Structural and Mechanistic Implications of Metal Binding inthe Small Heat-shock Protein �B-crystallin*□S

Received for publication, September 29, 2011, and in revised form, November 8, 2011 Published, JBC Papers in Press, November 15, 2011, DOI 10.1074/jbc.M111.309047

Andi Mainz‡§, Benjamin Bardiaux‡, Frank Kuppler¶, Gerd Multhaup¶, Isabella C. Felli�, Roberta Pierattelli�,and Bernd Reif‡§**1

From the ‡Leibniz-Institut für Molekulare Pharmakologie (FMP), Robert-Rössle-Strasse 10, Berlin-Buch 13125, Germany,§Helmholtz-Zentrum München, Deutsches Forschungszentrum für Gesundheit und Umwelt, Ingolstädter Landstrasse 1,Neuherberg 85764, Germany, the ¶Institut für Chemie und Biochemie, Freie Universität Berlin, Thielallee 63, Berlin 14195, Germany,the �Magnetic Resonance Center and Department of Chemistry, University of Florence, Via L. Sacconi 6, Sesto Fiorentino 50019,Italy, and the **Center for Integrated Protein Science Munich, Department of Chemie, Technische Universität München,Lichtenbergstrasse 4, Garching 85747, Germany

Background: �B-crystallin is an ATP-independent chaperone that prevents irreversible protein aggregation.Results:Cu(II) binds to the core domain of�B-crystallin, induces increased dynamics at the dimer interface, and thusmodulatesthe anti-aggregation properties of the chaperone.Conclusion: The small heat-shock protein �B-crystallin is a metal-regulated chaperone.Significance: The results open new perspectives in the field of protein homeostasis and oxidative stress resistance.

The human small heat-shock protein �B-crystallin (�B) res-cues misfolded proteins from irreversible aggregation duringcellular stress. Binding of Cu(II) was shown to modulate theoligomeric architecture and the chaperone activity of �B. How-ever, the mechanistic basis of this stimulation is so far notunderstood. We provide here first structural insights into thisCu(II)-mediated modulation of chaperone function using NMRspectroscopy and other biophysical approaches. We show thatthe �-crystallin domain is the elementary Cu(II)-binding unitspecifically coordinating one Cu(II) ion with picomolar bindingaffinity. Putative Cu(II) ligands are His83, His104, His111, andAsp109 at the dimer interface. These loop residues are conservedamong different metazoans, but also for human �A-crystallin,HSP20, andHSP27. The involvement of Asp109 has direct impli-cations for dimer stability, because this residue forms a saltbridge with the disease-related Arg120 of the neighboring mon-omer. Furthermore, we observe structural reorganization ofstrands �2-�3 triggered by Cu(II) binding. This N-terminalregion is known to mediate both the intermolecular arrangementin�Boligomers and the binding of client proteins. In the presenceof Cu(II), the size and the heterogeneity of �B multimers areincreased.At thesametime,Cu(II) increases thechaperoneactivityof �B toward the lens-specific protein �L-crystallin. We thereforesuggest that Cu(II) binding unblocks potential client binding sitesand alters quaternary dynamics of both the dimeric building blockas well as the higher order assemblies of �B.

Small heat-shock proteins (sHSPs)2 play a fundamental roleduring cellular stress by rescuing misfolded or partiallyunfolded proteins from final degradation. The ATP-indepen-dent sequestration of client proteins prevents their irreversibleaggregation and enables subsequent intervention of refolding-competent chaperone systems like HSP70/HSP90 (1–3).�B-crystallin (�B) was originally discovered in the mammalianeye lens as the B-subunit of �-crystallin (4). Its function is tomaintain the transparency and high refractive index of the eyelens, thereby counteracting cataract formation and visualimpairment (5). Besides this specific lenticular function, thebiological role of human �B is manifold. Likewise it is localizedin several other tissues such as the lung, kidney, brain, cardiac,and skeletal muscle (6). �B was shown to interact with a widerange of substrate proteins that either aggregate amorphouslyor form amyloid fibril structures (4, 7–9). As a consequence �Bwas found to be involved in multiple sclerosis, cancer, car-diomyopathy, and various neurodegenerative diseases like Par-kinson disease and Alzheimer disease (2, 10, 11).The 20-kDa protein �B assembles into highly dynamic and

polydisperse complexes of�600 kDa (24–32 subunits) (12). Asa general feature of sHSP, �B contains the �-crystallin domain(ACD), which is a conserved core domain of �90 residues (Fig.1A). The ACD is flanked by variable N-terminal (60 residues)andC-terminal (25 residues) domains that accomplish themul-timeric assembly. The dynamic nature of �B oligomers hin-dered their structural characterization.Only truncated variantsof �B, containing the ACD without the terminal anchors, wereamenable for x-ray crystallography and solution-state NMR(13–15). Electronmicroscopy of negatively stained �B revealed* This work was supported by the Leibniz-Gemeinschaft, by the Helmholtz-

Gemeinschaft, by the Deutsche Forschungsgemeinschaft (Grants Re1435,SFB449, SFB740, and SFB610), by access to research infrastructures activityin the 7th Framework Programme of the EC (project number 261863, Bio-NMR), and by the Center for Integrated Protein Science Munich.

□S This article contains supplemental Figs. S1–S9 and additional references.1 To whom correspondence should be addressed: Dept. Chemie, Center for

Integrated Protein Science Munich, Technische Universität München, Lich-tenbergstrasse 4, Garching 85747, Germany. Tel.: 49-89-28-91-36-67;E-mail: reif@tum.de.

2 The abbreviations used are: sHSP, small heat-shock protein; ACD, �-crystal-lin domain; AP, anti-parallel; CSP, chemical shift perturbation; FROSTY,freezing rotational diffusion of protein solutions at low temperature andhigh viscosity; HSQC, heteronuclear single-quantum coherence; ICP-MS,inductively coupled plasma mass spectrometry; ITC, isothermal titrationcalorimetry; MAS, magic-angle-spinning; PDSD, proton-driven spin diffu-sion; �B, �B-crystallin; �L, bovine �L-crystallin; rf, radiofrequency; T, tesla.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 287, NO. 2, pp. 1128 –1138, January 6, 2012© 2012 by The American Society for Biochemistry and Molecular Biology, Inc. Published in the U.S.A.

1128 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 2 • JANUARY 6, 2012

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

the envelope of a 24-mer with tetrahedral symmetry (16).Recently, solid-state magic-angle-spinning (MAS) NMRenabled structural studies of full-length �B and providedinsights into the oligomeric architecture of�B (15, 17, 18). In allatomic-level structures, theACD folds into a�-sandwich struc-ture comprising the two sheets�8-�9-�3-(�2) and�4-�5-�6�7 (Fig. 1B). The structurally variable �2 strand was refined onlyfor two of five molecules in the asymmetric unit (13). Twomonomers align in an anti-parallel fashion along the �6 � 7strands and form an extended �-sheet (bottom sheet). Thedimer interface has been termed the “AP” (anti-parallel) inter-face (14). A shared groove is located above the AP interface,which appears to be polymorphic, because three different align-ment registers were observed (13, 14). An extensive network ofionic interactions spans the interface. This network involvescharged residues of the strands �3 and �4, as well as �5 and�6 � 7 and the intervening loops. Notably, the conservedArg120 of one monomer is involved in a bidentate ion pair withAsp109 of the neighboring molecule. The missense mutationR120G causes an increase of the oligomeric size and a decreasein chaperone efficiency in vitro (19). Co-precipitates of desminand �B-R120G are found in cardiac muscle tissue of patientssuffering from desmin-related cardiomyopathy (11).The dimeric substructure represents the building block of

higher order oligomers (17, 18). Due to dynamic subunitexchange, the resulting �B ensemble is consequently heteroge-neous. The intrinsic subunit exchange and domain dynamicshave been shown to be essential for �B chaperone activity (20–22). Intriguingly, divalent metal ions like Cu(II) and Zn(II) canact as modulators of this chaperone property (23–25). It wasshown recently that �B can coordinate Cu(II) with a picomolaraffinity, whereas the interaction with other divalent metal ionsis significantly weaker (26). Cu(II) affects the secondary struc-ture and oligomeric size of �B (26, 27). Furthermore, �B is ableto reduce Cu(II)-mediated formation of reactive oxygen spe-

cies, thereby conferring cytoprotection to cells under condi-tions of oxidative stress (26).By using solution- and MAS solid-state NMR spectroscopy,

we demonstrate here how Cu(II) interacts with �B. Cu(II)-in-duced broadening of the oligomeric ensemble of �B is accom-panied by enhanced chaperone activity. Similar to full-length�B, the excised ACD (residues 64–152) shows picomolar bind-ing affinity with respect to Cu(II). Each ACD binds one Cu(II)ion at the outer edge of the AP interface. Cu(II) is coordinatedin the loop regions connecting the strands �3 and �4 as well as�5 and �6 � 7, involving residues His83, His104, His111, andAsp109 as potential ligands. Coordination of Cu(II) conse-quently competes with the intermolecular ion pair betweenAsp109 and Arg120 affecting the monomer-dimer equilibrium.Furthermore, Cu(II) binding triggers structural reorganizationof �2-�3, which is of importance for dimer-dimer contacts infull-length �B (17, 18, 28). Therefore, the metal ion Cu(II)potentially alters quaternary dynamics of both the dimeric sub-unit and the higher order assemblies of �B.

EXPERIMENTAL PROCEDURES

Cloning of Constructs, Protein Expression, and Purification—Expression and purification of human, full-length �B (UniProtaccession code: P02511) was performed as described elsewhere(29). Cloning of the truncated construct �B10m (�B residues64–152) was accomplished similarly to the procedure alreadydescribed for full-length �B (29). The following PCR primerswere used: CGTGGCCAGCATATGGGACTCTCAGAGAT-GCGCCTGG (forward-primer), CGCCTTCCATGGCTAGA-CCTGTTTCCTTGGTCCATTCACAGTG (reverse-primer).The obtained vector mod.pET30-�B10 was subject to theQuikChange site-directed mutagenesis protocol (Stratagene)to generate the point mutant �B10-N146D (�B10m). The vec-tor mod.pET30-�B10m in turn was used to introduce furtherpoint mutations. All constructs were verified by DNAsequencing.

�B10m and its variants were expressed in Escherichia coliBL21-DE3 cells for 4 h at 37 °C in LB medium (unlabeled pro-tein) or 16 h at 22 °C in M9 minimal medium (uniformly13C,15N-labeled protein). Due to the high susceptibility towardproteolysis, all purification steps were performed in the pres-ence of Complete protease inhibitor mixture (Roche AppliedScience). The following buffers were used: 50mMTris-HCl, pH8.5, 1 mM EDTA (buffer A); buffer A, 1 M sodium chloride(buffer B); 50 mM Tris-HCl, pH 7.5, 50 mM sodium chloride, 1mM EDTA (GF buffer); and 50 mM sodium phosphate, pH 7.5,50 mM sodium chloride (PBS buffer). After cell lysis and subse-quent centrifugation, the filtered lysate in buffer A was loadedonto a DEAE-Sepharose anion-exchange column. �B10m waseluted with a stepwise gradient of buffer B. The�B10m fractionwas pooled and subject to Q-Sepharose anion-exchange chro-matography. Finally, the �B10m fraction was pooled andloaded onto a Superdex-75 gel-filtration column with �B10meluting in GF buffer as a well separated and symmetrical peak.SDS-PAGE confirmed �98% purity of the �B10m pool, whichwas dialyzed against PBS buffer and concentrated to 1.1 mM.Aliquots were flash-frozen in liquid nitrogen and stored at

FIGURE 1. The core domain of the sHSP �B. A, domain organization of �B.The conserved �-crystallin domain (ACD) is flanked by variable N- and C-ter-minal domains involved in multimer assembly. B, x-ray structure of the �BACD dimer (residues 67–157) (13). The two monomers are colored cyan andgray, respectively. �-Strands are labeled. Strands �6 � 7 form the anti-parallel(AP) interface and an extended bottom sheet. The two top sheets (�2-�3-�9-�8) form a shared groove above the AP interface. The variable �2 strand isdisordered in one of the two monomers.

Metal Binding to the Molecular Chaperone �B-crystallin

JANUARY 6, 2012 • VOLUME 287 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1129

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

�80 °C. HSQC spectra of fresh and flash-frozen protein werefound to be identical.Dynamic Light Scattering—Dynamic light scattering meas-

urements were performed on a ZEN3500 Zetasizer NanoZSinstrument (Malvern Instruments) equipped with a 50-milli-watt laser operating at 532 nm. The instrument detects back-scattering at an angle of 173°. Samples of �B (50 �M) in PBSbufferwere incubatedwithCuCl2, ZnCl2, andCaCl2 at a 10-foldmolar excess of the respective metal ion. After filtration (0.1�m), 20-�l aliquots were measured at 25 °C in quartz cuvettesbuilt for small volumes. After temperature equilibration, threemeasurementswere performed for each sample. Eachmeasure-ment consisted of 60 runs with a duration of 5 s each, therebyaccumulating scattering data for 300 s. All samples revealedmonomodal size distributions for �Bmultimers. The hydrody-namic diameter Dh and the polydispersity index Pd% wereextracted from volume-weighted size distributions. Data anal-ysis and estimation of molecular weightsMw,est (supplementalFig. S1C), assuming a globular particle, was done using the soft-ware DTS 5.03 (Malvern Instruments).Chaperone Activity Assay—The Zetasizer NanoZS instru-

ment (Malvern Instruments) was used to follow light scatteringduring heat-induced aggregation of bovine �L-crystallin (�L).Final concentrations for the different components were as fol-lows: 2 �M �L (Sigma-Aldrich); 0.5, 1, and 2 �M full-length �B;1 �M CuCl2. Higher concentrations of CuCl2 induced aug-mented aggregation of�L. Lyophilized proteins andCuCl2weredissolved in PBS buffer and de-ionizedwater, respectively. Eachsample of 40 �l was incubated for 15 min at room temperaturebefore measurement. Three runs (5-s duration each) were per-formed tomeasure themean count rate of scattered light. Aftermeasuring the initial intensity at room temperature, the samplewas heated to 60 °C. Themean count rate was detected in 2minintervals for a period of 30 min. Each experiment was per-formed in triplicate.Notably,�B is stable at 60 °C in the absenceand presence of Cu(II) giving rise to a constant baseline. Thescattering intensities were normalized to the maximum inten-sity of the 2 �M �L sample in the absence of �B.ICP-MS—For inductively coupled plasma mass spectrome-

try (ICP-MS) experiments, samples of �B and �B10m (80 �M

monomer concentration) in PBS buffer were incubated withCuCl2 (160 �M) for 2 h with constant shaking at room temper-ature. The samples were subsequently dialyzed overnightagainst a 2000-fold volume of PBS buffer. After dialysis, proteinconcentrationswere quantitated. A 200-�l sample solutionwasmixed with 1 ml of HNO3 (65% v/v) and decomposed in amicrowave labstation (Ethos Plus, Milestone Inc.) at 170–210 °C and 6 bars for 30 min. ICP-MS was performed by usingan Element 2 ShieldTorch system instrument (Thermo-FischerScientific) in peak-hopping mode with spacing at 0.01 atomicmass units, three points per peak, three scans per replicate, andan integration time of 300ms per point. The rate of plasma flowwas 15 liters/min with an auxiliary flow of 0.9 liter/min and ablend gas flow rate of 0.1 liter/min. The radiofrequency (rf)power was 1.3 kW. The sample was introduced using a cross-flow nebulizer at a flow rate of 1.02 liters/min. The apparatuswas calibrated with a 6.5% HNO3 solution containing copperand zinc at 1, 5, 10, 25, 50, 100, and 200 parts per billion with

Rh-103 as internal standard for all isotopes. Samples weremeasured in triplicate.ITC—Isothermal titration calorimetry (ITC) experiments

were performed at 25 °C employing a VP-ITC instrument(MicroCal Inc.). Protein samples were dialyzed against a 500-fold volume of PBS buffer, and protein concentrations weredetermined by measuring the absorbance at 280 nm in tripli-cate. The protein solutionswere degassed for 10min before use.To stabilize Cu(II) in buffered solution and to minimize non-specific binding of Cu(II) in the titration experiment, we usedCu(Gly)2 as a weak chelator (26). The glycine solution was pre-pared by dissolving glycine in the corresponding buffer afterdialysis. CuCl2 was then dissolved in a 4-fold molar excess ofglycine to ensure complete chelation of Cu(II). All solutionswere prepared freshly. In the titration experiments 5-�l ali-quots of 2 mM Cu(II) were injected into the cell containing�B10m (110 �M). A spacing time of 300 s was used. In controlexperiments (i) Cu(Gly)2 was injected into buffer only and (ii)glycine was injected into the protein solution. Both experi-ments revealed very weak and constant heat changes that weresubtracted from the binding isotherm of the �B10m-Cu(II)titration. The curve was then fitted to a one-site binding modelusing Origin (MicroCal Inc.). The first three data points wereexcluded from the analysis, because the binding isothermrevealed an initial endothermic contribution. This observationmight arise from Cu(II)-induced dissociation of either dimersor higher oligomers. To account for the competition betweenprotein and glycine for Cu(II) binding, the apparent dissocia-tion constant Kd,app had to be corrected as described previ-ously (26, 30). Accordingly, the real dissociation constantKd forthe respective Cu(II)-protein interaction was obtained as theproduct of Kd,app and the first dissociation constant ofCu(Gly)2 (2.5 � 10�6 M).MAS NMR Spectroscopy—Instead of precipitating �B with

polyethylene glycol as published (15), we exploited the largemolecular weight, and thus the intrinsic very slow moleculartumbling, of the �B oligomer to retard its rotational diffusionduring MAS. The main advantage of this FROSTY MAS NMRapproach is that the target molecule can be studied in solution(29, 31). Sample conditions were as follows: 217 mg/ml uni-formly 13C,15N-labeled �B (�10 mM monomer) in PBS bufferand 20% (v/v) glycerol. For the metal-�B samples, CuCl2 andZnCl2 were added to the �B solutions at a 3-fold molar excess.40 �l of the viscous protein solutions were filled into 4-mmMAS rotors. Samples were measured at a magnetic fieldstrength of 16.4 teslas (T) (Bruker Biospin) and at an effectivetemperature of �5 °C. MAS was performed at 12 kHz. Cross-polarization from 1H to 13C was accomplished with contactpulses of 0.8-ms duration. The rf field strengths were matchedaccording to the n � 1 Hartmann-Hahn condition consideringthe ramp (75–100%) that was used for the 13C contact pulse (76kHz (1H), 56 kHz (13C)). 1H decoupling during indirect anddirect evolution periods was performedwith an rf field strengthof 76 kHz. The PDSD mixing time was set to 50 ms. A recycledelay of 3 s was used. The acquisition times amounted to 10msfor the indirect 13C dimension and 12 ms for the direct 13Cdimension.

Metal Binding to the Molecular Chaperone �B-crystallin

1130 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 2 • JANUARY 6, 2012

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Solution-state NMR Spectroscopy—Solution-state NMRexperiments were performed at 22 °C on 13C,15N-labeled�B10m and variants at monomer concentrations of 1 mM inPBS buffer containing 10% D2O. NMR spectrometers (BrukerBiospin)withmagnetic field strengths of 14.1 T and 17.6Twereequipped with cryogenically cooled probe heads. Data acquisi-tion and processing were done using TopSpin 2.0 (Bruker Bio-spin); further data analysis and resonance assignments wereperformed using the software Sparky.3 Backbone resonanceassignment of apo-�B10m and Cu(II)-�B10m was accom-plished by recording standard triple-resonance experiments(HNCA, HN(CO)CA, HNCO, and HN(CA)CO) (32–34). Allamide resonances were assigned for apo-�B10m except that ofSer139. In titration experiments, CuCl2 in de-ionized water wasadded stepwise to �B10m and its mutants, respectively. Signalintensities and chemical shifts were extracted from HSQCspectra in the absence and presence of Cu(II). Chemical shiftperturbations (CSPs) for 1H-15N correlations were calculatedas follows: CSP � ((0.2 � ��15N)2 � (��1H)2)�1/2 with ��15Nand ��1H being the absolute values of the chemical shift differ-ences in parts per million (ppm) for 15N and 1H, respectively(35).

15N T1 and T2 relaxation rates were determined using stan-dard pulse sequences with the following delays: 10, 50, 100, 150,200, 300, 400, 600, 900, 2000, and 5000 ms (T1 measurements)and 6, 10, 18, 26, 34, 42, 82, 122, 162, 202, and 242 ms (T2measurements) (36). Signal intensities were fitted with mono-exponential decays, and relaxation rates were extracted. Rota-tional correlation times, �c, for the protein core were obtainedas described elsewhere (37). Theoretical �c values were esti-mated with the program Hydro-NMR (38) based on the x-raystructure of the ACD of human �B (13). The calculations wererun for a temperature of 22 °C and a viscosity of 0.011 Poise.All the direct 13C-detected experiments were recorded at

22 °C on a 16.4-T Bruker AVANCE spectrometer, which wasequipped with a cryogenically cooled probe head optimized for13C direct detection. For mapping 13C chemical shifts, two-dimensional CACO and CBCACO spectra (39, 40) wererecorded on apo-, Cu(II)-, and Zn(II)-�B10m (1 mM). For theapo- and the 1:1 Cu(II) samples three-dimensional CBCACOand two-dimensional CCCO (41) experiments were also per-formed. The experiments enabled transfer of assignments forall 89 residues. The carrier frequencies were placed at 175.0 and54.0 ppm for C� and C�, respectively, at 120.0 ppm for 15N andat 4.0 ppm for 1H. Composite pulse decoupling was appliedduring acquisition and during some of the elements of the pulsesequences with an rf field strength of 1.7 kHz for 1H (waltz-16)(42) and 1.0 kHz for 15N (garp-4) (43). The commonparameterswere: recycle delay of 1.2 s, spectral widths of 50 ppm for C�, 40ppm for C�, 80 ppm for C�/Caliphatic. For the two-dimensionalexperiments 1024 � 256 data points in the direct and indirectacquisition dimensions, including the increments necessary forspin-state selection, were acquired. For the three-dimensionalexperiments 1024 � 40 � 256 data points were acquired. Thenumber of scans ranged from 16 to 128, depending on experi-

mental sensitivity. Among the different spin-state selectionmethods implemented for carbonyl direct detection, the IPAP(39) and the S3E (40) methods were applied to achieve virtualdecoupling in the direct acquisition dimension. For each timeincrement in the indirect dimension two Free InductionDecays(FIDs) were separately acquired and stored. The two FreeInduction Decays (FIDs) were then added and subtracted toseparate the two multiplet components. These were thenshifted to the center of the original multiplet (by JC�C�/2 Hz)and again added to obtain a singlet. The data were acquired andprocessed with the standard Bruker software TopSpin 1.3. TheCSP for the 13C-detected CBCACO spectrum represents thesum of absolute values of chemical shift differences in ppm forC�, C�, and C� resonances. Chemical shift indices for 13C (44)were generated using the software CcpNmr Analysis (45).Modeling of the Putative Cu(II) Coordination Sphere—The

structural model for the Cu(II)-bound ACD dimer of �B wasobtained by molecular dynamics simulation, and energy mini-mization was performedwith CNS 1.2 (46). Initial atomic coor-dinates for residues 66–150were taken fromProteinData Bankentry 2WJ7 (13). Tetrahedral coordination geometry of Cu(II)withN�2 of His83, His104, andHis111 andO�1 of Asp109 (mono-dentate) was constrained according to Harding (47) andincluded in the PARALLHDG force field (48). Correspond-ingly, the bond distances were constrained to 2.02 Å (Cu(II)-N�2) and 1.99Å (Cu(II)-O�1), respectively. The bond angles forN�2-Cu(II)-N�2 and N�2-Cu(II)-O�1 were constrained to109.5° (tetrahedral) and for C�-O�1-Cu(II) to 120° (monoden-tate Asp), respectively. The Cu(II) ion was fixed in the planes ofthe imidazole rings and the carboxylate moiety. All atoms,except for residues Val81 to Ser85 (�3-�4 loop) and Lys103 toPhe113 (�5-�6 � 7 loop), were harmonically restrained to theirinitial coordinates. Hence, the large CSP observed for residuesbelonging to�2-�3, indicating structural changes in this N-ter-minal region,were not taken into account. Additionally, torsionangle data base potential (49) was included for unrestrainedregions. The program PyMOL4 was used to generate figures ofmolecular structures.

RESULTS

Cu(II)-induced Increase of �B Hydrodynamic Diameter andPolydispersity—Dynamic light scattering experiments wereperformed to evaluate the effects of Cu(II) on �B quaternarystructure. Upon addition of Cu(II) the hydrodynamic diameterDh of �B particles increased from 14.7 0.2 nm to 19.0 0.5nm. This observation is in agreement with earlier studies (26,27). Furthermore, the augmented size was accompanied by anincrease of the polydispersity index Pd% from 32.9 0.9% to36.5 1.8%, implying a more heterogeneous oligomer ensem-ble for Cu(II)-�B. Zn(II) induced an even more pronouncedincrease of Dh (21.2 0.3 nm) and Pd% (40.4 0.9%). By con-trast, Ca(II) had no effect on Dh (14.7 0.1 nm) and Pd%(31.7 0.7%) of �B. Auto-correlation functions, size distribu-tions, and estimation of subunit stoichiometries are shown inthe supplement (supplemental Fig. S1).

3 T. D. Goddard and D. G. Kneller, University of California, San Francisco, CA.

4 DeLano, W. L. (2002) The PyMOL Molecular Graphics System, DeLano Scien-tific LLC, San Carlos, CA.

Metal Binding to the Molecular Chaperone �B-crystallin

JANUARY 6, 2012 • VOLUME 287 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1131

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Modulation of �B Chaperone Activity Triggered by Cu(II)—The increase of light scattering during heat-induced aggrega-tion of bovine �L at 60 °C was used to monitor the chaperoneability of human, full-length �B. �L aggregation reached a pla-teau after 30 min in the absence of �B and Cu(II) (Fig. 2). Addi-tion of the chaperone reduced protein aggregation in a con-centration-dependentmanner. At an equimolar ratio of�B and�L, the light-scattering intensity after 30 min was decreased to�35%with respect to the reference experiment. In the presenceof Cu(II), this level of protection was already reached at a 4-foldexcess of �L implying that the chaperone efficiency of �B wasincreased by the divalent metal ion (Fig. 2). Higher concentra-tions of �B reduced �L aggregation to �20%.Stoichiometry and Affinity of Cu(II) Binding—To study the

stoichiometry and affinity of the interaction between �B andCu(II), we performed ICP-MS and ITCmeasurements. In addi-tion to full-length �B, we used a truncated protein construct,which comprises only the conserved ACD (residues Gly64–Val152) with the point mutation N146D (15). The resulting10-kDa protein (�B10m) is unable to oligomerize but formsstable dimers accessible for solution-state NMR.ICP-MS experiments on full-length �B and the core domain

�B10m revealed comparable Cu(II):protein stoichiometries of1.18 0.12 and 1.13 0.12, respectively. Similarly, Zn(II)yielded stoichiometries of 1.03 0.10 and 0.79 0.08 for �Band �B10m, respectively. Apparently, both proteins sequesterone metal ion per monomeric subunit.In ITC experiments,�B10m revealed strong exothermic heat

changes upon injection of Cu(II) with extensive enthalpic con-tributions to the binding reaction (Fig. 3). The derived dissoci-ation constant Kd is on the order of 31 � 10�12 M for �B10m(see “Experimental Procedures”). Picomolar range affinity wasalready reported for full-length �B with a Kd of 11 � 10�12 M

and a stoichiometry, N, of 1.30 0.02 (26). The ITC-derivedstoichiometry for �B10m of 1.03 0.01 is in agreement withthe ICP-MSdata and demonstrates that oneCu(II) ion is boundper ACDmonomer. The isolated ACD is therefore sufficient tocoordinate Cu(II). This result legitimates solution-state NMRbinding studies on the truncated construct �B10m.Structural Investigations on Full-length �B Multimers—To

monitor the effects of Cu(II) on �B multimers, we performedMAS NMR experiments employing the FROSTY (freezing

rotational diffusion of protein solutions at low temperature andhigh viscosity) approach (29). In contrast to conventional solid-state MAS NMR, the protein was not precipitated but investi-gated in solution (see “Experimental Procedures”). Werecorded PDSD (proton-driven spin diffusion) spectra of apo-and Cu(II)-�B (see supplemental Fig. S2A for full spectra).Resolved correlations of residues Asp109, Gly112, and Ile114 dis-appeared in the presence of paramagnetic Cu(II) (Fig. 4, A–C).These residues are located at the AP interface of the ACDdimer. The resonances of Met68 (�2-�3 loop) revealed distinctCSPs for Cu(II)-�B (Fig. 4D). Binding of Cu(II) also affected theresonances of Leu89 (�3-�4 loop) and Ala57 (N-terminaldomain) (Fig. 4,E and F). The perturbations involvingAla57 andMet68 indicate structural changes at the N-terminal domainupon Cu(II) binding. These effects were also observed in thepresence of diamagnetic Zn(II) (supplemental Fig. S2B).Structural Investigations on the Dimeric �B10m—The

ICP-MS and ITC results demonstrate that theACD is sufficientto coordinate Cu(II). We therefore focus now on �B10m torefine the Cu(II) binding site.The 1H-15N HSQC (heteronuclear single-quantum coher-

ence) spectrum of apo-�B10m is well dispersed implicating astructured and folded�B core domain (supplemental Fig. S3A).Addition of paramagnetic Cu(II) causes tremendous changes inthe 1H-15N HSQC spectrum of �B10m (Fig. 5A and supple-mental Fig. S3A). SevereCSPof up to 1.2 ppmwere observed forthe holo-form of �B10m. Consistent with the ITC-derivedpicomolar binding affinity, the system is in the slow-exchangeregime. New resonances were fully populated at an equimolarratio of monomer and Cu(II), confirming that the monomericACD itself is the basic Cu(II)-binding unit. The top panel of Fig.6A illustrates the variations in the 1H-15N HSQC spectrum

FIGURE 2. Modulation of �B chaperone activity by Cu(II). Chaperone activ-ity of human full-length �B with respect to heat-induced aggregation ofbovine �L-crystallin (�L) in the absence (left) and presence of Cu(II) (right). Theclient protein �L (2 �M) was heated to 60 °C in the absence (f) and presenceof 0.5 �M (E), 1 �M (Œ), and 2 �M (�) �B. �, denotes the scattering profile for2 �M �B alone. Experiments were performed in triplicate.

FIGURE 3. Cu(II) binding to the truncated construct �B10m as observedby ITC. Curve-fitting analysis of the binding isotherm gave the followingapparent association constant (Ka,app), enthalpic change (�H), entropicchange (�S), and binding stoichiometry (N): Ka,app � 8.02 ( 0.41) � 104

M�1,

�H � �6987 107 cal mol�1, �S � �0.998 cal mol�1 K�1, N � 1.03 0.01.The correction for competition between �B10m and the chelator glycineyielded an effective dissociation constant Kd of 31 � 10�12

M.

Metal Binding to the Molecular Chaperone �B-crystallin

1132 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 2 • JANUARY 6, 2012

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

induced by Cu(II). The most intense CSP were observed in theN-terminal loop involving strands �2 and �3 (residues Glu71–Phe75) and at the C terminus around Pro148 (residues Gly147and Arg149). Only marginal differences occurred for residueslocated in strands �4, �8, and �9 at the outer edges of the ACD

dimer. By contrast, various resonances were broadened beyonddetection due to their spatial vicinity to paramagnetic Cu(II)(�50% of the resonances). Signals affected by paramagneticquenching comprised the N-terminal strand �2 (Gly64–Leu70),parts of strand�3 and the adjacent loop (Ser76–Ser85), as well asstrands �5 and �6� 7 and the intervening loop (Glu99–Arg123)(supplemental Fig. S4A). It is noteworthy that these regionsconstitute theAP interface and the adjacent shared groove. Thesame result is obtained when side-chain reporters like methylgroups are monitored (supplemental Fig. S3B).

1H nuclei are highly sensitive to paramagnetic relaxationenhancement. Paramagnetic relaxation enhancement isinduced by the unpaired electron of Cu(II) and depends on thesquare of the gyromagnetic ratio � of the NMR-observednucleus (50). Hence, bleaching of resonances occurs in a sphereof 12-Å radius around the Cu(II) ion for 1H-detected experi-ments. This sphere contracts to �6 Å employing 13C directdetection (51). To further restrict the potential binding regionof Cu(II), we performed 13C direct detected experiments. Sup-plemental Fig. S5 shows the CBCACO spectrum of apo-�B10m. The addition of Cu(II) induced the disappearance ofresonances as well as significant CSP. As expected, paramag-netic quenching of 13C resonances was less pronounced, and75% of all residues were still detectable. The locus of Cu(II)binding was thereby restricted to the interconnecting loopbetween strands �5 and �6 � 7 at the AP interface. This loopalso contains the residue Asp109, which is involved in the ionpair to the disease-related Arg120 of the neighboringmonomer.Resonances belonging to strand �6 � 7 (Phe113, Ile114, Arg116,Glu117, and His119–Lys121) and the preceding �5-�6 � 7 loop(His104 and Asp109–His111) disappeared in the presence ofCu(II) (Fig. 6A, bottom panel and supplemental Fig. S4A). Theresults for �B10m are thus in agreement with our findings forthe full-length protein. The �3-�4 loop (Val81–His83) and thevery N-terminal residues (Gly64 and Leu65) were also affectedby paramagnetic broadening. These flexible regions extend intothe shared groove and are in proximity to the �5-�6 � 7 loop.Potential Cu(II) ligands in this region are His83 (�3-�4 loop)and His104, Asp109, and His111 (�5-�6 � 7 loop) (Fig. 6B).Molecular dynamics simulations show that these residues canaccommodate a Cu(II) ion with metal-ligand distances of �2 Åand with tetrahedral geometry (Fig. 6C).The CSPs for C�, C�, and C� atoms are plotted for each

residue in the bottom panel of Fig. 6A. Again, the most intenseCSP occurred at the N-terminal �2-�3 region of �B10m,including residues Arg69–Ser76. Several residues at the end ofstrand �5 (Gly102–Lys103) and within the �5-�6 � 7 loop(Glu106–Gln108) were perturbed in the presence of Cu(II) aswell. The strongest effect though was observed for the loopresidue Phe84 (�3-�4 loop). Careful examination of the side-chain carbonyl correlations in the CBCACO spectrum revealedexplicit CSP for residues Glu88 and Gln108 as parts of the �3-�4loop and �5-�6 � 7 loop, respectively (supplemental Fig. S5).The data suggest structural reorganization of these loops dur-ing incorporation of Cu(II).

13C chemical shifts provide information on protein secondarystructure (44). Chemical shift indices for apo-�B10m are in goodagreement with the ACD crystal structure of �B except for the

FIGURE 4. Monitoring the effects of Cu(II) binding to �B multimers byMAS NMR. Selected regions of PDSD (proton-driven spin diffusion) spectra offull-length �B in the absence (black) and presence of Cu(II) (red). Correlationsof Gly (C�-C�) (A), Ile (C�-C�) (B), Asp/Asn (C�-C�) (C), Met (C�-C�) (D), Leu(C�-C�/C�, connected by dashed lines) (E), and Ala (C�-C�) (F) are shown.Assignments are taken from Ref. 15. See supplemental Fig. S2 for full spectra.

FIGURE 5. Binding of Cu(II) to dimeric �B10m and the essential role ofAsp109. Selected region of 1H-15N HSQC spectra of �B10m (wild type andmutants) in the absence (black) and presence of Cu(II) (red): �B10m (A),�B10m-D109A (B), and �B10m-R120G (C). Assignments from apo-�B10m areindicated in black. Assignments for residues, which show strong CSP, are indi-cated in red. See supplemental Figs. S3 and S7 for full spectra.

Metal Binding to the Molecular Chaperone �B-crystallin

JANUARY 6, 2012 • VOLUME 287 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1133

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

short strand �2 (disordered for three of five monomers in theasymmetric unit (13)). NMR predicts random-coil conformationfor this region. Cu(II)-induced changes in secondary structurewere restricted to the N-terminal stretch Met68–Glu71 revealing�-strand conformation (Fig. 6A and supplemental Fig. S6).Asp109 is involved in the intermolecular salt-bridge with

Arg120 and thus contributes significantly to dimer stability (twosalt bridges per AP interface). Given the important role ofAsp109, we investigated the respective point mutant D109A byNMR spectroscopy. �B10m-D109A completely lost the capa-bility to sequester Cu(II), reflected in the absence of paramag-netic quenching and CSP during Cu(II) titration (Fig. 5B andsupplemental Fig. S7B). Strikingly, the 1H-15N HSQC spectrumof apo-�B10m-D109A shows a pattern that is comparable to thespectrum obtained for Cu(II)-�B10m (supplemental Fig. S7D).Apart from that, themutation D109A causes structural heteroge-neity, which is manifested in increased line widths and multiplesets of resonances, e.g. for the side-chain amideofGln108 (�5-�6�7 loop) (supplemental Fig. S7C). This indicates diverse conforma-tional states for this loop residue. By contrast, when Arg120 as thesalt-bridge partner ofAsp109 is replaced by glycine,Cu(II)-bindingcompetence is not impaired. The ACD variant �B10m-R120Grevealed the characteristicwild-typeCSPpatternuponadditionofCu(II) (Fig. 5C and supplemental Fig. S7E).

15NT1/T2 ratios obtained for apo-�B10mat different proteinconcentrations and for Cu(II)-�B10m yield information on therotational correlation time �c and thus on the multimeric stateof the proteins. The theoretical tumbling correlation time�c,theo at 22 °C is on the order of 8 ns and 15 ns for the mono-meric and dimeric states, respectively (see “Experimental Pro-cedures”). At a monomer concentration of 1 mM, �c amountedto 18.9 ns. The correlation time at amonomer concentration of4 mM increased to 22.6 ns implying a tendency to form higheroligomers as already reported (15). By contrast, in the presenceof Cu(II) the rotational diffusion of 1 mM �B10m was acceler-ated significantly yielding a �c of 13.9 ns. This decrease relativeto the apo-form indicates that Cu(II) does not provoke associ-ation of �B10m dimers. Cu(II), by contrast, seems to inducepartial dissociation of the dimeric state.Titration of�B10mwith diamagnetic Zn(II) revealed amuch

weaker interaction and the extinction of resonances belongingto interfacial residues (supplemental Fig. S7A). The disappear-ance of signals in the spectra of Cu(II)-�B10m is therefore dueto both paramagnetic broadening and chemical exchange at theAP interface.To followmonomer-dimer exchange processes at theAP inter-

face, we analyzed the signal intensities of the 1H-15NHSQC spec-trum and the 15N T1/T2 of apo-�B10m (supplemental Fig. S8). A

FIGURE 6. Effects of Cu(II) binding to the core domain �B10m reveal the potential Cu(II) coordination sphere. A, chemical shift perturbations (CSPs) andparamagnetic bleaching observed in 1H-15N HSQC (top) and CBCACO spectra (bottom) of �B10m upon addition of Cu(II). CSPs (black bars) are plotted versus theprotein sequence. Non-observable residues are indicated by horizontal bars (white to red). Prolines and non-assigned residues (Ser139) are marked with anasterisk. Dashed lines indicate the average CSP. On top of the two panels, secondary structure elements derived from NMR secondary chemical shift analysis andas observed in the x-ray structure of the �B ACD (13) are shown. The short �2 strand (gray-shaded) is observed by NMR only in the presence of Cu(II). PotentialCu(II)-binding histidines are highlighted. B, the Cu(II)-binding region with the loops connecting strands �3 and �4 as well as �5 and �6 � 7 is illustrated for onemolecule in the ACD dimer of �B (13). The two monomers are colored cyan and gray, respectively. Potential Cu(II) ligands are labeled. Also shown is theconserved Arg120� of the opposing molecule forming the ion pair with Asp109 across the AP interface. C, structural model of the potential Cu(II) coordinationsphere based on the structure shown in B. The tetrahedral coordination sphere comprises residues His83, His104, His111, and Asp109 (monodentate interactionassumed). The intermolecular salt bridge between Asp109 and Arg120� is thereby disrupted. See also supplemental Figs. S4 –S6.

Metal Binding to the Molecular Chaperone �B-crystallin

1134 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 2 • JANUARY 6, 2012

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

decrease in signal intensities alongwith an increase of 15NT1/T2 isobserved in particular for the AP interface, including strands �5and�6� 7, but also for�3 and the subsequent loop. These resultsare inagreementwithachemical exchangeprocessat theAPinter-face involving association-dissociation processes between mono-mers in potentially different AP registers.

DISCUSSION

Biophysical studies have shown that �B is able to coordinateCu(II) with high affinity, thereby suppressing the redox chem-istry and the cytotoxic effects of the transition metal (26). Inaddition, the chaperone activity of �B was shown to be modu-lated by metal ions (23). These aspects suggest a role of �B incopper homeostasis and imply Cu(II) as a cofactor triggering�B chaperone function. No structural details concerning thisinteraction were available to date. We performed NMR exper-iments using full-length �B as well as the truncated ACD con-struct �B10m to investigate the Cu(II)-binding property.ITC revealed a picomolar range affinity for the core fragment

�B10m toward Cu(II) with a Kd of 31 � 10�12 M and a stoichi-ometry, N, of 1.03 0.01. This is in agreement with ourICP-MS data and with published results for full-length �B (26).Recently, it was proposed that�B can bind up to five Cu(II) ionsper subunit and that the binding sites are distributed over theentire sequence of �B (27). These results were obtained in arather indirect way. Our results show that the ACD is the basicunit specifically coordinating Cu(II). From our data, we do nothave any indication that the terminal sequences contribute tobinding in any respect. This finding was the basis for our NMRinvestigations on the isolated ACD.TheNMRexperiments show that Cu(II) coordination occurs

at the AP interface in the loop region between strands �5 and�6 � 7. This is valid for both the isolated ACD and full-length�B. Potential Cu(II) ligands are His104, Asp109, and His111

(�5-�6 � 7 loop) as well as His83 (�3-�4 loop). These residuesare conserved among different metazoans but also for human�A-crystallin (�A), HSP20, and HSP27 (see supplemental Fig.S9 for sequence alignment). EPR experiments employing full-length �B pointed toward a coordination involving three nitro-gen donor atoms (23). This is in agreement with our model inwhich the vicinity of His83, His104, His111, and Asp109 allows forconcerted Cu(II) binding. A tetrahedral Cu(II)-binding motifinvolving the imidazole rings of three histidines and the carbox-ylate moiety of an aspartic acid is observed in other proteinstructures as well, e.g. in Cu,Zn-superoxide dismutase or cal-granulin C (52, 53). However, it cannot be excluded that otherligands such as water or the nearby residues Glu105, Glu106, andGlu110 participate in binding of Cu(II). Hence, other coordina-tion geometries than tetrahedral might be adopted, e.g. tetrag-onal or bipyramidal arrangements (54). Although the exactgeometry of the metal site needs to be experimentally deter-mined, the present data allow for conclusions about the role ofCu(II) in the chaperone function of �B as discussed below. Itshould also be noted that our model of the Cu(II) coordinationsphere does not account for structural changes in the N-termi-nal region (�2-�3) and that it focuses exclusively on the �3-�4and �5-�6 � 7 loop.

Former molecular modeling studies predicted His101 andHis119 (AP interface) as well as His18 (N-terminal domain) aspossible residues participating in Cu(II) binding (23). Our find-ings show that neither residues from the N-terminal domainnor the interfacial residues His101 and His119 on the opposingface of the extended bottom sheet contribute to Cu(II) bindingof �B (supplemental Fig. S4B).The crystal structure of truncated �A (residues 59–163)

revealed an intermolecular Zn(II)-binding motif involvingthree monomers and the residues His100, Glu102 (monomer A),His107 (monomer B), and His154 (monomer C) (14). Thesequence alignment between �A and �B yields His104, Glu106,His111, and Gly154 as the corresponding residues in �B. Inter-estingly, the involvement of His104 and His111 is in agreementwith our findings. However, our 15N relaxation data clearlyexclude an intermolecular Cu(II)-binding motif for �B. Fur-thermore, a glycine is found in position 154 in �B, and theC-terminal sequence of �B does not contain additional histi-dine residues.Asp109 plays an essential role forCu(II) binding as themutant

�B10m-D109A is completely deprived of Cu(II)-binding com-petence. By contrast, �B10m-R120G retains its Cu(II)-bindingcapability. This finding suggests that Cu(II) competes with thedisease-related Arg120 of the neighboring monomer for inter-actions with Asp109. It is worth emphasizing that there are twoCu(II)-binding sites at the AP interface. The violation of saltbridges on both edges of the interface consequently affects sta-bility of the dimeric subunit. The crystal structure of the ACDmutant R120G shows the collapse of the bidentate ion pairsbetween Arg120 and Asp109 (55). In the mutant, a novel inter-molecular salt bridge is formed between residues Asp80 andHis83 (�3 strand) thereby inducing closure of the shared groove.The involvement of residues Asp109 and His83 in Cu(II) coordi-nation might impede formation of both intermolecular ionpairs. As a consequence, Cu(II) potentially weakens the mono-mer-monomer interaction and opens the shared groove,thereby entailing that client binding sites become more acces-sible. The destabilization of theAP interfacemight also result inan altered register. The variability of the AP interface was sug-gested to contribute to �B polydispersity (13, 14). Indeed,dynamic exchange broadening was observed for the entire setof interfacial residues in case of apo-�B10m, and even more sofor the mutants �B10m-D109A and �B10m-R120G, respec-tively. Moreover, Cu(II) substantially decreased the rotationalcorrelation time of �B10m. This suggests that Cu(II) has animpact on quaternary dynamics of the dimeric substructure.Previous studies revealed the monomer-dimer equilibrium of�B as an essential element for chaperone function (56, 57).

Upon addition of Cu(II), we found severe CSP for resonancesarising from N-terminal residues located proximal to the APinterface. This was observed for both full-length �B and�B10m. The�2-�3 region forms the edges of the shared grooveand is of functional and structural relevance. Residues Phe75–Lys82 (�3 strand) and Phe113–Arg120 (�6 � 7 strand) of �B arecrucial for chaperone activity (58). Protein constructs compris-ing residues Asp73–Lys92 (�3-�4) are known as mini-chaper-ones, which are able to bind and stabilize client proteins.Modification of residues Asp80 and His83 abolishes the anti-

Metal Binding to the Molecular Chaperone �B-crystallin

JANUARY 6, 2012 • VOLUME 287 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1135

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

aggregation property of thismini-chaperone (59). Further stud-ies highlighted the importance of residues Asn78, Lys82, andHis83 as well as Phe84 and Pro86 for the chaperone function of�B (60, 61). The dual role of strand �3 and the subsequent loopfor client and metal ion sequestration is supported by the factthat �B loaded with the substrate protein �-crystallin isimpaired in Cu(II) and Zn(II) binding (62). Intriguingly, resi-dues Phe75–Lys82 (�3 strand) were also found to be involved inintermolecular interactions between subunits and thus tomaintain the oligomeric network (28). A recent structuralmodel of �B multimers based on solid-state NMR, small-anglex-ray scattering, and electron microscopy data proposes a24-mer as the fundamental, though transient, oligomeric state(Fig. 7A) (16–18). According to this model, strand �2 (Trp60–Thr63) is either unoccupied (�2free) or hydrogen-bonded tostrand �3 of its ownACD (�2intra). Additional dimeric buildingblocks, which can be accommodated in the 24-mer scaffold,show intermolecular �2-�3 contacts in a 28-mer model(�2inter) (18). Hence, Met68 (�2-�3 loop) exists in diverse con-formational states and is involved in the intermolecular assem-bly (15, 18). Our MAS NMR results showed CSP for Met68 infull-length Cu(II)-�B. This observation implies Cu(II)-induced

structural changes in the N-terminal domain. Residue Ala57 islocated in a loop region between strands �1 (Tyr48–Pro51) and�2 (Trp60–Thr63) (18). The disappearance of its resonance inthe PDSD spectrum of Cu(II)-�B suggests that, in agreementwith the oligomermodel, Ala57 is located in close vicinity of theparamagnetic center. The severe broadening of the Ala57 reso-nance in the presence of diamagnetic Zn(II) indicates thatmetal binding alters the exchange dynamics in the N-terminaldomain.The interfacial strand �6 � 7 and the proximal strand �3 are

multifunctional elements in the core of the �B ACD dimer.They represent candidate binding sites for client proteins,maintain the stability of the dimeric building block, and con-tribute to its organization in multimer assembly. Our data pro-vide evidence that Cu(II) intervenes in these processes byrecruiting its ligands from the �3-�4 loop and the �5-�6 � 7loop. We propose a model in which Cu(II) affects dynamics ofthe dimeric substructure and, in a relayedmanner, of the native�B oligomer (Fig. 7B).

Consequently, the metal ion modulates �B morphology andplasticity as reflected in the augmented size and increased het-erogeneity of Cu(II)-�B oligomers. It was shown by mass spec-

FIGURE 7. The potential role of Cu(II) in the assembly and function of �B. A, surface representation of the current 24-mer model of human �B (18). Onehexameric ring containing three dimeric subunits is highlighted in white. The other three hexameric rings of the tetrahedral arrangement are colored in differentshades of green. The flexible C-terminal tails (residues 163–175) are excluded for clarity. Top, view onto a 3-fold axis. Bottom, side view of the hexameric ring(white) obtained after a clockwise 90° rotation. One ACD dimer (dashed box) shows the shared groove and the AP interface. Strands �6 � 7 of the two opposingmonomers are colored in orange and blue, respectively. B, model for the Cu(II)-induced modulation of �B assembly. The two monomers of the ACD dimer(schematic representation) are colored in light gray and dark gray, respectively. Strands �3 and �6 � 7 defining the AP interface and the shared groove arehighlighted for each monomer (orange and blue). For simplicity, only �-strands are depicted (with the exception of the intervening loops between strands �1and �2, �3 and �4, as well as �5 and �6 � 7). The intermolecular salt bridge between Asp109 and Arg120 is indicated. The small �1-�2 sheet interacting withstrand �3 of its own monomer or of a neighboring molecule is represented in red (18). Potential client proteins are also represented in red. Cu(II) ions areillustrated as small spheres. 1, intrinsic dynamics of the N-terminal domain allow release of the �1-�2 sheet thereby enhancing subunit exchange andunblocking the client binding site �3 within the shared groove. 2, each ACD sequesters one Cu(II) ion. The binding interferes with the intermolecularAsp109–Arg120 salt bridges and induces partial dissociation of the dimer. Residue Ala57 is part of the �1-�2 loop located proximal to the paramagnetic center.3, Cu(II) affects the intermolecular arrangement between the ACD and the N-terminal anchor. Quaternary structure dynamics and multimer plasticity are thusenhanced. Release of the N-terminal domain leaves the shared groove unoccupied, facilitating the accommodation of destabilized client proteins. 4, depen-dent on the client system, Cu(II) coordination might be impaired by the bound client protein as has been shown in the case of �-crystallin (62).

Metal Binding to the Molecular Chaperone �B-crystallin

1136 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 2 • JANUARY 6, 2012

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

trometry that the polydispersity of pea HSP18.1 mainly origi-nates from intrinsic quaternary dynamics and that thisplasticity confers the ability to adapt to a diverse range of clientproteins (63). In the case of HSP18.1, quaternary dynamics aretemperature-regulated. In addition to post-translational mod-ifications or changes of pH (21, 64, 65), the divalent metal ionCu(II) emerges as a further factor regulating �B chaperonefunction. Indeed, we observeCu(II)-enhanced chaperone activ-ity for �B with respect to the amorphous aggregation of lens-specific �L. By contrast, Cu(II) had no effect on the chaperoneability of �B in case of heat-induced aggregation of citrate syn-thase (26). On the other hand, Cu(II) yielded an enhancementof �B chaperone efficiency regarding the aggregation of insulin(23). �B provides variable substrate binding sites (58). Appar-ently, chaperone function of �B and the role of Cu(II) thereinare substrate-dependent and a function of the accessibility tothe respective binding site.

�B functions as a molecular chaperone in many human tis-sues, prevents the harmful aggregation of a diverse range ofproteins, and thus contributes significantly to stress toleranceof the cell (1, 4, 7). Its involvement in various pathologies likeAlzheimer disease, Parkinson disease, cancer, and cataracthighlights the essential role of �B as a key player for cellularviability (2, 5). Oxidative stress and free radical damage aretypical hallmarks of the above-mentioned pathologies (66–68).Due to its chemistry, the transition metal ion Cu(II) can con-tribute tremendously to the formation of reactive oxygen spe-cies and hence ultimately to the progression of diseases andaging (69). Lifelong exposure to high energetic irradiation ren-ders particularly the eye lens susceptible to free radical damage.Augmented expression levels of �B are reported for tissuesbeing subject to high oxidative stress aswell as for lens epithelialcells confronted with heavy metal ions like Cu(II) (6, 70). Thecytoprotective role of �Bmight arise not merely from its chap-eroning competence but, likewise, from its capability to seques-ter the toxic metal ion Cu(II). Given its high expression levelsunder stress conditions, �B might compete with other Cu(II)-binding proteins such as the Alzheimer disease �-amyloid pep-tide, which has a similar affinity to Cu(II) (71).Further structural investigations must elucidate the impact

of Cu(II) on the chaperonemechanismof�B. The physiologicalrelevance and the consequences of this strong interaction needto be explored further to better understand the role of �B incopper homeostasis and oxidative stress resistance.

Acknowledgments—We thank H. Oschkinat, S. Markovic, N. Jahnke,J. Buchner, M. Haslbeck, and S.Weinkauf for stimulating discussions.

REFERENCES1. Haslbeck,M., Franzmann, T.,Weinfurtner, D., and Buchner, J. (2005)Nat.

Struct. Mol. Biol. 12, 842–8462. Sun, Y., and MacRae, T. H. (2005) FEBS J. 272, 2613–26273. Veinger, L., Diamant, S., Buchner, J., and Goloubinoff, P. (1998) J. Biol.

Chem. 273, 11032–110374. Horwitz, J. (1992) Proc. Natl. Acad. Sci. U.S.A. 89, 10449–104535. Bloemendal, H., de Jong, W., Jaenicke, R., Lubsen, N. H., Slingsby, C., and

Tardieu, A. (2004) Prog. Biophys. Mol. Biol. 86, 407–4856. Iwaki, T., Kume-Iwaki, A., and Goldman, J. E. (1990) J. Histochem. Cy-

tochem. 38, 31–39

7. Nakamoto, H., and Vígh, L. (2007) Cell. Mol. Life Sci. 64, 294–3068. Narayanan, S., Kamps, B., Boelens, W. C., and Reif, B. (2006) FEBS Lett.

580, 5941–59469. Rekas, A., Adda, C. G., Andrew Aquilina, J., Barnham, K. J., Sunde, M.,

Galatis, D., Williamson, N. A., Masters, C. L., Anders, R. F., Robinson,C. V., Cappai, R., and Carver, J. A. (2004) J. Mol. Biol. 340, 1167–1183

10. Steinman, L. (2009) Nat. Rev. Immunol. 9, 440–44711. Vicart, P., Caron, A., Guicheney, P., Li, Z., Prévost, M. C., Faure, A., Cha-

teau, D., Chapon, F., Tomé, F., Dupret, J. M., Paulin, D., and Fardeau, M.(1998) Nat. Genet. 20, 92–95

12. Aquilina, J. A., Benesch, J. L., Bateman, O. A., Slingsby, C., and Robinson,C. V. (2003) Proc. Natl. Acad. Sci. U.S.A. 100, 10611–10616

13. Bagnéris, C., Bateman, O. A., Naylor, C. E., Cronin, N., Boelens, W. C.,Keep, N. H., and Slingsby, C. (2009) J. Mol. Biol. 392, 1242–1252

14. Laganowsky, A., Benesch, J. L., Landau,M., Ding, L., Sawaya,M. R., Cascio,D., Huang, Q., Robinson, C. V., Horwitz, J., and Eisenberg, D. (2010) Pro-tein Sci. 19, 1031–1043

15. Jehle, S., van Rossum, B., Stout, J. R., Noguchi, S. M., Falber, K., Rehbein,K., Oschkinat, H., Klevit, R. E., and Rajagopal, P. (2009) J. Mol. Biol. 385,1481–1497

16. Peschek, J., Braun, N., Franzmann, T. M., Georgalis, Y., Haslbeck, M.,Weinkauf, S., and Buchner, J. (2009) Proc. Natl. Acad. Sci. U.S.A. 106,13272–13277

17. Jehle, S., Rajagopal, P., Bardiaux, B., Markovic, S., Kühne, R., Stout, J. R.,Higman, V. A., Klevit, R. E., van Rossum, B. J., and Oschkinat, H. (2010)Nat. Struct. Mol. Biol. 17, 1037–1042

18. Jehle, S., Vollmar, B. S., Bardiaux, B., Dove, K. K., Rajagopal, P., Gonen, T.,Oschkinat, H., and Klevit, R. E. (2011) Proc. Natl. Acad. Sci. U.S.A. 108,6409–6414

19. Bova,M. P., Yaron, O., Huang, Q., Ding, L., Haley, D. A., Stewart, P. L., andHorwitz, J. (1999) Proc. Natl. Acad. Sci. U.S.A. 96, 6137–6142

20. Haley, D. A., Horwitz, J., and Stewart, P. L. (1998) J. Mol. Biol. 277, 27–3521. Liu, L., Ghosh, J. G., Clark, J. I., and Jiang, S. (2006) Anal. Biochem. 350,

186–19522. Srinivas, V., Raman, B., Rao, K. S., Ramakrishna, T., andRao, Ch.M. (2005)

Mol. Vis. 11, 249–25523. Ganadu, M. L., Aru, M., Mura, G. M., Coi, A., Mlynarz, P., and Kozlowski,

H. (2004) J. Inorg. Biochem. 98, 1103–110924. Biswas, A., and Das, K. P. (2008) Biochemistry 47, 804–81625. Reif, B., and Narayanan, S. (2007) in Topics in Current Chemistry (Peters,

T., ed) pp. 117–167, Springer, Berlin26. Ahmad,M. F., Singh, D., Taiyab, A., Ramakrishna, T., Raman, B., and Rao,

Ch. M. (2008) J. Mol. Biol. 382, 812–82427. Prabhu, S., Srinivas, V., Ramakrishna, T., Raman, B., and Rao, Ch. M.

(2011) Free Radic. Biol. Med. 51, 755–76228. Ghosh, J. G., and Clark, J. I. (2005) Protein Sci. 14, 684–69529. Mainz, A., Jehle, S., van Rossum, B. J., Oschkinat, H., and Reif, B. (2009)

J. Am. Chem. Soc. 131, 15968–1596930. Thompsett, A. R., Abdelraheim, S. R., Daniels,M., and Brown, D. R. (2005)

J. Biol. Chem. 280, 42750–4275831. Bertini, I., Luchinat, C., Parigi, G., Ravera, E., Reif, B., andTurano, P. (2011)

Proc. Natl. Acad. Sci. U.S.A. 108, 10396–1039932. Kay, L. E., Ikura, M., Tschudin, R., and Bax, A. (1990) J. Magn. Reson. 89,

496–51433. Clubb, R. T., Thanabal, V., and Wagner, G. (1992) J. Magn. Reson. 97,

213–21734. Bax, A., and Ikura, M. (1991) J. Biomol. NMR 1, 99–10435. Garrett, D. S., Seok, Y. J., Peterkofsky, A., Clore, G. M., and Gronenborn,

A. M. (1997) Biochemistry 36, 4393–439836. Kay, L. E., Torchia, D. A., and Bax, A. (1989) Biochemistry 28, 8972–897937. Fushman,D.,Ohlenschläger,O., andRüterjans,H. (1994) J. Biomol. Struct.

Dyn. 11, 1377–140238. García de la Torre, J., Huertas, M. L., and Carrasco, B. (2000) J. Magn.

Reson. 147, 138–14639. Bermel, W., Bertini, I., Duma, L., Felli, I. C., Emsley, L., Pierattelli, R., and

Vasos, P. R. (2005) Angew. Chem. Int. Ed. 44, 3089–309240. Bermel, W., Bertini, I., Felli, I. C., Matzapetakis, M., Pierattelli, R., Theil,

E. C., and Turano, P. (2007) J. Magn. Reson. 188, 301–310

Metal Binding to the Molecular Chaperone �B-crystallin

JANUARY 6, 2012 • VOLUME 287 • NUMBER 2 JOURNAL OF BIOLOGICAL CHEMISTRY 1137

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

41. Bermel,W., Bertini, I., Felli, I. C., Kümmerle, R., and Pierattelli, R. (2006) J.Magn. Reson. 178, 56–64

42. Shaka, A. J., Keeler, J., and Freeman, R. (1983) J. Magn. Reson. 53, 313–34043. Shaka, A. J., Barker, P. B., and Freeman, R. (1985) J. Magn. Reson. 64,

547–55244. Wishart, D. S., and Sykes, B. D. (1994) J. Biomol. NMR 4, 171–18045. Vranken, W. F., Boucher, W., Stevens, T. J., Fogh, R. H., Pajon, A., Llinas,

M., Ulrich, E. L., Markley, J. L., Ionides, J., and Laue, E. D. (2005) Proteins59, 687–696

46. Brunger, A. T. (2007) Nat. Protoc. 2, 2728–273347. Harding, M. M. (2006) Acta Crystallogr. Sect. D Biol. Crystallogr. 62,

678–68248. Linge, J. P., and Nilges, M. (1999) J. Biomol. NMR 13, 51–5949. Kuszewski, J., Gronenborn, A. M., and Clore, G. M. (1996) Protein Sci. 5,

1067–108050. Bertini, I., Luchinat, C., Parigi, G., and Pierattelli, R. (2005)ChemBioChem

6, 1536–154951. Bermel, W., Bertini, I., Felli, I. C., Kümmerle, R., and Pierattelli, R. (2003)

J. Am. Chem. Soc. 125, 16423–1642952. Hough, M. A., and Hasnain, S. S. (1999) J. Mol. Biol. 287, 579–59253. Moroz, O. V., Antson, A. A., Grist, S. J., Maitland, N. J., Dodson, G. G.,

Wilson, K. S., Lukanidin, E., and Bronstein, I. B. (2003) Acta Crystallogr.Sect. D Biol. Crystallogr. 59, 859–867

54. Bertini, I., Cavallaro, G., and McGreevy, K. S. (2010) Coord. Chem. Rev.254, 506–524

55. Clark, A. R., Naylor, C. E., Bagnéris, C., Keep,N.H., and Slingsby, C. (2011)J. Mol. Biol. 408, 118–134

56. Aquilina, J. A., Benesch, J. L., Ding, L. L., Yaron, O., Horwitz, J., and Rob-inson, C. V. (2004) J. Biol. Chem. 279, 28675–28680

57. Benesch, J. L., Ayoub,M., Robinson, C. V., andAquilina, J. A. (2008) J. Biol.Chem. 283, 28513–28517

58. Ghosh, J. G., Estrada, M. R., and Clark, J. I. (2005) Biochemistry 44,14854–14869

59. Bhattacharyya, J., Padmanabha Udupa, E. G., Wang, J., and Sharma, K. K.(2006) Biochemistry 45, 3069–3076

60. Ghosh, J. G., Estrada, M. R., Houck, S. A., and Clark, J. I. (2006) Cell StressChaperones 11, 187–197

61. Santhoshkumar, P., and Sharma, K. K. (2006) Protein Sci. 15, 2488–249862. Ghosh, K. S., Pande, A., and Pande, J. (2011) Biochemistry 50, 3279–328163. Stengel, F., Baldwin, A. J., Painter, A. J., Jaya, N., Basha, E., Kay, L. E.,

Vierling, E., Robinson, C. V., and Benesch, J. L. P. (2010) Proc. Natl. Acad.Sci. U.S.A. 107, 2007–2012

64. Ecroyd, H., Meehan, S., Horwitz, J., Aquilina, J. A., Benesch, J. L., Robin-son, C. V., Macphee, C. E., and Carver, J. A. (2007) Biochem. J. 401,129–141

65. Koretz, J. F., Doss, E.W., andLaButti, J. N. (1998) Int. J. Biol.Macromol.22,283–294

66. Truscott, R. J. W. (2005) Exp. Eye Res. 80, 709–72567. Jomova, K., Vondrakova, D., Lawson, M., and Valko, M. (2010)Mol. Cell.

Biochem. 345, 91–10468. Lawless, M. W., O’Byrne, K. J., and Gray, S. G. (2010) Expert Opin. Ther.

Targets 14, 1225–124569. Brewer, G. J. (2010) Chem. Res. Toxicol. 23, 319–32670. Hawse, J. R., Cumming, J. R., Oppermann, B., Sheets, N. L., Reddy, V. N.,

and Kantorow, M. (2003) Invest. Ophthalmol. Vis. Sci. 44, 672–67971. Sarell, C. J., Syme, C. D., Rigby, S. E., and Viles, J. H. (2009) Biochemistry

48, 4388–4402

Metal Binding to the Molecular Chaperone �B-crystallin

1138 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 2 • JANUARY 6, 2012

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Roberta Pierattelli and Bernd ReifAndi Mainz, Benjamin Bardiaux, Frank Kuppler, Gerd Multhaup, Isabella C. Felli,

B-crystallinαProtein Structural and Mechanistic Implications of Metal Binding in the Small Heat-shock

doi: 10.1074/jbc.M111.309047 originally published online November 15, 20112012, 287:1128-1138.J. Biol. Chem. 

  10.1074/jbc.M111.309047Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

Supplemental material:

  http://www.jbc.org/content/suppl/2011/11/15/M111.309047.DC1

  http://www.jbc.org/content/287/2/1128.full.html#ref-list-1

This article cites 70 references, 13 of which can be accessed free at

by guest on May 20, 2020

http://ww

w.jbc.org/

Dow

nloaded from