Monitoring Ligand Modulation of Protein-ProteinInteractions by Mass Spectrometry: EstrogenReceptor r-SRC1

Cedric Bovet,† Marc Ruff,‡ Sylvia Eiler,‡ Florence Granger,‡ Ryan Wenzel,§ Alexis Nazabal,§

Dino Moras,‡ and Renato Zenobi*,†

Department of Chemistry and Applied Biosciences, ETH Zurich, 8093 Zurich, Switzerland, IGBMC (Institut deGenetique et de Biologie Moleculaire et Cellulaire), Departement de Biologie et Genomique Structurales, UniversiteLouis Pasteur, U596 INSERM, UMR7104 CNRS, 1 rue Laurent Fries, 67404 Illkirch, France, and CovalX AG,Technoparkstrasse 1, 8005 Zurich, Switzerland

Many drugs and chemicals exert their biological effect bymodulating protein-protein interactions. In vitro ap-proaches to characterize these mechanisms are oftenbased on indirect measurements (e.g., fluorescence).Here, we used mass spectrometry (MS) to directly monitorthe effect of small-molecule ligands on the binding of acoactivator peptide (SRC1) by the human estrogen recep-tor r ligand binding domain (hERr LBD). Nanoelectro-spray mass spectrometry (nanoESI-MS) and high-massmatrix-assisted laser desorption/ionization mass spec-trometry (MALDI-MS) combined with chemical cross-linking were employed to follow these processes. Thechemical cross-linking protocol used prior to high-massMALDI analysis allows detection of intact noncovalentcomplexes. The binding of intact hERr LBD homodimerwith two coactivator peptides was detected with nanoESI-MS and high-mass MALDI-MS only in the presence of anagonist ligand. Furthermore, high-mass MALDI-MS re-vealed an increase of the homodimer abundance afterincubating the receptor with a ligand, independent of theligand character (i.e., agonist, antagonist). The bindingcharacteristics of the compounds tested by MS correlatevery well with their biological activity reported by cell-based assays. High-mass MALDI appears to be an efficientand simple tool for directly monitoring ligand regulationmechanisms involved in protein-protein interactions.Furthermore, the combination of both MS methods allowsidentifying and characterizing endocrine-disrupting com-pounds or new drug compounds in an efficient way.

The modulation of protein-protein interactions is crucial inmany biological processes, and such interactions are ofteninvolved in diseases as well. Therefore, the characterization ofthe regulation mechanisms involved is a key step in the discoveryof new drug candidates. Available biophysical methods (e.g.,surface plasmon resonance, analytical ultracentrifugation, fluo-

rescence spectroscopy) usually require one of the interactionpartners to be immobilized or labeled with a specific probe. Thisonly allows an indirect detection of the protein complex and mayalter the binding interactions. A label-free method that allows thedirect characterization of noncovalent complexes is nanoelectro-spray mass spectrometry (nanoESI-MS). Usually, nanoESI-MS ispreferred over matrix-assisted laser desorption/ionization massspectrometry (MALDI-MS) for studying noncovalent complexes,mainly because ESI is able to spray “nativelike” aqueous solutions.However, nanoESI-MS has limitations for monitoring ligandmodulation processes of protein-protein interactions. Discrimina-tion processes (e.g., mass-dependent ionization efficiency, mass-dependent ion transmission through the mass spectrometer, andnonuniform response of the detector) do not generally allowrelating the ion intensities of different species to their solutionconcentrations.1 The detection of intact noncovalent complexeswith MALDI-MS was reported in only few cases and required theuse of special matrixes, sample preparation methods, or collectingonly first laser shots on fresh sample spots.2–5 Recently, theapplication of MALDI-MS has been extended to the area of largenoncovalent complexes by combining cross-linking and high-massdetection.6,7 The more uniform sensitivity of the high-massdetection over a broad mass range allows relating the MALDIrelative intensities of observed species to their solution concentra-tions. Therefore, high-mass MALDI-MS might be a powerfulapproach for studying the modulation of ligand-dependentprotein-protein interactions and should complement nanoESI-MS very well.

This combination of methods is applied here to a complexsystem, the ligand-dependent recruitment of coactivators by anuclear receptor (NR). Using MS, we monitored the ligand

* To whom correspondence should be addressed. Phone: +41 44 632 43 76.Fax: +41 44 632 12 92. E-mail: zenobi@org.chem.ethz.ch.

† ETH Zurich.‡ Universite Louis Pasteur.§ CovalX AG.

(1) Gabelica, V.; Galic, N.; Rosu, F.; Houssier, C.; De Pauw, E. J. Mass Spectrom.2003, 38, 491–501.

(2) Cohen, L. R. H.; Strupat, K.; Hillenkamp, F. J. Am. Soc. Mass Spectrom.1997, 8, 1046–1052.

(3) Farmer, T. B.; Caprioli, R. M. J. Mass Spectrom. 1998, 33, 697–704.(4) Jespersen, S.; Niessen, W. M. A.; Tjaden, U. R.; van der Greef, J. J. Mass

Spectrom. 1998, 33, 1088–1093.(5) Sannes-Lowery, K. A.; Griffey, R. H.; Hofstadler, S. A. Anal. Biochem. 2000,

280, 264–271.(6) Nazabal, A.; Wenzel, R. J.; Zenobi, R. Anal. Chem. 2006, 78, 3562–3570.(7) Yanes, O.; Nazabal, A.; Wenzel, R.; Zenobi, R.; Aviles, F. X. J. Proteome

Res. 2006, 5, 2711–2719.

Anal. Chem. 2008, 80, 7833–7839

10.1021/ac8007169 CCC: $40.75 2008 American Chemical Society 7833Analytical Chemistry, Vol. 80, No. 20, October 15, 2008Published on Web 09/09/2008

regulation processes involved in the estrogen receptor (ER), aligand-activated transcription regulator. Upon ligand binding, ERundergoes a conformational change that modulates thedimerization8–10 and the recruitment of coactivators required fortranscriptional activation of the target genes.11,12 The ability ofER to bind the coactivator is dependent on the orientation of helix12 in the ligand binding domain (LBD), which is sensitive to thenature of the bound ligand (i.e., agonist or antagonist). Onlyagonist binding allows the LBD to interact with a conserved motifLXXLL (known as “NR box”, where L is leucine and X is any aminoacid) of coactivators, a motif necessary and sufficient to mediatethe binding of the coactivators to ligand-bound receptor.13,14

Therefore, once incubated with a ligand, the mass spectrometricdetection of an intact noncovalently bound NR box peptide to theER LBD can be used to discern agonist from antagonist ligands.Because the ER LBD dimerization is ligand-dependent, MS canalso identify ER ligands by measuring an increase of the ho-modimer abundance after ligand incubation.

By combining the advantages of nanoESI and MALDI-MS, weare able to clearly demonstrate and characterize the ligandregulation mechanisms of the ER at the molecular level. We showthat high-mass MALDI-MS can be used to study the ligand-dependent ER LBD dimerization in solution. Because a widevariety of nonsteroidal compounds, known as endocrine-disruptingcompounds (EDCs), interact with ER, molecules with a suspectedendocrine-disrupting activity may be efficiently identified with ourMS-based approach. These EDCs mimic the action of naturalhormones by binding with distinct affinities to the ER and mayinduce adverse health effects in humans and wildlife.15,16 There-fore, such a simple in vitro methodology will be a valuable toolfor industry and research to discover new drug candidates orevaluate biological effects of environmental contaminants, asrequired, for example, by the new European Union chemicalspolicy REACH (Registration, Evaluation and Authorisation ofChemicals).

EXPERIMENTAL SECTIONMaterials. The triple mutant human estrogen receptor R

ligand binding domain (hERR LBD) was overexpressed inEscherichia coli and purified as previously described.17 Aminoacids sequence 302-553 of hERR corresponding to the LBD(containing 3 cysteine to serine mutations at positions 381, 417,and 530) was generated in fusion with thioredoxin, six histidine

residues, and a thrombin cleavage sequence (Mr ) 43 947). Thetriple mutant hERR LBD without the fusion protein (Mr ) 30 025)was generated by thrombin cleavage. A steroid receptor coacti-vator peptide (CAP) was purchased from Anaspec (San Jose, CA)and had the amino acid sequence LTERHKILHRLLQE (Mr )1786.1), which contains the typical binding motif LXXLL. The CAPwas derived from the sequence of SRC-1 coactivator protein, amember of the p160 class of nuclear hormone receptor coactiva-tors, and contained the second NR box of SRC-1, which has thehighest affinity for ERR.18 Lyophilized CAP was dissolved in waterto a final concentration of 100 µM. Stock solutions of the differentligands were prepared in ethanol. 17�-Estradiol (E2), 4-hydrox-ytamoxifen (OHT), diethylstilbestrol (DES), hexestrol (HEX),dienestrol (DIE), and clomiphene citrate (CLO) were obtainedfrom Sigma (Buchs, Switzerland), genistein (GE) and zearalenone(ZEA) from Fluka (Buchs, Switzerland), and simazin (SIM) andnitrofen (NIT) from Riedel-de Haen Chemicals (Seelze, Germany).

Sample Preparation. Prior to nanoESI-MS measurements,the protein stock solution was desalted and incubated with 1 µMligand as previously described.19 After a 15-min incubation at roomtemperature, 1 µM CAP was added and the resultant mixturefurther incubated for 15 min. Prior to MALDI-MS measurements,the protein stock solution was diluted to 5 µM with water. Tenmicromolar ligand and CAP were incubated with the protein asdescribed above for nanoESI-MS. The cross-linking reactions wererealized using a solution containing different cross-linkers specificfor amino and sulfhydryl groups (K200 MALDI-MS analysis kit,CovalX AG, Zurich, Switzerland). The cross-linking reagentmixture was composed of suberic acid bis(3-sulfo-N-hydroxysuc-cinimide ester), suberic acid bis(N-hydroxysuccinimide ester),disuccinimidyl tartrate, and dithiobis(succinimidyl propionate).The protein complexes were stabilized with 1 µL of cross-linkingmixture in a total volume of 10 µL. The sample and the cross-linker were incubated for 60 min at room temperature to achievea complete reaction. After cross-linking, 1 µL of the samplecontaining the protein complex was directly mixed with 1 µL ofmatrix solution. The matrix solution contained sinapinic acid (10mg/mL) in acetonitrile/water (1:1, v/v) acidified with 0.1% (v/v)trifluoroacetic acid. After mixing, 1 µL of the sample was droppedonto the sample plate for MALDI-MS analysis. The final ethanolvolume of the ligand solution was kept constant at 2% of the totalvolume for nanoESI and MALDI measurements.

Mass Spectrometry. ESI mass spectra were acquired on aquadrupole time-of-flight mass spectrometer (Q-ToF Ultima,Waters, Manchester, UK) equipped with an automated chip-basednanoESI system (Nanomate 100, Advion Biosciences, Ithaca, NY).To prevent dissociation of the noncovalent complexes, the massspectrometer was tuned with gentle desolvation parameters aspreviously described.19 MALDI mass spectra were acquired on atime-of-flight mass spectrometer (Reflex IV, Bruker DaltonicsGmbH, Bremen, Germany) equipped with a high-mass detectionsystem (HM1, CovalX AG, Zurich, Switzerland). The high-massdetection system provides enhanced sensitivity in the high-massrange up to >1 MDa and avoids detector saturation due to intense

(8) Brandt, M. E.; Vickery, L. E. J. Biol. Chem. 1997, 272, 4843–4849.(9) Tamrazi, A.; Carlson, K. E.; Daniels, J. R.; Hurth, K. M.; Katzenellenbogen,

J. A. Mol. Endocrinol. 2002, 16, 2706–2719.(10) Wang, H.; Peters, G. A.; Zeng, X.; Tang, M.; Ip, W.; Khan, S. A. J. Biol.

Chem. 1995, 270, 23322–23329.(11) McKenna, N. J.; Lanz, R. B.; O’Malley, B. W. Endocr. Rev. 1999, 20, 321–

344.(12) Robyr, D.; Wolffe, A. P.; Wahli, W. Mol. Endocrinol. 2000, 14, 329–347.(13) Paige, L. A.; Christensen, D. J.; Gron, H.; Norris, J. D.; Gottlin, E. B.; Padilla,

K. M.; Chang, C. Y.; Ballas, L. M.; Hamilton, P. T.; McDonnell, D. P.;Fowlkes, D. M. Proc. Natl. Acad. Sci. U. S. A 1999, 96, 3999–4004.

(14) Heery, D. M.; Kalkhoven, E.; Hoare, S.; Parker, M. G. Nature 1997, 387,733–736.

(15) Colborn, T.; Saal, F. S. V.; Soto, A. M. Environ. Health Perspect. 1993,101, 378–384.

(16) Harrison, P. T. C.; Holmes, P.; Humfrey, C. D. N. Sci. Total Environ. 1997,205, 97–106.

(17) Gangloff, M.; Ruff, M.; Eiler, S.; Duclaud, S.; Wurtz, J. M.; Moras, D. J. Biol.Chem. 2001, 276, 15059–15065.

(18) Bramlett, K. S.; Wu, Y.; Burris, T. P. Mol. Endocrinol. 2001, 15, 909–922.(19) Bovet, C.; Wortmann, A.; Eiler, S.; Granger, F.; Ruff, M.; Gerrits, B.; Moras,

D.; Zenobi, R. Protein Sci. 2007, 16, 938–946.

7834 Analytical Chemistry, Vol. 80, No. 20, October 15, 2008

low-mass (e.g., matrix) species.20 The low saturation is critical toensure detection over a broad dynamic range. Standard MALDIinstrument conditions were used.

Data Processing. Before data processing, each MALDI andnanoESI mass spectrum was background subtracted and smoothedwith the Complex Tracker 1.1 (CovalX AG, Zurich, Switzerland)or the MassLynx 4.0 software (Waters, UK), respectively. Peakintegrations were performed with Igor Pro 4.09A software (Wave-Metrics, USA). For MALDI-MS and nanoESI-MS, the abundanceof the hERR LBD-CAP complex was evaluated on the singlycharged monomer ion and the 14+ monomer ion, respectively.The ligand-bound to free protein ratio (R) was calculated on the14+ monomer ion as previously described.19

RESULTS AND DISCUSSIONLigand Binding Regulates hERr LBD Dimerization. Cross-

linking prevents the noncovalent complex from dissociationinduced by standard MALDI protocols,6,7 and coupled with high-mass MALDI-MS, the direct detection of the native hERR LBDhomodimer becomes possible (Figure 1a). With standard MALDI,this complex would dissociate during sample preparation or duringthe ionization process. After incubating the receptor with E2, wemeasured a substantial increase in homodimer (Figure 1b,increase of 55 ± 2% relative to the aporeceptor). To study the effectof ligand binding on the dimerization, we incubated the receptorwith a selection of ER ligands with known or suspected estrogenicaction. In addition to E2, test compounds included the syntheticestrogens HEX, DIE, and DES; the antiestrogen drugs OHT and

CLO; the isoflavone GE found in the soybean; the mycoestrogenZEA found in a number of cereal crops; and the herbicide SIMand pesticide NIT (for abbreviations, see Table 1). A markedincrease in homodimer abundance was measured with high-massMALDI-MS after ligand incubation in all cases, except for SIMand NIT (Figure 1c). Judging from the lack of influence on thedimer abundance for these two compounds, we propose that SIMand NIT do not interact with the binding pocket of ER. In previousstudies, NIT was reported to be an antiestrogenic compound onlyat concentration above 100 µM;21 no potency was measured forSIM,22 in agreement with our findings.

The MALDI relative intensities of the homodimer and mono-mer provide a semiquantitative measurement of their solutionconcentrations. This is largely due to a more uniform sensitivityover a broad mass range of the high-mass detection versusconventional detection (e.g., a microchannel plate). The highrepeatability of the high-mass MALDI-MS measurements alsorenders them applicable for monitoring the effect of smallmolecules on protein-protein interactions. The increase of dimerabundance once incubated with a ligand suggests that one roleof ligand binding is to increase the dimer binding affinity. Becauseonly ER ligands enhance the homodimer abundance, suspectedcompounds with an endocrine-disrupting activity could easily beidentified by measuring the variation of the homodimer abundancewith high-mass MALDI-MS. To our knowledge, it is the first timethat a protein dimerization process is monitored with MS.

(20) Bahr, U.; Rohling, U.; Lautz, C.; Strupat, K.; Schurenberg, M.; Hillenkamp,F. Int. J. Mass Spectrom. Ion Processes 1996, 153, 9–21.

(21) Okubo, T.; Yokoyama, Y.; Kano, K.; Soya, Y.; Kano, I. Arch. Environ. Contam.Toxicol. 2004, 46, 445–453.

(22) Blair, R. M.; Fang, H.; Branham, W. S.; Hass, B. S.; Dial, S. L.; Moland,C. L.; Tong, W. D.; Shi, L. M.; Perkins, R.; Sheehan, D. M. Toxicol. Sci.2000, 54, 138–153.

Figure 1. High-mass MALDI-MS showing a ligand-dependent dimerization in solution for the mutant hERR LBD after cross-linking reaction. (a)MALDI mass spectrum of hERR LBD after cross-linking. The intact singly charged homodimer is labeled [2M + H]+, the singly charged monomer[M + H]+ (base peak), and the doubly charged monomer [M + 2H]2+. The asterisks indicate sample impurities. (b) After incubation with E2 andcross-linking, the MALDI mass spectrum clearly shows an increase of the homodimer. (c) After incubation with the test compounds and cross-linking, only SIM and NIT did not increase significantly the homodimer abundance relative to the aporeceptor.

7835Analytical Chemistry, Vol. 80, No. 20, October 15, 2008

Ligand-Dependent Recruitment of the Coactivator Pep-tide. Although the binding of agonist and antagonist ligands inthe LBD induce distinct conformational changes in the receptor,23–25

the dimer increase does not appear to reflect the ligand character

(given in Table 1) in agreement with previous data.8,10 We thusmonitored the ligand-dependent recruitment of a coactivatorpeptide by the receptor with nanoESI-MS and high-mass MALDI-MS. To validate an agonist-dependent interaction of CAP with themutant hERR LBD, we first incubated the receptor with a knownagonist and an antagonist ligand (E2 and OHT, respectively). Thechemical cross-linking reaction allows studying the ligand regula-tion of the hERR LBD-CAP interaction with MALDI-MS (Figure2a-c). After incubation with E2 and CAP, the MALDI massspectrum of hERR LBD clearly displays an additional peakassigned to the specific complex formed with the coactivatorpeptide (Figure 2b). The high-mass shoulder on the dimer peakin Figure 2b may indicate binding of two CAP, although theMALDI data do not allow a clear identification of such species.As expected, the presence of OHT largely suppressed the bindingof hERR LBD to CAP (Figure 2c).

Similar experiments were repeated with nondenaturing nano-ESI-MS (Figure 2d-f), which allows the detection of homodimerand intact ligand-bound receptor complexes.19 Because the ho-

Table 1. Characteristics of the Test Compoundsagainst ERr

ligand abbreviation character

17�-estradiol E2 agonistahexestrol HEX agonistbdienestrol DIE agonistbdiethylstilbestrol DES agonistc4-hydroxytamoxifen OHT antagonistcclomiphene citrate CLO antagonistbgenistein GE agonistdzearalenone ZEA f(concentration)e

simazin SIM unboundb

nitrofen NIT antagonist g100 µMf

a Reference 23. b Reference 22. c Reference 25. d Reference 24.e Reference 29. f Reference 21.

Figure 2. High-mass MALDI-MS and nanoESI-MS showing that an agonist ligand enhances the peptide recruitment by the mutant hERR LBD.Normalized MALDI mass spectra of hERR LBD after cross-linking reaction and incubation (a) with the agonist E2; (b) with E2 and CAP; (c) withthe antagonist OHT and CAP. Normalized nanoESI mass spectra of hERR LBD representing the monomer charge state 14+ after incubation(d) with the agonist E2; (e) with E2 and CAP; (f) with the antagonist OHT and CAP. ([) and (b) represent the bound E2 and OHT, respectively.Italic numbers indicate the measured m/z of the different species.

7836 Analytical Chemistry, Vol. 80, No. 20, October 15, 2008

modimer ion distribution (m/z 4000-5600) is not sufficientlyresolved to discern the ligand-bound from the apohomodimer, ouranalysis is based on the monomer range (m/z 2500-4000).19 Forsimplicity, only the monomer charge state 14+ is shown; the othercharge states exhibited identical behavior. After incubation withE2 and CAP, the apo and ligand-bound monomer were detectedwith a new pair of peaks appearing at higher m/z values (Figure2e, m/z 3250-3350). These new species correspond to the specifichERR LBD-CAP complex with and without ligand. Incubation withthe antagonist OHT only led to a very small intensity of the hERRLBD-CAP complex; i.e. there is very little binding of CAP tohERR LBD (Figure 2f). Therefore, both MS approaches revealhow a short coactivator peptide can be used to directly differentiateligand-induced conformational changes in the receptor. The ligandmodulation of the interactions with a coactivator has previouslybeen attributed to a ER LBD conformational change triggered byligand binding, which in antagonist-bound ER partially blocks thecoactivator binding site.25 MS allows the direct detection of thepeptide bound to hERR LBD in contrast to time-resolved fluores-cence assays, which measured a ligand-dependent affinity variationof CAP against the receptor.13,18,26,27 The ligand-dependentrecruitment of the coactivator peptide by the hERR LBD allowsdiscerning the agonist from antagonist character of uncharacter-ized ER ligand. The mass accuracy of high-mass MALDI-MS is

relatively low (±200 Da, 4550 ppm) compared to nondenaturingnanoESI (±2 Da, 50 ppm). This is mainly due to “decoration” ofthe proteins by cross-linker molecules, and to adduct formationduring the MALDI process, which both increase the molecularweight of the proteins in an uncontrolled fashion. For that reason,high-mass MALDI will probably never replace nanoESI for anaccurate measurement of the molecular weight and for the analysisof small molecules interacting with proteins.

To validate the use of the fusion receptor and to exclude anunspecific binding of CAP to the fusion protein (i.e., thioredoxin),we carried out analogous binding experiments with the receptorafter cleavage of the thioredoxin. Because the preparation of themutant receptor without fusion contained E2 (the protein lackingthe thioredoxin is not stable without ligand), we could not showthe influence of the ligand on dimerization with high-mass MALDI-MS. Nevertheless, the agonist E2 allows CAP to interact with thereceptor without the fusion, as suggested by the high-massMALDI mass spectrum and nanoESI-MS (Figure 3). After replac-ing the agonist E2 with the higher affinity antagonist OHT in acompetition experiment, the peptide-bound receptor disappearsfrom the high-mass MALDI mass spectrum. Therefore, the peptidespecifically binds to the hERR LBD, which means that the fusionreceptor is a valid model system to probe the ligand effect onthis interaction. Furthermore, a stoichiometry of two CAP per

Figure 3. Validation of the use of the mutant hERR LBD in fusion with thioredoxin for discerning agonist and antagonist ligand with MS. Aftercleavage of the fusion protein, agonist ligand still enhances the recruitment of CAP by the receptor and a stoichiometry of two CAP per homodimeris confirmed. Normalized MALDI mass spectra of hERR LBD without the fusion after cross-linking reaction and incubation (a) with the agonistE2, (b) with E2 and CAP, and (c) with the antagonist OHT and CAP. Normalized nanoESI mass spectra of hERR LBD without the fusionrepresenting the homodimer charge-state distribution of the hERR LBD (d) before incubation with CAP; (e) after incubation with CAP. Italicnumbers indicate the measured m/z of the different species.

7837Analytical Chemistry, Vol. 80, No. 20, October 15, 2008

homodimer was measured for the receptor without fusion withhigh-mass MALDI-MS and nanoESI-MS, a result consistent withthe crystal structure of ERR LBD complexed with an agonistligand and a coactivator peptide.25 It is then tenable that two NRboxes of the full-length coactivator interact with the ER.28 Thisclarifies the previously poorly defined homodimer-CAP stoichi-ometry measured with MALDI-MS for the receptor in fusion withthioredoxin, which, due to the large peak width, did not clearlyallow assigning an interaction with a second CAP (Figure 2b).The mutant receptor used in this study has shown near wild-typecharacteristics in ligand binding and a ∼60% lower transcriptionalactivity relative to the wild type,17 which indicates an equilibriumbetween an agonist and antagonist conformation. This implies thatthe mutant hERR LBD still retains its ability to bind CAP, asreflected by our mass spectrometry experiments. Although thepartial binding of CAP to the hERR LBD monomer measured withMALDI-MS may reflect this agonist-antagonist conformationalequilibrium in solution, the fraction of unfolded monomer mea-sured in solution by nanoESI-MS19 certainly contributes signifi-cantly to the loss of the hERR LBD’s ability to bind CAP.

With both MS methods, we next investigated the modulationof the coactivator peptide interaction with the mutant hERR LBDfor the different test compounds (Figure 4). Buffer only (apore-ceptor) was included as control. In the presence of E2, HEX, DIE,DES, and GE, the recruitment of the peptide by the receptor wasenhanced in contrast to OHT and CLO, a result that correlateswell with the known agonist and antagonist biological function ofthese ligands (Table 1). The absence of any influence on the dimerabundance previously measured by high-mass MALDI-MS for SIMand NIT allows us to classify them as nonbinding compounds(Figure 1c). ZEA was classified as partial agonist due to thesomewhat lower amount of bound peptide relative to the agonistligands. The biological functions of the tested compounds pre-dicted by MS agree well with the ones reported by cell-basedassays. Because in vitro coactivator peptide recruitment profileshave previously shown a good correlation with cell-based assays,30

the approach presented here can be an excellent complementarymethod for addressing the biological consequences of ligandbinding within a cell.

CAP Stabilizes Agonist-Bound hERr LBD Complex. Fora given charge state, the abundance of E2-bound hERR LBD oncebound with CAP measured with nanoESI-MS is considerably

higher than that of the receptor without CAP (Figure 2e). Forthe different agonist ligands, the agonist-bound to free proteinratio (R) of the monomer charge state 14+ is more than 3 timeshigher once bound with CAP (Figure 5), which suggests astabilization of the receptor-ligand interaction by the coactivatorpeptide. Such a comparison is only valid if the ionization efficien-cies and the extent of dissociation of all species are similar. Dueto the small masses of the ligand relative to the protein, a similarionization efficiency can be assumed for the ligand-bound and freeprotein.31 A number of factors influence the dissociation ofnoncovalent complexes in the gas phase, such as the MS transferpotentials, the nature of the collision gas, and the charge state.32

Because hERR LBD and hERR LBD-CAP ions have similar m/zvalues and the MS parameters remained constant during theacquisition, similar gas-phase dissociation efficiencies are expectedfor these species. Therefore, the nanoESI experiments reveal astabilization of the agonist-bound hERR LBD complex once boundby the coactivator peptide. This stabilization leads us to speculate

(23) Brzozowski, A. M.; Pike, A. C. W.; Dauter, Z.; Hubbard, R. E.; Bonn, T.;Engstrom, O.; Ohman, L.; Greene, G. L.; Gustafsson, J. A.; Carlquist, M.Nature 1997, 389, 753–758.

(24) Manas, E. S.; Xu, Z. B.; Unwalla, R. J.; Somers, W. S. Structure 2004, 12,2197–2207.

(25) Shiau, A. K.; Barstad, D.; Loria, P. M.; Cheng, L.; Kushner, P. J.; Agard,D. A.; Greene, G. L. Cell 1998, 95, 927–937.

(26) Norris, J. D.; Paige, L. A.; Christensen, D. J.; Chang, C. Y.; Huacani, M. R.;Fan, D.; Hamilton, P. T.; Fowlkes, D. M.; McDonnell, D. P. Science 1999,285, 744–746.

(27) Ozers, M. S.; Ervin, K. M.; Steffen, C. L.; Fronczak, J. A.; Lebakken, C. S.;Carnahan, K. A.; Lowery, R. G.; Burke, T. J. Mol. Endocrinol. 2005, 19,25–34.

(28) Gee, A. C.; Carlson, K. E.; Martini, P. G.; Katzenellenbogen, B. S.;Katzenellenbogen, J. A. Mol. Endocrinol. 1999, 13, 1912–1923.

(29) Mueller, S. O.; Simon, S.; Chae, K.; Metzler, M.; Korach, K. S. Toxicol. Sci.2004, 80, 14–25.

(30) Iannone, M. A.; Simmons, C. A.; Kadwell, S. H.; Svoboda, D. L.; Vanderwall,D. E.; Deng, S. J.; Consler, T. G.; Shearin, J.; Gray, J. G.; Pearce, K. H.Mol. Endocrinol. 2004, 18, 1064–1081.

(31) Peschke, M.; Verkerk, U. H.; Kebarle, P. J. Am. Soc. Mass Spectrom. 2004,15, 1424–1434.

(32) Nesatyy, V. J. J. Mass Spectrom. 2001, 36, 950–959.

Figure 4. hERR LBD-CAP complex abundance measured by (a)MALDI-MS and (b) nanoESI-MS for the test compounds. Thecomparison with the aporeceptor (-) suggests that E2, HEX, DIE,DES, and GE act as agonists, OHT and CLO act as antagonists,and ZEA acts as partial agonist. Because SIM and NIT did not affectthe hERR LBD dimerization (see Figure 1), they are classified asnonbinding compounds.

7838 Analytical Chemistry, Vol. 80, No. 20, October 15, 2008

that the binding of the peptide in the LBD hydrophobic groovecloses the binding pocket of the ligand and prevents furtherdissociation. The direct measurement of this effect with nanoESI-MS agrees with the marked reduction in ligand dissociation ratemeasured when the receptor was incubated with a coactivatorpeptide.28 Although the tissue distribution of the two ER subtypes,ERR and ER�, is thought to contribute to the tissue specificity ofsome ER ligands, the kinetic stabilization of the ligand bindingthrough coactivator has also been suggested to contribute to thisspecificity.28

CONCLUSIONSIn this study, MS was used for the first time to investigate

ligand modulation mechanisms involved in the ER LBD regulation

and created unique data to understand them at the molecular level.The ligand-dependent hERR LBD dimerization revealed by high-mass MALDI-MS combined with chemical cross-linking allowsidentifying ER ligands like EDCs. The ligand-dependent recruit-ment of the coactivator peptide by the hERR LBD further allowsdiscerning the agonist from antagonist character of uncharacter-ized ER ligands. For that purpose, high-mass MALDI-MS ispreferred over nanoESI-MS: analysis by MALDI-MS consumesonly very small amounts of sample (10 pmol of protein/ligand).Purification procedures and careful optimization of the experi-mental conditions are often required when working with nanoESI-MS, and therefore, somewhat larger sample quantities (90 pmolof protein/ligand) are necessary. The simple high-mass MALDImethodology will certainly promote this new approach for moni-toring modulation of protein-protein interactions induced throughsmall-molecule binding. These interactions are central to mostbiological processes, and the characterization of their modulationwith high-mass MALDI-MS can contribute to the discovery of newdrug candidates in combination with biological approaches.

ACKNOWLEDGMENTThis work was supported by the Swiss National Science

Foundation (projects 4050-104373 and 200020-111831) and theSPINE 1 and 2 European Project (FP6 Contracts QLG2-CT-2002-00988 and 031220). We are grateful to Novartis Pharma AG forpartial funding.

Received for review April 11, 2008. Accepted June 27,2008.

AC8007169

Figure 5. Variation of the agonist-bound to free protein ratios Rmeasured by nanoESI-MS once hERR LBD was incubated with CAPin 50 mM NH4OAc (pH 6). For each ligand, the R values werecalculated on the hERR LBD and hERR LBD-CAP charge states14+.

7839Analytical Chemistry, Vol. 80, No. 20, October 15, 2008