Revealing conformational substates of lipidatedN-Ras protein by pressure modulationShobhna Kapoora, Gemma Triolab,c, Ingrid R. Vetterd, Mirko Erlkampa, Herbert Waldmannb,c, and Roland Wintera,1

aPhysical Chemistry I—Biophysical Chemistry, Faculty of Chemistry, TU Dortmund University, Otto-Hahn-Strasse 6, D-44227 Dortmund, Germany; bChemicalBiology, Max Planck Institute of Molecular Physiology, Otto-Hahn-Strasse 11, D-44227 Dortmund, Germany; cChemical Biology, Faculty of Chemistry, TUDortmund University, Otto-Hahn-Strasse 6, D-44227 Dortmund, Germany; and dPhysical Biochemistry, Max Planck Institute of Molecular Physiology,Otto-Hahn-Strasse 11, D-44227 Dortmund, Germany

Edited by Michael L. Klein, Temple University, Philadelphia, PA, and approved November 18, 2011 (received for review June 30, 2011)

Regulation of protein function is often linked to a conformationalswitch triggered by chemical or physical signals. To evaluate suchconformational changes and to elucidate the underlying mole-cular mechanisms of subsequent protein function, experimentalidentification of conformational substates and characterizationof conformational equilibria are mandatory. We apply pressuremodulation in combination with FTIR spectroscopy to reveal equi-libria between spectroscopically resolved substates of the lipi-dated signaling protein N-Ras. Pressure has the advantage that itsthermodynamic conjugate is volume, a parameter that is directlyrelated to structure. The conformational dynamics of N-Ras in itsdifferent nucleotide binding states in the absence and presenceof a model biomembrane was probed by pressure perturbation.We show that not only nucleotide binding but also the presenceof the membrane has a drastic effect on the conformational dy-namics and selection of conformational substates of the protein,and a new substate appearing upon membrane binding could beuncovered. Population of this new substate is accompanied bystructural reorientations of the G domain, as also indicated by com-plementary ATR-FTIR and IRRAS measurements. These findingsthus illustrate that themembrane controls signaling conformationsby acting as an effective interaction partner, which has conse-quences for the G-domain orientation of membrane-associatedN-Ras, which in turn is known to be critical for its effector andmod-ulator interactions. Finally, these results provide insights into theinfluence of pressure on Ras-controlled signaling events in organ-isms living under extreme environmental conditions as they areencountered in the deep sea where pressures reach the kbar range.

Proteins generally and intrinsically can exist in a multitudeof conformational substates at ambient conditions and are

endowed with conformational fluctuations; i.e., a protein mole-cule in a given state spans a larger number of conformational sub-states (CS) (1). A prominent example is myoglobin, which hasbeen shown to exist in several functional substates important forligand binding (2). There are two types of inherent proteinmotions: equilibrium fluctuations, which are motions between CSin equilibrium, and nonequilibrium transitions known as func-tionally important motions, which are motions that lead from onestate to another state in a protein, respectively. The identificationand study of functionally relevant CS of proteins is not straight-forward, largely because the low fractional populations at ambi-ent conditions are not easily detectable by spectroscopic means(3). As a result, many events in protein conformational dynamicsgo unnoticed that might be decisively important for protein func-tion, such as binding, catalysis, and protein–protein or protein–nucleic acid interactions (2, 3).

In order to gain insight into the free energy landscape ofproteins, temperature and chemical perturbations (including pH,ionic strength, denaturing agents) are often employed to shiftthe population equilibrium and characterize the otherwise rareconformational substates. Temperature produces simultaneouschanges in internal energy and volume, whereas the effect of adenaturant depends on its binding properties (4, 5). Because

some of the CS might be separated by small energy differencesonly, separation by temperature or chemical perturbation can bedifficult to achieve. Conversely, pressure as the third fundamentalthermodynamic variable controls the population of protein CS byaffecting the equilibrium through volume differences (3, 5–11).Pressure favors states with smaller partial volumes, which areoften more solvated than the native state. Electrostriction, hydra-tion of newly exposed polar or nonpolar residues, and loss ofcavities, which arise due to packing defects in the folded ensem-ble of proteins, are some of the effects observed upon pressuriza-tion of proteins. (3, 9, 11). According to Akasaka et al. (11), low-lying excited CS generally have a lower molar volume, such thatapplication of pressure will populate such states that may beintimately involved in function. Thus, pressure provides an ele-gant means to populate excited states for spectroscopic studies.

Conformational selection has emerged as an alternative to the“induced fit model” and postulates that all CS of proteins preexistin solution in different regions of the energy landscape (featuringseveral minima)—not always easily detectable, however (12–14).Conformational selection has been observed for protein–ligand,protein–protein, and protein–DNA/RNA interactions, wherethe binding of different partners leads to a redistribution ofthe already existing conformations (i.e., a population shift). Con-formational selection is likely to be also the critical mechanismfor the functioning of signaling molecules such as Ras and RhoG-proteins (15). The conformational dynamics of these proteinsis linked to signaling pathways where many functional statescrucial for signaling may not be populated enough to be easilycharacterized. This seems to hold true also for members of theRas superfamily. The Ras proteins act as membrane-associatedmolecular switches in the early steps of signal transduction path-ways associated with cell growth or differentiation (16, 17). Forbiological activity, the Ras proteins need to be anchored to mem-branes, which is achieved by posttranslational lipidation. Thelipid anchors are connected to the G domain by a linker regioncomprising a hypervariable region (HVR) (Fig. 1A). Ras shuttlesbetween a GDP (inactive) and a GTP (active) bound state. Whenbound to GTP, it interacts with a variety of effectors. This cycle isregulated by guanine nucleotide exchange factors (GEFs) andGTPase activating proteins (GAPs), which stimulate GTP loadingand hydrolysis, respectively. Both these states of the protein aredistinguished by conformational differences mainly in the switch Iand II regions (Fig. 1 B and C) (18–22).

Ras proteins are thought to sample multiple conformations.Active Ras seems to exhibit diverse conformations of switch I

Author contributions: R.W. designed research; S.K. performed research; G.T. and H.W.contributed new reagents/analytic tools; S.K., I.R.V., and M.E. analyzed data; and S.K.,I.R.V., and R.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: roland.winter@tu-dortmund.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1110553109/-/DCSupplemental.

460–465 ∣ PNAS ∣ January 10, 2012 ∣ vol. 109 ∣ no. 2 www.pnas.org/cgi/doi/10.1073/pnas.1110553109

and II in solution, among which mainly two conformations ofswitch I have been characterized so far, denoted as state 1 andstate 2 (19). State 1 corresponds to a GTP-bound but inactivestate and interacts with GEFs. In contrast, state 2 can interactwith different effector proteins (20, 21). Notably, signaling ofthese lipidated proteins takes place in the presence of mem-branes, and the nucleotide binding state influences the partition-ing behavior (23) and orientation of the protein at the lipidinterface (24–28). Lipid interfaces and protein partitioning intolipid membranes are expected to have a marked effect on the CSof the membrane-associated signaling proteins, but this possibilityhas hardly been explored.

Here we employ hydrostatic pressure as a tool to shift the con-formational equilibrium of N-Ras to sufficiently populate low-ly-ing excited substates of Ras, which might be decisively importantin signaling. To this end, we have studied the lipidated proteinN-Ras HD/Far (hexadecyl/farnesyl) in its GDP and GTP analog(GppNHp) bound form (denoted as N-Ras (GDP) and N-Ras(GTP), respectively, hereafter) in the presence and absence of acanonical lipid raft membrane (29) by means of high pressureFTIR spectroscopy. Currently only few high pressure studies havebeen performed for probing conformational equilibria (6, 7, 11,21). Our results show that both the nucleotide binding state of theprotein and the presence of the lipid membrane have a drasticeffect on the conformational dynamics and selection of confor-mational substates of the protein. We reveal a conformationalsubstate appearing upon membrane binding that had not beendetected before. Complementary polarized attenuated totalreflection Fourier transform infrared (ATR-FTIR) and infrared

reflection absorption spectroscopy (IRRAS) data were taken toreveal orientational changes of the protein at the lipid interface.

ResultsEffect of Pressure on the Conformational Substates of N-Ras Modu-lated by the Bound Nucleotide State. We first measured high pres-sure FTIR spectra of N-Ras in its different nucleotide states inbulk solution in the pressure range from 1 bar to 10 kbar(1,000 MPa) and analyzed the pressure-induced changes in theamide I′ region of N-Ras (Fig. S1). At 1 bar, N-Ras (GTP/GDP)shows an asymmetric amide I′ band with predominant α-helicesand β-sheets, in good agreement with the X-ray crystal structuredata (Table S1) (30). The N-Ras protein retains its compacttertiary structure even at high pressures of 10 kbar, implying aremarkably high pressure stability of the protein, which makesit suitable for such kind of studies: Only small but significantchanges in spectral shape are visible upon pressurization. Thesecond derivative spectra of the amide I′ band allow a bettervisualization of the changes (Fig. 2 A and B). For N-Ras (GDP),the α-helical subband shows a splitting at around 3.5 kbar suggest-ing pressure-induced changes in the α-helical/turn region of theprotein forming solvent exposed turns/loops and solvated α-he-lices (9, 31, 32), thus leading to clearly developed subbands athigher pressures. Clearly, a slight decrease in the intensity ofintramolecular β-sheets with concomitant increase of a band atthe lower wavenumber side, which is typical for exposed β-sheets,is observed, the effect being less pronounced for the GTP-bound form.

The pressure-induced secondary structural changes can bemore easily monitored by means of a quantitative fit analysis ofthe amide I′ band region using mixed Gaussian and Lorentzianline shapes for deconvolution (9, 31, 32). As can be deduced fromFig. 3 A and C, even at high pressures, the N-Ras (GDP) and(GTP) still retain compact structures with only minor conforma-tional changes occurring. For N-Ras (GDP), a decrease is seenfor helices by about 4%, intramolecular β-sheets by 6%, turnsby 5%, and a concomitant increase for random coil (unordered)structures by 3%, and exposed β-sheets by 10%. Like the GDP-bound form, N-Ras (GTP) shows only minor changes in thesecondary structures—i.e., a decrease in helices by 2%, intramo-lecular β-sheets by 3% and turns by 3%, accompanied by anincrease of random coils by 3% and exposed β-sheets by 4%.These data indicate a pressure-induced shift toward more solventexposed β-sheet/disordered structures. The differences (albeitsmall) observed between the two forms apparently connote theinfluence of the bound nucleotide on the packing mode of theprotein, which determines the amplitude of pressure-dependentconformational changes. Assuming equilibrium between the twostates, the volume change ΔV 0 (1 bar till 10 kbar) accompanyingthe conformational change can be determined (see SI Text). AΔV 0 value of −24� 6 mL∕mol is obtained for N-Ras (GDP)from the pressure-dependent changes of turns/loops, random andexposed β-sheet structures. The pressure-induced changes aresimilar but smaller in N-Ras (GTP), for which a ΔV 0 value ofabout −14.7 mL∕mol is obtained. Such volume changes are afactor of 3–5 smaller than those observed upon pressure-inducedunfolding of proteins (9).

Because pressure generally causes compression of the chemi-cal bonds equivalent to changes of the force constant, linearpressure-induced frequency shifts (so-called elastic effects) arefrequently observed (9, 31, 32). The inactive form of N-Ras inbulk solution (Fig. 3E) shows a blue shift for turns/loops; all otherstructures exhibit red shifts. The α-helical subband observed at1;658 cm−1 remains intact up to 3.5 kbar, where it forms twoshoulders, one increasing up to 1;663 cm−1 (solvated turns andloops), the other decreasing to 1;654 cm−1 (solvated helices). Theposition of the band for intramolecular β-sheets observed at1;637 cm−1 remains constant up to 3.5 kbar, beyond which a sub-

Fig. 1. (A) Schematic representation of the switch regions, α-4 and α-5-helixand the HVR domain including the lipid anchors (gray) in the N-Ras protein inthe GDP (green) bound form. The structure of the G-domain was adoptedfrom the PDB entry 4Q21. Highlighted are the switch I [30–38; blue] andswitch II (residues 59–67; yellow) regions, the α-4-helix (residues 126–137;red) and the α-5-helix (residues 151–171, pink). (B and C) Representationof the flexibility of the switch regions in Ras-GTP forms (B) as comparedto Ras-GDP forms (C). It can be clearly seen that in the available X-ray struc-tures these loop regions tend to be in defined positions when fixed by theinfluence of the γ-phosphate of the GTP nucleotides, whereas the GDP formsfeature a large number of different conformations of the switch regions (theC terminus is in the back). Images prepared with PyMOL (http://www.pymol.org) using the PDB ID codes 5P21, 1P2T, 2RGE, 3lBH, 1CTQ, 3RAP, 1U8Y,2PMX, 3RAB, and 1YU9 for GTP and 4Q21, 1IOZ, 3CON, 1KAO, 1U8Z,1XTQ, 1X1R, 3KKQ, 1Z0A, and 2GF9 for the GDP forms.

Kapoor et al. PNAS ∣ January 10, 2012 ∣ vol. 109 ∣ no. 2 ∣ 461

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

band at lower wavenumbers, which is typical for solvent exposedβ-sheets, starts appearing (also shown in Fig. 2A). For activeN-Ras (GTP) in bulk solution (Fig. 3G), a similar scenario isobserved but with the exposed β-sheets appearing at higher pres-sures (>5 kbar), indicating a higher pressure stability of thisform, and the extent of exposed β-sheets arising with increasingpressure is lower for the N-Ras (GTP) (Fig. 3 A and C).

Effect of Membrane Binding on the Pressure-Dependent Conforma-tional Changes of N-Ras.Members of the Ras superfamily are foundattached to the inner side of the plasma membrane making themembrane interaction of the Ras proteins crucial for its signalingproperties so that the effect of membrane binding on the pres-sure-induced conformational changes needs to be explored andcompared with the bulk solution behavior. FTIR amide I′ spectrawere recorded for N-Ras (GDP) and N-Ras (GTP) bound to azwitterionic model raft membrane consisting of DOPC∶DPPC∶Chol 1∶2∶1 (29) (Fig. S1). As expected, no major secondary struc-tural changes or even unfolding are observed in the protein uponmembrane insertion, in agreement with previous reports (23, 33).Significant conformational changes are identified in the secondderivative spectra for N-Ras (GDP) and N-Ras (GTP) insertedinto the membrane upon pressurization, however (Figs. 2 Cand D). Changes are observed mainly in the α-helical, loop andβ-sheet region when compared with the bulk behavior. The α-he-lical subband is broader, suggesting existence of different spatialorientations of helices (mainly α-4 and α-5) and the hypervariableregion (HVR), and their involvement in membrane binding. Thisfinding is in accord with our previously published reports onmembrane interaction of K-Ras 4B (24). Pronounced splittingfor the α-helical band for N-Ras (GDP) is observed only above7 kbar, and thereafter both the band shoulders can be fairly dis-tinguished up to 10 kbar. For N-Ras (GTP) inserted into themembrane, definite splitting of the α-helical band is already ob-served at low pressures (clearly visible above 600 bar), forming ashoulder at 1;664 cm−1 (solvated turns/loops) and at 1;653 cm−1

(solvated helices). Likewise, for the membrane-bound N-Ras(GDP) and (GTP), the intensity of the intramolecular β-sheetschanges nonlinearly (see below), and the intensity at the lower

wavenumber side typical for exposed β-sheets increases signifi-cantly upon pressurization.

Fig. 3 B and D shows the variation of the secondary structuralcontents of membrane-bound N-Ras in the different nucleotidestates with pressure. For membrane-bound N-Ras (GDP), thehelix content decreases by about 3%, similar to the bulk, turns by6.0%, and increases are observed for random coils by 2% andexposed β-sheets by 6%. Similarly, membrane-bound N-Ras(GTP) also shows minor changes in the secondary structure only,such as a decrease in helices by 3%, turns by 6%, and increasesin random coils by 3% and exposed β-sheets by 6%. Most re-markably, the pressure dependence for intramolecular β-sheetsis strikingly different for both membrane-bound Ras formswhen compared with the bulk in that increasing pressure seemsto first stabilize the intramolecular β-sheets. At higher pressures(>4 kbar), a trend reversal is observed, and the intramolecularβ-sheet contribution is slightly decreased again. The occurrenceof the intensity maximum of this band indicates existence ofan additional conformational equilibrium—i.e., population of afurther, membrane-induced conformational substate of the pro-tein (denoted as state 3 hereafter). Fig. 3 F and H shows thecorresponding wavenumber shifts of the secondary structures ofmembrane-bound N-Ras (GDP) and (GTP). The behavior shownby the membrane-bound N-Ras (GDP) is similar to that of bulkN-Ras (GDP), differing only in the pressure value at which theα-helical subband displays a pronounced splitting (at approxi-mately 7 kbar compared with 3.5 kbar in the bulk). The clearlyseparated wavenumber for the exposed β-sheets also appears athigher pressures when compared with the bulk, suggesting a high-er conformational stability in the membrane-bound state. Differ-ently, for membrane-bound N-Ras (GTP), the splitting of theα-helical subband starts at rather low pressures (<600 bar) al-ready, suggesting significant involvement of the α-helical regionof the G domain upon membrane insertion in the GTP-boundstate. This would be in agreement with recent findings of Gorfeet al., where it is proposed that H-Ras (GDP) predominatelyexhibits membrane interaction through the HVR, and GTP load-ing triggers structural rearrangements in its switch regions thatare coupled to reorientations of helices (α-4, α-5) and their inter-action with the lipid interface (26, 27).

Because N-Ras exhibits 91% sequence identity to H-Ras, thismode of interaction might apply to N-Ras as well. To corroborateour findings, complementary polarized ATR-FTIR and IRRASexperiments were carried out, allowing us to yield additional in-formation on the orientation of the protein at the lipid interface.The dichroic ATR-FTIR spectra (Fig. S3) demonstrate distinctand different orientations of N-Ras GDP and GTP, respectively,with respect to the bilayer interface. A negative peak in the helixregion (at 1648–1665 cm−1) indicates an orientation of the ma-jority of the helices that is essentially parallel to the membranebilayer. The changes observed in the dichroic spectra imply minoryet significant differences in the orientation of N-Ras (GDP) andN-Ras (GTP) relative to each other and to the membrane bilayer.A more comprehensive and quantitative analysis of the orienta-tion of the protein at the lipid interface can be achieved by usingIRRAS on corresponding lipid monolayer films (for details seeSI Text and Fig. S4). Clearly, different angle dependent IRRAspectra are visible for the two bound nucleotide states of N-Ras.Simulations of the spectra were performed for two representativeorientations, denoted model 1 (GDP-bound) and model 2 (GTP-bound), adapted from ref. 26. It was found that the N-Ras (GDP)IRRA spectra are comparable to the simulated ones for model 1with respect to the intensity distribution over the whole amide I'vibrational wavenumber region, and the simulated spectra formodel 2 match reasonably well with the N-Ras (GTP) measuredones. These data reveal that N-Ras (GDP) is likely to adopt anorientation as depicted by Fig. S4A, Inset, which is similar to theconformations whose averaged structure is represented by model

Fig. 2. Pressure-induced changes in the second derivative amide I′ spectrafor (A) N-Ras GDP and (B) N-Ras GTP in bulk, (C) N-Ras GDP in the presenceof the lipid bilayer, and D) N-Ras GTP in the presence of the lipid bilayer.

462 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110553109 Kapoor et al.

1, where the protein mainly interacts with the membrane throughthe HVR region and N-terminal residues. N-Ras (GTP) showsan orientation (Fig. S4B, Inset) that would be consistent with theone described by model 2, which involves domain reorientationleading to a more extensive interaction of helices α-4 and α-5 withthe membrane.

To support the conclusions drawn from the high pressure FTIRdata, high pressure fluorescence spectroscopy was also carriedout on N-Ras (GTP) both in bulk and bound to membrane, bymapping the intrinsic tyrosine emission maxima as a function ofpressure (Fig. S5). No changes were observed for the protein in

the bulk, implying high pressure stability of the protein, whereasthe emission maximum exhibits a red shift upon pressurizationwhen bound to the membrane, implying increasing solvent expo-sure of the tyrosine residues upon pressurization starting as lowas 200 bar. These results also reveal that membrane binding hasa marked effect on the pressure-induced changes of the confor-mational substates of N-Ras (GTP).

In order to gain also information about the structural changesoccurring upon unfolding of N-Ras, temperature dependent stu-dies were carried out as well. The temperature-induced changesare pronounced and strikingly different from the pressure-induced structural changes (Fig. S2). Above 60 °C, unfolding ofN-Ras sets in, which is accompanied by a decrease of orderedstructural contents and a concomitant increase in conformationsoriginating from unordered structures and intermolecular β-sheetformation. Within the experimental error, we could not find anymarked differences in the thermal behavior of N-Ras (GTP) andN-Ras (GDP), both in the bulk and in the membrane-boundforms.

Discussion and ConclusionsIn an attempt to fully understand the structure, dynamics, andfunction of signaling proteins and their signaling networks, thevast conformational space exhibited by them needs to be ex-plored. For this endeavor, knowledge of not only the low-lyingnative conformation(s) but also of their high-energy conformersis required. These additional functionally relevant states are oftenpresent as relatively minor fractions of the ensemble under nor-mal conditions, making their detection and characterization anarduous task. The principle of conformational selection seems tobe the underlying mechanism for the functioning of Ras proteinsas well (15), and pressure-modulation is a rather new and effi-cient means for the exploration of the various conformationaland functional substates responsible for the various signaling out-puts of the protein. Pressure can shift the equilibrium already ex-isting between distinct substates at ambient pressure toward lowvolume conformers, which are generally more solvated, therebydecreasing the total volume of the protein system (3, 6, 7, 11, 34,35). As can be seen from the temperature dependent data onN-Ras (Fig. S2), temperature modulation to reveal low-lying ex-cited substates is less successful and results in full unfolding andsubsequent aggregation of the protein, only. On the contrary,pressure is a very mild perturbant and results in fully reversiblechanges, a prerequisite for determining thermodynamic proper-ties of the system.

It is generally assumed that Ras proteins in bulk solution havea less open conformation of the switch regions I and II and hencea smaller flexibility in the GTP-bound compared to the GDP-bound form (19, 21). This is also the case for N-Ras as evidencedfrom the amplitude of the frequency shifts in the second deriva-tive FT-IR spectra (Fig. 2 A and B) and from the overlay of dif-ferent Ras crystal structures in the GDP and GTP-bound form(Figs. 1 B and C). Pressure application to the conformationalensemble of N-Ras (GTP) shifts the equilibrium toward a moresolvent exposed and open conformation, in agreement with a pre-vious NMR report on H-Ras (22). It is a well-known effect of highhydrostatic pressure to weaken hydrophobic interactions, thusallowing water to penetrate the protein interior, in particular atplaces where voids and cavities are present (11, 36). As a result,pressurization is expected to shift the conformational equilibriumtoward the more open and solvent exposed conformations. Re-cent analysis of the protein structure database indicates that theresidues of β-strands have a higher tendency to line the cavitiescompared to others, probably due to a lower packing efficiency(37). This could explain the marked increase of exposed β-sheetsupon pressurization of both the GTP- and GDP-forms of N-Ras.

As indicated by our FTIR spectroscopic data, in N-Ras (GTP),pressure induces conformations with more solvent exposed struc-

Fig. 3. Secondary structural changes as a function of pressure at T ¼ 25 °Cfor N-Ras GDP and N-Ras GTP (A and C) in bulk, for N-Ras GDP in the presenceof lipid bilayer (B), and for N-Ras GTP in the presence of the lipid bilayer (D).Pressure-induced wavenumber shifts for the various secondary structure arealso shown for N-Ras GDP (E) and GTP (G) in bulk, for N-Ras GDP in thepresence of the lipid bilayer (F), and for N-Ras GTP in the presence of thelipid bilayer (H).

Kapoor et al. PNAS ∣ January 10, 2012 ∣ vol. 109 ∣ no. 2 ∣ 463

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

tures that could correspond to state 1 according to ref. 21, whichis known to have an enlarged protein surface due to the openingof the nucleotide binding pocket (necessary for stable interactionwith GEFs). The corresponding volume change (up to 10 kbar)driving the conformational transition amounts to −14.7 mL∕moland agrees very well with the 31 P NMR data (up to 2 kbar) on(unlipidated) Ras (GTP) in showing that with increasing pressurethe conformational equilibrium shifts from state 2 to state 1 (21).

The pressure-dependent conformational changes for bulk N-Ras (GDP) are significantly different fromN-Ras (GTP), with thesplitting of the subband in the α-helical region of the proteinstarting already at much lower pressures (Fig. 2A). Moreover, thepopulation of exposed β-sheets increases with pressure at a higherrate and reaches a higher level (10%) compared with N-Ras(GTP) (4%) at the highest pressure measured (Fig. 3 A and C).These results imply a higher pressure sensitivity of the GDP-bound N-Ras, which can be explained on the following grounds.Both the Ras proteins typically have free volumes (i.e., cavities)of less than 1% of their total volume in the available X-ray struc-tures, and the differences are probably not significant. If ever, theGTP-forms tend to have a slightly higher cavity volume thanthe GDP form, so that a (slightly) lower pressure sensitivity wouldbe foreseen for N-Ras (GDP). As this is not observed, the higherpredisposition to form open and solvent exposed conformationsin N-Ras (GDP) must be due to other reasons. One very likelypossibility is the easier opening of the highly flexible switch Iregion in N-Ras (GDP), which is clearly visible by comparingthe initial slopes, Δ½exposed sheets�∕Δp (up to 2 kbar) (Fig. 3 Aand C), which is markedly higher in the GDP-bound form. Thiswould also be in agreement with the 31P NMR data (21) wherethe maximum pressure of 2 kbar reached was able to completelypopulate state 1 in the—even less pressure sensitive—GTP-bound form. This explanation is further supported by ultrasoundvelocimetry studies that provide insight into the flexibility–com-pressibility relationship of proteins (38), suggesting higher com-pressibilities of the more flexible regions of the proteins. Highpressure induces state 1 in both Ras proteins, but due to thehigher flexibility in the GDP form, state 1 is populated moreeasily there.

As discussed above, pressure-dependent studies were per-formed for unlipidated H-Ras-GppNHp unlipidated in bulksolution and up to 2 kbar, only (21). Here, also the effects ofmembrane binding on the pressure-dependent conformationalchanges of both active (GTP form) and inactive (GDP form) fullylipidated N-Ras was investigated using FTIR spectroscopy up toabout 10 kbar, a pressure range that is currently not accessibleto NMR technology. As could be clearly demonstrated, not onlynucleotide binding but also membrane binding modulates theconformational dynamics of N-Ras markedly, and, interestinglyenough, brings about new and opposite changes in the conforma-tional properties of N-Ras (GDP) and (GTP), respectively. Uponmembrane binding, the pressure sensitivity of the two nucleotidestates of N-Ras is unexpectedly reversed. These differences couldarise from a different overall orientation of the G domain withrespect to the membrane as originally proposed by Gorfe et al.(26). Complementary results from polarized ATR-FTIR (Fig. S3)and IRRAS (Fig. S4) provide conclusive hints for the reorienta-tional changes in the Ras G-domain upon membrane insertionmediated by the bound nucleotide. The pressure-induced non-linear shift of the intramolecular β-sheet band in both membrane-bound N-Ras forms (Fig. 3 B and D) indicates that membranebinding modulates the conformational equilibrium of N-Ras(GDP/GTP) by inducing a new, membrane-associated conforma-tional substate, state 3.

A schematic representation of the pressure modulation ofconformational substates of Ras is depicted in Fig. 4. For the bulkN-Ras, increasing pressure populates the more open and solvatedconformer, which is state 1 in both cases, the effect being more

pronounced for the inactive N-Ras (GDP). Owing to the highintrinsic flexibility of the switch loops in N-Ras (GDP) and therapid interconversion between the different conformers, the en-ergy funnels of these substates are expected to be broader. Uponmembrane insertion, a new functional conformer, substate 3, isinduced in both cases, which is increasingly populated upon pres-surization. At higher pressures, the transition to state 1 is fosteredsince it then corresponds to the state with the lowest free energy.

To conclude, our data clearly show that, next to the nucleotidestate, the membrane also plays an important role in the confor-mational selection and dynamics of N-Ras. Membrane insertionitself at ambient pressure does not lead to significant secondarystructural changes in the G-protein. However, pressure applica-tion induces minor and local changes of certain secondary struc-tural elements and shifts the equilibrium toward conformationswith more open and solvated structures having smaller volumes.Membrane insertion leads to a marked population of a newsubstate, state 3. This state is increasingly populated upon pres-surization—and hence becomes clearly detectable—and is ac-companied by a significant structural reorientation of the Gdomain at the lipid interface, in particular in the GTP-loadedform of N-Ras. Efficient interaction with GEFs and effectors re-quires selection of conformations like states 1 and 2, respectively,and shifts the equilibrium toward these states. Our study indicatesthat upon membrane binding, such conformational selection isadditionally modulated, preselecting conformational substate 3with significant structural rearrangements. By these conforma-tional and orientational changes, the interaction with membrane-associated interaction partners, such as effectors and activitymodulators (e.g., galectins), may be influenced. In fact, recent invivo studies have shown that G-domain orientation of membrane-associated Ras proteins is important for effector/modulatorrecognition, such as of phosphoinositide-3-kinase and galectin-1, and G-domain orientation and nanoclustering—that has alsobeen observed in our AFM studies on model biomembrane

Fig. 4. Schematic representation of the modulation of conformational sub-states of active (GTP) and inactive (GDP) N-Ras by pressure in bulk solutionand in the presence of the lipid bilayer membrane. Small arrows indicatechanges at low pressure values, and longer arrows indicate changes occurringafter application of higher pressures. The membrane-associated substate 3 ismarked in red.

464 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110553109 Kapoor et al.

systems (23, 24)—generate Ras isoform specificity with respect toeffector interactions (27).

Finally, these results shed also some light on the effect of pres-sure on membrane-associated Ras-controlled signaling eventsunder extreme environmental conditions. Although pressure isan important environmental parameter, such as in the deep seawhere organisms have to cope with pressures up to the 1 kbarrange, the fundamental understanding of its effects remains lar-gely fragmentary. The effects of pressure on gene expression oron biomolecules such as lipid membranes, proteins and DNAhave been studied in recent years, the effects of pressure on thedynamics of metabolic events and signaling processes remainlargely unexplored, however (39–42). Our data indicate that in-creased hydrostatic pressure shifts the conformational equili-brium toward the more open water exposed state 1, which issupposed to interact most effectively with GEFs. Furthermore,high pressure leads to substantial stabilization of the otherwiselowly populated conformer 3 induced by membrane interactionin both Ras isoforms, which is expected to foster the interactionwith membrane-associated interaction partners and prevails up toa few kilobars.

Materials and MethodsMaterials and Sample Preparation. The phospholipids 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) and 1,2-dipalmitoyl-sn-glycero-3-phosphocholine(DPPC) were purchased from Avanti Polar Lipids. Cholesterol (Chol), otherreagents, and solvents were obtained from Sigma-Aldrich and Merck. Stocksolutions of 10 mgmL−1 lipids in chloroform were prepared and mixed to

obtain the desired composition of the DOPC/DPPC/Chol (1∶2∶1 molar ratio)lipid mixture.

N-Ras Protein Synthesis of Double Lipidated N-Ras (N-Ras HD/Far). Double lipi-dated N-Ras peptides were synthesized with an N-terminal maleimide groupvia a combination of solution and solid phase strategies. The maleimidopeptides were coupled to a bacterially expressed truncated N-Ras protein(amino acid residues 1–181) carrying a cysteine at the C terminus. Couplingwas achieved by conjugate addition of the only surface-accessible cysteine(181)-SH group of N-Ras to the maleimido function. Details on the nucleotideexchange on Ras proteins can be found in refs. 43 and 44.

High Pressure FTIR Spectroscopy. The dry lipid mixture was hydrated with Trisbuffer, pD 7.4. The hydrated lipid mixture was then subjected to five freeze-thaw-vortex cycles and brief sonication. Largemultilamellar vesicles obtainedherewith were transformed to large unilamellar vesicles of homogeneoussize by use of an extruder (Avanti Polar Lipids) with polycarbonate mem-branes of 100 nm pore size at 65 °C. For the pressure-dependent studies,the GDP- and GTP-loaded N-Ras protein (c ¼ 3 wt%) were lyophilized for2 h and mixed with extruded lipid mixture yielding a molar ratio of proteinto lipid of 1∶100. The pressure-dependent FTIR spectra were recorded with aMagna IR 550 spectrometer equipped with liquid nitrogen cooled MCT(HgCdTe) detector. For the measurements, the infrared light was focussedby a spectral bench onto the pin hole of a diamond anvil cell (with type IIadiamonds from Diamond Optics). For details of operation and data analysissee ref. 9.

ACKNOWLEDGMENTS. I.R.V., H.W., and R.W. would like to thank the DeutscheForschungsgemeinschaft (SFB 642, DFGWI 742/16) and the International MaxPlanck Research School in Chemical Biology for financial support.

1. Frauenfelder H, Parak F, Young RD (1988) Conformational substates in proteins. AnnuRev Biophys Biophys Chem 17:451–479.

2. Frauenfelder H, et al. (1990) Protein and pressure. J Phys Chem 94:1024–1037.3. Silva JL, Foguel D, Royer CA (2001) Pressure provides new insights into protein folding,

dynamics and structure. Trends Biochem Sci 26:612–618.4. Privalov PL, Gill SJ (1988) Stability of protein structure and hydrophobic interaction.

Adv Protein Chem 39:191–234.5. Ravindra R, Winter R (2003) On the temperature-pressure free-energy landscape of

proteins. ChemPhysChem 4:359–365.6. McCoy J, Hubbell WL (2011) High-pressure EPR reveals conformational equilibria and

volumetric properties of spin-labeled proteins. Proc Natl Acad Sci USA 108:1331–1336.7. Collins MD, Kim CU, Gruner SM (2011) High-pressure protein crystallography and NMR

to explore protein conformations. Annu Rev Biophys 40:81–98.8. Winter R, Dzwolak W (2004) Temperature-pressure configurational landscape of lipid

bilayers and proteins. Cell Mol Biol 50:397–417.9. Panick G, et al. (1998) Structural characterization of the pressure-denatured state and

unfolding/refolding kinetics of staphylococcal nuclease by synchrotron small-angleX-ray scattering and Fourier-transform infrared spectroscopy. J Mol Biol 275:389–402.

10. Schroer MA, et al. (2010) High pressure SAXS study of folded and unfolded ensemblesof proteins. Biophys J 99:3430–3437.

11. Akasaka K (2006) Probing conformational fluctuation of proteins by pressure pertur-bation. Chem Rev 106:1814–1835.

12. Frauenfelder H, Sligar SG, Wolynes PG (1991) The energy landscapes and motions ofproteins. Science 254:1598–1603.

13. Miller DW, Dill KA (1997) Ligand binding to proteins: The binding landscape model.Protein Sci 6:2166–2179.

14. Boehr DD, Nussinov R, Wright PE (2009) The role of dynamic conformational ensem-bles in biomolecular recognition. Nat Chem Bio 5:789–796.

15. Grant BJ, McCammon JA, Gorfe AA (2010) Conformational selection in G-proteins:Lessons from Ras and Rho. Biophys J 99:L87–89.

16. Wittinghofer A, Pai EF (1991) The structure of Ras protein: A model for a universalmolecular switch. Trends Biochem Sci 16:382–387.

17. Chiu VK, et al. (2002) Ras signalling on the endoplasmic reticulum and the golgi. NatCell Biol 4:343–350.

18. Vetter IR, Wittinghofer A (2001) The guanine nucleotide-binding switch in threedimensions. Science 294:1299–1304.

19. Geyer M, et al. (1996) Conformational transitions in p21ras and in its complexes witheffector protein Raf-RBD and the GTPase activating protein GAP. Biochemistry35:10308–10320.

20. Spoerner M, et al. (2001) Dynamic properties of the Ras switch I region and itsimportance for binding to effectors. Proc Natl Acad Sci USA 98:4944–4949.

21. Kalbitzer HR, et al. (2009) Fundamental Link between folding states and functionalstates of proteins. J Am Chem Soc 131:16714–16719.

22. Gorfe AA, Grant BJ, McCammon JA (2008) Mapping the nucleotide and isoform-dependent structural and dynamical features of Ras proteins. Structure 16:885–896.

23. Weise K, et al. (2009) Influence of the lipidation motif on the partitioning and associa-tion of N-Ras in model membrane subdomains. J Am Chem Soc 131:1557–1564.

24. Weise K, et al. (2011) Membrane-mediated induction and sorting of K-Ras micro-domain signaling platforms. J Am Chem Soc 133:880–887.

25. Rotblat B, et al. (2004) Three separable domains regulate GTP-dependent associationof H-ras with the plasma membrane. Mol Cell Bio 24:6799–6810.

26. Gorfe AA, et al. (2007) Structure and dynamics of full-length lipid-modified H-Rasprotein in a 1,2-Dimyristoyglycero-3-phosphocholine bilayer. J Med Chem 50:674–684.

27. Abankwa D, Gorfe AA, Inder K, Hancock JF (2010) Ras membrane orientation andnanodomain localization generate isoform diversity. Proc Natl Acad Sci USA107:1130–1135.

28. Abankwa D, et al. (2008) A novel switch region regulates H-ras membrane orientationand signal output. EMBO J 27:727–735.

29. Jeworrek C, PühseM,Winter R (2008) X-ray kinematography of phase transformationsof three-component lipid mixtures: A time-resolved synchrotron X-ray scattering studyusing the pressure-jump relaxation technique. Langmuir 24:11851–11859.

30. Milburn MV, et al. (1990) Molecular switch for signal transduction: Structural differ-ences between active and inactive forms of protooncogenic ras proteins. Science247:939–945.

31. Heremans K, Smeller L (1998) Protein structure and dynamics at high pressure. BiochimBiophys Acta 1386:353–370.

32. Dzwolak W, Kato M, Taniguchi Y (2002) Fourier transform infrared spectroscopy inhigh-pressure studies on proteins. Biochim Biophys Acta 1595:131–144.

33. Güldenhaupt J, et al. (2008) Seconday structure of lipidated Ras bound to a lipidbilayer. FEBS 275:5910–5918.

34. Baxter NJ, Hosszu LLP, Waltho JP, Williamson MP (1998) Characterisation of low free-energy excited states of folded proteins.. J Mol Biol 284:1625–1639.

35. Inoue K, et al. (2000) Pressure-induced local unfolding of the Ras binding domain ofRalGDS. Nat Struct Biol 7:547–550.

36. Hummer G, et al. (1998) The pressure dependence of hydrophobic interactions is con-sistent with the observed pressure denaturation of proteins. Proc Natl Acad Sci USA95:1552–1555.

37. Sonavane S, Chakarbarti P (2008) Cavities and atomic packing in protein structures andinterfaces. PLOS Comp Biol 4:1–15.

38. Gekko K, Kamiyama T, Ohmae E, Katayanagi K (2000) Single Amino acid substitutionsin flexible loops can induce large compressibility changes in dihydrofolate reductase.J Biochem 128:21–27.

39. Abe F, Kato C, Horikoshi K (1999) Pressure-regulated metabolism in microorganisms.Trends Microbiol 7:447–453.

40. Daniel I, Oger P, Winter R (2006) Origins of life and biochemistry under high- pressureconditions. Chem Soc Rev 35:858–875.

41. Linke K, et al. (2008) Influence of high pressure on the dimerization of ToxR, a proteininvolved in bacterial signal transduction. Appl Environm Microbiol 74:7821–7823.

42. Siebenaller JF, Garrett DJ (2002) The effects of the deep-sea environment on trans-membrane signaling. Comp Biochem Physiol B 131:675–694.

43. Brunsveld L, et al. (2006) Lipidated Ras and Rab peptides and proteins—synthesis,structure, and function.. Angew Chem Int Ed 45:6622–6646.

44. Nägele E, Schelhaas M, Kuder N, Waldmann H (1998) Chemoenzymatic synthesis ofN-Ras lipopeptides.. J Am Chem Soc 120:6889–6902.

Kapoor et al. PNAS ∣ January 10, 2012 ∣ vol. 109 ∣ no. 2 ∣ 465

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY