Research ArticleCarborane-Based Carbonic Anhydrase Inhibitors Insight intoCAIICAIX Specificity from a High-Resolution Crystal StructureModeling and Quantum Chemical Calculations

Pavel Mader12 Adam Pecina3 Petr Ciacutegler3 Martin Lepšiacutek3 Vaacuteclav Šiacutecha4 Pavel Hobza35

Bohumiacuter Gruumlner4 Jindlich Fanfrliacutek3 Jiliacute Brynda13 and Pavliacutena kezaacuteIovaacute13

1 Institute of Molecular Genetics Academy of Sciences of the Czech Republic Vıdenska 1083 140 00 Prague 4 Czech Republic2 Structural Genomics Consortium University of Toronto Toronto ON Canada M5G 1L73 Institute of Organic Chemistry and Biochemistry Academy of Sciences of the Czech Republic Gilead Sciences andIOCB Research Center Flemingovo nam 2 166 10 Prague 6 Czech Republic

4 Institute of Inorganic Chemistry Academy of Sciences of the Czech Republic vvi Hlavnı 1001250 68 Rez near Prague Czech Republic

5 Regional Center of Advanced Technologies and Materials Department of Physical Chemistry Palacky University77146 Olomouc Czech Republic

Correspondence should be addressed to Jindrich Fanfrlık fanfrlikuochbcascz Jirı Brynda bryndaimgcasczand Pavlına Rezacova rezacovauochbcascz

Received 24 April 2014 Accepted 8 June 2014 Published 18 September 2014

Academic Editor Mariya Al-Rashida

Copyright copy 2014 Pavel Mader et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Carborane-based compounds are promising lead structures for development of inhibitors of carbonic anhydrases (CAs) Herewe report structural and computational analysis applicable to structure-based design of carborane compounds with selectivitytoward the cancer-specific CAIX isoenzyme We determined the crystal structure of CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane at 10 A resolution and used this structure to model the 1-methylenesulfamide-12-dicarba-closo-dodecaborane interactions with CAIX A virtual glycine scan revealed the contributions of individual residues to the energy ofbinding of 1-methylenesulfamide-12-dicarba-closo-dodecaborane to CAII and CAIX respectively

1 Introduction

Carbonic anhydrases (CAs) play important roles in manyphysiological and pathophysiological processes For exampleextracellular CAs participate in tumor growth and progres-sion [1] CAIX which is selectively expressed in a range ofhypoxic tumors is a validated diagnostic and therapeutictarget (recently reviewed in [2ndash4]) There are 15 human CAisoenzymes and due to the ubiquity of these enzymes inhuman tissues selective inhibition is a very important aspectof drug design

Three main classes of CA inhibitors have been describedto date (reviewed in [5]) (i) metal ion binders (sulfonamidessulfamides sulfamates dithiocarbamates thiols and hydro-xamates) (ii) compounds that anchor the zinc-coordinated

water moleculehydroxide ion (phenols carboxylates poly-amines esters and sulfocoumarins) and (iii) coumarins andrelated compounds that bind further away from themetal ion

CA inhibitors from the first class (metal ion binders) con-tain specific functional groups that interact with the catalyticZn2+ ion in the CA active siteThese metal-binding function-alities are typically joined to a ldquoringrdquo structureThis moiety isnot necessarily aromatic however it is usually consisting of a5- or 6-membered hydrocarbon ring or conjugated ring sys-tem containing nitrogen oxygen andor sulfur Numerousfunctional groups have been added to the ring structure scaf-fold to modify inhibitor properties such as specificity towarda particular CA isoenzyme pKa or solubility (reviewed in[6]) Recently we reported design ofCA inhibitors containing

2 BioMed Research International

space-filling carborane clusters in place of the typical ringstructure [7] We showed that various carborane clustersact as CA inhibitors and that modifying these clusters withan appropriately attached sulfamide group and other sub-stituents leads to compounds with selectivity toward thecancer-specific CAIX isoenzyme

Boron is one of few chemical elements that can formbinary hydrides composed of more than two atoms whichleads to formation of boron cluster compounds (boronhydrides or boranes) Their basic structural feature is forma-tion of a polyhedronwith triangular facets held together by 3-center 2-electron bondswith an extensive electron delocaliza-tion [8] A typical structural archetype is represented by thedivalent closo-B

12H12

2minus anion an extremely stable compoundwith a symmetrical 12-vertex icosahedron structure [9]Replacement of one or more BHminus in borane cage with CHleads to series of carboranes and removal of BH vertices leadsto various open-cage (nido-) species Carboranes thus offer alarge variety of structural archetypes that provide interestingcounterparts to organic compounds [10]

Many features of icosahedral 12-vertex carboranes areuseful in the design of biologically active compounds Carbo-ranes have high thermal and chemical stability thereforethey generally do not undergo catabolism and are nontoxicto the host organism [11 12] The basic closo-C

2B10H12

carborane cluster is highly hydrophobic [13] however itscontrolled deboronation can generate water soluble 11-vertexnido-C

2B9H12

minus These anions represent important interme-diates in the synthesis of a family of mainly anionic metalbis(dicarbollides) accessible via metal insertion Incorpora-tion of carborane cages into the structures of certain sub-stances of medicinal interest can enhance hydrophobic inter-actions between the boron cluster-coupled pharmaceuticalsand their protein targets increase in vivo stability and facili-tate uptake through cellular membranes [14 15]The success-ful use of boron clusters as hydrophobic pharmacophores hasrecently been increasing [16 17] Examples of carborane phar-macophores include boron-containing antifolates [18] HIVprotease inhibitors [19 20] and estrogen receptor agonistsand antagonists [21] among others [16 22 23]

Drug design efforts benefit greatly from knowledge of the3D structures of protein-ligand complexes X-ray crystallog-raphy has contributed considerably to the development ofCA inhibitors more than 500 structures of human CA isoen-zymes (wild-type andmutant forms) in complex with variousinhibitors have offered unprecedented insight into inhibitorbinding modes (reviewed in [24]) Structural informationcoupled with experimental inhibition data can be used tovalidate various computational approaches to assess inhibitorbinding strength Once a particular theoretical approachreproduces the knowndatawell it can be used for prospectivedesign For studies involving metal ions and unusual com-pounds such as boranes the use of quantum chemistry (QM)is warranted [25 26] Indeed we recently used a quantummechanicsmolecular mechanics (QMMM) methodologyto quantitatively describe the binding of two carborane-based sulfamides to CAII [7] and to explain fundamentaldifferences in the binding modes of closo- and nido-cages[27]

Here we report the X-ray structure of CAII with bound 1-methylenesulfamide-12-dicarba-closo-dodecaborane (com-pound 1 Figure 1(a)) determined at 10 A resolution Thisatomic-level resolution allowed us to assess in detail thepositions of carbon and boron atoms in the carborane cageof 1 Additionally we modeled the complex of 1 with CAIXWe employed a virtual glycine scan to analyze the differencesbetween the interactions of 1 with CAII and CAIX

2 Materials and Methods

21 Protein Crystallization and Diffraction Data CollectionFor crystallization of human CAII (Sigma catalogue numberC6165) in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (compound 1) we adapted a previ-ously described procedure [28] CAII (at a concentration of4mgsdotmLminus1 dissolved in water) was incubated in aqueoussolution containing a 2-fold molar excess of p-hydroxymer-curibenzoate (Sigma catalogue number 55540) The proteinwas concentrated to 10mgsdotmLminus1 and unbound p-hydroxy-mercuribenzoate was removed with AmiconUltra-4 concen-trators (Merck-Millipore MWCO 10 kDa)

The complex of CAII with 1 was prepared by adding a 11-fold molar excess of 1 (in DMSO) to the 10mgsdotmLminus1 solutionof CAII in water without pH adjustment (the final DMSOconcentration did not exceed 5 vv)

The best diffracting crystals were obtained using thehanging-drop vapor diffusion method under the followingconditions 2120583L protein-inhibitor complex solution wasmixed with 2 120583L precipitant solution [25M (NH

4)2SO4

03M NaCl and 100mM Tris-HCl pH 82] and equilibratedover a reservoir containing 1mL of precipitant solution at18∘C Crystals with dimensions of 03mm times 01mm times 01mmgrew within 7 days

For cryoprotection the crystals were incubated inmotherliquor supplemented with 25 glycerol for approximately30 s flash-frozen and stored in liquid nitrogen Diffractiondata for the CAII complex were collected at 100K at theX142 BESSY beamline in Berlin Germany [29] Data werecollected in two passes the high-resolution range (1175ndash100 A) and the low-resolution range (2108ndash120 A) The twodatasets were integrated with iMOSFLM [30] and mergedand scaled with SCALA [31] Data collection and refinementstatistics are summarized in Table 1

22 Structure Determination Refinement and AnalysisCrystal structures were solved by difference Fourier methodusing the CAII structure (PDB code 3IGP [34]) as a startingmodel The model was refined using REFMAC5 [35] part ofthe CCP4 program suite [36]Themodel was initially refinedwith isotropic atomic displacement parameters (ADPs)hydrogen atoms in riding positions were added later Forthe final rounds of refinement we used a mixed isotropic-anisotropic model of ADPs anisotropic ADPs were used forall atoms and only atoms in alternative conformations wererefined isotropically Atomic coordinates for the structureof 1 were generated by quantum mechanics computationwith DFT-D methodology [37] using the B-LYP functionaland SVP basis set [38] in the Turbomole program [39]

BioMed Research International 3

1

2

34

5 6

789 10 11

12

3

1

1

NH2

NH

OS

H

O

C

C

21

2

(a) (b)

Figure 1 (a) Structural formula of 1 with atom numbers used in the crystal structure coordinate file The vertices in carborane clusterrepresent BH groups (b) Crystal structure of CAII in complex with 1 The CAII active site is shown in cartoon representation residuesinvolved in interactions with the Zn2+ ion (purple sphere) and 1 are shown in stick representation with carbon atoms colored green Boronatoms are colored pink and other heteroatoms are colored according to standard color coding oxygen red nitrogen blue sulfur yellowThe2Fo-Fc electron density map for 1 is contoured at 1 120590

A geometric library for 1 was generated using the Libcheckprogram from the CCP4 suite Coot [40] was used forrebuilding The quality of the refined model was assessedusing MolProbity [33] The coordinates and structure factorswere deposited in the PDB under accession code 4Q78 Finalrefinement statistics are summarized in Table 1 All structuralfigures were prepared using PyMOL 141 [41]

23 Model of CAIX-1 Complex The complex of CAIX and 1was modeled by aligning the existing crystal structures of theCAIX catalytic domain (PDB code 3IAI [42]) with the CAII-1 complex (PDB code 4MDG [7]) using PyMOL version12 [43] Preparation of structure coordinate files for furthercalculations was performed as described before for CAII [27]

The complex was fully optimized using a QMMM pro-cedure We used ONIOM-like subtractive scheme [44] withlink atoms and mechanical embedding to be consistent withour previous studies [27 45ndash48] The QM part is describedat the DFT-D TPSSTZVPBLYPSVP level of theory [39]and comprises 218 atoms including the atoms present in1 and 8 amino acids (Trp5 Asn62 His64 Gln67 Gln92Val131 Leu135 and Pro202) The MM part constituted theremainder of the protein and the surrounding solvent wasapproximated by a generalized Born (GB) implicit modelDetailed description of the procedure was published in [27]One crystal water molecule (Wat272) bridging the inhibitorsand CAII residues Thr199 Glu106 and Tyr7 was retainedto maintain the integrity of the active site Other watermolecules present in the crystal structures were omitted

The positions of the added hydrogen atoms 1 and 15amino acids surrounding the ligand (Trp5 Asn62 Gly63His64 Gln67 Leu91 Gln92 Leu123 Val131 Leu135 Leu141Thr200 Pro201 Pro202 and Ala204) were relaxed in a GBimplicit solvent model using the FIRE algorithm followed

by 10 ps annealing from 100K or 150K to 0K using theBerendsen thermostat [49] in the SANDER module of theAMBER 10 package [50]

24 Virtual Glycine Scan The contribution of the active siteamino acids to inhibitor binding was examined by virtualglycine scanning Individual amino acids in contact with1 in the CAIX-1 model and CAII-1 crystal structure weresubstituted with glycine The energy contributions (ΔΔG1015840int)were calculated as the difference between the original ΔG1015840intat the QMMM level with the wild-type amino acid and thenew ΔG1015840int with the mutated glycine residue [27]

3 Results and Discussion

31 Crystal Structure of CAII in Complex with 1 at Atomic Res-olution The overall structure of CAII in complex with 1 wasrefined to 10 A resolutionThis high resolution allowed us toobserve details that could not be fully resolved in the complexstructure determined previously at lower resolution Atomicresolution was achieved by derivatization of CAII using the4-(hydroxymercury)benzoic acid (abbreviated MBO in thecif library of small molecules) method described by [28]Themercury atom ofMBO covalently binds to S120574 of Cys206Thismodification allows formation of a hydrogen bond betweenthe OZ1 oxygen of the MBO carboxyl group and the main-chain amino group of Tyr40 in the neighboring proteinmolecule reinforcing the crystal lattice and increasing thediffraction quality of the crystal In our structure MBO ismodeled in two alternative conformations with occupanciesof 06 and 02

When our atomic resolution structure is compared withthe structure of the CAII-1 complex determined at 135 Aresolution (PDB code 4MDG [7]) the RMSD value for

4 BioMed Research International

Table 1 Data collection and refinement statisticsData collection statistics

Space group 11987521

Cell parameters (A ∘) 4220 4173 7216900 1044 900

Wavelength (A) 09184Resolution (A) 2108ndash100 (105ndash100)Number of unique reflections 108781 (15490)Multiplicity 35 (25)Completeness () 831 (814)119877merge

a 0056 (0375)Average 119868120590(119868) 108 (23)Wilson B (A2) 65

Refinement statisticsResolution range (A) 6990ndash100 (103ndash100)No of reflections in working set 97856 (7831)No of reflections in test set 5426 (412)119877 value ()b 175 (244)119877free value ()c 200 (262)RMSD bond length (A) 0011RMSD angle (∘) 153Number of atoms in AU 2297Number of protein atoms in AU 2081Number of water molecules in AU 176Mean ADP value proteininhibitor (A2) 120176

Ramachandran plot statisticsd

Residues in favored regions () 9656Residues in allowed regions () 344

The data in parentheses refer to the highest-resolution shella119877merge = sumℎ119896119897 sum119894 119868119894(ℎ119896119897) minus ⟨119868(ℎ119896119897)⟩ | sumℎ119896119897 sum119894 119868119894(ℎ119896119897) where 119868119894(ℎ119896119897) is theindividual intensity of the 119894th observation of reflection ℎ119896119897 and ⟨119868(ℎ119896119897)⟩ isthe average intensity of reflection ℎ119896119897 with summation over all datab119877 value = ||119865119900| minus |119865119888|||119865119900| where 119865119900 and 119865119888 are the observed and calculatedstructure factors respectively119888119877free is equivalent to 119877 value but is calculated for 5 of reflections chosen atrandom and omitted from the refinement process [32]das determined by Molprobity [33]

superposition of the C120572 atoms of residues 4ndash261 is 0142 Aa value typical for superposition of identical structures [51]The N-terminal residue His3 is traced differently in the twostructures double conformations of numerous side chains(eg Glu14 His64 and Gln74) are resolved in the atomicresolution structure We found an additional difference inthe loop formed by amino acid residues 124ndash139 with amaximum difference of 0738 A for the position of Gln136C120572 Gln136 forms van der Waals contacts with the MBOcovalently attached to Cys206 The positions of Phe131 andVal135 which form a hydrophobic rim at the active site arealso influenced by MBO binding This results in a subtlepositional shift of the inhibitor with an RMSD of 0145 A forsuperposition of 12 atoms in the carborane cage of 1 bound toCAII and CAII derivatized by MBO This value is below thevalue observed for superposition of identical structures [51]

Atomic-level resolution allowed us to resolve the carbonand boron atom positions in the symmetrical carborane

Phe131

Gln92Val121

Leu198

His94

His96

His119Thr200

Thr199

Figure 2 Interactions of 1 with CAII The protein is shown incartoon representation residues involved in interactions with theZn2+ ion (gray sphere) and 1 are shown in stick representationPolar interactions are represented by blue dashed lines Zn2+ ioncoordination is shown as black dashed lines

cage of 1 When analyzing the values of the electron densitymap at positions of atoms bonded to the C1 atom we canassume that positions with higher density levels are morelikely to be carbon than boron atoms Similar analysis wasdone by others for boron-containing inhibitor of humandihydrofolate reductase [18] The C2 atom of the carboranecage (Figure 1(a)) was modeled into the position with anelectron density value of 116 eA3 which was approximately015 eA3 higher than those for the B3 B4 B5 and B6 atomsTo exclude the possibility that higher density is caused bymodel bias we altered the composition of the cage byreplacing the C2 atom with a boron atom Electron densityvalues did not change significantly after several rounds ofrefinement cycles

Thus we can conclude that the most probable positionof the second carbon atom in the carborane cage of 1 is theposition assigned to the C2 atom in our crystal structureThisis in good agreement with the recently published QMMMmodeling study [27]

32 Detailed Analysis of Inhibitor Interactions with CAIIThe crystal structure of human CAII in complex with 1determined at 10 A resolution confirmed the key interactionsthat our group observed previously [7] The compound fitsvery well into the CAII active site cavity andmakes numerouspolar and nonpolar interactions with the residues in theenzyme active site The sulfamide moiety which forms keypolar interactions with the active site Zn2+ ion also makespolar interactions with Thr199 typical of other sulfamideinhibitors of CAII (Figure 2) The linker NH group formsan additional polar interaction with O120574 of Thr200 Thecompound makes several van der Waals interactions withresidues Gln92 His94 His96 His119 Val121 Phe131 Leu198andThr200 (Figure 2) All interactions between the inhibitorand protein are summarized in Table 2

BioMed Research International 5

2 3

H3C CH3

H2NS

OO

OO

N

OO

SH2N

H3C

OO CH3H3C

N

(a)

Phe131Val121

Gln92

His94Asn67

Asn62

Thr200Pro201

Pro202

Leu198

(b)

Phe131

Gln92

His94

Thr200

Leu198

(c) (d)

Figure 3 (a) Structural formulas of 2 and 3 (b) Interactions of 2 and 3 with the CAII active site Compound 2 is represented with goldencarbon atoms while the carbon atoms of 3 are colored turquoise Surface of residues making contacts with the isoquinoline moiety of 2 and 3are highlighted in yellow and blue respectively Surface of residues colored orange make contacts with both compounds Atoms involvedin contacts with the sulfonamide groups are not highlighted (c) Interactions of 1 with the CAII active site Surface of residues makingcontacts with the carborane and linker moiety of 1 are highlighted in green Atoms involved in contacts with the sulfonamide groups arenot highlighted (d) Superposition of binding poses of 1 2 and 3 in the CAII active site Superposition of the complex structures was basedon the best fit for C120572 atoms of CAII residues 6ndash261

The idea of designing CA inhibitors containing a carbo-rane cluster moiety originated from our previous structuralstudies of isoquinoline-containing sulfonamide inhibitors(Figure 3(a)) Structural analysis of CAII in complex with 67-dimethoxy-1234-tetrahydroisoquinolin-2-ylsulfonamide (2PDB code 3IGP [34]) and 67-dimethoxy-1-methyl-1234-tetrahydroisoquinolin-2-ylsulfonamide (3 PDB code 3PO6[52]) revealed two distinct binding modes that engage twoopposite sides of the enzyme active site cavity (Figure 3(b))Following this analysis we hypothesized that the bindingspace within the enzyme active site cavity could be effectivelyfilled with a bulky hydrophobic molecule with a sphericalstructure This led to design of 1 which exhibited inhibitoryproperty to CAII and CAIX with Ki values in submicro-molar range Structural analysis of CAII-1 indicates that ourstructure-based design was sound We found that the carbo-rane cluster interacts with both sides of the enzyme active siteas predicted (Figure 3(c) Table 3) and that the position of 1

in the CAII active site superposes well with the two bindingmodes observed for 2 and 3 (Figure 3(d))

33 Model of the CAIX-1 Complex The CAII-1 crystal struc-ture was used tomodel binding of compound 1 into the CAIXactive site using QMMMmethods (Figure 4)

The substrate binding sites of CAII and CAIX differ byonly six amino acids Asn67 of CAII is replaced by Gln inCAIX Ile91 by Leu Trp123 by Leu Phe131 by Val Val135 byLeu andLeu204 byAlaThese variations result in a differentlyshaped active site cavity which accommodated 1 in a slightlydifferent pose (Figure 4) While the position of the sulfamideanchor remained unchanged the carborane cluster shifted by21 A (expressed as a difference in the position of B12) awayfrom the central 120573-sheet In CAIX-1 the carborane interactsmore with the opposite site of the active site specifically withamino acid residues His94 His96 Glu106 Leu198 Thr199Thr200 and Pro201 (Figure 4 Table 3) All polar and van der

6 BioMed Research International

Table 2 List of contacts between CAII and 1CAII 1

Residue Atom Atoma Distance [A]b

Zn ZN N2 187c

Zn ZN S 304Zn ZN O2 305

92 Gln OE1 B6 34792 Gln OE1 B11 35292 Gln CD B6 38494 His CE1 O2 29794 His NE2 N2 32394 His NE2 O2 33194 His CE1 C3 36794 His NE2 S 38194 His CE1 N2 38294 His CE1 S 38494 His NE2 C3 39496 His NE2 N2 31496 His CE1 N2 356119 His ND1 N2 339119 His ND1 O2 388119 His CE1 N2 396121 Val CG2 O2 382131 Phe CZ B8 383131 Phe CZ B7 397198 Leu CA O1 309198 Leu C O1 336198 Leu CB O1 360198 Leu CD2 O1 363198 Leu CD1 B3 386199 Thr N O1 270199 Thr OG1 N2 274199 Thr OG1 O1 358199 Thr OG1 S 378199 Thr N S 383199 Thr CA O1 383199 Thr CB N2 398200 Thr OG1 N1 302200 Thr OG1 C3 314200 Thr OG1 B4 336200 Thr OG1 B3 356200 Thr OG1 C1 366aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance between ligand and protein (or Zn) atoms lessthan or equal to 4 A are listedcPolar interactions are highlighted in bold

Waals interactions between CAIX and 1 are summarized inTable 4

We used a virtual glycine scan to study the roles of indi-vidual amino acid side chains in the active sites of CAIIand CAIX in binding of 1 The changes in free energy of

Trp5Glu106

His96His119

Val121

Trp209

Leu198Thr199

Thr200

Pro201

His94

Figure 4 Interactions of 1with the CAIX active site Atomsmakingcontacts with the carborane and linkermoiety of 1 are highlighted inmagenta Atoms involved in contacts with the sulfonamide groupsare not highlighted Superposition of the binding pose of 1 in CAIIis shown as black lines Superposition is based on the best fit for C120572atoms of all residues of CAII onto CAIX

Table 3 CAII or CAIX residues interacting with 1 2 and 3

CAII CAIX1a 2b 3c 1d

Trp5Asn62Asn67e

Gln92 Gln92 Gln92His94 His94 His94 His94His96 His96 His96 His96

Glu106His119 His119 His119 His119Val121 Val121 Val121Phe131 Phe131 Phe131

Val143 Val143Leu198 Leu198 Leu198 Leu198Thr199 Thr199 Thr199 Thr199Thr200 Thr200 Thr200 Thr200

Pro201 Pro201Pro202Trp209 Trp209

Interacting residues were identified from acrystal structure 4Q78 (this work)bcrystal structure 3IGP [34] ccrystal structure 3PO6 [52] dcomputationalmodel (this work) eresidues making polar interactions are highlighted inbold

interaction (ΔΔG1015840int) upon mutation of a given amino acidresidue to glycine are shown in Figure 5

The largest energy change (26 kcalmol) occurred forTrp5 which is positioned closer to 1 in CAIX-1 than in CAII-1 The side chain of Trp5 forms several dihydrogen bondswith the carborane cage of 1 The shortest one has a H sdot sdot sdotHdistance of 23 A The other major contributor to strong

BioMed Research International 7

Table 4 Interactions between CAIX and 1

CAIX 1Residue Atom Atoma Distance [A]b

Zn ZN N2 21c

Zn ZN S 33Zn ZN O2 35

5 Trp CZ2 B5 3745 Trp CZ2 B10 38194 His CE1 O2 31594 His CE1 C3 37494 His NE2 N2 33694 His NE2 S 38894 His NE2 O2 34594 His NE2 C3 37696 His CE1 N2 39996 His NE2 N2 349106 Glu OE2 N2 371119 His ND1 N2 337119 His CE1 N2 383121 Val CG2 O2 358198 Leu CA O1 304198 Leu CB O1 34198 Leu CD2 O1 343198 Leu C O1 338199 Thr N S 388199 Thr N O1 279199 Thr CA O1 396199 Thr CB N2 385199 Thr OG1 N2 263199 Thr OG1 S 369199 Thr OG1 O1 365200 Thr OG1 C1 377200 Thr OG1 B5 356200 Thr OG1 N1 313200 Thr OG1 C3 331200 Thr OG1 B4 364201 Pro O B4 36201 Pro O B10 349201 Pro O B8 396209 Trp CZ2 O1 374aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance less than or equal to 4 A between ligand andprotein (and Zn) atoms are listedcPolar interactions are highlighted in bold

CAIX-1 binding was Asn62 the energy of binding exceededthat in CAII-1 by nearly 1 kcalmol These contributions werecancelled out by differences in binding energy contributionsof amino acid residues 131 (PheVal) and 135 (ValLeu) whichwere lower in CAIX by 07 and 09 kcalmol respectivelyTheenergy changes of other residues were small

When we compared binding of 1 to CAII and CAIX wenoted that the favorable energy changes in CAIX-1 due tothe binding of residues Trp5 Asn62 and His64 were slightly

0

1

Trp5

Asn

62

His6

4

Asn

Gln

67

Gln

92

Phe

Val1

31

ValL

eu13

5

Pro2

02

CAIICAIX amino acid residues

CAIICAIX

minus1

minus2

minus3

ΔΔ

G998400 in

t(k

calm

ol)

Figure 5 Results of virtual glycine scan showing contributions ofindividual residues to the energy of binding of 1 to CAII and CAIXrespectively

larger than the unfavorable changes in binding caused bythe different amino acids at residues 131 and 135 This is inqualitative agreement with the experimental119870

119894values which

are 700plusmn141 nm for inhibition of CAII and 380plusmn111 nM forinhibition of CAIX [7]

4 Conclusions

We determined to atomic resolution the crystal structureof CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (1) a parent compound of a recentlyreported series of CA inhibitors containing carborane cages[7] Comparing this crystal structure with those of CAIIcomplexes with conventional organic inhibitors showed thatthe three-dimensional cluster fills the enzyme active sitecavity Atomic-level resolution allowed us to distinguishthe positions of carbon and boron atoms in the carboranecage The crystal structure also served as a model forconstruction of the CAIX-1 computational model Virtualglycine scan enabled us to quantify the contributions ofindividual residues to the energy of binding of 1 to CAIIand CAIX and uncover differences of the enzyme active sitecavities Structural and computational analysis will be used infuture structure-based design of carborane compounds withselectivity toward the cancer-specific CAIX isoenzyme

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

Pavel Mader and Adam Pecina contributed equally to thiswork

Acknowledgments

This work was supported by the Technology Agency ofthe Czech Republic (Project TE01020028) and in part byResearch Projects RVO 68378050 and 61388963 awarded by

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

2 BioMed Research International

space-filling carborane clusters in place of the typical ringstructure [7] We showed that various carborane clustersact as CA inhibitors and that modifying these clusters withan appropriately attached sulfamide group and other sub-stituents leads to compounds with selectivity toward thecancer-specific CAIX isoenzyme

Boron is one of few chemical elements that can formbinary hydrides composed of more than two atoms whichleads to formation of boron cluster compounds (boronhydrides or boranes) Their basic structural feature is forma-tion of a polyhedronwith triangular facets held together by 3-center 2-electron bondswith an extensive electron delocaliza-tion [8] A typical structural archetype is represented by thedivalent closo-B

12H12

2minus anion an extremely stable compoundwith a symmetrical 12-vertex icosahedron structure [9]Replacement of one or more BHminus in borane cage with CHleads to series of carboranes and removal of BH vertices leadsto various open-cage (nido-) species Carboranes thus offer alarge variety of structural archetypes that provide interestingcounterparts to organic compounds [10]

Many features of icosahedral 12-vertex carboranes areuseful in the design of biologically active compounds Carbo-ranes have high thermal and chemical stability thereforethey generally do not undergo catabolism and are nontoxicto the host organism [11 12] The basic closo-C

2B10H12

carborane cluster is highly hydrophobic [13] however itscontrolled deboronation can generate water soluble 11-vertexnido-C

2B9H12

minus These anions represent important interme-diates in the synthesis of a family of mainly anionic metalbis(dicarbollides) accessible via metal insertion Incorpora-tion of carborane cages into the structures of certain sub-stances of medicinal interest can enhance hydrophobic inter-actions between the boron cluster-coupled pharmaceuticalsand their protein targets increase in vivo stability and facili-tate uptake through cellular membranes [14 15]The success-ful use of boron clusters as hydrophobic pharmacophores hasrecently been increasing [16 17] Examples of carborane phar-macophores include boron-containing antifolates [18] HIVprotease inhibitors [19 20] and estrogen receptor agonistsand antagonists [21] among others [16 22 23]

Drug design efforts benefit greatly from knowledge of the3D structures of protein-ligand complexes X-ray crystallog-raphy has contributed considerably to the development ofCA inhibitors more than 500 structures of human CA isoen-zymes (wild-type andmutant forms) in complex with variousinhibitors have offered unprecedented insight into inhibitorbinding modes (reviewed in [24]) Structural informationcoupled with experimental inhibition data can be used tovalidate various computational approaches to assess inhibitorbinding strength Once a particular theoretical approachreproduces the knowndatawell it can be used for prospectivedesign For studies involving metal ions and unusual com-pounds such as boranes the use of quantum chemistry (QM)is warranted [25 26] Indeed we recently used a quantummechanicsmolecular mechanics (QMMM) methodologyto quantitatively describe the binding of two carborane-based sulfamides to CAII [7] and to explain fundamentaldifferences in the binding modes of closo- and nido-cages[27]

Here we report the X-ray structure of CAII with bound 1-methylenesulfamide-12-dicarba-closo-dodecaborane (com-pound 1 Figure 1(a)) determined at 10 A resolution Thisatomic-level resolution allowed us to assess in detail thepositions of carbon and boron atoms in the carborane cageof 1 Additionally we modeled the complex of 1 with CAIXWe employed a virtual glycine scan to analyze the differencesbetween the interactions of 1 with CAII and CAIX

2 Materials and Methods

21 Protein Crystallization and Diffraction Data CollectionFor crystallization of human CAII (Sigma catalogue numberC6165) in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (compound 1) we adapted a previ-ously described procedure [28] CAII (at a concentration of4mgsdotmLminus1 dissolved in water) was incubated in aqueoussolution containing a 2-fold molar excess of p-hydroxymer-curibenzoate (Sigma catalogue number 55540) The proteinwas concentrated to 10mgsdotmLminus1 and unbound p-hydroxy-mercuribenzoate was removed with AmiconUltra-4 concen-trators (Merck-Millipore MWCO 10 kDa)

The complex of CAII with 1 was prepared by adding a 11-fold molar excess of 1 (in DMSO) to the 10mgsdotmLminus1 solutionof CAII in water without pH adjustment (the final DMSOconcentration did not exceed 5 vv)

The best diffracting crystals were obtained using thehanging-drop vapor diffusion method under the followingconditions 2120583L protein-inhibitor complex solution wasmixed with 2 120583L precipitant solution [25M (NH

4)2SO4

03M NaCl and 100mM Tris-HCl pH 82] and equilibratedover a reservoir containing 1mL of precipitant solution at18∘C Crystals with dimensions of 03mm times 01mm times 01mmgrew within 7 days

For cryoprotection the crystals were incubated inmotherliquor supplemented with 25 glycerol for approximately30 s flash-frozen and stored in liquid nitrogen Diffractiondata for the CAII complex were collected at 100K at theX142 BESSY beamline in Berlin Germany [29] Data werecollected in two passes the high-resolution range (1175ndash100 A) and the low-resolution range (2108ndash120 A) The twodatasets were integrated with iMOSFLM [30] and mergedand scaled with SCALA [31] Data collection and refinementstatistics are summarized in Table 1

22 Structure Determination Refinement and AnalysisCrystal structures were solved by difference Fourier methodusing the CAII structure (PDB code 3IGP [34]) as a startingmodel The model was refined using REFMAC5 [35] part ofthe CCP4 program suite [36]Themodel was initially refinedwith isotropic atomic displacement parameters (ADPs)hydrogen atoms in riding positions were added later Forthe final rounds of refinement we used a mixed isotropic-anisotropic model of ADPs anisotropic ADPs were used forall atoms and only atoms in alternative conformations wererefined isotropically Atomic coordinates for the structureof 1 were generated by quantum mechanics computationwith DFT-D methodology [37] using the B-LYP functionaland SVP basis set [38] in the Turbomole program [39]

BioMed Research International 3

1

2

34

5 6

789 10 11

12

3

1

1

NH2

NH

OS

H

O

C

C

21

2

(a) (b)

Figure 1 (a) Structural formula of 1 with atom numbers used in the crystal structure coordinate file The vertices in carborane clusterrepresent BH groups (b) Crystal structure of CAII in complex with 1 The CAII active site is shown in cartoon representation residuesinvolved in interactions with the Zn2+ ion (purple sphere) and 1 are shown in stick representation with carbon atoms colored green Boronatoms are colored pink and other heteroatoms are colored according to standard color coding oxygen red nitrogen blue sulfur yellowThe2Fo-Fc electron density map for 1 is contoured at 1 120590

A geometric library for 1 was generated using the Libcheckprogram from the CCP4 suite Coot [40] was used forrebuilding The quality of the refined model was assessedusing MolProbity [33] The coordinates and structure factorswere deposited in the PDB under accession code 4Q78 Finalrefinement statistics are summarized in Table 1 All structuralfigures were prepared using PyMOL 141 [41]

23 Model of CAIX-1 Complex The complex of CAIX and 1was modeled by aligning the existing crystal structures of theCAIX catalytic domain (PDB code 3IAI [42]) with the CAII-1 complex (PDB code 4MDG [7]) using PyMOL version12 [43] Preparation of structure coordinate files for furthercalculations was performed as described before for CAII [27]

The complex was fully optimized using a QMMM pro-cedure We used ONIOM-like subtractive scheme [44] withlink atoms and mechanical embedding to be consistent withour previous studies [27 45ndash48] The QM part is describedat the DFT-D TPSSTZVPBLYPSVP level of theory [39]and comprises 218 atoms including the atoms present in1 and 8 amino acids (Trp5 Asn62 His64 Gln67 Gln92Val131 Leu135 and Pro202) The MM part constituted theremainder of the protein and the surrounding solvent wasapproximated by a generalized Born (GB) implicit modelDetailed description of the procedure was published in [27]One crystal water molecule (Wat272) bridging the inhibitorsand CAII residues Thr199 Glu106 and Tyr7 was retainedto maintain the integrity of the active site Other watermolecules present in the crystal structures were omitted

The positions of the added hydrogen atoms 1 and 15amino acids surrounding the ligand (Trp5 Asn62 Gly63His64 Gln67 Leu91 Gln92 Leu123 Val131 Leu135 Leu141Thr200 Pro201 Pro202 and Ala204) were relaxed in a GBimplicit solvent model using the FIRE algorithm followed

by 10 ps annealing from 100K or 150K to 0K using theBerendsen thermostat [49] in the SANDER module of theAMBER 10 package [50]

24 Virtual Glycine Scan The contribution of the active siteamino acids to inhibitor binding was examined by virtualglycine scanning Individual amino acids in contact with1 in the CAIX-1 model and CAII-1 crystal structure weresubstituted with glycine The energy contributions (ΔΔG1015840int)were calculated as the difference between the original ΔG1015840intat the QMMM level with the wild-type amino acid and thenew ΔG1015840int with the mutated glycine residue [27]

3 Results and Discussion

31 Crystal Structure of CAII in Complex with 1 at Atomic Res-olution The overall structure of CAII in complex with 1 wasrefined to 10 A resolutionThis high resolution allowed us toobserve details that could not be fully resolved in the complexstructure determined previously at lower resolution Atomicresolution was achieved by derivatization of CAII using the4-(hydroxymercury)benzoic acid (abbreviated MBO in thecif library of small molecules) method described by [28]Themercury atom ofMBO covalently binds to S120574 of Cys206Thismodification allows formation of a hydrogen bond betweenthe OZ1 oxygen of the MBO carboxyl group and the main-chain amino group of Tyr40 in the neighboring proteinmolecule reinforcing the crystal lattice and increasing thediffraction quality of the crystal In our structure MBO ismodeled in two alternative conformations with occupanciesof 06 and 02

When our atomic resolution structure is compared withthe structure of the CAII-1 complex determined at 135 Aresolution (PDB code 4MDG [7]) the RMSD value for

4 BioMed Research International

Table 1 Data collection and refinement statisticsData collection statistics

Space group 11987521

Cell parameters (A ∘) 4220 4173 7216900 1044 900

Wavelength (A) 09184Resolution (A) 2108ndash100 (105ndash100)Number of unique reflections 108781 (15490)Multiplicity 35 (25)Completeness () 831 (814)119877merge

a 0056 (0375)Average 119868120590(119868) 108 (23)Wilson B (A2) 65

Refinement statisticsResolution range (A) 6990ndash100 (103ndash100)No of reflections in working set 97856 (7831)No of reflections in test set 5426 (412)119877 value ()b 175 (244)119877free value ()c 200 (262)RMSD bond length (A) 0011RMSD angle (∘) 153Number of atoms in AU 2297Number of protein atoms in AU 2081Number of water molecules in AU 176Mean ADP value proteininhibitor (A2) 120176

Ramachandran plot statisticsd

Residues in favored regions () 9656Residues in allowed regions () 344

The data in parentheses refer to the highest-resolution shella119877merge = sumℎ119896119897 sum119894 119868119894(ℎ119896119897) minus ⟨119868(ℎ119896119897)⟩ | sumℎ119896119897 sum119894 119868119894(ℎ119896119897) where 119868119894(ℎ119896119897) is theindividual intensity of the 119894th observation of reflection ℎ119896119897 and ⟨119868(ℎ119896119897)⟩ isthe average intensity of reflection ℎ119896119897 with summation over all datab119877 value = ||119865119900| minus |119865119888|||119865119900| where 119865119900 and 119865119888 are the observed and calculatedstructure factors respectively119888119877free is equivalent to 119877 value but is calculated for 5 of reflections chosen atrandom and omitted from the refinement process [32]das determined by Molprobity [33]

superposition of the C120572 atoms of residues 4ndash261 is 0142 Aa value typical for superposition of identical structures [51]The N-terminal residue His3 is traced differently in the twostructures double conformations of numerous side chains(eg Glu14 His64 and Gln74) are resolved in the atomicresolution structure We found an additional difference inthe loop formed by amino acid residues 124ndash139 with amaximum difference of 0738 A for the position of Gln136C120572 Gln136 forms van der Waals contacts with the MBOcovalently attached to Cys206 The positions of Phe131 andVal135 which form a hydrophobic rim at the active site arealso influenced by MBO binding This results in a subtlepositional shift of the inhibitor with an RMSD of 0145 A forsuperposition of 12 atoms in the carborane cage of 1 bound toCAII and CAII derivatized by MBO This value is below thevalue observed for superposition of identical structures [51]

Atomic-level resolution allowed us to resolve the carbonand boron atom positions in the symmetrical carborane

Phe131

Gln92Val121

Leu198

His94

His96

His119Thr200

Thr199

Figure 2 Interactions of 1 with CAII The protein is shown incartoon representation residues involved in interactions with theZn2+ ion (gray sphere) and 1 are shown in stick representationPolar interactions are represented by blue dashed lines Zn2+ ioncoordination is shown as black dashed lines

cage of 1 When analyzing the values of the electron densitymap at positions of atoms bonded to the C1 atom we canassume that positions with higher density levels are morelikely to be carbon than boron atoms Similar analysis wasdone by others for boron-containing inhibitor of humandihydrofolate reductase [18] The C2 atom of the carboranecage (Figure 1(a)) was modeled into the position with anelectron density value of 116 eA3 which was approximately015 eA3 higher than those for the B3 B4 B5 and B6 atomsTo exclude the possibility that higher density is caused bymodel bias we altered the composition of the cage byreplacing the C2 atom with a boron atom Electron densityvalues did not change significantly after several rounds ofrefinement cycles

Thus we can conclude that the most probable positionof the second carbon atom in the carborane cage of 1 is theposition assigned to the C2 atom in our crystal structureThisis in good agreement with the recently published QMMMmodeling study [27]

32 Detailed Analysis of Inhibitor Interactions with CAIIThe crystal structure of human CAII in complex with 1determined at 10 A resolution confirmed the key interactionsthat our group observed previously [7] The compound fitsvery well into the CAII active site cavity andmakes numerouspolar and nonpolar interactions with the residues in theenzyme active site The sulfamide moiety which forms keypolar interactions with the active site Zn2+ ion also makespolar interactions with Thr199 typical of other sulfamideinhibitors of CAII (Figure 2) The linker NH group formsan additional polar interaction with O120574 of Thr200 Thecompound makes several van der Waals interactions withresidues Gln92 His94 His96 His119 Val121 Phe131 Leu198andThr200 (Figure 2) All interactions between the inhibitorand protein are summarized in Table 2

BioMed Research International 5

2 3

H3C CH3

H2NS

OO

OO

N

OO

SH2N

H3C

OO CH3H3C

N

(a)

Phe131Val121

Gln92

His94Asn67

Asn62

Thr200Pro201

Pro202

Leu198

(b)

Phe131

Gln92

His94

Thr200

Leu198

(c) (d)

Figure 3 (a) Structural formulas of 2 and 3 (b) Interactions of 2 and 3 with the CAII active site Compound 2 is represented with goldencarbon atoms while the carbon atoms of 3 are colored turquoise Surface of residues making contacts with the isoquinoline moiety of 2 and 3are highlighted in yellow and blue respectively Surface of residues colored orange make contacts with both compounds Atoms involvedin contacts with the sulfonamide groups are not highlighted (c) Interactions of 1 with the CAII active site Surface of residues makingcontacts with the carborane and linker moiety of 1 are highlighted in green Atoms involved in contacts with the sulfonamide groups arenot highlighted (d) Superposition of binding poses of 1 2 and 3 in the CAII active site Superposition of the complex structures was basedon the best fit for C120572 atoms of CAII residues 6ndash261

The idea of designing CA inhibitors containing a carbo-rane cluster moiety originated from our previous structuralstudies of isoquinoline-containing sulfonamide inhibitors(Figure 3(a)) Structural analysis of CAII in complex with 67-dimethoxy-1234-tetrahydroisoquinolin-2-ylsulfonamide (2PDB code 3IGP [34]) and 67-dimethoxy-1-methyl-1234-tetrahydroisoquinolin-2-ylsulfonamide (3 PDB code 3PO6[52]) revealed two distinct binding modes that engage twoopposite sides of the enzyme active site cavity (Figure 3(b))Following this analysis we hypothesized that the bindingspace within the enzyme active site cavity could be effectivelyfilled with a bulky hydrophobic molecule with a sphericalstructure This led to design of 1 which exhibited inhibitoryproperty to CAII and CAIX with Ki values in submicro-molar range Structural analysis of CAII-1 indicates that ourstructure-based design was sound We found that the carbo-rane cluster interacts with both sides of the enzyme active siteas predicted (Figure 3(c) Table 3) and that the position of 1

in the CAII active site superposes well with the two bindingmodes observed for 2 and 3 (Figure 3(d))

33 Model of the CAIX-1 Complex The CAII-1 crystal struc-ture was used tomodel binding of compound 1 into the CAIXactive site using QMMMmethods (Figure 4)

The substrate binding sites of CAII and CAIX differ byonly six amino acids Asn67 of CAII is replaced by Gln inCAIX Ile91 by Leu Trp123 by Leu Phe131 by Val Val135 byLeu andLeu204 byAlaThese variations result in a differentlyshaped active site cavity which accommodated 1 in a slightlydifferent pose (Figure 4) While the position of the sulfamideanchor remained unchanged the carborane cluster shifted by21 A (expressed as a difference in the position of B12) awayfrom the central 120573-sheet In CAIX-1 the carborane interactsmore with the opposite site of the active site specifically withamino acid residues His94 His96 Glu106 Leu198 Thr199Thr200 and Pro201 (Figure 4 Table 3) All polar and van der

6 BioMed Research International

Table 2 List of contacts between CAII and 1CAII 1

Residue Atom Atoma Distance [A]b

Zn ZN N2 187c

Zn ZN S 304Zn ZN O2 305

92 Gln OE1 B6 34792 Gln OE1 B11 35292 Gln CD B6 38494 His CE1 O2 29794 His NE2 N2 32394 His NE2 O2 33194 His CE1 C3 36794 His NE2 S 38194 His CE1 N2 38294 His CE1 S 38494 His NE2 C3 39496 His NE2 N2 31496 His CE1 N2 356119 His ND1 N2 339119 His ND1 O2 388119 His CE1 N2 396121 Val CG2 O2 382131 Phe CZ B8 383131 Phe CZ B7 397198 Leu CA O1 309198 Leu C O1 336198 Leu CB O1 360198 Leu CD2 O1 363198 Leu CD1 B3 386199 Thr N O1 270199 Thr OG1 N2 274199 Thr OG1 O1 358199 Thr OG1 S 378199 Thr N S 383199 Thr CA O1 383199 Thr CB N2 398200 Thr OG1 N1 302200 Thr OG1 C3 314200 Thr OG1 B4 336200 Thr OG1 B3 356200 Thr OG1 C1 366aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance between ligand and protein (or Zn) atoms lessthan or equal to 4 A are listedcPolar interactions are highlighted in bold

Waals interactions between CAIX and 1 are summarized inTable 4

We used a virtual glycine scan to study the roles of indi-vidual amino acid side chains in the active sites of CAIIand CAIX in binding of 1 The changes in free energy of

Trp5Glu106

His96His119

Val121

Trp209

Leu198Thr199

Thr200

Pro201

His94

Figure 4 Interactions of 1with the CAIX active site Atomsmakingcontacts with the carborane and linkermoiety of 1 are highlighted inmagenta Atoms involved in contacts with the sulfonamide groupsare not highlighted Superposition of the binding pose of 1 in CAIIis shown as black lines Superposition is based on the best fit for C120572atoms of all residues of CAII onto CAIX

Table 3 CAII or CAIX residues interacting with 1 2 and 3

CAII CAIX1a 2b 3c 1d

Trp5Asn62Asn67e

Gln92 Gln92 Gln92His94 His94 His94 His94His96 His96 His96 His96

Glu106His119 His119 His119 His119Val121 Val121 Val121Phe131 Phe131 Phe131

Val143 Val143Leu198 Leu198 Leu198 Leu198Thr199 Thr199 Thr199 Thr199Thr200 Thr200 Thr200 Thr200

Pro201 Pro201Pro202Trp209 Trp209

Interacting residues were identified from acrystal structure 4Q78 (this work)bcrystal structure 3IGP [34] ccrystal structure 3PO6 [52] dcomputationalmodel (this work) eresidues making polar interactions are highlighted inbold

interaction (ΔΔG1015840int) upon mutation of a given amino acidresidue to glycine are shown in Figure 5

The largest energy change (26 kcalmol) occurred forTrp5 which is positioned closer to 1 in CAIX-1 than in CAII-1 The side chain of Trp5 forms several dihydrogen bondswith the carborane cage of 1 The shortest one has a H sdot sdot sdotHdistance of 23 A The other major contributor to strong

BioMed Research International 7

Table 4 Interactions between CAIX and 1

CAIX 1Residue Atom Atoma Distance [A]b

Zn ZN N2 21c

Zn ZN S 33Zn ZN O2 35

5 Trp CZ2 B5 3745 Trp CZ2 B10 38194 His CE1 O2 31594 His CE1 C3 37494 His NE2 N2 33694 His NE2 S 38894 His NE2 O2 34594 His NE2 C3 37696 His CE1 N2 39996 His NE2 N2 349106 Glu OE2 N2 371119 His ND1 N2 337119 His CE1 N2 383121 Val CG2 O2 358198 Leu CA O1 304198 Leu CB O1 34198 Leu CD2 O1 343198 Leu C O1 338199 Thr N S 388199 Thr N O1 279199 Thr CA O1 396199 Thr CB N2 385199 Thr OG1 N2 263199 Thr OG1 S 369199 Thr OG1 O1 365200 Thr OG1 C1 377200 Thr OG1 B5 356200 Thr OG1 N1 313200 Thr OG1 C3 331200 Thr OG1 B4 364201 Pro O B4 36201 Pro O B10 349201 Pro O B8 396209 Trp CZ2 O1 374aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance less than or equal to 4 A between ligand andprotein (and Zn) atoms are listedcPolar interactions are highlighted in bold

CAIX-1 binding was Asn62 the energy of binding exceededthat in CAII-1 by nearly 1 kcalmol These contributions werecancelled out by differences in binding energy contributionsof amino acid residues 131 (PheVal) and 135 (ValLeu) whichwere lower in CAIX by 07 and 09 kcalmol respectivelyTheenergy changes of other residues were small

When we compared binding of 1 to CAII and CAIX wenoted that the favorable energy changes in CAIX-1 due tothe binding of residues Trp5 Asn62 and His64 were slightly

0

1

Trp5

Asn

62

His6

4

Asn

Gln

67

Gln

92

Phe

Val1

31

ValL

eu13

5

Pro2

02

CAIICAIX amino acid residues

CAIICAIX

minus1

minus2

minus3

ΔΔ

G998400 in

t(k

calm

ol)

Figure 5 Results of virtual glycine scan showing contributions ofindividual residues to the energy of binding of 1 to CAII and CAIXrespectively

larger than the unfavorable changes in binding caused bythe different amino acids at residues 131 and 135 This is inqualitative agreement with the experimental119870

119894values which

are 700plusmn141 nm for inhibition of CAII and 380plusmn111 nM forinhibition of CAIX [7]

4 Conclusions

We determined to atomic resolution the crystal structureof CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (1) a parent compound of a recentlyreported series of CA inhibitors containing carborane cages[7] Comparing this crystal structure with those of CAIIcomplexes with conventional organic inhibitors showed thatthe three-dimensional cluster fills the enzyme active sitecavity Atomic-level resolution allowed us to distinguishthe positions of carbon and boron atoms in the carboranecage The crystal structure also served as a model forconstruction of the CAIX-1 computational model Virtualglycine scan enabled us to quantify the contributions ofindividual residues to the energy of binding of 1 to CAIIand CAIX and uncover differences of the enzyme active sitecavities Structural and computational analysis will be used infuture structure-based design of carborane compounds withselectivity toward the cancer-specific CAIX isoenzyme

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

Pavel Mader and Adam Pecina contributed equally to thiswork

Acknowledgments

This work was supported by the Technology Agency ofthe Czech Republic (Project TE01020028) and in part byResearch Projects RVO 68378050 and 61388963 awarded by

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

BioMed Research International 3

1

2

34

5 6

789 10 11

12

3

1

1

NH2

NH

OS

H

O

C

C

21

2

(a) (b)

Figure 1 (a) Structural formula of 1 with atom numbers used in the crystal structure coordinate file The vertices in carborane clusterrepresent BH groups (b) Crystal structure of CAII in complex with 1 The CAII active site is shown in cartoon representation residuesinvolved in interactions with the Zn2+ ion (purple sphere) and 1 are shown in stick representation with carbon atoms colored green Boronatoms are colored pink and other heteroatoms are colored according to standard color coding oxygen red nitrogen blue sulfur yellowThe2Fo-Fc electron density map for 1 is contoured at 1 120590

A geometric library for 1 was generated using the Libcheckprogram from the CCP4 suite Coot [40] was used forrebuilding The quality of the refined model was assessedusing MolProbity [33] The coordinates and structure factorswere deposited in the PDB under accession code 4Q78 Finalrefinement statistics are summarized in Table 1 All structuralfigures were prepared using PyMOL 141 [41]

23 Model of CAIX-1 Complex The complex of CAIX and 1was modeled by aligning the existing crystal structures of theCAIX catalytic domain (PDB code 3IAI [42]) with the CAII-1 complex (PDB code 4MDG [7]) using PyMOL version12 [43] Preparation of structure coordinate files for furthercalculations was performed as described before for CAII [27]

The complex was fully optimized using a QMMM pro-cedure We used ONIOM-like subtractive scheme [44] withlink atoms and mechanical embedding to be consistent withour previous studies [27 45ndash48] The QM part is describedat the DFT-D TPSSTZVPBLYPSVP level of theory [39]and comprises 218 atoms including the atoms present in1 and 8 amino acids (Trp5 Asn62 His64 Gln67 Gln92Val131 Leu135 and Pro202) The MM part constituted theremainder of the protein and the surrounding solvent wasapproximated by a generalized Born (GB) implicit modelDetailed description of the procedure was published in [27]One crystal water molecule (Wat272) bridging the inhibitorsand CAII residues Thr199 Glu106 and Tyr7 was retainedto maintain the integrity of the active site Other watermolecules present in the crystal structures were omitted

The positions of the added hydrogen atoms 1 and 15amino acids surrounding the ligand (Trp5 Asn62 Gly63His64 Gln67 Leu91 Gln92 Leu123 Val131 Leu135 Leu141Thr200 Pro201 Pro202 and Ala204) were relaxed in a GBimplicit solvent model using the FIRE algorithm followed

by 10 ps annealing from 100K or 150K to 0K using theBerendsen thermostat [49] in the SANDER module of theAMBER 10 package [50]

24 Virtual Glycine Scan The contribution of the active siteamino acids to inhibitor binding was examined by virtualglycine scanning Individual amino acids in contact with1 in the CAIX-1 model and CAII-1 crystal structure weresubstituted with glycine The energy contributions (ΔΔG1015840int)were calculated as the difference between the original ΔG1015840intat the QMMM level with the wild-type amino acid and thenew ΔG1015840int with the mutated glycine residue [27]

3 Results and Discussion

31 Crystal Structure of CAII in Complex with 1 at Atomic Res-olution The overall structure of CAII in complex with 1 wasrefined to 10 A resolutionThis high resolution allowed us toobserve details that could not be fully resolved in the complexstructure determined previously at lower resolution Atomicresolution was achieved by derivatization of CAII using the4-(hydroxymercury)benzoic acid (abbreviated MBO in thecif library of small molecules) method described by [28]Themercury atom ofMBO covalently binds to S120574 of Cys206Thismodification allows formation of a hydrogen bond betweenthe OZ1 oxygen of the MBO carboxyl group and the main-chain amino group of Tyr40 in the neighboring proteinmolecule reinforcing the crystal lattice and increasing thediffraction quality of the crystal In our structure MBO ismodeled in two alternative conformations with occupanciesof 06 and 02

When our atomic resolution structure is compared withthe structure of the CAII-1 complex determined at 135 Aresolution (PDB code 4MDG [7]) the RMSD value for

4 BioMed Research International

Table 1 Data collection and refinement statisticsData collection statistics

Space group 11987521

Cell parameters (A ∘) 4220 4173 7216900 1044 900

Wavelength (A) 09184Resolution (A) 2108ndash100 (105ndash100)Number of unique reflections 108781 (15490)Multiplicity 35 (25)Completeness () 831 (814)119877merge

a 0056 (0375)Average 119868120590(119868) 108 (23)Wilson B (A2) 65

Refinement statisticsResolution range (A) 6990ndash100 (103ndash100)No of reflections in working set 97856 (7831)No of reflections in test set 5426 (412)119877 value ()b 175 (244)119877free value ()c 200 (262)RMSD bond length (A) 0011RMSD angle (∘) 153Number of atoms in AU 2297Number of protein atoms in AU 2081Number of water molecules in AU 176Mean ADP value proteininhibitor (A2) 120176

Ramachandran plot statisticsd

Residues in favored regions () 9656Residues in allowed regions () 344

The data in parentheses refer to the highest-resolution shella119877merge = sumℎ119896119897 sum119894 119868119894(ℎ119896119897) minus ⟨119868(ℎ119896119897)⟩ | sumℎ119896119897 sum119894 119868119894(ℎ119896119897) where 119868119894(ℎ119896119897) is theindividual intensity of the 119894th observation of reflection ℎ119896119897 and ⟨119868(ℎ119896119897)⟩ isthe average intensity of reflection ℎ119896119897 with summation over all datab119877 value = ||119865119900| minus |119865119888|||119865119900| where 119865119900 and 119865119888 are the observed and calculatedstructure factors respectively119888119877free is equivalent to 119877 value but is calculated for 5 of reflections chosen atrandom and omitted from the refinement process [32]das determined by Molprobity [33]

superposition of the C120572 atoms of residues 4ndash261 is 0142 Aa value typical for superposition of identical structures [51]The N-terminal residue His3 is traced differently in the twostructures double conformations of numerous side chains(eg Glu14 His64 and Gln74) are resolved in the atomicresolution structure We found an additional difference inthe loop formed by amino acid residues 124ndash139 with amaximum difference of 0738 A for the position of Gln136C120572 Gln136 forms van der Waals contacts with the MBOcovalently attached to Cys206 The positions of Phe131 andVal135 which form a hydrophobic rim at the active site arealso influenced by MBO binding This results in a subtlepositional shift of the inhibitor with an RMSD of 0145 A forsuperposition of 12 atoms in the carborane cage of 1 bound toCAII and CAII derivatized by MBO This value is below thevalue observed for superposition of identical structures [51]

Atomic-level resolution allowed us to resolve the carbonand boron atom positions in the symmetrical carborane

Phe131

Gln92Val121

Leu198

His94

His96

His119Thr200

Thr199

Figure 2 Interactions of 1 with CAII The protein is shown incartoon representation residues involved in interactions with theZn2+ ion (gray sphere) and 1 are shown in stick representationPolar interactions are represented by blue dashed lines Zn2+ ioncoordination is shown as black dashed lines

cage of 1 When analyzing the values of the electron densitymap at positions of atoms bonded to the C1 atom we canassume that positions with higher density levels are morelikely to be carbon than boron atoms Similar analysis wasdone by others for boron-containing inhibitor of humandihydrofolate reductase [18] The C2 atom of the carboranecage (Figure 1(a)) was modeled into the position with anelectron density value of 116 eA3 which was approximately015 eA3 higher than those for the B3 B4 B5 and B6 atomsTo exclude the possibility that higher density is caused bymodel bias we altered the composition of the cage byreplacing the C2 atom with a boron atom Electron densityvalues did not change significantly after several rounds ofrefinement cycles

Thus we can conclude that the most probable positionof the second carbon atom in the carborane cage of 1 is theposition assigned to the C2 atom in our crystal structureThisis in good agreement with the recently published QMMMmodeling study [27]

32 Detailed Analysis of Inhibitor Interactions with CAIIThe crystal structure of human CAII in complex with 1determined at 10 A resolution confirmed the key interactionsthat our group observed previously [7] The compound fitsvery well into the CAII active site cavity andmakes numerouspolar and nonpolar interactions with the residues in theenzyme active site The sulfamide moiety which forms keypolar interactions with the active site Zn2+ ion also makespolar interactions with Thr199 typical of other sulfamideinhibitors of CAII (Figure 2) The linker NH group formsan additional polar interaction with O120574 of Thr200 Thecompound makes several van der Waals interactions withresidues Gln92 His94 His96 His119 Val121 Phe131 Leu198andThr200 (Figure 2) All interactions between the inhibitorand protein are summarized in Table 2

BioMed Research International 5

2 3

H3C CH3

H2NS

OO

OO

N

OO

SH2N

H3C

OO CH3H3C

N

(a)

Phe131Val121

Gln92

His94Asn67

Asn62

Thr200Pro201

Pro202

Leu198

(b)

Phe131

Gln92

His94

Thr200

Leu198

(c) (d)

Figure 3 (a) Structural formulas of 2 and 3 (b) Interactions of 2 and 3 with the CAII active site Compound 2 is represented with goldencarbon atoms while the carbon atoms of 3 are colored turquoise Surface of residues making contacts with the isoquinoline moiety of 2 and 3are highlighted in yellow and blue respectively Surface of residues colored orange make contacts with both compounds Atoms involvedin contacts with the sulfonamide groups are not highlighted (c) Interactions of 1 with the CAII active site Surface of residues makingcontacts with the carborane and linker moiety of 1 are highlighted in green Atoms involved in contacts with the sulfonamide groups arenot highlighted (d) Superposition of binding poses of 1 2 and 3 in the CAII active site Superposition of the complex structures was basedon the best fit for C120572 atoms of CAII residues 6ndash261

The idea of designing CA inhibitors containing a carbo-rane cluster moiety originated from our previous structuralstudies of isoquinoline-containing sulfonamide inhibitors(Figure 3(a)) Structural analysis of CAII in complex with 67-dimethoxy-1234-tetrahydroisoquinolin-2-ylsulfonamide (2PDB code 3IGP [34]) and 67-dimethoxy-1-methyl-1234-tetrahydroisoquinolin-2-ylsulfonamide (3 PDB code 3PO6[52]) revealed two distinct binding modes that engage twoopposite sides of the enzyme active site cavity (Figure 3(b))Following this analysis we hypothesized that the bindingspace within the enzyme active site cavity could be effectivelyfilled with a bulky hydrophobic molecule with a sphericalstructure This led to design of 1 which exhibited inhibitoryproperty to CAII and CAIX with Ki values in submicro-molar range Structural analysis of CAII-1 indicates that ourstructure-based design was sound We found that the carbo-rane cluster interacts with both sides of the enzyme active siteas predicted (Figure 3(c) Table 3) and that the position of 1

in the CAII active site superposes well with the two bindingmodes observed for 2 and 3 (Figure 3(d))

33 Model of the CAIX-1 Complex The CAII-1 crystal struc-ture was used tomodel binding of compound 1 into the CAIXactive site using QMMMmethods (Figure 4)

The substrate binding sites of CAII and CAIX differ byonly six amino acids Asn67 of CAII is replaced by Gln inCAIX Ile91 by Leu Trp123 by Leu Phe131 by Val Val135 byLeu andLeu204 byAlaThese variations result in a differentlyshaped active site cavity which accommodated 1 in a slightlydifferent pose (Figure 4) While the position of the sulfamideanchor remained unchanged the carborane cluster shifted by21 A (expressed as a difference in the position of B12) awayfrom the central 120573-sheet In CAIX-1 the carborane interactsmore with the opposite site of the active site specifically withamino acid residues His94 His96 Glu106 Leu198 Thr199Thr200 and Pro201 (Figure 4 Table 3) All polar and van der

6 BioMed Research International

Table 2 List of contacts between CAII and 1CAII 1

Residue Atom Atoma Distance [A]b

Zn ZN N2 187c

Zn ZN S 304Zn ZN O2 305

92 Gln OE1 B6 34792 Gln OE1 B11 35292 Gln CD B6 38494 His CE1 O2 29794 His NE2 N2 32394 His NE2 O2 33194 His CE1 C3 36794 His NE2 S 38194 His CE1 N2 38294 His CE1 S 38494 His NE2 C3 39496 His NE2 N2 31496 His CE1 N2 356119 His ND1 N2 339119 His ND1 O2 388119 His CE1 N2 396121 Val CG2 O2 382131 Phe CZ B8 383131 Phe CZ B7 397198 Leu CA O1 309198 Leu C O1 336198 Leu CB O1 360198 Leu CD2 O1 363198 Leu CD1 B3 386199 Thr N O1 270199 Thr OG1 N2 274199 Thr OG1 O1 358199 Thr OG1 S 378199 Thr N S 383199 Thr CA O1 383199 Thr CB N2 398200 Thr OG1 N1 302200 Thr OG1 C3 314200 Thr OG1 B4 336200 Thr OG1 B3 356200 Thr OG1 C1 366aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance between ligand and protein (or Zn) atoms lessthan or equal to 4 A are listedcPolar interactions are highlighted in bold

Waals interactions between CAIX and 1 are summarized inTable 4

We used a virtual glycine scan to study the roles of indi-vidual amino acid side chains in the active sites of CAIIand CAIX in binding of 1 The changes in free energy of

Trp5Glu106

His96His119

Val121

Trp209

Leu198Thr199

Thr200

Pro201

His94

Figure 4 Interactions of 1with the CAIX active site Atomsmakingcontacts with the carborane and linkermoiety of 1 are highlighted inmagenta Atoms involved in contacts with the sulfonamide groupsare not highlighted Superposition of the binding pose of 1 in CAIIis shown as black lines Superposition is based on the best fit for C120572atoms of all residues of CAII onto CAIX

Table 3 CAII or CAIX residues interacting with 1 2 and 3

CAII CAIX1a 2b 3c 1d

Trp5Asn62Asn67e

Gln92 Gln92 Gln92His94 His94 His94 His94His96 His96 His96 His96

Glu106His119 His119 His119 His119Val121 Val121 Val121Phe131 Phe131 Phe131

Val143 Val143Leu198 Leu198 Leu198 Leu198Thr199 Thr199 Thr199 Thr199Thr200 Thr200 Thr200 Thr200

Pro201 Pro201Pro202Trp209 Trp209

Interacting residues were identified from acrystal structure 4Q78 (this work)bcrystal structure 3IGP [34] ccrystal structure 3PO6 [52] dcomputationalmodel (this work) eresidues making polar interactions are highlighted inbold

interaction (ΔΔG1015840int) upon mutation of a given amino acidresidue to glycine are shown in Figure 5

The largest energy change (26 kcalmol) occurred forTrp5 which is positioned closer to 1 in CAIX-1 than in CAII-1 The side chain of Trp5 forms several dihydrogen bondswith the carborane cage of 1 The shortest one has a H sdot sdot sdotHdistance of 23 A The other major contributor to strong

BioMed Research International 7

Table 4 Interactions between CAIX and 1

CAIX 1Residue Atom Atoma Distance [A]b

Zn ZN N2 21c

Zn ZN S 33Zn ZN O2 35

5 Trp CZ2 B5 3745 Trp CZ2 B10 38194 His CE1 O2 31594 His CE1 C3 37494 His NE2 N2 33694 His NE2 S 38894 His NE2 O2 34594 His NE2 C3 37696 His CE1 N2 39996 His NE2 N2 349106 Glu OE2 N2 371119 His ND1 N2 337119 His CE1 N2 383121 Val CG2 O2 358198 Leu CA O1 304198 Leu CB O1 34198 Leu CD2 O1 343198 Leu C O1 338199 Thr N S 388199 Thr N O1 279199 Thr CA O1 396199 Thr CB N2 385199 Thr OG1 N2 263199 Thr OG1 S 369199 Thr OG1 O1 365200 Thr OG1 C1 377200 Thr OG1 B5 356200 Thr OG1 N1 313200 Thr OG1 C3 331200 Thr OG1 B4 364201 Pro O B4 36201 Pro O B10 349201 Pro O B8 396209 Trp CZ2 O1 374aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance less than or equal to 4 A between ligand andprotein (and Zn) atoms are listedcPolar interactions are highlighted in bold

CAIX-1 binding was Asn62 the energy of binding exceededthat in CAII-1 by nearly 1 kcalmol These contributions werecancelled out by differences in binding energy contributionsof amino acid residues 131 (PheVal) and 135 (ValLeu) whichwere lower in CAIX by 07 and 09 kcalmol respectivelyTheenergy changes of other residues were small

When we compared binding of 1 to CAII and CAIX wenoted that the favorable energy changes in CAIX-1 due tothe binding of residues Trp5 Asn62 and His64 were slightly

0

1

Trp5

Asn

62

His6

4

Asn

Gln

67

Gln

92

Phe

Val1

31

ValL

eu13

5

Pro2

02

CAIICAIX amino acid residues

CAIICAIX

minus1

minus2

minus3

ΔΔ

G998400 in

t(k

calm

ol)

Figure 5 Results of virtual glycine scan showing contributions ofindividual residues to the energy of binding of 1 to CAII and CAIXrespectively

larger than the unfavorable changes in binding caused bythe different amino acids at residues 131 and 135 This is inqualitative agreement with the experimental119870

119894values which

are 700plusmn141 nm for inhibition of CAII and 380plusmn111 nM forinhibition of CAIX [7]

4 Conclusions

We determined to atomic resolution the crystal structureof CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (1) a parent compound of a recentlyreported series of CA inhibitors containing carborane cages[7] Comparing this crystal structure with those of CAIIcomplexes with conventional organic inhibitors showed thatthe three-dimensional cluster fills the enzyme active sitecavity Atomic-level resolution allowed us to distinguishthe positions of carbon and boron atoms in the carboranecage The crystal structure also served as a model forconstruction of the CAIX-1 computational model Virtualglycine scan enabled us to quantify the contributions ofindividual residues to the energy of binding of 1 to CAIIand CAIX and uncover differences of the enzyme active sitecavities Structural and computational analysis will be used infuture structure-based design of carborane compounds withselectivity toward the cancer-specific CAIX isoenzyme

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

Pavel Mader and Adam Pecina contributed equally to thiswork

Acknowledgments

This work was supported by the Technology Agency ofthe Czech Republic (Project TE01020028) and in part byResearch Projects RVO 68378050 and 61388963 awarded by

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

4 BioMed Research International

Table 1 Data collection and refinement statisticsData collection statistics

Space group 11987521

Cell parameters (A ∘) 4220 4173 7216900 1044 900

Wavelength (A) 09184Resolution (A) 2108ndash100 (105ndash100)Number of unique reflections 108781 (15490)Multiplicity 35 (25)Completeness () 831 (814)119877merge

a 0056 (0375)Average 119868120590(119868) 108 (23)Wilson B (A2) 65

Refinement statisticsResolution range (A) 6990ndash100 (103ndash100)No of reflections in working set 97856 (7831)No of reflections in test set 5426 (412)119877 value ()b 175 (244)119877free value ()c 200 (262)RMSD bond length (A) 0011RMSD angle (∘) 153Number of atoms in AU 2297Number of protein atoms in AU 2081Number of water molecules in AU 176Mean ADP value proteininhibitor (A2) 120176

Ramachandran plot statisticsd

Residues in favored regions () 9656Residues in allowed regions () 344

The data in parentheses refer to the highest-resolution shella119877merge = sumℎ119896119897 sum119894 119868119894(ℎ119896119897) minus ⟨119868(ℎ119896119897)⟩ | sumℎ119896119897 sum119894 119868119894(ℎ119896119897) where 119868119894(ℎ119896119897) is theindividual intensity of the 119894th observation of reflection ℎ119896119897 and ⟨119868(ℎ119896119897)⟩ isthe average intensity of reflection ℎ119896119897 with summation over all datab119877 value = ||119865119900| minus |119865119888|||119865119900| where 119865119900 and 119865119888 are the observed and calculatedstructure factors respectively119888119877free is equivalent to 119877 value but is calculated for 5 of reflections chosen atrandom and omitted from the refinement process [32]das determined by Molprobity [33]

superposition of the C120572 atoms of residues 4ndash261 is 0142 Aa value typical for superposition of identical structures [51]The N-terminal residue His3 is traced differently in the twostructures double conformations of numerous side chains(eg Glu14 His64 and Gln74) are resolved in the atomicresolution structure We found an additional difference inthe loop formed by amino acid residues 124ndash139 with amaximum difference of 0738 A for the position of Gln136C120572 Gln136 forms van der Waals contacts with the MBOcovalently attached to Cys206 The positions of Phe131 andVal135 which form a hydrophobic rim at the active site arealso influenced by MBO binding This results in a subtlepositional shift of the inhibitor with an RMSD of 0145 A forsuperposition of 12 atoms in the carborane cage of 1 bound toCAII and CAII derivatized by MBO This value is below thevalue observed for superposition of identical structures [51]

Atomic-level resolution allowed us to resolve the carbonand boron atom positions in the symmetrical carborane

Phe131

Gln92Val121

Leu198

His94

His96

His119Thr200

Thr199

Figure 2 Interactions of 1 with CAII The protein is shown incartoon representation residues involved in interactions with theZn2+ ion (gray sphere) and 1 are shown in stick representationPolar interactions are represented by blue dashed lines Zn2+ ioncoordination is shown as black dashed lines

cage of 1 When analyzing the values of the electron densitymap at positions of atoms bonded to the C1 atom we canassume that positions with higher density levels are morelikely to be carbon than boron atoms Similar analysis wasdone by others for boron-containing inhibitor of humandihydrofolate reductase [18] The C2 atom of the carboranecage (Figure 1(a)) was modeled into the position with anelectron density value of 116 eA3 which was approximately015 eA3 higher than those for the B3 B4 B5 and B6 atomsTo exclude the possibility that higher density is caused bymodel bias we altered the composition of the cage byreplacing the C2 atom with a boron atom Electron densityvalues did not change significantly after several rounds ofrefinement cycles

Thus we can conclude that the most probable positionof the second carbon atom in the carborane cage of 1 is theposition assigned to the C2 atom in our crystal structureThisis in good agreement with the recently published QMMMmodeling study [27]

32 Detailed Analysis of Inhibitor Interactions with CAIIThe crystal structure of human CAII in complex with 1determined at 10 A resolution confirmed the key interactionsthat our group observed previously [7] The compound fitsvery well into the CAII active site cavity andmakes numerouspolar and nonpolar interactions with the residues in theenzyme active site The sulfamide moiety which forms keypolar interactions with the active site Zn2+ ion also makespolar interactions with Thr199 typical of other sulfamideinhibitors of CAII (Figure 2) The linker NH group formsan additional polar interaction with O120574 of Thr200 Thecompound makes several van der Waals interactions withresidues Gln92 His94 His96 His119 Val121 Phe131 Leu198andThr200 (Figure 2) All interactions between the inhibitorand protein are summarized in Table 2

BioMed Research International 5

2 3

H3C CH3

H2NS

OO

OO

N

OO

SH2N

H3C

OO CH3H3C

N

(a)

Phe131Val121

Gln92

His94Asn67

Asn62

Thr200Pro201

Pro202

Leu198

(b)

Phe131

Gln92

His94

Thr200

Leu198

(c) (d)

Figure 3 (a) Structural formulas of 2 and 3 (b) Interactions of 2 and 3 with the CAII active site Compound 2 is represented with goldencarbon atoms while the carbon atoms of 3 are colored turquoise Surface of residues making contacts with the isoquinoline moiety of 2 and 3are highlighted in yellow and blue respectively Surface of residues colored orange make contacts with both compounds Atoms involvedin contacts with the sulfonamide groups are not highlighted (c) Interactions of 1 with the CAII active site Surface of residues makingcontacts with the carborane and linker moiety of 1 are highlighted in green Atoms involved in contacts with the sulfonamide groups arenot highlighted (d) Superposition of binding poses of 1 2 and 3 in the CAII active site Superposition of the complex structures was basedon the best fit for C120572 atoms of CAII residues 6ndash261

The idea of designing CA inhibitors containing a carbo-rane cluster moiety originated from our previous structuralstudies of isoquinoline-containing sulfonamide inhibitors(Figure 3(a)) Structural analysis of CAII in complex with 67-dimethoxy-1234-tetrahydroisoquinolin-2-ylsulfonamide (2PDB code 3IGP [34]) and 67-dimethoxy-1-methyl-1234-tetrahydroisoquinolin-2-ylsulfonamide (3 PDB code 3PO6[52]) revealed two distinct binding modes that engage twoopposite sides of the enzyme active site cavity (Figure 3(b))Following this analysis we hypothesized that the bindingspace within the enzyme active site cavity could be effectivelyfilled with a bulky hydrophobic molecule with a sphericalstructure This led to design of 1 which exhibited inhibitoryproperty to CAII and CAIX with Ki values in submicro-molar range Structural analysis of CAII-1 indicates that ourstructure-based design was sound We found that the carbo-rane cluster interacts with both sides of the enzyme active siteas predicted (Figure 3(c) Table 3) and that the position of 1

in the CAII active site superposes well with the two bindingmodes observed for 2 and 3 (Figure 3(d))

33 Model of the CAIX-1 Complex The CAII-1 crystal struc-ture was used tomodel binding of compound 1 into the CAIXactive site using QMMMmethods (Figure 4)

The substrate binding sites of CAII and CAIX differ byonly six amino acids Asn67 of CAII is replaced by Gln inCAIX Ile91 by Leu Trp123 by Leu Phe131 by Val Val135 byLeu andLeu204 byAlaThese variations result in a differentlyshaped active site cavity which accommodated 1 in a slightlydifferent pose (Figure 4) While the position of the sulfamideanchor remained unchanged the carborane cluster shifted by21 A (expressed as a difference in the position of B12) awayfrom the central 120573-sheet In CAIX-1 the carborane interactsmore with the opposite site of the active site specifically withamino acid residues His94 His96 Glu106 Leu198 Thr199Thr200 and Pro201 (Figure 4 Table 3) All polar and van der

6 BioMed Research International

Table 2 List of contacts between CAII and 1CAII 1

Residue Atom Atoma Distance [A]b

Zn ZN N2 187c

Zn ZN S 304Zn ZN O2 305

92 Gln OE1 B6 34792 Gln OE1 B11 35292 Gln CD B6 38494 His CE1 O2 29794 His NE2 N2 32394 His NE2 O2 33194 His CE1 C3 36794 His NE2 S 38194 His CE1 N2 38294 His CE1 S 38494 His NE2 C3 39496 His NE2 N2 31496 His CE1 N2 356119 His ND1 N2 339119 His ND1 O2 388119 His CE1 N2 396121 Val CG2 O2 382131 Phe CZ B8 383131 Phe CZ B7 397198 Leu CA O1 309198 Leu C O1 336198 Leu CB O1 360198 Leu CD2 O1 363198 Leu CD1 B3 386199 Thr N O1 270199 Thr OG1 N2 274199 Thr OG1 O1 358199 Thr OG1 S 378199 Thr N S 383199 Thr CA O1 383199 Thr CB N2 398200 Thr OG1 N1 302200 Thr OG1 C3 314200 Thr OG1 B4 336200 Thr OG1 B3 356200 Thr OG1 C1 366aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance between ligand and protein (or Zn) atoms lessthan or equal to 4 A are listedcPolar interactions are highlighted in bold

Waals interactions between CAIX and 1 are summarized inTable 4

We used a virtual glycine scan to study the roles of indi-vidual amino acid side chains in the active sites of CAIIand CAIX in binding of 1 The changes in free energy of

Trp5Glu106

His96His119

Val121

Trp209

Leu198Thr199

Thr200

Pro201

His94

Figure 4 Interactions of 1with the CAIX active site Atomsmakingcontacts with the carborane and linkermoiety of 1 are highlighted inmagenta Atoms involved in contacts with the sulfonamide groupsare not highlighted Superposition of the binding pose of 1 in CAIIis shown as black lines Superposition is based on the best fit for C120572atoms of all residues of CAII onto CAIX

Table 3 CAII or CAIX residues interacting with 1 2 and 3

CAII CAIX1a 2b 3c 1d

Trp5Asn62Asn67e

Gln92 Gln92 Gln92His94 His94 His94 His94His96 His96 His96 His96

Glu106His119 His119 His119 His119Val121 Val121 Val121Phe131 Phe131 Phe131

Val143 Val143Leu198 Leu198 Leu198 Leu198Thr199 Thr199 Thr199 Thr199Thr200 Thr200 Thr200 Thr200

Pro201 Pro201Pro202Trp209 Trp209

Interacting residues were identified from acrystal structure 4Q78 (this work)bcrystal structure 3IGP [34] ccrystal structure 3PO6 [52] dcomputationalmodel (this work) eresidues making polar interactions are highlighted inbold

interaction (ΔΔG1015840int) upon mutation of a given amino acidresidue to glycine are shown in Figure 5

The largest energy change (26 kcalmol) occurred forTrp5 which is positioned closer to 1 in CAIX-1 than in CAII-1 The side chain of Trp5 forms several dihydrogen bondswith the carborane cage of 1 The shortest one has a H sdot sdot sdotHdistance of 23 A The other major contributor to strong

BioMed Research International 7

Table 4 Interactions between CAIX and 1

CAIX 1Residue Atom Atoma Distance [A]b

Zn ZN N2 21c

Zn ZN S 33Zn ZN O2 35

5 Trp CZ2 B5 3745 Trp CZ2 B10 38194 His CE1 O2 31594 His CE1 C3 37494 His NE2 N2 33694 His NE2 S 38894 His NE2 O2 34594 His NE2 C3 37696 His CE1 N2 39996 His NE2 N2 349106 Glu OE2 N2 371119 His ND1 N2 337119 His CE1 N2 383121 Val CG2 O2 358198 Leu CA O1 304198 Leu CB O1 34198 Leu CD2 O1 343198 Leu C O1 338199 Thr N S 388199 Thr N O1 279199 Thr CA O1 396199 Thr CB N2 385199 Thr OG1 N2 263199 Thr OG1 S 369199 Thr OG1 O1 365200 Thr OG1 C1 377200 Thr OG1 B5 356200 Thr OG1 N1 313200 Thr OG1 C3 331200 Thr OG1 B4 364201 Pro O B4 36201 Pro O B10 349201 Pro O B8 396209 Trp CZ2 O1 374aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance less than or equal to 4 A between ligand andprotein (and Zn) atoms are listedcPolar interactions are highlighted in bold

CAIX-1 binding was Asn62 the energy of binding exceededthat in CAII-1 by nearly 1 kcalmol These contributions werecancelled out by differences in binding energy contributionsof amino acid residues 131 (PheVal) and 135 (ValLeu) whichwere lower in CAIX by 07 and 09 kcalmol respectivelyTheenergy changes of other residues were small

When we compared binding of 1 to CAII and CAIX wenoted that the favorable energy changes in CAIX-1 due tothe binding of residues Trp5 Asn62 and His64 were slightly

0

1

Trp5

Asn

62

His6

4

Asn

Gln

67

Gln

92

Phe

Val1

31

ValL

eu13

5

Pro2

02

CAIICAIX amino acid residues

CAIICAIX

minus1

minus2

minus3

ΔΔ

G998400 in

t(k

calm

ol)

Figure 5 Results of virtual glycine scan showing contributions ofindividual residues to the energy of binding of 1 to CAII and CAIXrespectively

larger than the unfavorable changes in binding caused bythe different amino acids at residues 131 and 135 This is inqualitative agreement with the experimental119870

119894values which

are 700plusmn141 nm for inhibition of CAII and 380plusmn111 nM forinhibition of CAIX [7]

4 Conclusions

We determined to atomic resolution the crystal structureof CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (1) a parent compound of a recentlyreported series of CA inhibitors containing carborane cages[7] Comparing this crystal structure with those of CAIIcomplexes with conventional organic inhibitors showed thatthe three-dimensional cluster fills the enzyme active sitecavity Atomic-level resolution allowed us to distinguishthe positions of carbon and boron atoms in the carboranecage The crystal structure also served as a model forconstruction of the CAIX-1 computational model Virtualglycine scan enabled us to quantify the contributions ofindividual residues to the energy of binding of 1 to CAIIand CAIX and uncover differences of the enzyme active sitecavities Structural and computational analysis will be used infuture structure-based design of carborane compounds withselectivity toward the cancer-specific CAIX isoenzyme

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

Pavel Mader and Adam Pecina contributed equally to thiswork

Acknowledgments

This work was supported by the Technology Agency ofthe Czech Republic (Project TE01020028) and in part byResearch Projects RVO 68378050 and 61388963 awarded by

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

BioMed Research International 5

2 3

H3C CH3

H2NS

OO

OO

N

OO

SH2N

H3C

OO CH3H3C

N

(a)

Phe131Val121

Gln92

His94Asn67

Asn62

Thr200Pro201

Pro202

Leu198

(b)

Phe131

Gln92

His94

Thr200

Leu198

(c) (d)

Figure 3 (a) Structural formulas of 2 and 3 (b) Interactions of 2 and 3 with the CAII active site Compound 2 is represented with goldencarbon atoms while the carbon atoms of 3 are colored turquoise Surface of residues making contacts with the isoquinoline moiety of 2 and 3are highlighted in yellow and blue respectively Surface of residues colored orange make contacts with both compounds Atoms involvedin contacts with the sulfonamide groups are not highlighted (c) Interactions of 1 with the CAII active site Surface of residues makingcontacts with the carborane and linker moiety of 1 are highlighted in green Atoms involved in contacts with the sulfonamide groups arenot highlighted (d) Superposition of binding poses of 1 2 and 3 in the CAII active site Superposition of the complex structures was basedon the best fit for C120572 atoms of CAII residues 6ndash261

The idea of designing CA inhibitors containing a carbo-rane cluster moiety originated from our previous structuralstudies of isoquinoline-containing sulfonamide inhibitors(Figure 3(a)) Structural analysis of CAII in complex with 67-dimethoxy-1234-tetrahydroisoquinolin-2-ylsulfonamide (2PDB code 3IGP [34]) and 67-dimethoxy-1-methyl-1234-tetrahydroisoquinolin-2-ylsulfonamide (3 PDB code 3PO6[52]) revealed two distinct binding modes that engage twoopposite sides of the enzyme active site cavity (Figure 3(b))Following this analysis we hypothesized that the bindingspace within the enzyme active site cavity could be effectivelyfilled with a bulky hydrophobic molecule with a sphericalstructure This led to design of 1 which exhibited inhibitoryproperty to CAII and CAIX with Ki values in submicro-molar range Structural analysis of CAII-1 indicates that ourstructure-based design was sound We found that the carbo-rane cluster interacts with both sides of the enzyme active siteas predicted (Figure 3(c) Table 3) and that the position of 1

in the CAII active site superposes well with the two bindingmodes observed for 2 and 3 (Figure 3(d))

33 Model of the CAIX-1 Complex The CAII-1 crystal struc-ture was used tomodel binding of compound 1 into the CAIXactive site using QMMMmethods (Figure 4)

The substrate binding sites of CAII and CAIX differ byonly six amino acids Asn67 of CAII is replaced by Gln inCAIX Ile91 by Leu Trp123 by Leu Phe131 by Val Val135 byLeu andLeu204 byAlaThese variations result in a differentlyshaped active site cavity which accommodated 1 in a slightlydifferent pose (Figure 4) While the position of the sulfamideanchor remained unchanged the carborane cluster shifted by21 A (expressed as a difference in the position of B12) awayfrom the central 120573-sheet In CAIX-1 the carborane interactsmore with the opposite site of the active site specifically withamino acid residues His94 His96 Glu106 Leu198 Thr199Thr200 and Pro201 (Figure 4 Table 3) All polar and van der

6 BioMed Research International

Table 2 List of contacts between CAII and 1CAII 1

Residue Atom Atoma Distance [A]b

Zn ZN N2 187c

Zn ZN S 304Zn ZN O2 305

92 Gln OE1 B6 34792 Gln OE1 B11 35292 Gln CD B6 38494 His CE1 O2 29794 His NE2 N2 32394 His NE2 O2 33194 His CE1 C3 36794 His NE2 S 38194 His CE1 N2 38294 His CE1 S 38494 His NE2 C3 39496 His NE2 N2 31496 His CE1 N2 356119 His ND1 N2 339119 His ND1 O2 388119 His CE1 N2 396121 Val CG2 O2 382131 Phe CZ B8 383131 Phe CZ B7 397198 Leu CA O1 309198 Leu C O1 336198 Leu CB O1 360198 Leu CD2 O1 363198 Leu CD1 B3 386199 Thr N O1 270199 Thr OG1 N2 274199 Thr OG1 O1 358199 Thr OG1 S 378199 Thr N S 383199 Thr CA O1 383199 Thr CB N2 398200 Thr OG1 N1 302200 Thr OG1 C3 314200 Thr OG1 B4 336200 Thr OG1 B3 356200 Thr OG1 C1 366aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance between ligand and protein (or Zn) atoms lessthan or equal to 4 A are listedcPolar interactions are highlighted in bold

Waals interactions between CAIX and 1 are summarized inTable 4

We used a virtual glycine scan to study the roles of indi-vidual amino acid side chains in the active sites of CAIIand CAIX in binding of 1 The changes in free energy of

Trp5Glu106

His96His119

Val121

Trp209

Leu198Thr199

Thr200

Pro201

His94

Figure 4 Interactions of 1with the CAIX active site Atomsmakingcontacts with the carborane and linkermoiety of 1 are highlighted inmagenta Atoms involved in contacts with the sulfonamide groupsare not highlighted Superposition of the binding pose of 1 in CAIIis shown as black lines Superposition is based on the best fit for C120572atoms of all residues of CAII onto CAIX

Table 3 CAII or CAIX residues interacting with 1 2 and 3

CAII CAIX1a 2b 3c 1d

Trp5Asn62Asn67e

Gln92 Gln92 Gln92His94 His94 His94 His94His96 His96 His96 His96

Glu106His119 His119 His119 His119Val121 Val121 Val121Phe131 Phe131 Phe131

Val143 Val143Leu198 Leu198 Leu198 Leu198Thr199 Thr199 Thr199 Thr199Thr200 Thr200 Thr200 Thr200

Pro201 Pro201Pro202Trp209 Trp209

Interacting residues were identified from acrystal structure 4Q78 (this work)bcrystal structure 3IGP [34] ccrystal structure 3PO6 [52] dcomputationalmodel (this work) eresidues making polar interactions are highlighted inbold

interaction (ΔΔG1015840int) upon mutation of a given amino acidresidue to glycine are shown in Figure 5

The largest energy change (26 kcalmol) occurred forTrp5 which is positioned closer to 1 in CAIX-1 than in CAII-1 The side chain of Trp5 forms several dihydrogen bondswith the carborane cage of 1 The shortest one has a H sdot sdot sdotHdistance of 23 A The other major contributor to strong

BioMed Research International 7

Table 4 Interactions between CAIX and 1

CAIX 1Residue Atom Atoma Distance [A]b

Zn ZN N2 21c

Zn ZN S 33Zn ZN O2 35

5 Trp CZ2 B5 3745 Trp CZ2 B10 38194 His CE1 O2 31594 His CE1 C3 37494 His NE2 N2 33694 His NE2 S 38894 His NE2 O2 34594 His NE2 C3 37696 His CE1 N2 39996 His NE2 N2 349106 Glu OE2 N2 371119 His ND1 N2 337119 His CE1 N2 383121 Val CG2 O2 358198 Leu CA O1 304198 Leu CB O1 34198 Leu CD2 O1 343198 Leu C O1 338199 Thr N S 388199 Thr N O1 279199 Thr CA O1 396199 Thr CB N2 385199 Thr OG1 N2 263199 Thr OG1 S 369199 Thr OG1 O1 365200 Thr OG1 C1 377200 Thr OG1 B5 356200 Thr OG1 N1 313200 Thr OG1 C3 331200 Thr OG1 B4 364201 Pro O B4 36201 Pro O B10 349201 Pro O B8 396209 Trp CZ2 O1 374aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance less than or equal to 4 A between ligand andprotein (and Zn) atoms are listedcPolar interactions are highlighted in bold

CAIX-1 binding was Asn62 the energy of binding exceededthat in CAII-1 by nearly 1 kcalmol These contributions werecancelled out by differences in binding energy contributionsof amino acid residues 131 (PheVal) and 135 (ValLeu) whichwere lower in CAIX by 07 and 09 kcalmol respectivelyTheenergy changes of other residues were small

When we compared binding of 1 to CAII and CAIX wenoted that the favorable energy changes in CAIX-1 due tothe binding of residues Trp5 Asn62 and His64 were slightly

0

1

Trp5

Asn

62

His6

4

Asn

Gln

67

Gln

92

Phe

Val1

31

ValL

eu13

5

Pro2

02

CAIICAIX amino acid residues

CAIICAIX

minus1

minus2

minus3

ΔΔ

G998400 in

t(k

calm

ol)

Figure 5 Results of virtual glycine scan showing contributions ofindividual residues to the energy of binding of 1 to CAII and CAIXrespectively

larger than the unfavorable changes in binding caused bythe different amino acids at residues 131 and 135 This is inqualitative agreement with the experimental119870

119894values which

are 700plusmn141 nm for inhibition of CAII and 380plusmn111 nM forinhibition of CAIX [7]

4 Conclusions

We determined to atomic resolution the crystal structureof CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (1) a parent compound of a recentlyreported series of CA inhibitors containing carborane cages[7] Comparing this crystal structure with those of CAIIcomplexes with conventional organic inhibitors showed thatthe three-dimensional cluster fills the enzyme active sitecavity Atomic-level resolution allowed us to distinguishthe positions of carbon and boron atoms in the carboranecage The crystal structure also served as a model forconstruction of the CAIX-1 computational model Virtualglycine scan enabled us to quantify the contributions ofindividual residues to the energy of binding of 1 to CAIIand CAIX and uncover differences of the enzyme active sitecavities Structural and computational analysis will be used infuture structure-based design of carborane compounds withselectivity toward the cancer-specific CAIX isoenzyme

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

Pavel Mader and Adam Pecina contributed equally to thiswork

Acknowledgments

This work was supported by the Technology Agency ofthe Czech Republic (Project TE01020028) and in part byResearch Projects RVO 68378050 and 61388963 awarded by

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

6 BioMed Research International

Table 2 List of contacts between CAII and 1CAII 1

Residue Atom Atoma Distance [A]b

Zn ZN N2 187c

Zn ZN S 304Zn ZN O2 305

92 Gln OE1 B6 34792 Gln OE1 B11 35292 Gln CD B6 38494 His CE1 O2 29794 His NE2 N2 32394 His NE2 O2 33194 His CE1 C3 36794 His NE2 S 38194 His CE1 N2 38294 His CE1 S 38494 His NE2 C3 39496 His NE2 N2 31496 His CE1 N2 356119 His ND1 N2 339119 His ND1 O2 388119 His CE1 N2 396121 Val CG2 O2 382131 Phe CZ B8 383131 Phe CZ B7 397198 Leu CA O1 309198 Leu C O1 336198 Leu CB O1 360198 Leu CD2 O1 363198 Leu CD1 B3 386199 Thr N O1 270199 Thr OG1 N2 274199 Thr OG1 O1 358199 Thr OG1 S 378199 Thr N S 383199 Thr CA O1 383199 Thr CB N2 398200 Thr OG1 N1 302200 Thr OG1 C3 314200 Thr OG1 B4 336200 Thr OG1 B3 356200 Thr OG1 C1 366aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance between ligand and protein (or Zn) atoms lessthan or equal to 4 A are listedcPolar interactions are highlighted in bold

Waals interactions between CAIX and 1 are summarized inTable 4

We used a virtual glycine scan to study the roles of indi-vidual amino acid side chains in the active sites of CAIIand CAIX in binding of 1 The changes in free energy of

Trp5Glu106

His96His119

Val121

Trp209

Leu198Thr199

Thr200

Pro201

His94

Figure 4 Interactions of 1with the CAIX active site Atomsmakingcontacts with the carborane and linkermoiety of 1 are highlighted inmagenta Atoms involved in contacts with the sulfonamide groupsare not highlighted Superposition of the binding pose of 1 in CAIIis shown as black lines Superposition is based on the best fit for C120572atoms of all residues of CAII onto CAIX

Table 3 CAII or CAIX residues interacting with 1 2 and 3

CAII CAIX1a 2b 3c 1d

Trp5Asn62Asn67e

Gln92 Gln92 Gln92His94 His94 His94 His94His96 His96 His96 His96

Glu106His119 His119 His119 His119Val121 Val121 Val121Phe131 Phe131 Phe131

Val143 Val143Leu198 Leu198 Leu198 Leu198Thr199 Thr199 Thr199 Thr199Thr200 Thr200 Thr200 Thr200

Pro201 Pro201Pro202Trp209 Trp209

Interacting residues were identified from acrystal structure 4Q78 (this work)bcrystal structure 3IGP [34] ccrystal structure 3PO6 [52] dcomputationalmodel (this work) eresidues making polar interactions are highlighted inbold

interaction (ΔΔG1015840int) upon mutation of a given amino acidresidue to glycine are shown in Figure 5

The largest energy change (26 kcalmol) occurred forTrp5 which is positioned closer to 1 in CAIX-1 than in CAII-1 The side chain of Trp5 forms several dihydrogen bondswith the carborane cage of 1 The shortest one has a H sdot sdot sdotHdistance of 23 A The other major contributor to strong

BioMed Research International 7

Table 4 Interactions between CAIX and 1

CAIX 1Residue Atom Atoma Distance [A]b

Zn ZN N2 21c

Zn ZN S 33Zn ZN O2 35

5 Trp CZ2 B5 3745 Trp CZ2 B10 38194 His CE1 O2 31594 His CE1 C3 37494 His NE2 N2 33694 His NE2 S 38894 His NE2 O2 34594 His NE2 C3 37696 His CE1 N2 39996 His NE2 N2 349106 Glu OE2 N2 371119 His ND1 N2 337119 His CE1 N2 383121 Val CG2 O2 358198 Leu CA O1 304198 Leu CB O1 34198 Leu CD2 O1 343198 Leu C O1 338199 Thr N S 388199 Thr N O1 279199 Thr CA O1 396199 Thr CB N2 385199 Thr OG1 N2 263199 Thr OG1 S 369199 Thr OG1 O1 365200 Thr OG1 C1 377200 Thr OG1 B5 356200 Thr OG1 N1 313200 Thr OG1 C3 331200 Thr OG1 B4 364201 Pro O B4 36201 Pro O B10 349201 Pro O B8 396209 Trp CZ2 O1 374aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance less than or equal to 4 A between ligand andprotein (and Zn) atoms are listedcPolar interactions are highlighted in bold

CAIX-1 binding was Asn62 the energy of binding exceededthat in CAII-1 by nearly 1 kcalmol These contributions werecancelled out by differences in binding energy contributionsof amino acid residues 131 (PheVal) and 135 (ValLeu) whichwere lower in CAIX by 07 and 09 kcalmol respectivelyTheenergy changes of other residues were small

When we compared binding of 1 to CAII and CAIX wenoted that the favorable energy changes in CAIX-1 due tothe binding of residues Trp5 Asn62 and His64 were slightly

0

1

Trp5

Asn

62

His6

4

Asn

Gln

67

Gln

92

Phe

Val1

31

ValL

eu13

5

Pro2

02

CAIICAIX amino acid residues

CAIICAIX

minus1

minus2

minus3

ΔΔ

G998400 in

t(k

calm

ol)

Figure 5 Results of virtual glycine scan showing contributions ofindividual residues to the energy of binding of 1 to CAII and CAIXrespectively

larger than the unfavorable changes in binding caused bythe different amino acids at residues 131 and 135 This is inqualitative agreement with the experimental119870

119894values which

are 700plusmn141 nm for inhibition of CAII and 380plusmn111 nM forinhibition of CAIX [7]

4 Conclusions

We determined to atomic resolution the crystal structureof CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (1) a parent compound of a recentlyreported series of CA inhibitors containing carborane cages[7] Comparing this crystal structure with those of CAIIcomplexes with conventional organic inhibitors showed thatthe three-dimensional cluster fills the enzyme active sitecavity Atomic-level resolution allowed us to distinguishthe positions of carbon and boron atoms in the carboranecage The crystal structure also served as a model forconstruction of the CAIX-1 computational model Virtualglycine scan enabled us to quantify the contributions ofindividual residues to the energy of binding of 1 to CAIIand CAIX and uncover differences of the enzyme active sitecavities Structural and computational analysis will be used infuture structure-based design of carborane compounds withselectivity toward the cancer-specific CAIX isoenzyme

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

Pavel Mader and Adam Pecina contributed equally to thiswork

Acknowledgments

This work was supported by the Technology Agency ofthe Czech Republic (Project TE01020028) and in part byResearch Projects RVO 68378050 and 61388963 awarded by

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

BioMed Research International 7

Table 4 Interactions between CAIX and 1

CAIX 1Residue Atom Atoma Distance [A]b

Zn ZN N2 21c

Zn ZN S 33Zn ZN O2 35

5 Trp CZ2 B5 3745 Trp CZ2 B10 38194 His CE1 O2 31594 His CE1 C3 37494 His NE2 N2 33694 His NE2 S 38894 His NE2 O2 34594 His NE2 C3 37696 His CE1 N2 39996 His NE2 N2 349106 Glu OE2 N2 371119 His ND1 N2 337119 His CE1 N2 383121 Val CG2 O2 358198 Leu CA O1 304198 Leu CB O1 34198 Leu CD2 O1 343198 Leu C O1 338199 Thr N S 388199 Thr N O1 279199 Thr CA O1 396199 Thr CB N2 385199 Thr OG1 N2 263199 Thr OG1 S 369199 Thr OG1 O1 365200 Thr OG1 C1 377200 Thr OG1 B5 356200 Thr OG1 N1 313200 Thr OG1 C3 331200 Thr OG1 B4 364201 Pro O B4 36201 Pro O B10 349201 Pro O B8 396209 Trp CZ2 O1 374aAtom labels correspond to those shown in Figure 1(a)bAll contacts with a distance less than or equal to 4 A between ligand andprotein (and Zn) atoms are listedcPolar interactions are highlighted in bold

CAIX-1 binding was Asn62 the energy of binding exceededthat in CAII-1 by nearly 1 kcalmol These contributions werecancelled out by differences in binding energy contributionsof amino acid residues 131 (PheVal) and 135 (ValLeu) whichwere lower in CAIX by 07 and 09 kcalmol respectivelyTheenergy changes of other residues were small

When we compared binding of 1 to CAII and CAIX wenoted that the favorable energy changes in CAIX-1 due tothe binding of residues Trp5 Asn62 and His64 were slightly

0

1

Trp5

Asn

62

His6

4

Asn

Gln

67

Gln

92

Phe

Val1

31

ValL

eu13

5

Pro2

02

CAIICAIX amino acid residues

CAIICAIX

minus1

minus2

minus3

ΔΔ

G998400 in

t(k

calm

ol)

Figure 5 Results of virtual glycine scan showing contributions ofindividual residues to the energy of binding of 1 to CAII and CAIXrespectively

larger than the unfavorable changes in binding caused bythe different amino acids at residues 131 and 135 This is inqualitative agreement with the experimental119870

119894values which

are 700plusmn141 nm for inhibition of CAII and 380plusmn111 nM forinhibition of CAIX [7]

4 Conclusions

We determined to atomic resolution the crystal structureof CAII in complex with 1-methylenesulfamide-12-dicarba-closo-dodecaborane (1) a parent compound of a recentlyreported series of CA inhibitors containing carborane cages[7] Comparing this crystal structure with those of CAIIcomplexes with conventional organic inhibitors showed thatthe three-dimensional cluster fills the enzyme active sitecavity Atomic-level resolution allowed us to distinguishthe positions of carbon and boron atoms in the carboranecage The crystal structure also served as a model forconstruction of the CAIX-1 computational model Virtualglycine scan enabled us to quantify the contributions ofindividual residues to the energy of binding of 1 to CAIIand CAIX and uncover differences of the enzyme active sitecavities Structural and computational analysis will be used infuture structure-based design of carborane compounds withselectivity toward the cancer-specific CAIX isoenzyme

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Authorsrsquo Contribution

Pavel Mader and Adam Pecina contributed equally to thiswork

Acknowledgments

This work was supported by the Technology Agency ofthe Czech Republic (Project TE01020028) and in part byResearch Projects RVO 68378050 and 61388963 awarded by

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

8 BioMed Research International

the Academy of Sciences of the Czech Republic The authorsacknowledge the financial support of the Czech ScienceFoundation (P20812G016) and the operational programResearch and Development for Innovations of the EuropeanSocial Fund (CZ 1052100030058)They also thank GileadSciences and IOCB Research Centre for financial supportThe use of MX142 operated by the Helmholtz-ZentrumBerlin at the BESSY II electron storage ring (Berlin-AdlershofGermany) to collect diffraction data is acknowledged

References

[1] P Swietach S Patiar C T Supuran A L Harris and R DVaughan-Jones ldquoThe role of carbonic anhydrase 9 in regulatingextracellular and intracellular pH in three-dimensional tumorcell growthsrdquo The Journal of Biological Chemistry vol 284 no30 pp 20299ndash20310 2009

[2] P C McDonald J Winum C T Supuran and S DedharldquoRecent developments in targeting carbonic anhydrase IX forcancer therapeuticsrdquo Oncotarget vol 3 no 1 pp 84ndash97 2012

[3] F E Lock P C McDonald Y Lou et al ldquoTargeting carbonicanhydrase IX depletes breast cancer stem cells within thehypoxic nicherdquo Oncogene vol 32 no 44 pp 5210ndash5219 2013

[4] N K Tafreshi M C Lloyd M M Bui R J Gillies and D LMorse ldquoCarbonic anhydrase IX as an imaging and therapeutictarget for tumors and metastasesrdquo in Carbonic AnhydraseMechanism Regulation Links to Disease and Industrial Appli-cations vol 75 of Subcellular Biochemistry pp 221ndash254 2014

[5] S M Monti C T Supuran and G de Simone ldquoAnticancercarbonic anhydrase inhibitors a patent review (2008ndash2013)rdquoExpert Opinion on Therapeutic Patents vol 23 no 6 pp 737ndash749 2013

[6] RMcKenna and C T Supuran ldquoCarbonic anhydrase inhibitorsdrug designrdquo Subcellular Biochemistry vol 75 pp 291ndash323 2014

[7] J Brynda P Mader V Sicha et al ldquoCarborane-based carbonicanhydrase inhibitorsrdquo Angewandte Chemie International Edi-tion in English vol 52 pp 13760ndash13763 2013

[8] W N Lipscomb Boron Hydrides WA Benjamin New YorkNY USA 1963

[9] P R Schleyer and K Najafian ldquoStability and three-dimensionalaromaticity of closo-monocarbaborane anions Cb

119899minus1H119899

minus andcloso-dicarboranes C

2B119899minus2

H119899rdquo Inorganic Chemistry vol 37 pp

3454ndash3470 1998[10] R E Williams in The Borane Carborane and Carbocation

Continuum J Casanova Ed pp 3ndash57 JohnWiley amp Sons NewYork NY USA 1997

[11] J F Valliant K J Guenther A S King et al ldquoThe medicinalchemistry of carboranesrdquo Coordination Chemistry Reviews vol232 no 1-2 pp 173ndash230 2002

[12] Z J Lesnikowski ldquoBoron units as pharmacophoresmdashnew appli-cations andopportunities of boron cluster chemistryrdquoCollectionof Czechoslovak Chemical Communications vol 72 no 12 pp1646ndash1658 2007

[13] J Plesek ldquoPotential applications of the boron cluster com-poundsrdquo Chemical Reviews vol 92 no 2 pp 269ndash278 1992

[14] J Seidler S L McGovern T N Doman and B K ShoichetldquoIdentification and prediction of promiscuous aggregatinginhibitors among known drugsrdquo Journal of Medicinal Chem-istry vol 46 no 21 pp 4477ndash4486 2003

[15] M Sibrian-Vazquez E Hao T J Jensen and M G H VicenteldquoEnhanced cellular uptake with a cobaltacarborane-porphyrin-HIV-1 Tat 48-60 conjugaterdquo Bioconjugate Chemistry vol 17 no4 pp 928ndash934 2006

[16] M ScholzM Steinhagen J THeiker AG Beck-Sickinger andE Hey-Hawkins ldquoAsborin Inhibits AldoKeto Reductase 1A1rdquoChemMedChem vol 6 no 1 pp 89ndash93 2011

[17] F Issa M Kassiou and L M Rendina ldquoBoron in drug dis-covery carboranes as unique pharmacophores in biologicallyactive compoundsrdquo Chemical Reviews vol 111 no 9 pp 5701ndash5722 2011

[18] R C Reynolds S R Campbell R G Fairchild et al ldquoNovelboron-containing nonclassical antifolates synthesis and pre-liminary biological and structural evaluationrdquo Journal ofMedic-inal Chemistry vol 50 no 14 pp 3283ndash3289 2007

[19] P Cigler M Kozisek P Rezacova et al ldquoFrom nonpeptidetoward noncarbon protease inhibitors metallacarboranes asspecific and potent inhibitors of HIV proteaserdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 pp 15394ndash15399 2005

[20] P Rezacova J Pokorna J Brynda et al ldquoDesign ofHIVproteaseinhibitors based on inorganic polyhedral metallacarboranesrdquoJournal of Medicinal Chemistry vol 52 no 22 pp 7132ndash71412009

[21] Y Endo T Iijima Y Yamakoshi et al ldquoPotent estrogen agonistsbased on carborane as a hydrophobic skeletal structure a newmedicinal application of boron clustersrdquo Chemistry amp Biologyvol 8 no 4 pp 341ndash355 2001

[22] R L Julius O K Farha J Chiang L J Perry and M FHawthorne ldquoSynthesis and evaluation of transthyretin amyloi-dosis inhibitors containing carborane pharmacophoresrdquo Pro-ceedings of the National Academy of Sciences of the United Statesof America vol 104 no 12 pp 4808ndash4813 2007

[23] S Fujii H Masuno Y Taoda et al ldquoBoron cluster-based devel-opment of potent nonsecosteroidal vitamin D receptor ligandsdirect observation of hydrophobic interaction between proteinsurface and carboranerdquo Journal of the American ChemicalSociety vol 133 no 51 pp 20933ndash20941 2011

[24] V M Krishnamurthy G K Kaufman A R Urbach et al ldquoCar-bonic anhydrase as a model for biophysical and physical-organic studies of proteins and protein-ligand bindingrdquo Chem-ical Reviews vol 108 no 3 pp 946ndash1051 2008

[25] K Raha M B Peters B Wang et al ldquoThe role of quantummechanics in structure-based drug designrdquo Drug DiscoveryToday vol 12 no 17-18 pp 725ndash731 2007

[26] M Lepsık J Rezac M Kolar A Pecina P Hobza and J Fan-frlık ldquoThe semiempirical quantummechanical scoring functionfor in silico drug designrdquo ChemPlusChem vol 78 pp 921ndash9312013

[27] A Pecina M Lepsik J Rezac et al ldquoQMMM calculationsreveal the different nature of the interaction of two carborane-based sulfamide inhibitors of human carbonic anhydrase IIrdquoThe Journal of Physical Chemistry B vol 117 pp 16096ndash161042013

[28] C A Behnke I Le Trong J W Godden et al ldquoAtomic resolu-tion studies of carbonic anhydrase IIrdquoActa Crystallographica DBiological Crystallography vol 66 no 5 pp 616ndash627 2010

[29] U Mueller N Darowski M R Fuchs et al ldquoFacilities formacromolecular crystallography at the Helmholtz-ZentrumBerlinrdquo Journal of Synchrotron Radiation vol 19 no 3 pp 442ndash449 2012

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011

BioMed Research International 9

[30] T G G Battye L Kontogiannis O Johnson H R Powelland A G Leslie ldquoiMOSFLM a new graphical interface fordiffraction-image processing with MOSFLMrdquo Acta Crystallo-graphica D vol 67 no 4 pp 271ndash281 2011

[31] P Evans ldquoScaling and assessment of data qualityrdquo Acta Crystal-lographica Section D Biological Crystallography vol 62 no 1pp 72ndash82 2006

[32] A T Brunger ldquoFree R value a novel statistical quantity forassessing the accuracy of crystal strucutresrdquo Nature vol 355no 6359 pp 472ndash475 1992

[33] S C Lovell I W Davis W B Arendall III et al ldquoStructurevalidation byC120572 geometry120593120595 andC120573 deviationrdquo Proteins vol50 no 3 pp 437ndash450 2003

[34] R Gitto S Agnello S Ferro et al ldquoIdentification of 34-dihy-droisoquinoline-2(1H)-sulfonamides as potent carbonic anhy-drase inhibitors synthesis biological evaluation and enzyme-ligand X-ray studiesrdquo Journal of Medicinal Chemistry vol 53no 6 pp 2401ndash2408 2010

[35] G N Murshudov A A Vagin and E J Dodson ldquoRefinementof macromolecular structures by the maximum-likelihoodmethodrdquo Acta Crystallographica D vol 53 no 3 pp 240ndash2551997

[36] ldquoThe CCP4 suite programs for protein crystallographyrdquo ActaCrystallographica D vol 50 pp 760ndash763 1994

[37] P Jurecka J Cerny P Hobza and D R Salahub ldquoDensityfunctional theory augmented with an empirical dispersionterm Interaction energies and geometries of 80 noncovalentcomplexes compared with ab initio quantum mechanics calcu-lationsrdquo Journal of Computational Chemistry vol 28 no 2 pp555ndash569 2007

[38] FWeigend andR Ahlrichs ldquoBalanced basis sets of split valencetriple zeta valence and quadruple zeta valence quality for Hto Rn design and assessment of accuracyrdquo Physical ChemistryChemical Physics vol 7 no 18 pp 3297ndash3305 2005

[39] R Ahlrichs M Bar M Haser H Horn and C KolmelldquoElectronic structure calculations on workstation computersthe program system turbomolerdquo Chemical Physics Letters vol162 no 3 pp 165ndash169 1989

[40] P Emsley and K Cowtan ldquoCoot model-building tools formolecular graphicsrdquo Acta Crystallographica D Biological Crys-tallography vol 60 no 12 pp 2126ndash2132 2004

[41] W L DeLanoThe PyMOLMolecular Graphics System DeLanoScientific LLC San Carlos Calif USA 2002 httpwwwpymolorg

[42] V Alterio M Hilvo A Di Fiore et al ldquoCrystal structure ofthe catalytic domain of the tumor-associated human carbonicanhydrase IXrdquo Proceedings of the National Academy of Sciencesof the United States of America vol 106 no 38 pp 16233ndash162382009

[43] W L DeLano The PymoL Molecular Graphics System DeLanoScientific Palo Alto Calif USA 2002

[44] M Svensson S Humbel R D J Froese T Matsubara SSieber and K Morokuma ldquoONIOM a multilayered integratedMO + MM method for geometry optimizations and singlepoint energy predictions A test for Diels-Alder reactions andPt(P(t-Bu)

3)2+ H2oxidative additionrdquo The Journal of Physical

Chemistry vol 100 no 50 pp 19357ndash19363 1996[45] A Pecina O Prenosil J Fanfrlık et al ldquoOn the reliability

of the corrected semiempirical quantum chemical method(PM6-DH2) for assigning the protonation states in HIV-1 pro-teaseinhibitor complexesrdquo Collection of Czechoslovak ChemicalCommunications vol 76 no 5 pp 457ndash479 2011

[46] P S Brahmkshatriya P Dobes J Fanfrlık et al ldquoQuantummechanical scoring structural and energetic insights intocyclin-dependent kinase 2 inhibition by pyrazolo[15-a]pyrimi-dinesrdquo Current ComputermdashAided Drug Design vol 9 no 1 pp118ndash129 2013

[47] J Fanfrlık M Kolar M Kamlar et al ldquoModulation of aldosereductase inhibition by halogen bond tuningrdquo ACS ChemicalBiology vol 8 pp 2484ndash2492 2013

[48] J Fanfrlık P S Brahmkshatriya J Rezac et al ldquoQuantummechanics-based scoring rationalizes the irreversible inactiva-tion of parasitic Schistosoma mansoni cysteine peptidase byvinyl sulfone inhibitorsrdquo Journal of Physical Chemistry B vol117 pp 14973ndash14982 2013

[49] H J C Berendsen J P M Postma W F Van Gunsteren ADinola and J R Haak ldquoMolecular dynamics with coupling toan external bathrdquoThe Journal of Chemical Physics vol 81 no 8pp 3684ndash3690 1984

[50] D A Case T E Cheatham III T Darden et al ldquoThe Amberbiomolecular simulation programsrdquo Journal of ComputationalChemistry vol 26 no 16 pp 1668ndash1688 2005

[51] M J Betts and M J E Sternberg ldquoAn analysis of conforma-tional changes on protein-protein association implications forpredictive dockingrdquo Protein Engineering vol 12 no 4 pp 271ndash283 1999

[52] P Mader J Brynda R Gitto et al ldquoStructural Basis for theInteraction between Carbonic Anhydrase and 1234-tetrahy-droisoquinolin-2-ylsulfonamidesrdquo Journal of Medicinal Chem-istry vol 54 no 7 pp 2522ndash2526 2011