Biochem. Biophys. Res. Commun. , 1985, , 701.11 S. Linse, C. Cabaleiro-Lago, W.-F. Xue, I. Lynch,S. Lindman, E. Thulin, S. E. Radford and K. A. Dawson,
Proc. Natl. Acad. Sci. U. S. A. , 2007, , 8691.12 S. Valleix, J. D. Gillmore, F. Bridoux, P. P. Mangione,A. Dogan, B. Nedelec, M. Boimard, G. Touchard,J.-M. Goujon, C. Lacombe, P. Lozeron, D. Adams,C. Lacroix, T. Maisonobe, V. Planté-Bordeneuve,J. A. Vrana, J. D. Theis, S. Giorgetti, R. Porcari,S. Ricagno, et al. , Engl. J. Med. , 2012, ,2276.13 V. Bellotti, M. Gallieni, S. Giorgetti and D. Brancaccio,
Semin. Dial. , 2001, , 117.14 T. Eichner, A. P. Kalverda, G. S. Thompson, S. W. Homansand S. E. Radford, Mol. Cell , 2011, , 161.15 P. Mangione, G. Esposito, A. Relini, S. Raimondi,R. Porcari, S. Giorgetti, A. Corazza, F. Fogolari, A. Penco,Y. Goto, Y.-H. Lee, H. Yagi, C. Cecconi, M. M. Naqvi,J. D. Gillmore, P. N. Hawkins, F. Chiti, R. Rolandi, Nanoscale Paper
This journal is © The Royal Society of Chemistry 2018
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..
Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce .. View Article Online . W. Taylor, M. B. Pepys, M. Stoppini and V. Bellotti,
J. Biol. Chem. , 2013, , 30917.16 F. Iori, R. Di Felice, E. Molinari and S. Corni,
J. Comput.Chem. , 2009, , 1465.17 M. Martinez, N. J. Bruce, J. Romanowska, D. B. Kokh,M. Ozboyaci, X. Yu, M. A. Äztürk, S. Richter andR. C. Wade, J. Comput. Chem. , 2015, , 1631.18 J. C. Gordon, J. B. Myers, T. Folta, V. Shoja, L. S. Heath andA. Onufriev, Nucleic Acids Res. , 2005, , W368.19 A. H. Elcock, R. R. Gabdoulline, R. C. Wade andJ. A. McCammon, J. Mol. Biol. , 1999, , 149.20 D. B. Kokh, S. Corni, P. J. Winn, M. Hoefling,K. E. Gottschalk and R. C. Wade,
J. Chem. Theory Comput. ,2010, , 1753.21 R. R. Gabdoulline and R. C. Wade, J. Phys. Chem. , 1996, , 3868.22 Y. Lin, G. Pan, G. J. Su, X. H. Fang, L. J. Wan and C. L. Bai,
Langmuir , 2003, , 10000.23 J. Kunze, I. Burgess, R. Nichols, I. Buess-Herman andJ. Lipkowski, J. Electroanal. Chem. , 2007, , 147.24 M. Hoefling, F. Iori, S. Corni and K. E. Gottschalk,
Langmuir , 2010, , 8347.25 D. van der Spoel, E. Lindahl, B. Hess, G. Groenhof,A. E. Mark and H. J. C. Berendsen, J. Comput. Chem. , 2005, , 1701.26 W. L. Jorgensen, D. S. Maxwell and J. TiradoRives, J. Am.Chem. Soc. , 1996, , 11225.27 B. Hess and N. F. van der Vegt,
J. Phys. Chem. B , 2006, ,17616.28 S. Giorgetti, S. Raimondi, K. Pagano, A. Relini,M. Bucciantini, A. Corazza, F. Fogolari, L. Codutti,M. Salmona, P. Mangione, L. Colombo, A. De Luigi,R. Porcari, A. Gliozzi, M. Stefani, G. Esposito, V. Bellottiand M. Stoppini,
J. Biol. Chem. , 2011, , 2121. 29 G. Esposito, R. Michelutti, G. Verdone, P. Viglino,H. Hernández, C. V. Robinson, A. Amoresano, F. Dal Piaz,M. Monti, P. Pucci, P. Mangione, M. Stoppini, G. Merlini,G. Ferri and V. Bellotti,
Protein Sci. , 2000, , 831.30 D. Gumral, F. Fogolari, A. Corazza, P. Viglino, S. Giorgetti,M. Stoppini, V. Bellotti and G. Esposito, Magn. Reson.Chem. , 2013, , 795.31 G. Esposito, M. Garvey, V. Alverdi, F. Pettirossi, A. Corazza,F. Fogolari, M. Polano, P. P. Mangione, S. Giorgetti,M. Stoppini, A. Rekas, V. Bellotti, A. J. Heck andJ. A. Carver, J. Biol. Chem. , 2013, , 17844.32 C. Cantarutti, S. Raimondi, G. Brancolini, A. Corazza,S. Giorgetti, M. Ballico, S. Zanini, G. Palmisano,P. Bertoncin, L. Marchese, P. Mangione, V. Bellotti,S. Corni, F. Fogolari and G. Esposito,
Nanoscale , 2017, ,3941.33 G. Esposito, A. Corazza and V. Bellotti, Subcell. Biochem. ,2012, , 1917.34 K. F. Morris and C. S. Johnson Jr., J. Am. Chem. Soc. , 1992, , 3130.35 G. Bodenhausen and D. J. Ruben,
Chem. Phys. Lett. , 1980, , 185.36 J. Keeler, R. T. Clowes, A. L. Davis and E. D. Laue, MethodsEnzymol. , 1994, , 145.37 S. Grzesiek and A. Bax,
J. Am. Chem. Soc. , 1993, , 12593.38 T. D. Goddard and D. G. Kneller,
SPARKY 3 , University ofCalifornia, San Francisco.39 F. A. A. Mulder, D. Schipper, R. Bott and R. Boelens,
J. Mol.Biol. , 1999, , 111.40 A. Jerschow and N. Müller,
J. Magn. Reson. , 1998, , 13.41 M. Piotto, V. Saudek and V. Sklenar,
J. Biomol. NMR , 1992, , 661.42 T.-L. Hwang and A. J. Shaka, J. Magn. Reson., Ser. A , 1995, , 275.
Paper Nanoscale | Nanoscale , 2018, , 4793 – O p e n A cce ss A r ti c l e . P ub li s h e d on F e b r u a r y . D o w n l o a d e d on / / : : P M . T h i s a r ti c l e i s li ce n s e d und e r a C r ea ti v e C o mm on s A tt r i bu ti on . U npo r t e d L i ce n ce ..