Platinum-based inhibitors of amyloid-� as therapeuticagents for Alzheimer’s diseaseKevin J. Barnham*†‡§, Vijaya B. Kenche*†‡, Giuseppe D. Ciccotosto*†‡, David P. Smith*¶, Deborah J. Tew*†‡, Xiang Liu*†,Keyla Perez*†‡, Greg A. Cranston*†‡, Timothy J. Johanssen*†‡, Irene Volitakis*†, Ashley I. Bush*†, Colin L. Masters*†�,Anthony R. White*†�, Jeffrey P. Smith*†, Robert A. Cherny*†, and Roberto Cappai*†‡

*Department of Pathology, †The Mental Health Research Institute, ‡Bio21 Institute of Molecular Science and Biotechnology, and �Centre for Neuroscience,University of Melbourne, Parkville, Victoria, 3010, Australia

Edited by Peter J. Sadler, University of Warwick, Coventry, United Kingdom, and accepted by the Editorial Board March 2, 2008 (received for reviewJanuary 23, 2008)

Amelyoid-� peptide (A�) is a major causative agent responsible forAlzheimer’s disease (AD). A� contains a high affinity metal bindingsite that modulates peptide aggregation and toxicity. Therefore,identifying molecules targeting this site represents a valid thera-peutic strategy. To test this hypothesis, a range of L-PtCl2 (L �

1,10-phenanthroline derivatives) complexes were examined andshown to bind to A�, inhibit neurotoxicity and rescue A�-inducedsynaptotoxicity in mouse hippocampal slices. Coordination of thecomplexes to A� altered the chemical properties of the peptideinhibiting amyloid formation and the generation of reactive oxy-gen species. In comparison, the classic anticancer drug cisplatin didnot affect any of the biochemical and cellular effects of A�. Thisimplies that the planar aromatic 1,10-phenanthroline ligands Lconfer some specificity for A� onto the platinum complexes. Thepotent effect of the L-PtCl2 complexes identifies this class ofcompounds as therapeutic agents for AD.

A significant body of data indicates the amyloid-� peptide(A�) is a key mediator of the neurodegeneration that occurs

in Alzheimer’s disease (AD) (1), although the neurotoxic mech-anism(s) and pathway(s) involved remain unresolved (2). Givenits central role in AD, diverse therapeutic strategies that targetthe generation, disaggregation, and clearance of A� are beingpursued. Our sequence activity studies have shown that alteringthe metal binding activity of A� inhibits its neurotoxic activity(3, 4). Methylation of the imidazole side chains of His-6, -13, and-14, which constitute the high affinity metal binding site (5),changed A�–metal interactions and A�:cell binding and ren-dered the peptide nontoxic (3, 4). An important conclusion fromthese studies is that agents, which can target the metal bindingsite and alkylate the imidazole side chains of A� have therapeuticpotential.

Platinum anticancer drugs have been used in the clinic for overthirty years, and annual sales exceed one billion U.S. dollars. Thestructure of the prototypic anti-cancer drug, cisplatin [cis-Pt(NH3)2Cl2] (1) is shown in Fig. 1a. Despite this success, thereis scant evidence that Pt-based drugs are viable therapeutics forother diseases. The anticancer Pt compounds have been termed‘‘DNA alkylators,’’ because they bind to the nucleobases of DNA(guanine in particular). These interactions apparently depend onthe specific formation of hydrogen bonds between the am(m)ineligands bound to the Pt and the DNA (6).

Given the importance of the histidine residues in the A� metalbinding site (7), we targeted the imidazole side chains as astrategy to inhibit A�’s neurotoxic activity. Imidazole side chainsare excellent ligands for a variety of metal ions, including Pt(II).Our strategy was to design a specific class of ligand that wouldexplicitly target Pt(II) to the histidine residues of A�. Pt(II)complexes with 1,10-phenanthroline ligands (Fig. 1a) have beenwell characterized. The ligands themselves have an intrinsic,albeit very weak (approximately millimolar) affinity for A� (8).Because the ligands 2,9-dimethyl-4,7-diphenyl-1,10-phenanthro-line and 4,7-diphenyl-1,10-phenanthroline interact predomi-

nantly with the aromatic residues Phe-4, Tyr-10, and Phe-19 (8),it was proposed that the ligand-A� contacts were mediatedthrough �–� stacking interactions (8). The A� amino acidsequence (Fig. 1b) shows that the histidine residues (6, 13, and14) and the aromatic residues are located in the hydrophilicN-terminal domain. The residue most perturbed by the ligandbinding, Tyr-10, is located in the middle of the sequence thatspans the metal binding site of A�, whereas Phe-4 and Phe-19flank the metal binding site. Therefore, we have postulated thatPt(II)-1,10-phenanthroline complexes will also target the N-terminal domain, allowing the Pt to coordinate specifically to thisregion and so modify the behavior of the A� peptide.

We have evaluated three platinum phenanthroline derivatives,Pt(1,10-phenanthroline)Cl2 (2), Pt(4,7-diphenyl-1,10-phenanth-

Author contributions: K.J.B. designed research; K.J.B., V.B.K., G.D.C., D.P.S., D.J.T., X.L., K.P.,G.A.C., T.J.J., I.V., A.R.W., and J.P.S. performed research; A.I.B., C.L.M., R.A.C., and R.C.contributed new reagents/analytic tools; K.J.B., R.A.C., and R.C. analyzed data; and K.J.B.wrote the paper.

Conflict of interest statement: K.J.B., A.I.B., C.L.M., R.A.C., and R.C. are consultants to PranaBiotechnology.

This article is a PNAS Direct Submission. P.J.S. is a guest editor invited by the Editorial Board.

§To whom correspondence should be addressed. E-mail: kbarnham@unimelb.edu.au.

¶Present address: Institute of Molecular and Cellular Biology, University of Leeds, Leeds LS29JT, United Kingdom.

© 2008 by The National Academy of Sciences of the USA

Fig. 1. Pt complexes and A� sequences. (a) The structures of L-PtCl2 com-plexes. (b) Amino acid sequence of A�42. Histidine residues His-6, -13, and -14,proposed to be targeted by L-PtCl2, are highlighted, as are the aromaticresidues Phe-4, -19, and -20 and Tyr-10, which are critical to the interactionswith the polyaromatic ligands L.

www.pnas.org�cgi�doi�10.1073�pnas.0800712105 PNAS � May 13, 2008 � vol. 105 � no. 19 � 6813–6818

CHEM

ISTR

Y

roline)Cl2 (3), and Pt(4,7-diphenyl-1,10-phenanthroline disul-fonate)Cl2 (4) (Fig. 1a) ability to inhibit the metal-dependentbiochemical and cellular actions of A�, while using the proto-typic anticancer drug cisplatin (1) (Fig. 1a) as a negative control.The work demonstrates that Pt(II)-1,10-phenanthroline com-plexes coordinate to the histidine residues of A� and act aspotent inhibitors of A�–metal chemistry and synaptotoxicactivity. These complexes have potential as therapeutic agentsfor AD.

ResultsSelection of Platinum Complexes to Target A�. Pt compounds(L-PtCl2; L � 1,10-phenanthroline ligand) 2–4 (Fig. 1a) wereselected for testing, because, after loss of chloro ligands, they arepredicted to coordinate to the histidine residues of the A�peptide and alter its physiochemical and biological properties.Cisplatin (1) lacks the organic scaffold predicted to be necessaryto target the Pt(II) metal center to the histidine residues of A�and should have little effect on A� properties.

Platinum Complexes Bind to A�. Mass spectrometry. To determinewhether the platinum complexes bind to A�, the L-PtCl2 2–4 andcisplatin compounds were incubated with A�1–42 and theproduct solutions analyzed by surface-enhanced laser desorp-tion/ionization time-of-f light (SELDI-TOF) mass spectrometry,using 4G8 as the capture antibody. The A� peptide alone gavea single peak at 4,515 � 1 Da, corresponding to its expected mass(4,515 Da) (Fig. 2a). Incubation of A�42 overnight with 2produced a second peak at 4,890 � 1 Da (Fig. 2a). The increasein mass of 375-Da units corresponds to the formation of aL-Pt-A� adduct in which the two chloro ligands of 2 have beendisplaced as the compound coordinated to A�. The mass spectrafrom the incubation of A� with cisplatin, 3 and 4 showed thatsimilar adducts are formed by these compounds (data notshown).NMR spectroscopy. As a ‘‘soft’’ metal, Pt has a preference forligands with ‘‘soft’’ donor atoms. Within the A� sequence,potential Pt binding sites include the ‘‘soft’’ sulfur atom ofMet-35 and the imidazole nitrogen atoms of the three histidineside chains that are considered an intermediate between ‘‘hard’’and ‘‘soft’’ ligand. In keeping with these preferences, cisplatinhas a high affinity for sulfur containing ligands, such as gluta-

thione, but, because of its relatively slow kinetics, it is still ableto form complexes with nitrogen-based ligands, such as with thenucleobases of DNA (9). To confirm that the complexes L-PtCl2coordinated the histidine residues of A�, their interaction withA�40 was monitored by 1H NMR. A�40 was used in theseexperiments, because its greater solubility and slower aggrega-tion rates over A�42 allowed NMR experiments to be per-formed. Spectra recorded before and after the addition of 2 tosolutions of A�40 revealed a strong perturbation of the peakscorresponding to the C4H and C2H of the imidazole side chainsof His-6, -13, and -14, consistent with the complex coordinatingto these residues (Fig. 2b). The assignment of these peaks asbeing due to the imidazole side chains was confirmed by 2D 1HTOCSY NMR spectra and are consistent with data published inref. 10. There was little perturbation of the peak due to C�H3 ofMet-35, indicating no significant interaction with this residue.Although the SELDI-TOF MS data indicated that cisplatin wasalso able to form adducts with A�; the 1H NMR spectrum ofsolutions of 1 with A� showed that the nature of these adductsis different from those formed with 2. A reduction in intensityof the peak due C�H3 of Met-35 was observed together with abroadening of the spectrum (Fig. 2b). These observations areconsistent with the formation of A� oligomers and multipleother products.

L-PtCl2 Complexes Alter A� Secondary Structure: Circular DichroismSpectroscopy. Aging aqueous solutions of A�42 in the presenceof equimolar Cu2� results in a conformational change from apredominately random coil conformation to a �-sheet confor-mation (Fig. 3) (3, 11). The presence of compounds 2–4 in agingA�42/Cu2� solutions induced a change in peptide conformation(Fig. 3) with a blue shift away from the 215-nm minimumnormally associated with classic �-sheet structures. Although thespectra of the A�42/Cu2�/L-PtCl2 solutions could not be fittedto any of the classic secondary structure peptide conformations,the observed differences are consistent with different confor-mations. An alternative explanation for the changes observed isthat compounds 2–4 have aromatic ligands and, as such, willabsorb in the UV region and when coordinated to the peptidewill be in a chiral environment and, therefore, being opticallyactive, give rise to a circular dichroism (CD) signal also in the UV

Fig. 2. Pt complexes coordinate A�. (a) SELDI-TOF Mass spectrum of A�42 incubated (2). The mass spectrum shows the formation of an A� drug adduct. (b)1H NMR Spectra of A�40 before and after the addition of 1 and 2. The spectra show that 2 is able to perturb the resonances because of the C4H protons of theimidazole side chains of the three histidine residues, whereas cisplatin does not (see box centered at 7.15 ppm). Conversely, the peak due to the S-CH3 protonsof Met 25 at 2.12 ppm was significantly reduced in intensity after incubation with cisplatin, whereas incubation with 2 does not significantly affect this peak.The peak marked with asterisk was due to acetate impurity present in the peptide samples.

6814 � www.pnas.org�cgi�doi�10.1073�pnas.0800712105 Barnham et al.

region. Either explanation is consistent with the compoundscoordinating to A�.

L-PtCl2 Complexes Inhibit A� Aggregation. A� toxicity is stronglycorrelated with peptide aggregation. The time-dependent ag-gregation of A� into amyloid fibrillar structures can be followedby using ThT, which gives a characteristic f luorescence signalwhen bound to amyloid (12). Compounds 2–4 all inhibited ThTfluorescence in a dose-dependent manner (Fig. 4), indicatingthat the formation of amyloidogenic structures was inhibited.This was substantiated by negative staining EM, which showedthe generation of amphorous aggregates rather than amyloidfibrils (data not shown). Both ThT and EM showed that cisplatin(1) did not inhibit amyloid formation (data not shown).

L-PtCl2 Complexes Inhibit Copper Redox Chemistry. We showed thatA� in the presence of Cu2� is able to generate H2O2 catalytically(13, 14). To ascertain whether the Pt complexes could inhibitA�:Cu2�-mediated redox chemistry, the complexes were titratedinto an A�:Cu preparation, and H2O2 generation was measuredwith a fluorimetric assay (13). The IC50 values for compounds1–4 are reported in Table 1. Although cisplatin had no observ-able effect, compounds 2–4 inhibited Cu2�-mediated H2O2

production with an IC50 in the nanomolar range. This level ofinhibition is comparable with that of clioquinol (5-chloro-7-iodo-8-hydroxyquinoline), which inhibits A�:Cu redox chemistry by

directly targeting the Cu rather than the peptide component ofthe A�–metal complex (14).

L-PtCl2 Complexes Inhibit A� Neurotoxicity. Having demonstratedthat the L-PtCl2 complexes could bind to the A� peptide andchange its chemical and structural properties, we assessed theirability to inhibit A� toxicity in primary mouse cortical neuronalcell cultures. Treatment of the neurons with 10 �M A�42 for 4days reduced cell viability to 65% as measured by the MTS assay(Fig. 5). Coincubation of 10 �M A�42 with 2–4 at either 10 �Mor 5 �M significantly increased cell viability (Fig. 5). Compound3 completely restored neuronal viability, whereas cisplatin 1 wasinactive at either 5 or 10 �M. Compounds 2–4 were not toxic atthe concentrations tested (data not shown).

L-PtCl2 Complexes Rescue A�: Inhibition of Long-Term Potentiation.Long-term potentiation (LTP) in the rodent hippocampal sliceis a measure of synaptic plasticity that focuses on activity-dependent persistent increases in synaptic strength and is con-sidered to be the biochemical basis of learning and memory (15,16). Synthetic and cell derived A� can inhibit LTP in vitro andin vivo and supports the role of A� in promoting the learning andmemory loss that occurs in AD (17, 18). The high-frequencystimulation of a mouse hippocampal slice gives an LTP rangingfrom 148% (Fig. 6b) to 135% (Fig. 6c). Incubating the hip-pocampal slice with 2 �M A�42 for 30 min before the stimulusfor LTP significantly reduced LTP from 148% to 124% (Fig. 6b)and from 135% to 105% (Fig. 6c). Compound 3 was chosen asthe L-PtCl2 complex to be tested in the LTP assay as it had thebest handling properties with respect to high solubility and lowheterogeneity. Conversely, 2 has low solubility, and 4 is heter-ogeneous because of the multiple positions occupied by thesulfonate groups on the bathophenanthroline ligand. The addi-tion of 4 �M compound 3 to the A�42 solution completelyreversed the A�42-inhibition of LTP (Fig. 6 a and b). Incomparison, 4 �M cisplatin did not affect A�42-dependentinhibition of LTP (Fig. 6c), supporting the specificity of the

Fig. 3. Synchrotron radiation circular dichroism spectra. A�42 was incubatedat 37°C for 17 h with equimolar Cu alone (solid line) or in the presence of 2(dotted line), 3 (dashed line), or 4 (dot-dash line).

Fig. 4. Inhibition of amyloid formation. The ability of the Pt complexes toinhibit amyloid formation by A�42 was monitored by ThT fluorescence. All threePt 1,10-phenanthroline complexes inhibit amyloid formation in a dose-dependentmanner.ThTfluorescencedataforeachsamplesetwasnormalizedbysubtraction of the blank and by taking A�42 in the absence of Pt complexes asunity. Data are displayed in arbitrary units. Open squares, 2, �1 arbitrary unit;open circles, 3, �0 arbitrary units; open triangles, 4, �1 arbitrary unit.

Table 1. IC50 values for the inhibition of copper-mediatedhydrogen peroxide generation by A�

Compound IC50, �M

1 �102 0.33 0.54 0.6CQ 0.5

Fig. 5. Inhibition of A�42 induced neurotoxicity. Primary cortical neuronswere grown at low density (150,000 cells:cm2) for 6 days and then treated with10 �M A�42 for 4 more days. The platinum compounds 2, 3, and 4 but not 1inhibited the toxicity of A�42 at both 5 and 10 �M concentrations. Cell viabilitywas determined by measuring the inhibition of MTS reduction. *, P � 0.05 and#, P � 0.001 versus A�42. n � 3–6 samples per group. Each sample group wasdone in triplicate. Results are shown as mean � SE.

Barnham et al. PNAS � May 13, 2008 � vol. 105 � no. 19 � 6815

CHEM

ISTR

Y

effect of 3 being mediated through its organic scaffold. Com-pound 3 alone did not affect LTP, consistent with specifictargeting of the A�42-dependent inhibition of LTP.

DiscussionThe pathological accumulation of A� in the brain is a majorhallmark of AD. Genetic studies from early onset cases of ADidentify alterations in A� metabolism as being directly linked tothe disease (19). Although the mechanism of A� neurotoxicityis still unknown, it is generally accepted that the A� peptide isa valid target for therapeutic development. Because it is un-structured in the native state, the de novo design of effectiveinhibitors of A� is problematic. We have adopted a strategy ofusing metal compounds that target A� specifically by takingadvantage of its intrinsic affinity for metal ions. The threehistidine residues His-6, -13, and -14 are the A� metal bindingligands (5). Methylation of the imidazole side chains altered Cubinding and inhibited A� toxicity (3, 4).

We used Pt(II) complexes to target the A�–metal binding site.Pt(II) compounds are stable and essentially redox inert whenpresent in biological systems. The slow kinetics associated withsubstitution reactions at the Pt(II) center means that, oncebound to a target, the Pt(II) metal is difficult to displace. Thespecificity of the interaction between Pt anticancer drugs andDNA has been attributed largely to the ability of the am(m)ineligands to form hydrogen-bonds to guanine nucleotides of DNA.To promote specific binding to A� by L-PtCl2 2–4 complexes,the 1,10-phenanthroline ligand L was designed to target theN-terminal domain of A�. This was based on the observation

that polyaromatic compounds bind to A� and inhibit its aggre-gation (20). Moreover, the classic amyloid-binding fluorescentdyes Congo red and thioflavin T are also polyaromatic com-pounds. L-PtCl2 complexes are highly stable and the likelihoodof the chelating ligand, L, dissociating from the metal is remote.We demonstrated that the metal free ligands, L, bind weakly toA� via interactions with the aromatic residues Phe-4, Tyr-10 andPhe-19 on A� (8). Importantly for our strategy, these residuesspan the metal binding residues His-6, -13, and -14.

To establish that the 1,10-phenanthroline ligands were con-ferring the necessary specificity of action on the Pt complexes,we tested the compound’s ability to inhibit key activities of A�and compared them with cisplatin, which lacks the polyaromaticligand (Fig. 1a). The SELDI-TOF mass spectra and NMRspectra (Fig. 2) indicated that the L-PtCl2 complexes bound toA� as a stable adduct primarily at the histidine residues. Incontrast, although cisplatin (1) did form adducts with A�, itcoordinated predominantly at the sulfur atom of Met-35. Theability of the L-PtCl2 complexes to promote coordination to A�via the histidine residues demonstrates that the 1,10-phenanthroline ligands target these residues as predicted.

The coordination of the L-PtCl2 complexes to A� significantlyaltered the chemical and biophysical properties of the peptide.This is reflected in the altered secondary structure indicated byCD spectra (Fig. 3), inhibition of amyloid formation (Fig. 4), andA�:Cu2�-mediated redox chemistry (Table 1). An interestingobservation from these biophysical/chemical studies is that,although the Pt-free 1,10-phenanthroline ligands have a very lowmillimolar affinity for A� (8), the corresponding Pt complexes

Fig. 6. Rescue of A� synaptotoxicity. (a) Inhibition of LTP by A� was reversed by 3 in mouse hippocampal slices. When compared with DMSO-vehicle alone (closedcircles), LTP was significantly inhibited by treating the slices for 30 min with 2 �M A� (open circles). Time-matched treatment of control slices in 4 �M 3 alonedid not affect LTP (closed triangles), but inhibition of LTP by A� was completely reversed by cotreatment with 4 �M 3 (open triangles). (b) Bar graphs showingLTP levels quantified as the baseline-percentage of the fEPSP averaged from 55 and 60 min after tetanus � SE. Treatment in ACSF alone produced LTP levels of148 � 10%, whereas treatment with A� significantly reduced these levels to 122 � 7%. Treatment with 3 gave LTP of 155 � 10% that was not significantlydifferent from control but was significantly greater than slices treated with A� alone. Most importantly, treatment with A� and 3 together restored the levelsof LTP to 145 � 9% and was not significantly different from controls. (c) Bar graphs from a similar set of experiments showing that cisplatin does not affect LTPor its inhibition by A�. ACSF alone produced LTP levels of 135 � 10%. Treatment with 2 �M A� significantly reduced these levels to 104 � 10% of baseline.Treatment with 4 �M cisplatin did not affect LTP levels compared with control and were measured at 134 � 7%. Cotreatment of slices with A�and cisplatintogether did not affect the inhibition of LTP by A�, producing LTP of 96 � 6%.

6816 � www.pnas.org�cgi�doi�10.1073�pnas.0800712105 Barnham et al.

are potent inhibitors of A�. This reflects that Pt chemistry iskinetically, rather than thermodynamically, controlled. There-fore, even though the low affinity of L for A� should result inshort residency times with the intended target, the kineticinertness of Pt complexes means that, once bound to the target,the L-Pt:A� adducts are very stable and will not dissociate. Thesynergistic effect of coupling the low affinity ligand and kinet-ically stable Pt is highlighted by the lack of activity by thecisplatin in inhibiting either amyloid formation or A�:Cu2�

redox chemistry. The inability of cisplatin to inhibit A�:Cu2�-mediated redox chemistry is interesting, because cisplatin willcoordinate to Met-35, and this residue has been implicated in theredox chemistry of A� (21, 22).

A key test of the L-PtCl2 complexes potential efficacy is theirability to alter the biological properties of A� in a cellular system.We examined the compounds ability to rescue A� inducedtoxicity in primary neuronal cell cultures and A� inducedinhibition of LTP in hippocampal neuronal slices. All threeL-PtCl2 complexes rescued A� induced toxicity in the primarycortical neurons, whereas cisplatin was inactive. The mechanismof A� toxicity is still unclear, with proposed mechanisms de-pendent on a variety of A� biophysical and chemical properties,such as peptide aggregation, and the ability of A� to coordinatemetal ions, such as copper and zinc (2). Changes in the CDprofile and the inhibition of amyloid formation indicated that theL-PtCl2 complexes altered the structural properties of A�.Because the L-PtCl2 complexes coordinate to the histidineresidues of A�, they occupy the zinc and copper binding site onA� and so inhibit metal-mediated phenomena, such as ROSgeneration.

Having established that the L-PtCl2 complexes inhibit A�neurotoxicity, we investigated whether an example of this classof compound could rescue A�-induced inhibition of LTP.Compound 3 was chosen as the L-PtCl2 complex to be tested inthis assay, because it had the best handling properties withrespect to high solubility and low heterogeneity. Conversely, 2has low solubility, and 4 is heterogeneous because of the multiplepositions occupied by the sulfonate groups on the bathophenan-throline ligand. Electrophysiological recordings of LTP in thehippocampal slice measure the level of synaptic strengtheningafter high-frequency stimulation of a population of synapticallyconnected neurons. Because changes in synaptic strength arethought to underlie the learning and memory processes, the LTPassay and A� inhibition of LTP is an established method forinvestigating A� synaptotoxicity (17, 18) and assessing putativeAD-therapeutic compounds. Compound 3 completely reversedthe A� induced inhibition of LTP, whereas, in the absence of A�,it did not significantly alter the levels of LTP (Fig. 6). Impor-tantly, cisplatin did not affect either LTP or rescue A�-inhibitionof LTP, thus supporting the specificity of 3. These cellular resultsdemonstrate the potency, specificity, and efficacy of this com-pound in reducing the neurotoxic and synaptotoxic activitiesof A�.

The data presented here indicate that the 1,10-phenanthrolinecomplexes of Pt(II) coordinate to the histidine imidazole sidechains of A� and alter the biochemical and biophysical proper-ties of the peptide. Importantly, these alterations to the physicalproperties of A� potently inhibit the peptides neurotoxic andsynaptotoxic actions. The inactivity of cisplatin indicates that thearomatic 1,10-phenanthroline scaffold coordinated to Pt(II)conferred the necessary specific targeting of the Pt to thehistidine residues of A�. The results achieved with the L-PtCl2complexes support the future development of this class ofcompound as therapeutic agents for AD to ensure they effi-ciently cross the blood–brain barrier. The potent effects of thesePt complexes define the histidine residues of A� as a viabletherapeutic target to inhibit the neurotoxic and synaptotoxicactions of A�.

Materials and MethodsPt complexes 1 and 2 were purchased from Aldrich; 3 and 4 were prepared asdescribed in ref. 23. Peptides were obtained from AusPep and from the W. M.Keck Laboratory (Yale University, New Haven, CT). 2H2O was obtained fromCambridge Isotope Laboratories. Stock solutions of 2, 3, and 4 were preparedby dissolving known amounts of the compounds in DMSO to give a finalconcentration of 4 mM.

A� Peptide Preparation. Dry A�1-42 or A�1-40 peptide was weighed anddissolved in hexafluro-2-isopropanol (HFIP) and incubated at 25°C for 1 h toremove any preformed aggregates. It was then aliquotted into knownamounts and dried by using a speed-vac. The dry peptide was stored at �80°C.Aliquots were dissolved in 20 mM NaOH and diluted 1:10 with PBS (pH 7.4) andsonicated in a water bath containing ice for 15 min. The solution was thencentrifuged in a bench-top centrifuge at 16,000 � g for 20 min, and thesupernatant was stored on ice until used. Initial peptide concentrations weredetermined by spectrophotometry at 214 nm, using an extinction coefficientof 75,887 liters�mol�1�cm�1.

NMR Spectroscopy. Samples for NMR were run in aqueous PBS with 10% 2H2Oadded. Samples containing A�40 were run at 0.3 mM. The compounds wereincubated with A� at 30°C for 2 h. NMR spectra were recorded on BrukerDRX-600 and AMX-500 spectrometers as described in ref. 5.

SELDI-TOF Mass Spectrometry. A� Pt drug adducts were analyzed by using PS10ProteinChip arrays (Ciphergen Biosystems). Two microliters of antibody (4G8)in PBS (0.25 mg/ml) was added to the spots of the PS10 chip, which wasincubated in the humidified chamber at 4°C overnight. The antibody wasremoved, and blocking buffer (0.5 M ethanolamine in PBS) was added (5 �l).The array was incubated for 30 min. The blocking buffer was removed, andeach spot was washed with 5 �l of 0.5% Triton X-100/PBS (wash buffer) for 5min. The solvent was removed, and the spots were washed with 5 �l of PBS for5 min. A 60-�l sample was added to each spot, and the array was incubated atroom temperature for 3 h. The samples were removed, and each spot waswashed twice with 60 �l of wash-buffer for 5 min. Each spot was washed with60 �l of PBS twice for 5 min and then washed with 60 �l of 1 mM Hepes twicefor 1 min. The array was air-dried. One microliter of sinapinic acid (SPA) [50%saturated in 50% (vol/vol) acetonitrile and 0.5% in TFA] was applied to eachspot twice. The array was air-dried between each application. All incubationsand washes were performed on a shaking table. Chips were analyzed in aPBSIIC protein chip reader; SELDI-TOF MS and peaks were analyzed by usingCiphergen ProteinChip software, Version 3.1.

CD Spectroscopy. A� peptide was prepared as described above and incubatedat 37°C. The final peptide concentration was determined to be 26 �M, and itwas incubated in the presence of equimolar Cu2� alone or equimolar Cu2� plus26 �M 2, 3, or 4.

Synchrotron radiation circular dichroism spectra were collected on station12.1 at the Synchrotron Radiation Source, Daresbury Laboratory (Warrington,U.K.). Data were collected at 37°C in a 0.02-cm fused silica cell with a peptideconcentration of 26 �M. Data were collected between 195 and 260 nm in0.5-nm increments with 1-s accumulation at each wavelength. Backgroundreadings of buffer in the absence of peptide were subtracted. Spectra weresmoothed by using a Fourier transform.

Thioflavin T Assay for A� Aggregation. Compounds to be tested were dissolvedat 50, 10, 5, 1, 0.5, and 0.1 �M in 100% DMSO to create a 100� stock solution.The final DMSO concentration was 1%. Compound was added to 200 �l ofA�1-42 at 10 �M in PBS (prepared as described above) and incubated at 37°Cwith agitation at 200 oscillations per minute for 24 h. Samples where run intriplicate. For determination of fibril growth end points, peptide samples (20�l) were removed from incubation and added to 200 �l of a 20 �M thioflavin-T(ThT) solution at pH 7.4 in PBS. The ThT signal was quantified by averaging thefluorescence emission at 500 nm over 10 sec when excited at 442 nm, using aHidex Oy Plate CHAMELEMON II plate reader in 96-well black plates.

Inhibition of H2O2 Generated by A�:Cu. H2O2 production by A� peptides wasmeasured by using a fluorimetric assay described in ref. 13. Dichlorofluores-cein diacetate (DCF) (Molecular Probes) was dissolved (5 mM) in 100% di-methyl sulfoxide (argon purged for 2 h at 20°C), deacetylated with 0.25 MNaOH for 30 min, and neutralized (pH 7.4) to a final concentration of 1 mM.Horseradish peroxidase (HRP) stock solution was prepared to 1 �M in PBS (pH7.4). The reactions were carried out in PBS (pH 7.4) under ambient conditionsin a 96-well plate (250 �l per well) containing freshly prepared synthetic

Barnham et al. PNAS � May 13, 2008 � vol. 105 � no. 19 � 6817

CHEM

ISTR

Y

peptide (up to 1 �M), Cu-Gly (up to 2 �M), reducing agents (up to 10 �M),deacetylated DCF (100 �M), and HRP (0.1 �M) incubated at 37°C. EDTA (10 �M)was included to prevent reactions with contaminating concentrations (�0.2�M) of free Cu2�. The concentrations of A� used varied to bring readoutvalues into a convenient target range. A� was used at 200 nM. Studies werecompleted on the day of reagent preparation. Reactions were conducted inthe dark to avoid photodynamic effects. The signal specific for H2O2 was thedecrease in fluorescence of parallel samples coincubated with catalase (4,000units/ml; 10 �M). Fluorescent readings were recorded by a Packard Fluoro-count plate reader (485-nm excitation, 530-nm emission) against a standardcurve of reagent-grade H2O2 in PBS (pH 7.4).

A� Neurotoxicity. Primary neuronal cultures and Cell viability assay wereprepared and carried out as described in ref. 24. Data are shown as mean � SE.Statistical comparisons between groups were done with Student’s t test.

Long-Term Potentiation. 14 to 40-day-old C57Bl6 mice were decapitated underanesthesia by halothane inhalation in accordance with University of Mel-bourne animal ethics guidelines. Brains were rapidly removed and chilled inice-cold artificial cerebral spinal fluid (ACSF). ACSF contained 124 mM NaCl, 2.5mM KCl, 2 mM MgSO4, 2 mM CaCl2, 10 mM D-glucose, 1.25 mM NaH2PO4, and26 mM NaHCO3 and was gassed with 95% O2 and 5% CO2 (pH 7.35)(HCl�NaHCO3). A�1–42 (3) and cisplatin were dissolved in DMSO and prein-cubated on the slices for 30 min in ACSF vehicle with DMSO levels controlledat 0.3%. All experiments were interleaved and conducted at room tempera-ture. Field potential recordings were made in 350-�m transverse sections ofthe hippocampus by stimulating and recording in the stratum radiatum of theCA1 region. Recordings were made with an NPI microelectrode amplifier inbridge-mode, connected to a 20� preamplifier with a 10-kHz low-pass eight-pole Bessel filtration (Krohn–Hite; model 3381 filter/amplifier). Signals weredigitized with a Digidata 1322A A/D converter (Axon Instruments) and stored

in pClamp software, Versions 8.2 or 9.0 (Axon Instruments). Baseline stimula-tion intensity was calibrated at the beginning of each experiment to produceresponses of 20–30% of the maximum slope of the field excitatory postsynaptic potential (fEPSP). Tetanic stimulation was a 1-s 100-Hz pulse deliveredat the test intensity in substitution for the test stimulus. Data were analyzedoffline, using pClamp 8.2, Microsoft Excel, and GraphPad Prism 3.03 programs.The slope of each fEPSP was determined by linear regression of the data pointsbetween the peak of the presynaptic fiber volley and the peak of the fEPSP.Normalized fEPSP slopes were expressed as the percentage of the averageslope of the baseline between 30 to 20 min before tetanus. Long-termpotentiation was quantified by averaging the normalized data 55–60 minafter tetanus for each slice. Results were presented as the means � SE, andstatistical significance was determined by using an unpaired t test at the 95%confidence interval.

Experimental Treatment Protocol for Brain Slices. Slices were transferred to a24-well plate and preincubated in gassed ACSF at room temperature with orwithout experimental compounds for 30 min immediately before electrophys-iological recording. A�42 was dissolved in HFIP, dried, resuspended in DMSO,and stored in small aliquots at �20°C. A�42 aliquots were rapidly thawed andused immediately. Compound 3 and cisplatin were dissolved in DMSO imme-diately before use. Final DMSO concentrations were controlled at 0.3% in allpreincubations.

ACKNOWLEDGMENTS. We thank Tony Wedd and Paul Donnelly for helpfuldiscussions with this manuscript. Synchrotron radiation circular dichroism wascarried out with the support of the Daresbury Synchrotron Radiation Sourceand the assistance of David Clarke. This work was funded by the NationalHealth and Medical Research Council, Wellcome Trust Grant WT069851MA,and Prana Biotechnology. Travel was funded by the Australian Nuclear Scienceand Technology Organisation through the Access to Major Resource FacilitiesProgram.

1. Hardy J, Selkoe DJ (2002) The amyloid hypothesis of Alzheimer’s disease: Progress andproblems on the road to therapeutics. Science 297:353–356.

2. Cappai R, Barnham KJ (2007) Molecular determinants of Alzheimer’s disease Abetapeptide neurotoxicity. Future Neurol 2:397–409.

3. Tickler AK, et al. (2005) Methylation of the imidazole side chains of the Alzheimerdisease amyloid-beta peptide results in abolition of superoxide dismutase-like struc-tures and inhibition of neurotoxicity. J Biol Chem 280:13355–13363.

4. Smith DP, et al. (2006) Copper-mediated amyloid-beta toxicity is associated with anintermolecular histidine bridge. J Biol Chem 281:15145–15154.

5. Curtain CC, et al. (2001) Alzheimer’s disease amyloid-beta binds copper and zinc togenerate an allosterically ordered membrane-penetrating structure containing super-oxide dismutase-like subunits. J Biol Chem 276:20466–20473.

6. Takahara PM, Rosenzweig AC, Frederick CA, Lippard SJ (1995) Crystal structure ofdouble-stranded DNA containing the major adduct of the anticancer drug cisplatin.Nature 377:649–652.

7. Atwood CS, et al. (2000) Characterization of copper interactions with Alzheimeramyloid beta peptides: Identification of an attomolar-affinity copper binding site onamyloid beta1–42. J Neurochem 75:1219–1233.

8. Yao S, Cherny RA, Bush AI, Masters CL, Barnham KJ (2004) Characterizing bathocu-proine self-association and subsequent binding to Alzheimer’s disease amyloid beta-peptide by NMR. J Pept Sci 10:210–217.

9. Sherman SE, Lippard SJ (1987) Structural aspects of platinum anticancer drug-interactions with DNA. Chem Rev 87:1153–1181.

10. Lee JP, et al. (1995) 1H NMR of A beta amyloid peptide congeners in water solution.Conformational changes correlate with plaque competence. Biochemistry 34:5191–5200.

11. Simmons LK, et al. (1994) Secondary structure of amyloid beta peptide correlates withneurotoxic activity in vitro. Mol Pharmacol 45:373–379.

12. Burdick D, et al. (1992) Assembly and aggregation properties of synthetic Alzheimer’sA4/beta amyloid peptide analogs. J Biol Chem 267:546–554.

13. Opazo C, et al. (2002) Metalloenzyme-like activity of Alzheimer’s disease beta-amyloid.Cu-dependent catalytic conversion of dopamine, cholesterol, and biological reducingagents to neurotoxic H2O2. J Biol Chem 277:40302–40308.

14. Barnham KJ, et al. (2004) Tyrosine gated electron transfer is key to the toxic mechanismof Alzheimer’s disease beta-amyloid. FASEB J 18:1427–1429.

15. Lisman J (2003) Long-term potentiation: Outstanding questions and attempted syn-thesis. Philos Trans R Soc Lond B 358:829–842.

16. Tsodyks M (2002) Spike-timing-dependent synaptic plasticity—the long road towardsunderstanding neuronal mechanisms of learning and memory. Trends Neurosci25:599–600.

17. Walsh DW, et al. (2002) Naturally secreted oliogmers of amyloid beta protein potentlyinhibit hippocampal long-term potentiation in vivo. Nature 416:535–539.

18. Townsend M, Shankar GM, Mehta T, Walsh DM, Selkoe DJ (2006) Effects of secretedoligomers of amyloid beta-protein on hippocampal synaptic plasticity: A potent rolefor trimers. J Physiol 572:477–492.

19. Hardy J (1997) Amyloid, the presenilins and Alzheimer’s disease. Trends Neurosci20:154–159.

20. Porat Y, Abramowitz A, Gazit E (2006) Inhibition of amyloid fibril formation bypolyphenols: Structural similarity and aromatic interactions as a common inhibitionmechanism. Chem Biol Drug Des 67:27–37.

21. Varadarajan S, Kanski J, Aksenova M, Lauderback C, Butterfield DA (2001) Differentmechanisms of oxidative stress and neurotoxicity for Alzheimer’s A�(1-42) and A�(25-35). J Am Chem Soc 123:5625–5631.

22. Barnham KJ, et al. (2003) Neurotoxic, redox-competent Alzheimer’s �-amyloid is releasedfrom lipid membrane by methionine oxidation. J Biol Chem 278:42959–42965.

23. Fanizzi FP, et al. (1996) Steric crowding and redox reactivity in platinum(II) and platinu-m(IV) complexes containing substituted 1,10-phenanthrolines. Inorg Chem 35:3173–3182.

24. Ciccotosto GD, et al. (2004) Enhanced toxicity and cellular binding of a modifiedamyloid beta peptide with a methionine to valine substitution. J Biol Chem 279:42528–42534.

6818 � www.pnas.org�cgi�doi�10.1073�pnas.0800712105 Barnham et al.