1521-009X/42/9/1438–1446$25.00 http://dx.doi.org/10.1124/dmd.114.059295DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 42:1438–1446, September 2014Copyright ª 2014 by The American Society for Pharmacology and Experimental Therapeutics

Time- and NADPH-Dependent Inhibition of Cytochrome P450 3A4 bythe Cyclopentapeptide Cilengitide: Significance of the Guanidine

Group and Accompanying Spectral Changes s

Mirza Boji�c, Luca Barbero, Hugues Dolgos, Achim Freisleben, Dieter Gallemann, Simona Riva,and F. Peter Guengerich

Department of Biochemistry and Center in Molecular Toxicology, Vanderbilt University School of Medicine, Nashville, Tennessee(M.B., F.P.G.); Merck-Serono, RBM S.p.A. Istituto di Ricerche Biomediche A. Marxer, Colleretto Giacosa, Torino, Italy (L.B., S.R.);Merck-Serono, Global Early Development, Darmstadt, Germany (H.D.); and Merck-Serono, Global Early Development, Darmstadt/

Global DMPK, Grafing, Germany (A.F., D.G.)

Received May 29, 2014; accepted July 1, 2014

ABSTRACT

Cilengitide is a stable cyclic pentapeptide containing an Arg-Gly-Asp motif responsible for selective binding to aVb3 and aVb5integrins. The candidate drug showed unexpected inhibition ofcytochrome P450 (P450) 3A4 at high concentrations, that is, a15-mM concentration caused attenuation of P450 3A4 activity (de-pending on the probe substrate): 15–19% direct inhibition, 10–23%time-dependent inhibition (30-minute preincubation), and 54–60%metabolism-dependent inhibition (30-minute preincubation). Theinactivation efficiency determined with human liver microsomeswas 0.003 6 0.001 min21 mM21 and was 0.04 6 0.01 min21 mM21

with baculovirus-based microsomes containing recombinant P4503A4. Neither heme loss nor covalent binding to apoprotein couldexplain the observed reductions in residual activity. Slowly formingtype II difference spectra were observed, with maximum spectral

changes after 2 hours. Binding to both reduced and oxidized P4503A4 was observed, with apparent Kd values of 0.66 mM and 6 mM.The significance of the guanidine group in inhibition was demon-strated using ligand binding spectral changes and inactivationassays with guanidine analogs (debrisoquine, N-acetylarginine-O-methyl ester) and the acetylated ornithine derivative of cilengitide.The observed inhibition could be explained by direct inhibition, plusby formation of stable complexes with both ferric and ferrous formsof heme iron and to some extent by the formation of reactive speciescapable to react to the protein or heme. Formation of the complexrequired time and NADPH and is attributed to the guanidino group.Thus, the NADPH-dependent inhibition is considered to be mainlydue to the formation of a stable complex rather than the formation ofreactive species.

Introduction

Cilengitide is a stable cyclic pentapeptide containing the Arg-Gly-Asp(RGD; Fig. 1) motif responsible for selective binding to aVb3 andaVb5 integrins (Dechantsreiter et al., 1999). The blocking of integrinsprevents tumor angiogenesis, providing the prospect of a broaderspectra of indications for cilengitide (e.g., newly diagnosed and re-current glioblastoma, advanced solid tumors, and pancreatic, prostate,non-small-cell lung, and head and neck cancers) (Goodman and Picard,2012). Cilengitide is not metabolized in vitro, neither in recombinantcytochrome P450 (P450) microsomes, human liver microsomes, orhuman hepatocytes. About 70% of a 2 g dose administered intra-venously in healthy volunteers undergoes renal excretion unchangedwhile the rest is eliminated mainly unchanged via biliary secretion infeces. Only very minor metabolites originating from cleavage have beenobserved. The maximal plasma concentration and total area under thecurve increased in proportion to the dose, while no change in clearance,volume of distribution, or half-life was observed, indicating linearpharmacokinetics (Eskens et al., 2003). Cilengitide does not accumulate

after repeated dosing in that no significant changes were observed inpharmacokinetic parameters (Reardon et al., 2011). Although unique inits mode of action and having favorable pharmacokinetic properties,cilengitide did not meet its primary end point of significantly increasingoverall survival when added to the current standard chemoradiotherapyregimen for brain tumors (i.e., temozolomide and radiotherapy), inthe most advanced phase III clinical trial (CENTRIC) (Soffietti et al.,2014).During the course of the drug development, cilengitide showed

unexpected inhibition of P450 3A4 at recommended therapeuticconcentrations, associated with nonextractable radioactivity during invivo experiments on human volunteers. Physiologically based phar-macokinetic modeling predicted a 1.3-fold increase of the area underthe curve of the probe substrate midazolam after administration ofclinically recommended dose of 2 g of cilengitide. Although poten-tially classified as a weak inhibitor, inactivation of P450 3A4 couldcontribute in relevant drug-drug interactions during multitherapybecause P450 3A4 is involved in the metabolism of more than 50% ofdrugs (Williams et al., 2004; Guengerich, 2005a).More than 100 peptide drugs are on the market (Craik et al., 2013),

out of which only cyclosporine has shown clinically significant in-teractions (Amundsen et al., 2012). A predominant metabolic pathwayfor peptides is hydrolysis of peptide bonds (Bernkop-Schnurch and

This work was financed by Merck Serono [Project 4-04-250-6975].dx.doi.org/10.1124/dmd.114.059295.s This article has supplemental material available at dmd.aspetjournals.org.

ABBREVIATIONS: P450, cytochrome P450.

1438

http://dmd.aspetjournals.org/content/suppl/2014/07/01/dmd.114.059295.DC1Supplemental material to this article can be found at:

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

Schmitz, 2007). Based on the structural features of cilengitide (Fig. 1),potential metabolic reactions catalyzed by P450s could include aromatichydroxylation of phenylalanine, aliphatic hydroxylation of valine,N-demethylation of the valine N-methyl group, and deamination anddealkylation of arginine (Guengerich, 2001).Our study characterized the inhibition of P450 3A4 by cilengitide

using human liver microsomes and recombinant P450 3A4, includingthe type and mechanism of inactivation, and we explain the observednonextractable radioactivity observed in vivo.

Materials and Methods

Chemicals and Enzymes. Cilengitide, [14C-guanidine]-labeled cilengitide,and the ornithine homolog of cilengitide were synthesized by Merck SeronoGmbH (Darmstadt, Germany) (Jonczyk et al., 1999). Nifedipine was obtainedfrom Sigma-Aldrich (St. Louis, MO) and recrystallized from C2H5OH in amberglass (Guengerich et al., 1986). Testosterone was obtained from Steraloids(Newport, RI). Baculosomes (microsomes from baculovirus-infected P450recombinant insect cells) with human P450 3A4 or P450 3A5, NADPH-P450reductase, and cytochrome b5 were purchased from Life Technologies (Carlsbad,CA). Bovine serum albumin and ovalbumin were obtained from Sigma-Aldrich.The guanidine-based compounds debrisoquine and N-acetylarginine-O-methylester were purchased from Sigma-Aldrich and Bachem (Bubendorf, Switzerland),respectively. All other reagents and solvents were obtained from general commercialsuppliers and were used without further purification.

Ten liver samples from a stock (Schadt et al., 2008) in our laboratory(Vanderbilt) were used to prepare human liver microsomes (four males, fivefemales, one unknown sex; median age 27 years). Recombinant P450 3A4 witha C-terminal (His)5 tag (Gillam et al., 1993; Hosea et al., 2000) was expressedin Escherichia coli and purified as described previously elsewhere (Hosea et al.,2000). E. coli recombinant rat NADPH-P450 reductase (Hanna et al., 1998) andhuman liver cytochrome b5 (Guengerich, 2005b) were prepared as describedelsewhere. All other proteins and enzymes were purchased from Sigma-Aldrich.

Incubations and Residual Activity. Incubations were conducted at 37°C in100 ml incubation mixtures containing 50 mM potassium phosphate buffer (pH 7.4for testosterone and pH 7.85 for nifedipine), an NADPH-generating system(10 mM glucose 6-phosphate, 0.5 mg/ml NADP+, and 2 mg/ml yeast glucose6-phosphate dehydrogenase), and diagnostic substrate (testosterone 210 mM,nifedipine 20 mM). Depending on the P450 enzyme system used in incubations,final concentrations were 1 mM (P450) in human liver microsomes, 0.05 mMbaculovirus recombinant P450 3A4 or 3A5, or 2 mM reconstituted recombinantP450 3A4. The reconstituted P450 3A4 system contained: P450 3A4 (2 mM),NADPH-P450 reductase (4 mM), cytochrome b5 (2 mM), sodium cholate (0.5 mM),and a lipid mixture (L-a-1,2-dioleoyl-sn-glycero-3-phosphocholine, L-a-1,2-dilauroyl-sn-glycero-3-phosphocholine, and bovine brain phosphatidylserine

in a ratio of 1:1:1 (w/w/w), 40 mg/ml). Cilengitide solution in phosphate bufferwas prepared ex tempore; the highest concentration used was 15 mM.

In general, preincubations were performed for 30 minutes with or without anNADPH-generating system. Residual activity was determined by coincubatingwith a marker substrate for an additional 5 minutes (Sohl et al., 2009). Briefly,each reaction mixture was quenched by the addition of four volumes of CH2Cl2and centrifuged at 1900g for 10 minutes. The organic layer was transferred,and the solvent was evaporated under a stream of nitrogen; the dried samplewas dissolved in CH3OH for analysis. 6b-Hydroxytestosterone and oxidizednifedipine were determined using a liquid chromatography photodiode arrayAcquity system (Waters, Milford, MA) and an octadecylsilane (C18) column(6.2 mm � 80 mm, 3 mm; Agilent Technologies, Santa Clara, CA) for thenifedipine assays and a similar octadecylsilane column (4.6 mm � 250 mm,5 mm; Phenomenex, Torrance, CA) for the testosterone assays with CH3OH–H2O (64/36, v/v) for isocratic elution in both cases.

For determining inactivation kinetics with cilengitide, concentrations of 0,0.75, 1, 3, 5, 7.5, and 15 mM were preincubated (in duplicate) with pooledhuman liver microsomes and an NADPH-generating system for 0, 3, 6, 9, 15,and 30 minutes. After the preincubation, an aliquot of marker substrate wasadded, with the procedures of isolation and quantitation described earlier.

Determination of Reversible and Quasi-irreversible Inhibition. Cilengitide(final concentration 15 mM) was preincubated for 30 minutes with humanliver microsomes (500 pmol P450) at 37 6 1°C in 500 ml mixtures containing50 mM potassium phosphate buffer (pH 7.4) and an NADPH-generating system.Preincubation and solvent controls were treated with potassium ferricyanide(final concentration 2 mM), after which samples and control were dialyzed for24 hours against 50 mM phosphate buffer (pH 7.4). Three replicates of eachincubation were used to measure residual enzyme activity with nifedipine andtestosterone as substrates. When reversible or quasi-irreversible inhibition wasobserved with human liver microsomes, the results were confirmed using re-combinant (baculovirus) P450 3A4.

Pyridine Hemochrome Spectrophotometric Assay.We added H2O (950 ml)to 250 ml of a P450 preincubation mixture (see above). After the addition of0.20 ml of pyridine, the mixture was mixed using a vortex device, and 0.10 mlof 1 M NaOH was added, followed by more vortex mixing. The solution wassplit in two parts, reference and sample, and placed in cuvettes (capped) thatwere scanned from 500 to 600 nm to establish a baseline with an AmincoDW2a/OLIS spectrophotometer (On-Line Instrument Systems, Bogart, GA).A few crystals of solid sodium dithionite and 10 ml of H2O were added to thesample cuvette. Potassium ferricyanide (10 ml of a 1 M solution) was added tothe reference cuvette. Samples were scanned within 1 minute after the addition ofthe alkaline solution because of the instability of the pyridine hemochromogenunder basic conditions. The heme concentration was calculated based on anextinction coefficient of 20.7 mM21 cm21 for the difference in absorption be-tween peak at 557 nm and the trough at 541 nm (Flink and Watson, 1942; Paulet al., 1953).

Radioactivity Assays. The reconstituted P450 3A4 system containing0.5 nmol P450 3A4, 1.0 nmol NADPH-P450 reductase, and 0.5 nmol cyto-chrome b5 was incubated with 7.5 mM 14C-labeled cilengitide (specific activity8 Ci/mol). Two incubations, with and without (control) an NAPDH-generatingsystem, were performed. The reactions were stopped by chilling the tubes onice. Enzymes were separated by size-exclusion chromatography (Micro Bios-Spin column P-6; Bio-Rad Laboratories, Hercules, CA), and cilengitide wasrecovered from the column by washing the column with a C2H5OH–H2Omixture (1:1, v/v, 4°C). Recovered enzyme samples were loaded on a gradientpolyacrylamide electrophoresis gel (NuPAGE 4–12% (w/v) Bis-Tris Gel; LifeTechnologies, Grand Island, NY). To test the specificity of binding, incubationsincluded bovine serum albumin (4 mM) and ovalbumin (4 mM); proteins wereseparated on 7.5% (w/v) SDS-PAGE gels. After development, gels werestained with Colloidal Coomassie Blue (Kang et al., 2002).

For liquid scintillation counting, bands containing individual proteins werecut from the gel and were treated with 1.0 ml of 15% H2O2 (w/v) for 2 hours at70°C in the dark. Scintillation fluid (10 ml of ScintiVerse II Cocktail; FisherScientific, Waltham, MA) was added, and the samples were left in the darkovernight to diminish luminescence, after which radioactivity was recorded onan LS6500 Multipurpose Scintillation Counter (Beckman Coulter, Pasadena,CA). Percentages of covalently modified proteins were determined based on thedifference in counts of incubations with and without NADPH.

Fig. 1. Structure of cilengitide. The ornithine derivative has an amine substituted forthe guanidine group.

Cilengitide Inhibition of P450 3A4 1439

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

One gel (loaded with proteins from the incubations with and without anNADPH-generating system) was treated with Autofluor (National Diagnostics,Atlanta, GA). Autofluor enables impregnation with phosphors, which convertradioactive b emission into more photons. After the treatment, the gel wasdried, placed on an imaging screen (Kodak; Bio-Rad Laboratories), and left

at 270°C. The imagining screen was visualized after 7 to 10 days on aMolecular Imager Pharos FX Plus Systems (Bio-Rad Laboratories).

Spectrophotometry. Spectral changes resulting from 2 mM ligand (cilengitide,cyclopeptide analogs of cilengitide, and guanidine derivatives) binding to 2.5 mMrecombinant P450 3A4 were monitored for 8 hours. The binding affinity of

Fig. 2. Inactivation kinetics determined with humanliver microsomes and recombinant P450 3A4 sys-tems using testosterone and nifedipine assays. (A)The cilengitide inactivation constant determined withhuman liver microsomes (testosterone assay), KI,d was.15 mM (highest concentration analyzed; if extrapo-lated, KI = 216 1 mM and kinact = 0.076 0.01 min21).(B) For the nifedipine assay, KI = 6.3 6 1.6 mM andinactivation rate kinact = 0.017 6 0.002 min21. Forinactivation of recombinant P450 3A4 testosterone assay(C), KI = 1.1 6 0.2 mM and kinact = 0.06 6 0.03 min21

and for the nifedipine assay (D), KI = 5.0 6 1.1 mMand kinact = 0.21 6 0.03 min21. Thus, with bothtestosterone and nifedipine assays, kinact/KI � 0.0036 0.001 min21 mM21 for recombinant P450 3A4.Concentrations of cilengitide: d 0, j 0.75, m 1,. 3, ♦ 5, s 7.5, u 15 mM.

1440 Boji�c et al.

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

cilengitide to both reduced and oxidized CYP3A4 was estimated by monitoringthe (absolute) spectral changes of 2.5 mM enzyme with each addition ofcilengitide (every 10 minutes) in a total volume of 1.0 ml of 50 mM potassiumphosphate buffer (pH 7.4) at 23°C. Binding to the ferrous form of enzyme wasdetermined under anaerobic conditions [achieved under argon atmosphere andaddition of protocatechuate-3,4-dioxygenase with 3,4-dihydoxybenzoic acid forremoval of oxygen traces (Bull et al., 1981; Guengerich et al., 2004)] afterphotoreduction (halogen lamp 500W, 1 minute) in the presence of 1mM5-deazaflavin(Massey and Hemmerich, 1978).

Absorbance spectra were recorded from 350 to 500 nm using an AmincoDW2a/OLIS (for monitoring binding to oxidized P450 3A4) or a Cary 14/OLISspectrophotometer (for monitoring binding to reduced P450 3A4; On-LineInstrument Systems). Changes in the difference spectra (at the wavelengthmaximum and minimum) were plotted against the cilengitide concentrations toestimate affinity.

Synthesis of Acetylated Ornithine Derivative of Cilengitide. Theornithine derivative of cilengitide (Merck Serono) was acetylated with aceticanhydride in pyridine at room temperature for 1 hour (Karanewsky et al.,1988). The organic solvent was removed by coevaporation with (C2H5)2O in

vacuo. Residual traces of solvents were removed in vacuo after addition ofCH2Cl2 with (C2H5)3N. A cation exchange column was used for purification;that is, a Dowex AG 50W-X2 ion-exchange column (H+ form) was equilibratedwith acetic buffer (pH 4.5) before application of an aqueous solution of theresidue (after evaporation). Fractions were monitored with an Aminco DW2a/OLIS spectrophotometer (200–400 nm). Structure and purity of the product(55% yield) were confirmed by liquid chromatography with mass spectrometryand NMR (Supplemental Fig. 1).

Data Analysis. Data from assays were processed in GraphPad Prismsoftware (GraphPad Software, San Diego, CA). As inhibition was monitored,a one-sided t-test (a = 0.05) was used for the assessment of statisticalsignificance in the differences between samples and controls, based onmeasurements of residual activity. A nonlinear three parameter sigmoidal-logistic equation was used for IC50 calculations: Residual activity = Bottom +(Top-Bottom)/(1 + 10exp((X 2 logIC50))). The Michaelis-Menten equationwas used for determination of inactivation constants (KI) and inactivation rateconstants (kinact) of inhibitor (I): kobs = kinact � [I]/(KI + [I]). A quadraticequation was used to determine dissociation constants (Kd) of ligands (L) withthe enzyme (E): DA = A0 + (Bmax/2[E])((Kd + [E] + [L]) 2 ((Kd + [E] + [L])2 24[E][L])1/2). A0 and Bmax were constants in each analysis. DynaFit simulationsoftware (Biokin, Pullman, WA) was used with a one-step enzyme-ligandbinding model (Kuzmic, 1996), yielding similar results obtained from thequadratic nonlinear regression analyses in GraphPad Prism.

Results

Inhibition Assays and Kinetics. Cilengitide was tested with humanliver microsomes for direct inhibition (no preincubation) and afterpreincubation with and without an NADPH-generating system. In boththe nifedipine and testosterone assays, direct inhibition was observedwhen the highest concentrations of cilengitide were used. Cilengitide(15 mM) decreased enzyme activity by 15% (P = 0.033) and 19% (P =0.034) in testosterone and nifedipine assays, respectively. When cilengitidewas preincubated without the addition of NADPH, reductions of en-zyme activity by 23% (testosterone assay, P = 0.031) and 10% (nifed-ipine assay, P = 0.008), respectively, were observed. Greater inhibitionwas observed when cilengitide was preincubated with NADPH: bothnifedipine and testosterone assays showed statistically significant de-clines in enzyme activity: 60% (P = 0.007) and 54% (P = 0.007),respectively.

Fig. 3. Heme assays. After incubation for 30 minutes with an NADPH-generatingsystem in the presence of catalase, no statistically significant difference in pyridinehemochrome complex formation was observed in comparing a control incubation,a cilengitide incubation, and reconstituted P450 3A4 without incubation (enzymesmixture).

Fig. 4. Binding of cilengitide to proteins. (A) An SDS-polyacrylamidegradient (4–12%, w/v) gel showed binding of cilengitide to NAPDH-P450 reductase and P450 3A4 in a comparison of incubations ofa reconstituted P450 3A4 system (P450 3A4, reductase, cytochrome b5,phospholipids) with and without an NADPH-generating system. (B) Todetermine whether nonspecific covalent binding to other proteinsoccurred, bovine serum albumin (BSA) and ovalbumin (OVA) wereadded to the incubation mixtures and separated on an SDS-polyacrylamide(7.5%, w/v) gel. [B1] Autoradiography was performed after 10 daysof exposure. [B2] The same gel after staining with CoomassieBrilliant Blue R-250. More intensive binding of [14C] cilengitide tobovine serum albumin was observed in incubations with the NADPH-generating system. B2: cytochrome b5: b5; NADPH-P450 reductase,NPR; P450 3A4: 3A4. (C) Nonspecific binding was observed whenthe incubation of cilengitide was preformed with human livermicrosomes (C1, autoradiography; C2, Coomassie Brilliant BlueR-250 staining). The percentages of binding to P450 3A4, NADPH-P450 reductase, and bovine serum albumin were determined afterH2O2 digestion of gels by scintillation counting.

Cilengitide Inhibition of P450 3A4 1441

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

For the NADPH-dependent inhibition, inactivation kinetics weredetermined (Fig. 2). In the nifedipine assay, the inactivation constantKI was 6.3 6 1.6 mM and the inactivation rate (kinact) was 0.017 60.002 min21, corresponding to an inactivation efficiency (kinact/KI)of 0.003 6 0.001 min21 mM21. In the testosterone assay the appar-ent KI value was greater than the highest concentration analyzed(KI . 15 mM).In that nifedipine and testosterone are substrates for both P450 3A4

and P450 3A5, recombinant baculovirus-based microsomes were usedto confirm the inhibitory effect of cilengitide. When 15 mM cilengitidewas used, complete inhibition of P450 3A4 was observed in bothassays. Preincubation of P450 3A5 with NADPH caused 9% (P =0.063) and 23% (P = 0.006) decreases of activity in the nifedipineand testosterone assays, respectively.The values of inactivation parameters determined for P450 3A4

were KI = 1.1 6 0.2 mM and kinact = 0.06 6 0.03 min21 in thetestosterone assay and KI = 5.06 1.1 mM and kinact = 0.216 0.03 min21

in the nifedipine assay (Fig. 2). The inactivation efficiency (kinact/KI) inboth assays with recombinant P450 3A4 was 0.046 0.01 min21 mM21.Determination of Inactivation Mechanism. The pyridine hemo-

chrome spectrophotometric assay was used to determine possiblecovalent modification of heme (Flink and Watson, 1942; Paul et al.,1953) after preincubation of 15 mM cilengitide and P450 3A4 withNADPH. When the assay was performed without the addition of anoxidizing agent (ferricyanide) to the reference cuvette, an apparent72% (P , 0.001) decrease in heme concentration was observed.In the presence of ferricyanide (to reoxidize the P450 heme), theheme loss was reduced to 51% (P, 0.001). However, when preincubationwas performed in the presence of catalase, no heme loss was ob-served (Fig. 3), in accordance with a lack of statistically significantchanges in concentrations of (spectrally determined) P450 (P =0.12) and cytochrome b5 (P = 0.11), as well as residual activity(P = 0.070).Autoradiography of a reconstituted P450 3A4 system showed

binding of 14C-labeled cilengitide to P450 3A4 and NADPH-P450reductase (Fig. 4A). This result was confirmed with scintillationcounting after digestion of gels with H2O2: 8% binding to P450 3A4and 6% to the reductase. Binding was observed not only to P450 3A4but also the reductase, and therefore incubations were performed

with bovine serum albumin and ovalbumin as “traps” (Fig. 4, B1and B2). Significant binding of cilengitide to bovine serum albuminwas observed when compared with control (without NADPH) (4%).In human liver microsomes, binding of 14C-labeled cilengitidewas observed to proteins with higher molecular weights (Fig. 4,C1 and C2).The reversibility of cilengitide binding was investigated by

dialyzing incubations for 24 hours and testing residual activity. Noreversibility was observed in the testosterone and nifedipine assays(Fig. 5). To determine possible quasi-irreversible inhibition (Silverman,1995), preincubations were treated with potassium ferricyanide anddialyzed for 24 hours. Residual activity was assayed; no recovery ofenzyme activity was observed in the nifedipine assay (Fig. 5). Recoveryin the testosterone assay on human liver microsomes was observed(Fig. 5) but not confirmed when preincubations were performed withrecombinant P450 3A4.Binding of Cilengitide to Oxidized and Reduced Forms of P450

3A4. When monitoring an incubation of cilengitide with reconstitutedP450 3A4 for 2 hours, slow changes in absorbance were observed at426 nm (increase) and at 394 and 410 nm (decrease). To assess thebinding of cilengitide to P450 3A4, spectral changes were monitoredfor 8 hours. Slow type II spectral changes were observed (Fig. 6, Aand C). The maximal spectral change was achieved after 2 hours,indicating slow inhibitor-enzyme complex formation. This was furtherconfirmed by testing residual activity after a 2-hour preincubation ofcilengitide with human liver microsomes and without NADPH. Highervalues of inhibition were observed in both the testosterone (46%, P =0.003) and nifedipine (38%, P = 0.001) assays when compared with30 minutes of preincubation (see Inhibition Assays and Kinetics).A very low apparent second-order rate of cilengitide-P450 3A4

complex formation (130 M21 s21) was estimated on the basis of aone-ligand/two-state binding model, although it is not necessarilyappropriate in that the binding of ligands to P450 3A4 involvesmultiple states/reactions after a diffusion-limited encounter (Isin andGuengerich, 2007).A complex with the ferric form of P450 3A4 was formed at much

lower concentrations (Kd = 0.66 mM; Fig. 6E) compared with themillimolar concentrations at which enzyme inhibition was observed.Opposite to the ferric system, the inhibitor-enzyme complex with ferrous

Fig. 5. Reversibility of cilengitide binding to P450 3A4. To test the reversibility of cilengitide binding to P450 3A4, incubations with human liver microsomes wereperformed with the addition of an NAPDH-generating system for 30 minutes, after which the samples were dialyzed for 24 hours at 4°C. Negative controls contained noNADPH-generating system. Residual activity was tested with both the testosterone (left) and nifedipine assays (right). To test quasi-irreversibility, potassium ferricyanide(K3Fe(CN)6) was added after preincubation and before dialysis for 1 hour.

1442 Boji�c et al.

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

P450 3A4 formed quickly (Fig. 6, B and D). A Kd for ferrous P4503A4 (determined under anaerobic conditions) was also in the micromolarrange (Kd = 6 mM; Fig. 6F), explaining the observed NADPH-dependentinhibition.Spectral Changes and Inhibition Assays of Guanidine Deriva-

tives. The guanidine group was hypothesized to be involved incilengitide inhibition of P450 3A4. Debrisoquine and N-acetylarginine-O-methyl ester were tested for inhibitory effects and spectral changes.Binding constants were determined using a one-ligand/two-state binding

model: Kd = 1.0 mM for debrisoquine and Kd = 0.14 mM for thearginine derivative (Fig. 7, A and B). The arginine derivative (finalconcentration 15 mM) caused statistically significant inhibition of P4503A4 only in the nifedipine assay; the residual activity was 71 6 4% (nopreincubation), 68 6 8% (preincubation without NADPH), 85 6 6%(preincubation with NADPH) (Fig. 7D). Debrisoquine (final concen-tration 15 mM) caused significant reduction of P450 3A4 activity inboth assays. A greater decrease was observed in the testosterone assay,with residual activity being 13 6 5% (no preincubation), 24 6 3%

Fig. 6. Binding of cilengitide to the ferric and ferrous forms of P450 3A4. (A) Type II spectral changes (maximum at 435 nm and minimum at 415 nm) when cilengitide wasadded to the oxidized (ferric) form of enzyme (A), and the appearance of a maximum at 417 nm and minimums at 403 and 447 nm were observed for binding to the ferrous(reduced) form of the enzyme (B). Binding to the oxidized (ferric) form was slow and reached a maximum after 2 to 3 hours (C), in contrast to the rapid binding to thereduced form of P450 3A4 (,2 minutes, D). Binding of cilengitide to both ferric (E) and ferrous (F) P450 3A4, fit to a one-ligand quadratic model: Kd,Fe3+ = 0.666 0.09 mMand Kd,Fe2+ = 6 6 3 mM.

Cilengitide Inhibition of P450 3A4 1443

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

(preincubation without NADPH), and 19 6 2% (preincubation withNADPH) (Fig. 7C). In the nifedipine assay, the residual activity was 4664% (no preincubation), 38 6 4% (preincubation without NADPH),and 31 6 1% (preincubation with NADPH) (Fig. 7D).Spectral Changes and Inhibition Assays with Cyclopeptide

Analogs. Spectral changes were monitored to determine the bindingconstants of cilengitide analogs. Based on the one-ligand/one-sitebinding model, the dissociation constants Kd were 0.52 mM for theornithine derivative and 0.44 mM for the acetylated ornithinederivative. Complete inhibition of P450 3A4 activity was observedwith the ornithine derivative in the nifedipine assay whereas 17 6 2%residual activity was observed in the testosterone assay (Fig. 8C). Theacetylated derivative caused only a 6 6 2% decrease of activity in thenifedipine assay (P = 0.015); a more potent decrease was observed inthe testosterone assay (39 6 6%, P = 0.003; Fig. 8C).

Discussion

In preliminary single-point inhibition assays, cilengitide (15 mM)caused direct (no preincubation), time-dependent (30 minute preincuba-tion without NADPH-generating system), and also NADPH-dependentinhibition (30 minute preincubation with NADPH-generating system) of

P450 3A4 in human liver microsomes, the most prominent being the latter(54–60% depending on the probe substrate used). Complete inactivationwas confirmed with a baculosome-based recombinant P450 3A4 systemfor both substrate probes, and the maximal inhibition of P450 3A5 was23% under the same conditions.Inactivation kinetics for metabolism (NADPH)-dependent inhibi-

tion were studied with both human liver microsomes and recombinantbaculosomes (Fig. 2). As the values of the inhibition constant are inthe millimolar range and the maximal plasma concentrations in hu-mans are in a low micromolar range, this would indicate low potentialfor drug-drug interactions.In principle, a P450 inactivator can modify prosthetic heme, bind

to apoprotein, or do both (Wienkers and Heath, 2005). The spec-trophotometric assay (pyridine hemochrome) did not confirm anyheme loss when incubations were performed in the presence of cat-alase to destroy any H2O2 generated (Fig. 3). This assay is based ondetermining heme iron in ferrous form as a complex with pyridine inbasic media [versus a reference to which ferricyanide is added (Flinkand Watson, 1942; Paul et al., 1953)]. During the optimization ofassay, greater heme loss was observed when ferricyanide was notadded (70 versus 27%), indicating the possible existence of a stableFe2+-cilengitide complex.

Fig. 7. Spectral changes and inhibitory effects of guanidines. Spectral changes were recorded every 10 minutes, and difference spectra are shown (A and B). Type II spectralchanges were observed with (A) debrisoquine and (B) N-acetyl-arginine O-methyl ester (Ac-Arg-O-Me). Residual activity was measured in testosterone and nifedipineassays with human liver microsomes, with or without preincubation and an NADPH-generating system. (C) No statistically significant inhibition was observed with 15 mMAc-Arg-O-Me. (D) In the nifedipine assay, 15 mM debrisoquine caused a significant reduction of activity (up to 70%), most prominent being preincubation with NADPH-generating system. Ac-Arg-O-Me caused a 32% decrease of activity.

1444 Boji�c et al.

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

[14C] Cilengitide was bound not only to P450 3A4 but also toNADPH-P450 reductase (Fig. 4). Nonspecific binding was also confirmedwhen radioactive cilengitide was coincubated with bovine serum albuminand ovalbumin present. However, only 8% binding to P450 3A4 wasobserved (i.e., 0.08 nmol cilengitide label/nmol P450 3A4), which doesnot explain the complete inhibition of P450 3A4 observed with thebaculovirus system under the same conditions.In that covalent binding did not explain the observed inhibition, possible

quasi-irreversible inhibition was analyzed. In this case, the heme ironwould form a complex with cilengitide that appears irreversible but can bedestabilized and enzyme activity recovered by subsequent incubation withferricyanide and dialysis (Silverman, 1995). However, no recovery ofenzyme activity was observed (Fig. 5). A disadvantage of this approach isthe assumption that ligand binds to the ferric form (Ortiz de Montellanoand De Voss, 2005; Sevrioukova and Poulos, 2013) and forms a morestable enzyme-ligand complex after one-electron reduction (Correia andOrtiz de Montellano, 2005). However, cilengitide formed a more stablecomplex with the oxidized form of P450 3A4 (Kd,Fe3+ = 0.66 mM versusKd,Fe2+ = 6 mM).The quasi-irreversibility experiment suggested possible formation of

a time-dependent cilengitide-P450 3A4 complex (i.e., slow tight-bindinginhibition). This phenomenon has not been reported with P450s but hasbeen reported for nonsteroid anti-inflammatory drug inhibition of cyclo-oxygenase (Prusakiewicz et al., 2004). Drug-enzyme complexes shouldeventually dissociate when dialyzed (Silverman, 1995; Szedlacsek andDuggleby, 1995), but no recovery of enzyme activity was observed afterdialysis for 1 day (Fig. 5). Negative results for this experiment cannoteliminate the possibility of slow, tight binding, as has been shown forproteasome inhibitors (Manam et al., 2008).Based on spectral changes, we postulated that the guanidine group

of cilengitide was responsible for the observed inhibition. To test thishypothesis, we used guanidine-containing analogs (debrisoquine andN-acetyl-arginine-O-methyl ester) as well as the N-acetyl ornithine

derivative of cilengitide (substitution for Arg). For both debrisoquineand N-acetyl-arginine-O-methyl ester, slow ligand-enzyme formationwas observed, taking more than 1 and 6 hours to achieve themaximum type II spectral change. Inhibition assays also confirmed thelow residual activity of P450 3A4 preincubated with debrisoquine inboth the testosterone and nifedipine assays, while inhibition with thearginine derivative was only observed in the nifedipine assay (Fig. 7).The ornithine derivative has an amino group on a residue chain and,

as might be expected, caused type II spectral changes characteristic forinhibitors (Schenkman et al., 1967); strong inhibition was confirmedin inhibition assays with both testosterone and nifedipine as markersubstrates. However, similar spectral changes were observed with theN-acetyl ornithine derivative of cilengitide. Nonetheless, the inhibitionassays showed a much lower reduction of catalytic activity comparedwith cilengitide and the ornithine derivative (Fig. 8).Binding of ligands to P450 3A4 is not a one-step process and requires

time (Isin and Guengerich, 2007; Isin and Guengerich, 2008). Recently,unusual type II spectral changes of bicalutamide (an antiandrogen forprostate cancer treatment) binding to P450 46A1 were reported and at-tributed to binding involving a water molecule as an intermediate (Mastet al., 2013). This phenomenon could provide an explanation for theobserved changes, even with the protected amino derivative of cilengitide.A guanidine group has not been reported to cause clinically significant

interactions, at least not with P450s. Commonly used drugs containing aterminal guanidino group include famotidine (an antiulcer drug) and met-formin (an oral hypoglycemic). Famotidine was developed as a histamineH2-receptor antagonist with low interaction potential compared with ci-metidine and ranitidine (Humphries, 1987). Metformin exhibits interactionswith organic cation transporters (the guanidino group is positively charged)(Somogyi et al., 1987).To date, the drug-drug interaction package of P450 inhibition includes

an assessment of the direct inhibition caused by the candidate drug.In addition, the indirect inhibition caused by the formation of a reactive

Fig. 8. Spectral changes and inhibitory effectof cyclopeptides. Spectral changes were re-corded every 10 minutes, and difference spectraare shown (A and B). (A) The ornithine deri-vative (cilengitide with arginine replaced withornithine) caused type II spectral changes thatresemble cilengitide binding to P450 3A4. (B)Similar spectral changes were observed forbinding of acetylated derivative to P450 3A4.(C) Residual activity was measured in testos-terone and nifedipine assays; preincubation wasperformed with human liver microsomes and anNADPH-generating system. The ornithine de-rivative caused complete inhibition of P450 3A4in the nifedipine assay and a decrease of 82% inactivity in the testosterone assay. Acetylation ofthe ornithine derivative reduced the inhibitoryeffect (testosterone assay: 39%, nifedipine as-say: 6%).

Cilengitide Inhibition of P450 3A4 1445

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from

metabolite or more potent metabolite is studied by preincubating thecandidate drug with the test system (e.g., human liver microsomes) andNADPH- or NADPH-re/generating system. Our finding with cilengitideindicates that other relevant inhibition mechanisms such as time- but notNADPH-dependent (e.g., slow and tight binding) might also exist. Beforeincluding such a new evaluation into a drug-drug interaction package, it isimportant to learn how often this mechanism occurs in the more classicchemical space for drugs.In conclusion, cilengitide causes time- and NADPH-dependent in-

hibition of P450 3A4. The observed inhibition can be explained by directinhibition and by formation of stable complexes with both ferric andferrous forms of heme iron. Formation of these stable complexes is time-and NADPH-dependent and is attributed to the guanidino group. Onlya small part of the NADPH-dependent inhibition could be attributed to anyreactive products capable of reacting with the protein or heme, explainingthe nonextractable radioactivity observed in vivo. To date, the chemicalmechanism of the adduct formation is not known.

Acknowledgments

The authors thank Leslie D. Nagy and Eric Gonzalez for technical assistancewith some of the experiments and Drs. Florian Huber and Markus Fluck forproviding the 14C material and for valuable comments in preparing the manuscript.

Authorship ContributionsParticipated in research design: Guengerich, Barbero, Dolgos, Boji�c.Conducted experiments: Boji�c, Guengerich.Contributed new reagents or analytic tools: Barbero, Riva.Performed data analysis: Boji�c, Guengerich, Barbero, Dolgos, Gallemann.Wrote or contributed to the writing of the manuscript: Boji�c, Guengerich,

Barbero, Dolgos, Gallemann, Freisleben.

References

Amundsen R, Åsberg A, Ohm IK, and Christensen H (2012) Cyclosporine A- and tacrolimus-mediated inhibition of CYP3A4 and CYP3A5 in vitro. Drug Metab Dispos 40:655–661.

Bernkop-Schnürch A and Schmitz T (2007) Presystemic metabolism of orally administeredpeptide drugs and strategies to overcome it. Curr Drug Metab 8:509–517.

Bull C, Ballou DP, and Otsuka S (1981) The reaction of oxygen with protocatechuate 3,4-dioxygenase from Pseudomonas putida. Characterization of a new oxygenated intermediate.J Biol Chem 256:12681–12686.

Correia MA and Ortiz de Montellano PR (2005) Inhibition of cytochrome P450 enzymes, inCytochrome P450: Structure, Mechanism, and Biochemistry, 3rd ed (Ortiz de Montellano PRed) pp 247–322, Kluwer Academic/Plenum Press, New York.

Craik DJ, Fairlie DP, Liras S, and Price D (2013) The future of peptide-based drugs. Chem BiolDrug Des 81:136–147.

Dechantsreiter MA, Planker E, Mathä B, Lohof E, Hölzemann G, Jonczyk A, Goodman SL, and KesslerH (1999) N-Methylated cyclic RGD peptides as highly active and selective alpha(V)beta(3) integrinantagonists. J Med Chem 42:3033–3040.

Eskens FA, Dumez H, Hoekstra R, Perschl A, Brindley C, Böttcher S, Wynendaele W, DrevsJ, Verweij J, and van Oosterom AT (2003) Phase I and pharmacokinetic study of continuoustwice weekly intravenous administration of cilengitide (EMD 121974), a novel inhibitor ofthe integrins avb3 and avb5 in patients with advanced solid tumours. Eur J Cancer 39:917–926.

Flink EB and Watson CJ (1942) A method for the quantitative determination of hemoglobin andrelated heme pigments in feces, urine, and blood plasma. J Biol Chem 146:171–178 http://www.jbc.org/content/146/1/171.citation.

Gillam EM, Baba T, Kim BR, Ohmori S, and Guengerich FP (1993) Expression of modifiedhuman cytochrome P450 3A4 in Escherichia coli and purification and reconstitution of theenzyme. Arch Biochem Biophys 305:123–131.

Goodman SL and Picard M (2012) Integrins as therapeutic targets. Trends Pharmacol Sci 33:405–412.Guengerich FP (2001) Common and uncommon cytochrome P450 reactions related to metabo-lism and chemical toxicity. Chem Res Toxicol 14:611–650.

Guengerich FP (2005a) Human cytochrome P450 enzymes, in Cytochrome P450: Structure,Mechanism, and Biochemistry, 3rd ed (Ortiz de Montellano PR ed) pp 377–530, KluwerAcademic/Plenum Press, New York

Guengerich FP (2005b) Reduction of cytochrome b5 by NADPH-cytochrome P450 reductase.Arch Biochem Biophys 440:204–211.

Guengerich FP, Krauser JA, and Johnson WW (2004) Rate-limiting steps in oxidations catalyzedby rabbit cytochrome P450 1A2. Biochemistry 43:10775–10788.

Guengerich FP, Martin MV, Beaune PH, Kremers P, Wolff T, and Waxman DJ (1986) Char-acterization of rat and human liver microsomal cytochrome P-450 forms involved in nifedipineoxidation, a prototype for genetic polymorphism in oxidative drug metabolism. J Biol Chem261:5051–5060.

Hanna IH, Teiber JF, Kokones KL, and Hollenberg PF (1998) Role of the alanine at position 363of cytochrome P450 2B2 in influencing the NADPH- and hydroperoxide-supported activities.Arch Biochem Biophys 350:324–332.

Hosea NA, Miller GP, and Guengerich FP (2000) Elucidation of distinct ligand binding sites forcytochrome P450 3A4. Biochemistry 39:5929–5939.

Humphries TJ (1987) Famotidine: a notable lack of drug interactions. Scand J Gastroenterol Suppl 134:55–60.

Isin EM and Guengerich FP (2007) Multiple sequential steps involved in the binding of inhibitorsto cytochrome P450 3A4. J Biol Chem 282:6863–6874.

Isin EM and Guengerich FP (2008) Substrate binding to cytochromes P450. Anal Bioanal Chem392:1019–1030.

Jonczyk A, Goodman S, Diefenbach B, Sutter A, Holzemann G, Kessler H, and Dechantsreiter M(1999) inventors, Merck Patent Gesellschaft Mit Beschrankter Haftung, assignee. Cyclic ad-hesion inhibitors. U.S. patent 6,001,961 A. 1996 Sep 16.

Kang D, Gho YS, Suh M, and Kang C (2002) Highly sensitive and fast protein detection withCoomassie Brilliant Blue in sodium dodecyl sulfate-polyacrylamide gel electrophoresis. BullKorean Chem Soc 23:1511–1512 http://post.queensu.ca/~chinsang/lab-protocols/handbooks-and-manuals/kang-et-al-2002-new-coomass.pdf.

Karanewsky DS, Badia MC, Cushman DW, DeForrest JM, Dejneka T, Loots MJ, Perri MG,Petrillo EW, Jr, and Powell JR (1988) (Phosphinyloxy)acyl amino acid inhibitors of angiotensinconverting enzyme (ACE). 1. Discovery of (S)-1-[6-amino-2-[[hydroxy(4-phenylbutyl)phosphinyl]oxy]-1-oxohexyl]-L-proline a novel orally active inhibitor of ACE. J Med Chem 31:204–212.

Kuzmic P (1996) Program DYNAFIT for the analysis of enzyme kinetic data: application to HIVproteinase. Anal Biochem 237:260–273.

Manam RR, McArthur KA, Chao TH, Weiss J, Ali JA, Palombella VJ, Groll M, Lloyd GK,Palladino MA, and Neuteboom ST, et al. (2008) Leaving groups prolong the duration of 20Sproteasome inhibition and enhance the potency of salinosporamides. J Med Chem 51:6711–6724.

Massey V and Hemmerich P (1978) Photoreduction of flavoproteins and other biological com-pounds catalyzed by deazaflavins. Biochemistry 17:9–16.

Mast N, Zheng W, Stout CD, and Pikuleva IA (2013) Binding of a cyano- and fluoro-containingdrug bicalutamide to cytochrome P450 46A1: unusual features and spectral response. J BiolChem 288:4613–4624.

Ortiz de Montellano PR and De Voss JJ (2005) Substrate oxidation by cytochrome P450enzymes, in Cytochrome P450: Structure, Mechanism, and Biochemistry, 3rd ed (Ortiz deMontellano PR ed) pp 183–245, Kluwer Academic/Plenum Press, New York.

Paul KG, Theorell H, and Åkeson Å (1953) The molar light absorption of pyridine ferroproto-porphrin (pyridine haemochromogen). Acta Chem Scand 7:1284–1287 DOI: 10.3891/acta.chem.scand.07-1284.

Prusakiewicz JJ, Felts AS, Mackenzie BS, and Marnett LJ (2004) Molecular basis of the time-dependent inhibition of cyclooxygenases by indomethacin. Biochemistry 43:15439–15445.

Reardon DA, Neyns B, Weller M, Tonn JC, Nabors LB, and Stupp R (2011) Cilengitide: an RGDpentapeptide avb3 and avb5 integrin inhibitor in development for glioblastoma and othermalignancies. Future Oncol 7:339–354.

Schadt EE, Molony C, Chudin E, Hao K, Yang X, Lum PY, Kasarskis A, Zhang B, Wang S, and SuverC, et al. (2008) Mapping the genetic architecture of gene expression in human liver. PLoS Biol 6:e107.

Schenkman JB, Remmer H, and Estabrook RW (1967) Spectral studies of drug interaction withhepatic microsomal cytochrome. Mol Pharmacol 3:113–123.

Sevrioukova IF and Poulos TL (2013) Understanding the mechanism of cytochrome P450 3A4:recent advances and remaining problems. Dalton Trans 42:3116–3126.

Silverman RB (1995) Mechanism-based enzyme inactivators. Methods Enzymol 249:240–283.Soffietti R, Trevisan E, and Rudà R (2014) What have we learned from trials on antiangiogenicagents in glioblastoma? Expert Rev Neurother 14:1–3.

Sohl CD, Cheng Q, and Guengerich FP (2009) Chromatographic assays of drug oxidation byhuman cytochrome P450 3A4. Nat Protoc 4:1252–1257.

Somogyi A, Stockley C, Keal J, Rolan P, and Bochner F (1987) Reduction of metformin renaltubular secretion by cimetidine in man. Br J Clin Pharmacol 23:545–551.

Szedlacsek SE and Duggleby RG (1995) Kinetics of slow and tight-binding inhibitors. MethodsEnzymol 249:144–180.

Wienkers LC and Heath TG (2005) Predicting in vivo drug interactions from in vitro drugdiscovery data. Nat Rev Drug Discov 4:825–833.

Williams JA, Hyland R, Jones BC, Smith DA, Hurst S, Goosen TC, Peterkin V, Koup JR,and Ball SE (2004) Drug-drug interactions for UDP-glucuronosyltransferase substrates:a pharmacokinetic explanation for typically observed low exposure (AUCi/AUC) ratios. DrugMetab Dispos 32:1201–1208.

Address correspondence to: F. Peter Guengerich, Department of Biochemistry,Vanderbilt University School of Medicine, 638 Robinson Research Building, 2200Pierce Avenue, Nashville, TN 37232. E-mail: f.guengerich@vanderbilt.edu

1446 Boji�c et al.

at ASPE

T Journals on June 7, 2020

dmd.aspetjournals.org

Dow

nloaded from