`ANACOR EX. 2025 - 1/15
`
`
`
`Journal of
`Medicinal
`Chemistry
`
`Journal of Medicinal Chemistry (ISSN 0022-
`2623) is published biweekly by the American
`Chemical Society at 1155 16th St., N.W.,
`Washington, DC 20036. Periodicals postage
`paid at Washington, DC, and additional
`mailing offices. POSTMASTER: Send
`address changes to Journal of Medicinai
`Chemistry, Member & Subscriber Services,
`P.O. Box 3337, Columbus, OH 43210.
`
`American Chemical Society
`1155 16th St., N.W.
`Washington, DC 20036
`(202) 872-4614
`TDD (202) 872-6076
`Fax (202) 776-6264
`
`Journal Publications
`
`American Chemical Society
`2540 Olentangy River Road
`PO. Box 3330, Columbus, OH 43210
`(614) 447-3665
`Fax (614) 447-3745; Telex 6842086
`E-mail acsproof@acs.org
`
`Member & Subscriber Services
`
`American Chemical Society
`PO. Box 3337, Columbus, OH 43210
`(614) 447-3776; (800) 333-9511
`Fax (614) 447-3671
`E-rnail service@acs.org
`
`Advertising Office
`Centcom, Ltd.
`676 East Swedesford Road
`
`Suite 202, Wayne, PA 19087-1612
`(610) 964-8061
`
`Publications Division
`Robert D. Bovenschulte, Director
`
`Publishing Operations
`Mary E. Scanlan, Director
`
`Jotuncii Publications
`
`Anne C. O'Melia, Manager; Diane E.
`Needharn, Journals Editing Manager;
`Loretta
`Yam, Associate Editor;
`Tracy A. Pritchard, "Production Associ-
`ate
`
`Composition Services
`
`Teresa K. Lewanclowski, Manager
`
`Sales & Marketing
`Justin R. Spence, Director
`Dean J . Smith, Sales
`Matthew J. Price, Marketing
`
`Copyright Permission: See copyright status form for certain rights (http::7pubs.acs.org]_
`Reprographic copying beyond that permitted by Section 107 or 108 of the U.S. Copyright Act is
`allowed, provided that the current per‘ article fee is paid to the Copyright Clearance Cente-1-_
`Tel:
`(978) 750-8400. Republication or reproduction for sale of articles or abstracts in this joui-.
`nal is permitted only by written permission from the Copyright Oflice, ACS, Washington, DC,
`Tel: ( 202) 872-4363. Fax: (202) 776-8112. E—mail: copyright@acs.org.
`Editorial Information
`
`Instructions for Authors and Copyright Status Form: See the first issue of each volume
`or visit the Publications Division Web site (http:.-7pubs.acs.org) for instructions on preparing
`and submitting manuscripts and copyright forms to the Editor.
`
`Manuscript Submission: Send manusc1'ipts to the Editor, Philip S. Portoghese, Department
`of Medicinal Chemistry, College of Pharmacy, WDH 8-114, University of Minnesota, 308
`Harvard St. SE, Minneapolis, MN 55455. European contributors may send manuscripts to
`Povl Krogsgaard-Larsen, Department of Medicinal Chemistry, The Royal Danish School of
`Pharmacy, 2 Universitetsparken, DK-2100 Copenhagen, Denmark. Letters may be submitted
`over the Web at http:a7'pubs.acs.orgJ'JMC.
`
`Accepted Papers and Proofs: Direct correspondence to Journal Publications, Columbus, OH_
`Tel: (614) 447-3665. Fax: (614) 447-3745. E-mail: acsproof@acs.org.
`
`Journal Policies: The American Chemical Society and its Editors assume no responsibility
`for the statements and opinions advanced by contributors. Registered names and trademarks,
`etc., used in this publication, even without specific indication thereof, are not to be considered
`unprotected by law.
`
`Document Number: At the end of each document is a 9-character code that serves as a link
`between the printed and electronic products and facilitates retrieval of the document in elec-
`tronic form.
`
`Digital Object Identifier (DOI): The DOI identification system for digital media has bwn
`designed to provide persistent and reliable identification of digital objects. Information on the
`D01 and its governing body, the International DOI Foundation, can be found at httrn:l;'
`w'ww.doi.org. In the Web editions of ACS journals, the DOI appears at the top of the HTML
`version of an article and at the bottom of the first page in its PDF version; in the print editions,
`the DOI appears in the same location as in the PDF version.
`
`2002 Subscription and Ordering Information
`
`North
`America
`
`Outside
`North America"‘
`
`Printed Edition
`
`Web Edition‘'
`
`Members
`Student Members
`Nonmembers
`Institutional
`Members
`
`$ 118
`33
`89
`$ 276
`$1419
`
`$
`
`50
`
`3% 259
`3; 230
`$ 417
`$1560
`
`$
`
`50
`
`$ 309
`$ 168
`Printfweb Combinationi
`* Air service included. i For Institutional rates, call Member & Subscriber Services.
`
`Members
`
`Web Edition: This journal is available to subscribers via the Internet. Contact Member &
`Subscriber Services [Tel: (614) 447-3776 or (800) 333-9511. Fax: (614) 447-3671. E-mail:
`service@acs.org] or visit the Publications Division Web site (http'.{J'pubs.acs.org) for additional
`details.
`'
`
`New and Renewal Subscriptions: Send with payment to American Chemical Society.
`P.O. Box 182426, Columbus, OH 43218-2426.
`Institutional subscribers in Japan must enter
`subscription orders with Maruzen Company Ltd., 3-10 Nihonbashi 2—chome, Chuo-ku, Tol-'._'.'o
`103, Japan. Tel:
`(03) 272-7211.
`
`Subscription Donations: Members may donate.-‘share their personal subscriptions toiwith
`libraries but only after 5 years from publication.
`
`Change of Address: Notify Member & Subscriber Services, ACS, Columbus. OH. Tel: (614)
`447-3776 or (800) 333-9511. Fax: (614) 447-3671. E-mail: address@acs.org_ Include both old and
`new addresses and a mailing label from a recent issue.
`
`Microfihn, llilicrofiche, Back Issue, and Printed Edition Single Issue Orders: Sercl
`requests to Publications Support Services, ACS, Washington, DC. Tel: (202) 872-4376. Fax:
`(202) 872-6325. E-mail: pss@acs.org. Printed edition not available prior to 1998.
`
`Bulk Reprint Orders: For quotes and information, contact Cadmus Reprints. Tel: (888) 257-
`2134 or (410) 819-3995. Fax: (410) 820-9765.
`
`Claims for Issues Not Received: Claims will be honored only if submitted within 90 days of
`the issue date (subscribers in North America) or within 180 days of the issue date (all other
`subscribers). Send requests to Member & Subscriber Services, ACS, Columbus, OH. Tel: (61-13
`447-3776 or (800) 333-9511. Fax: (614) 447-3671. E-mail: service@acs.org.
`Supporting Information (SI): S] from 1995 to the present is available free of charge via the
`Internet (l1ttp:i':’pubs.acs.o1'gl from the journal’s home page. For information on electronic
`access, send e-mail to si—help@acs.org. SI prior to 1995 is available, for a fee, from Publication:-'
`Support Services. Tel: (202) 872-4376. Fax: ( 202) 872-6325. E-mail: pss@acs.org.
`
`Canadian GST Reg. No. 127571347
`Printed in the USA
`
`© Copyright 2002 American Chemical Society
`
`ANACOR EX. 2025 - 2/15
`
`
`
`3222
`
`J. Med. Chem. 2002, 45, 3222-3234
`
`Structure-Based Approach for Binding Site Identification on AmpC
`
`,8-Lactamase
`
`Rachel A. Powers and Brian K. Shoichet“‘
`
`Department of Moieciiioi‘ Plicirmocology and Biological Chemistry, Noi't/iwestervi University, 303 East Chicago Avenue,
`Chicago, Illinois SOGII
`
`Received January 2, 2002
`
`/3—Lactamases are the most widespread resistance mechanism to f3—lactam antibiotics and are
`an increasing menace to public health. Several /3-lactamase structures have been determined,
`making this enzyme an attractive target for structure-based drug design. To facilitate inhibitor
`design for the class C /3-lactamase AmpC, binding site “hot spots” on the enzyme were identified
`using experimental and computational approaches. Experimentally, X—ray crystal structures
`of AmpC in complexes with four boronic acid inhibitors and a higher resolution (1.72 A) native
`apo structure were determined. Along with previously determined structures of AmpC in
`complexes with five other boronic acid inhibitors and four (3-lactams, consensus binding sites
`were identified. Computationally, the programs GRID, MCSS, and X-SITE were used to predict
`potential binding site hot spots on AmpC. Several consensus binding sites were identified from
`the crystal structures. An amide recognition site was identified by the interaction between the
`carbonyl oxygen in the R1 side chain of [)‘—lactams and the atom N62 of the conserved AS11152.
`Surprisingly, this site also recognizes the aryl rings of arylboronic acids, appearing to form
`quadrupole-dipole interactions with Asn 152. The highly conserved “oxyanion” hole defines a
`site that recognizes both carbonyl and hydroxyl groups. A hydroxyl binding site was identified
`by the O2 hydroxyl in the boronic acids, which hydrogen bonds with 'Iyr150 and a conserved
`water. A hydrophobic site is formed by Leu119 and Leu293. A carboxylate binding site was
`identified by the ubiquitous C3(4) carboxylate of the [J'—lactams, which interacts with Asn346
`and Arg349. Four water sites were identified by ordered waters observed in most of the
`structures; these waters form extensive hydrogen—bonding networks with AmpC and occasion-
`ally the ligand. Predictions by the computational programs showed some correlation with the
`experimentally observed binding sites. Several sites were not predicted, but novel binding sites
`were suggested. Taken together, a map of binding site hot spots found on ArnpC, along with
`information on the functionality recognized at each site, was constructed. This map may be
`useful for structure—based inhibitor design against AinpC.
`
`Introduction
`
`,6-Lactamases are the most widespread resistance
`mechanism to /3-lactam antibiotics, hydrolyzing and
`inactivating penicillins, cephalosporins, and related
`molecules. Class C ,8-lactamases, such as AmpC, are
`among the most problematic of these enzymes. Not only
`are they widely expressed among nosocomial pathogens
`but also class C ,8-lactamases are not significantly
`inhibited by clinically used ,8-lactamase inhibitors, such
`as clavulanatel Moreover’ they naturally have a bread
`spectrum of action and can hydrolyze “,6-lactamase
`resistant” ,6—lactams, such as the third-generation ceph—
`alosporins. Indeed, ,6-lactam-based inhibitors and ,.6—lac—
`tsmsse resistant Jglsetsms can upregelste the eXp1.es_
`sion of class C fi—lactamases, thereby defeating the1n—
`se]ves_1—3 There is e pressing need fer nave], nen_fi_
`]sstsm_bssedjehibite1.s sf these enzymes
`Since the X-ray crystal structures of class C ,6-lacta-
`msses were first deteI.mined}4—7 these enzymes have
`been attractive targets for novel inhibitor discovery
`using structure—based methods. There are well-defined
`peekets in the binding sites that leek as though they
`
`3 To whom correspondence should be addressed. Tel.: 312603-0081,
`Fax: 312-503-5349. E—mail: b—shoichet@r1o1‘thwe5tern.ed1i.
`
`might accommodate inhibitor functionality, and exten-
`sive structure—fiinction studies suggest a role for many
`of these sites.3‘11 The structures also present challenges
`that are typical of many enzyme targets. The active sites
`of these enzymes are much larger than their substrates,
`and some of the observed sites, however well-defined
`structurally, have no known function (Figure 1). Water
`molecules are observed to bind in some of these sites,
`and it is unclear whether they should be treated as
`_
`_
`_
`displaceable or structurally integral components of the
`pmtein'12'13
`f
`_
`“
`”
`,
`d
`The Challenge 0 findmg the hot Spots
`for 1133“
`binding is Ccfmmon to m°5t_ Stm°t"“'e'baSed efi'°rt?"
`Several techniques have been introduced to address this
`P1"'3'l31'?m- Binding Sites may be explored experimentally
`by using ligand molecules as probes. This can be done
`crystallographically, using solvent” or even larger
`Incfleculesils or by nuclear magmetm resonance (NMR "
`using small molecule fragments that can probe the
`ab1ht_y °f_a particular 51159
`t°_bmd_ F0 3 partlculm
`functlflnallty 01" 3T0‘-lp of functlonalmes-16 Compute‘-
`tional approaches are also used to identify hot spots for
`ligand binding. Computer programs such as GRID" and
`M08818 use functional groups to probe the energy
`potential of protein binding sites and identify favorable
`
`10.1021:'j1n020002p CCC: $22.00 © 2002 American Chemical Society
`Published on Web 06f19i'2002
`
`ANACOR EX. 2025 - 3/15
`
`
`
`Idmtifyirig Biridirig Sites on AmpC
`
`Journal ofllfedicinoi Ch,em.ist1y, 2002, Vol. 45, No. 15 3223
`
`Table 1. Ligands Used to Probe the Active Site of A1npC
`Resoiulion of
`crystal stluclurc
`
`Compound
`
`Ligand
`
`R-f:u:inr.FR,,“
`
`11133 [D
`
`I(,- (1,: M}
`
`17.42 I .1
`
`1|(DS“
`
`1.7“
`
`I 9.9.-"245
`
`19.3i"23.?
`
`Figure 1. Molecular surface“ of the active site region of
`AmpC. The catalytic residue Ser64 is colored green. Several
`conserved residues are labeled. Interesting features of AmpC
`that appear to be preorganized to bind a ligand are shown. A
`tunnel begins behind the catalytic Ser64 and extends for
`approximately 15 A through the interior of AmpC and two
`pockets observed structurally that bind waters. The surface
`contributed by nitrogen atoms is colored blue, oxygen atoms
`are colored red, and carbon atoms are colored gray. Cyan
`spheres represent ordered water molecules. All figures were
`generated with MidasPlus,59 unless otherwise noted.
`
`positions for different ligand functionalities. Programs
`such as X-SITE19 use knowledge-based potentials to find
`favorable positions for similar probe groups.
`Structure-based inhibitor discovery in class C {Hac-
`tamases has proceeded without much formal effort to
`map the binding site for hot spots but has instead relied
`more on substrate and substrate—analogue approaches
`to lead discovery. Thus, the structure of AmpC has been
`determined in complex with inhibitors such as [S'-lac-
`tams,“»2”‘22 transition state analogues,23'24 and arylbo-
`ronic acids,5»25'25 the leads for which were known before
`the first structures were determined.” There are now
`
`18 inhibitor complex structures with class C ,6-lacta-
`inases that have been published. This wealth of struc-
`tural information allows us to consider a consensus view
`
`of where the hot spots for ligand binding are on class C
`,8-lactamases and what sort of ligand groups they
`recognize.
`Here, we use ni.ne previously determined X-ray crystal
`structures to partly construct a consensus map of hot
`spots for functional group binding on A1npC ,8-lactamase
`(compounds 5-13, Table 1). To further enrich the
`functionality explored, we determined the structures of
`four new inhibitor complexes by X—ray crystallography
`(compounds 1-4, Tables 1-3). As a reference, we have
`also determined the structure of native apo AmpC to
`higher resolution (1.72 A) and with better crystal-
`lographic statistics (Table 2} than was previously de-
`termined (Protein Data Bank (PDB) entry 2BLS; 2.00
`ll). Functionalities displayed on these nine boronic acids
`and four {3—1actams act as probes of the AmpC site. By
`overlaying these 13 experimental structures, we ask
`whether hot spots emerge that were not apparent when
`considering the individual structures separately. We
`also use the GRID, MOSS, and X-SITE computer
`programs to further probe AmpC for functional group
`binding sites. This allows us to investigate regions that
`
`lS.2i'22.6
`
`2o.?;25.5
`
`11'-cM“"
`
`21325.3
`
`11=Co’°
`
`16_&,.20_fi
`
`H5022
`
`NA
`
`NA
`
`moxaiactam
`
`" This work. 5 K; values were reported in Weston et al., 1998.
`"K; values were reported in Caselli et al., 2001. ‘’ NA, 11ot ap-
`plicable.
`
`are not explored in the crystal structures and compare
`the computational predictions with the experimentally
`determined complexes.
`
`Results
`
`X—ray Crystallographic Structure Determina-
`tion. The crystal structures of AmpC in complexes with
`four different arylboronic acid inhibitors were deter-
`mined for this study (compounds 1-4; Tables 1-2,
`Figure 2A-D). The resolution of these complexes ranged
`from 2.15 to 2.30 A. In each complex, the location of the
`inhibitor was unambiguously identified in the initial
`Fo—Fc difference maps when contoured at a level of 3
`0. Simulated annealing omit maps of the refined models
`agree well with the conformation of the inhibitors in the
`active sites (not shown). The quality of each of the
`models was analyzed with the program Procheck.23 For
`the AmpC.=’1 complex, 91.5% of the nonproline, nong1y—
`
`ANACOR EX. 2025 - 4/15
`
`
`
`3224 Jourual o_f'Med.icinnl Chemistry, 2002, Vol. 45, No. 15
`
`Powers and Shoichet
`
`Table 2. Data Collection and Refinement Statistics
`
`cell constants (:1; deg)
`
`resolution (1-ts)
`total observations
`unique observations
`Rmergc
`completeness (%)"’
`(l))'(crp<92)
`resolution range for refinement (A)
`no. of protein residues
`no. of water molecules
`RMSD bond lengths (A)
`RMSD bond angles (deg)
`R factor ('16)
`Rfrm (%)L'
`average B factor, protein atoms (A?)
`average B factor, inhibitor (A2)
`
`AmpC;’I
`a = 119.11,
`b = 77.73,
`c = 99.01;
`[3 = 115.86
`2.15 (2.25—2.15)"
`112 903
`43 507
`7.6 (22.2)
`98.0 (94.7)
`9.3 (3.3)
`10-215
`716
`140
`0.003
`1.5
`17.4
`21.1
`22.1
`23.3
`
`AmpC}'2
`a = 118.55,
`b = 77.38,
`c = 97.33;
`,9 = 115.30
`2.23 (2.37—2.28)
`90 630
`33 450
`7.5 (19.0)
`91.0 (62.6)
`16.2 (3.5)
`20-223
`716
`263
`0.010
`1.5
`18.6
`
`28.1
`30.5
`
`AmpCJ'3
`o: = 118.62,
`b = 77.43,
`c = 97.37;
`,6 = 115.32
`2.30 (2.35-2.30)
`63 591
`33 436
`7.5 (23.8)
`94.2 (96.6)
`12.2 (3.9)
`20-2.30
`716
`139
`0.010
`1.6
`19.9
`24.5
`25.9
`35.7
`
`AmpC.’4
`ct = 117.80,
`b = 78.36,
`c = 97.38;
`)5’ = 115.93
`2.15 (2.21-2.15)
`115 882
`42 306
`7.5 (33.2)
`97.7 (96.3)
`15.2 (3.7)
`20-2.15
`716
`307
`0.010
`1.6
`19.3
`23.7
`30.0
`33.6
`
`native apo ArnpC
`a = 118.49,
`E) = 76.97,
`c = 97.67;
`,8 = 115.90
`1.72 (1.76— 1.72)
`322 742
`81 989
`5.7 (37.2)
`98.2 (97.8)
`23.2 (5.1)
`20-1.72
`710
`377
`0.012
`1.7
`19.0
`21.4
`27.7
`NA“
`
`" Values in parentheses are for the highest resolution shell. 5 Fraction of theoretically possible reflections observed. " Rpm was calculated
`with 10% of reflections set aside randomly, except for native A1npC where 5% was used. 9' Not applicable.
`
`Table 3. Interactions in Cornplexed and Native apo AmpC ,8-Lactamase
`
`interaction
`
`S641‘-I-0 1"
`A318N—01
`A318N-01
`Y150OH-O2“
`Wat402—02
`Y150OH—K315N’r,'
`Y1500H-S640}!
`Y1500H—K67N(‘,‘
`K67NZ_j-A22 00
`K67N<‘;—S640y
`Wat402—T3160}*1
`Wat402-Wat403
`Wat403-N3460<§1
`Wat403—R34 9N if 1
`N1520d1—K67N§
`N152Nr)2-Q1200e1
`N15Nd2-centroid aryl ring
`centroid Y221-nearest carbon of
`inhibitor aryl ring
`
`An-1pC;"1"
`3.0
`2.8
`2.3
`2.8
`3.0
`2.9
`3.0
`3.5
`3.0
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`AmpC."2(’
`2.8
`2.7
`2.8
`2.8
`2.6
`2.9
`2.7
`3.6
`2.8
`2.6
`3.0
`2.9
`2.8
`3.3
`2.7
`7.3
`3.7
`4.4
`
`distance (A)
`
`AmpC.-’35
`3.0
`2.7
`2.9
`2.9
`2.6
`2.9
`2.9
`3.5
`2.9
`2.7
`3.1
`2.6
`2.9
`3.1
`2.7
`7.2
`3.7
`4.2
`
`AmpC.-‘4*"
`2.8
`2.7
`2.8
`2.6
`2.6
`2.9
`2.8
`3.5
`
`:'5F'°."-‘E°§°F°!‘3F-°.N.l°
`
`I005!-*-JI---2|U'AOO':m
`
`native apo AmpC
`molecule 1
`molecule 2
`3.6”’
`3.5“
`2.8”
`3.0"
`3.3”’
`3.2“
`NPe
`NP
`NP
`NP
`2.8
`2.8
`3.0
`2.9
`3.4
`3.3
`3.0
`2.9
`2.9
`2.8
`3.31"
`3.0
`2.51"
`2.9
`2.77
`2.6
`2.9!’
`2.9
`2.6
`2.6
`7.2
`2.8
`NP
`NP
`NP
`NP
`
`“ Distances are for molecule 2 of the asymmetric unit. “’ Distances are for molecule 1 of the asymmetric unit. *‘ Atoms 01 and O2 refer
`to the boronic acid hydroxyls. ‘( In the native structure, atom 03 of a phosphate ion (molecule 1) and Wat768 (molecule 2) are in equivalent
`positions to the O1 hydroxyl of the inhibitors. The distances reported here are to these analogous atoms. 9 Not present. -" In molecule 1 of
`the native structure, Wat402 is displaced by ()4 of the phosphate ion, and Wat4D3 is numbered Wat401. Distances to these atoms are
`reported here.
`
`cine residues are in the most favored region of the
`Ramachandran plot (8.5% in the additionally allowed
`region); for AmpC/2, 91.0% of the nonproline, nonglycine
`residues are in the most favored region (9.0% in the
`additionally allowed region); for AmpCf3, 89.5% of the
`nonproline, nonglycine residues are in the most favored
`region (10.5% in the additionally allowed region); and
`for AmpCf4, 92.0% of the nonproline, nonglycine resi-
`dues are in the most favored region (8.0% in the
`additionally allowed region). The structures have been
`deposited with the PDB as IKDS, IKDW, HCEO, and
`IKE3 (complexes of AmpC with 1—4, respectively).
`The unique portions of the inhibitors make relatively
`few interactions with the protein. The nitro group of 1
`does not make any interactions with AmpC. The car-
`boxylate group of 2 is 3.3 A from Ne2 of Gln120 in
`molecule 1 of the asymmetric unit, slightly long for a
`hydrogen bond. However, in molecule 2, this distance
`
`is 2.9 A due to the different conformation of Gln120.
`The carboxylate group found in 3 interacts with a single
`water molecule in the site. In contrast, the structure of
`AmpCf4 shows that specific polar interactions can be
`observed between the inhibitor and the enzyme. One
`hydroxyl of the terminal boronic acid group interacts
`with atom N52 of Gln120 and atom 0:51 of Asp123 (2.5
`and 2.9 A, respectively).
`To provide a reference for our comparisons, the native
`apo structure of AmpC was determined to 1.72 A
`resolution, and the coordinates and structure factors
`have been deposited with the PDB as IKE4 (Tables 2-3,
`Figure 2E). This is a higher resolution structure than
`the apo structure previously determined (PDB entry
`2BLS), with better crystallographic statistics, allowing
`for more reliable placement of side chains and water
`molecules. The quality of the final model was analyzed
`with Procheck.23 For the nonproline, nouglycine resi-
`
`ANACOR EX. 2025 - 5/15
`
`
`
`Irientifyzlng Binding Sites on AmpC
`
`Journal.’ ofllafedicinai Chemistry, 2002, Vol. 45, No. 15 3225
`
`B.
`
`N12
`
`Figure 2. Stereoview of the active site region of both apo and complexed AmpC, focusing on representative electron density.
`2Fo-Fc electron density maps are shown in blue, contoured at 1.0 0, except for D and E, which are contoured at 0.9 or. (A) An1pCI'1
`complex; (B) AmpC/'2 complex; (C) AmpC!3 complex; (D) A1-npCl4 complex; (E) the apo structure, determined to 1.72 A; and (F)
`interactions observed in the apo active site. Dashed yellow lines indicate hydrogen bonds. Carbon atoms are colored gray, nitrogens
`are colored blue, and oxygens are colored red. Water molecules are represented with cyan spheres. Figure 3A—E was made with
`SETOR.5°
`
`dues, 91.8% are in the most favored region of the
`Itarnachandran plot (8.2% in the additionally allowed
`region}.
`
`Overall, the higher resolution structure (IKE4) re-
`sembles the lower resolution structure (ZBLS). The
`RMSD in all Cu atom positions of molecule 2 from each
`structure is 0.29 A (for molecule 1, 0.70 A}. The RMSD
`in all atoms of active site residues (Ser64, Lys67,
`'l‘yr150, Asn152, and Lys315) between the two struc-
`tures is 1.37 A for molecule 1 and 1.41 A for molecule
`2. Some significant differences were observed. For
`instance, in the higher resolution structure, a phosphate
`ion was bound near Ser64 of molecule 1 (Table 3), and
`many more water molecules were identified (377 vs 83).
`The observation of an ordered phosphate ion in the
`active site is consistent with observations by Pratt and
`colleagues that class C ,8-lactamases are inhibited by
`phosphate ions at millimolar concentrations.” The
`presumed deacylating water, Wat402, was observed only
`in molecule 2 of 2BLS, and Wat-403, a well-ordered
`water in the active site that hydrogen bonds with
`Wat402, was not observed in either molecule of the
`ZBLS structure. In the higher resolution structure, the
`completely conserved Gln120 adopts a different confor-
`mation in each molecule. In molecule 1, Gln120 is
`flipped out of the active site and interacts with residues
`A1a4 and Asn9 of a symmetry mate (Gln1200e1—
`
`Asn9N(52 = 2.8 A; Gln120Ne2-Ala4O = 3.1 A). In
`molecule 2, Gln 120 points into the active site and
`interacts with Asn152, as observed in the 2]?-LS struc-
`ture (Table 3, Figure 2F). Residues 285-290 of molecule
`1 were excluded from the final model of the higher
`resolution structure, due to poor electron density.
`Conensus Binding Sites. Along with these newly
`determined structures, the previously published struc-
`tures ofAmpC in complexes with five other boronic acids
`and four ,8-lactams were used to identify consensus
`binding sites (compounds 5-13, Table 1). Individually,
`the structures provided little information about binding
`hot spots on AmpC. However, when all 13 complexes
`were superimposed, clear patterns emerged. Several
`consensus binding sites were identified in this way: an
`amide site, a carbonyl/hydroxyl site, a hydroxyl site, a
`hydrophobic site, an aryl group site, a carboxylate site,
`and several conserved water sites (Figure 3). Addition-
`ally, the computer programs GRID, MCSS, and X-SITE
`were used to identify binding sites on AmpC. We were
`interested to know how their predictions corresponded
`to the experimentally determined binding sites and also
`to learn of new binding sites identified by the programs
`but not seen crystallographically.
`Amide Binding Site. An amide recognition site on
`AmpC was identified based on superposition of the
`complexes with 7 ~13 (Figure 3A). The amide groups of
`
`ANACOR EX. 2025 - 6/15
`
`
`
`3226 Journal of Medicinal Chemistry, 2002, Vol. 45, No. 15
`
`Powers and Shoichct
`
`A
`
`B.
`
`N152
`
`compound 8
`
`OHO
`
`C1
`
`2
`NO I
`
`fi=OO
`
`cornpounfia
`
`fi’\P.oHH0
`
`R349
`
`Figure 3. Binding sites identified experimentally on ArnpC. (A) An amide binding site identified by superimposing all amide-
`containing compounds from Table 1. For clarity, only the structures of 8-12 are shown in this figure; these positions are
`representative of all other ligands determined. The amide group is circled in black. (B) 'I\evo hyclroxyl binding sites identified by
`superimposing all boronic acid inhibitors. The hydroxyl sites are circled and labeled O1 and 02. (C) A hydrophobic binding site
`identified by superimposing the structures of G, 8, 9, 11, and 12; f'or clarity, only 6, 8, and 9 are shown. The hydrophobic binding
`site is composed of residues Leul19 and Leu293, and the hydrophobic portions of the compounds interacting with these residues
`are circled. (D) An aryl binding site identified by superimposing the structures of compounds 1-4. Dashed yellow lines indicate
`the quadrupole interactions with AS11152 and Tyr221. (E) The carboxylate binding site identified by superimposing the structures
`of the ,6—lactarns (10-13). The C3(4) carboxylate group is circled. Nitrogen atoms are colored blue, oxygens are colored red, sulfurs
`are colored yellow, chlorines are colored magenta, borons are colored purple, and carbons are colored gray, except f'or carbon
`atoms of the ligands, which are colored as follows: coral for 1, indigo for 2, sea green for 3, violet for 4, tan for 5, lime green for
`6, gold for 7, orange for 8, magenta for 9, green for 10, brown for 11, pink for 12, and cyan for 13.
`
`ANACOR EX. 2025 - 7/15
`
`
`
`Iricrltifyiiig Biiidirig Sites on AmpC
`
`Journat ofilfediciiiai Chemistry, 2002, Vol. 45, N0. 15 3227
`
`Table 4. Comparison of Computationally Predicted Sites with Experimentally Determined Binding Sites in the AmpC Active Site
`GRID“
`X-SITE”
`MOSS”
`
`site predicted?
`
`site predicted?
`
`site predicted?
`
`experimental binding site
`R1 amide
`0 1 hydroxyl
`O2 hydroxyl
`Leul19.’l_.eu293 patch
`'l‘yr22 1
`Wat402
`Wat403
`water in upper pocket (Wat-404)
`water in lower pocket (Wat405)
`
`(apo)
`}''ES
`110
`1'10
`YES
`3'85
`H0
`1'10
`yes
`YES
`
`(complex)
`3535
`I10
`110
`YES
`}"EES
`I10
`YES
`YES
`3595
`
`(apo)
`yes
`110
`no
`yes
`1'10
`110
`no
`yes
`no
`
`(complex)
`3'85
`110
`1'10
`yes
`_YBS
`B0
`1.10
`yes
`110
`
`(complex)
`T10
`yes
`1'10
`1'10
`I10
`I10
`1'10
`YES
`YES
`
`“ The contour levels for dete1'mi11ing whether a site was predicted by GRID we1'e -10 kcalfmol for all p1'obes except the hydrophobic
`probe where a contour level of -1.0 kcalfmol was used. ’’ A site was successfully predicted by X—SITE if the closest probe position was
`<1.5
`A from the experimental site. " A site was successfully predicted by MCSS if a probe molecule ranked in the top 20% was also within
`1.5 A of the experimental site.
`
`the _6—lactams, 10-13, in the AmpCfacyl—enzyme com-
`plexes inade interactions characteristic of this site in
`[3-lactam-recognizing enzymes.2”'3°‘34 Atom N62 of
`Asn152 hydrogen bonds with the ligand amide oxygen
`in all of the complexes, although this distance is slightly
`long in the Ampcna complex (3.3 A). Atom N52 of
`Gln120 interacts with the ligand amide oxygen only in
`the AmpCJ'13 complex. The main chain oxygen of Ala3 18
`interacts with the ligand amide nitrogen in the AmpC
`complexes with 10 and 13. Three acylglycineboronic acid
`inhibitors, 7-9, also contain an amide group in an
`equivalent position as the one found in the fl-lactams.
`The amide groups of these compounds also made these
`characteristic interactions. Atom N852 of Asn 152 hydro-
`gen bonds with the ligand amide oxygen in all of the
`complexes. The interaction between atom N52 of Gln120
`and the ligand amide oxygen is only observed in the
`AmpCf'7 complex. The interaction between the main
`chain oxygen of Ala318 and the ligand amide nitrogen
`is not observed in any of the boronic acid structures;
`howgver,
`in the AmpCf8 complex,
`this distance is
`3.3
`.
`How well is the amide recognition site predicted by
`the computational programs? GRID and X-SITE were
`tested with different starting conformations of AmpC,
`an apo and a cornplexed form. In the apo structure, the
`conformation of Gln 120 is such that it hydrogen bonds
`with Asn152 and is better positioned to interact with
`the R1 amide group. In the complexed structure, Gh1120
`is swung out of the active site and no longer interacts
`with Asn152. In GRID, the trans amide probe success-
`fully predicted this site in both conformations (Table 4).
`Starting with a complexed form of AmpC, a peak is
`observed near the nitrogen atom of the amide group at
`a contour level of -10 kcalfmol. Viewed at a slightly
`lower contour level (—9 kcalfmol), another peak appears
`near the position of the oxygen atom of the R1 amide
`group. Starting with the apo form of AmpC, both the
`nitrogen and the oxygen atoms of the R1 amide are
`predicted at a contour of -10 kcalfmol, and this site is
`the lowest energy minima overall (- 15.5 kcalfmol). For
`X-SITE, the backbone N and backbone O probes were
`used to predict the amide site. The backbone N probe
`identified the amide nitrogen in both forms of AmpC.
`The amide oxygen was predicted by the backbone 0
`probe only in the apo form. MCSS provided a large
`number of predictions (N = 1018) using the amide group
`probes (acw1, acw2, and acw3). A probe with an energy
`
`of -7.18 kcal/mol did predict the correct position and
`orientation of the amide, but this was not one of the
`most energetically favorable predictions; it ranked 846,
`where a rank of 1 indicates the best energy probe
`position (Table 4).
`Carbonyllflydroxyl Binding Site. A binding site
`that recognizes both carbonyl oxygens and hydroxyls
`was identified; this site corresponds to the “oxyanion”35
`or “9lectrophilic”5 hole of AmpC. Two hydroxyl groups,
`01 and 02, are displayed as part of the boronic acid
`portion of 1-9 (Figure SB), and in each of the structures,
`the hydroxyls make characteristic interactions with the
`enzyme (Table 3). The O1 hydroxyl always interacts
`with the main chain nitrogens of Ser64 and Ala318,
`acting as an acceptor. 01 also appears to donate its
`proton to form a hydrogen bond with the main chain
`oxygen of Ala318 (Figure 3B). The carbonyl oxygen of
`the ,8-lactams (10-13) was also observed to bind in this
`site. For the complexes with 10 and 13, the lactam
`carbonyl oxygen hydrogen bonds with the main chain
`nitrogens of Ser64 and Ala318. In addition to interac-
`tions with these nitrogens, a close contact to the main
`chain oxygen of Ala318 is observed in the complexes
`with 11 and 12.
`
`Hydroxyl Binding Site. An additional binding site
`for hydroxyl groups was also identified (Figure 3B). The
`other boronic acid hydroxyl, 02, of 1-9 also makes
`characteristic interactions with the enzyme (Table 3).
`The O2 hydroxyl hydrogen bonds with Tyr150 and a
`water molecule, Wat402.
`Neither hydroxyl binding site was identified compu-
`tationally by GRID or X-SITE, owing to the close
`proximity of Serfi-40}: to the O1 and 02 hydroxyls of the
`boronic acid (Table 4). Contouring at very low levels in
`GRID resulted in a cage forming around the Oy atom,
`showing that probes would not be predicted to bind
`within the van der Waals radius of this atom. MOSS
`
`did not identify the 02 site but did correctly identify
`the O1 hydroxyl site. A methanol probe (ranked 36 out
`of 199 possible) was positioned such that it interacted
`with the main chain nitrogen and oxygen atoms of
`Ala318, and its RMSD from the experimental position
`of the O1 hydroxyl was 0.6 A.
`Hydrophobic Binding Site. Another binding site
`identified on AInpC recognizes hydrophobic groups. This
`site is composed of residues Leul 19 and Leu293, which
`form a hydrophobic patch in the active site. Phenyl rings
`were observed to form van der Waals interactions with
`
`ANACOR EX. 2025 - 8/15
`
`
`
`3228 Journal of Medicinal Chemistry, 2002, Vol. 45, No. 15
`
`Powers and Shoiehet
`
`these residues, as observed in the complexes of AmpC
`with 6, 8, and 129°'24'9‘”’ [Table 1; Figure 3C). In the
`complexes of AmpC with the ,6-lactams 10 and 13, the
`carbacephem and oxacephem rings, respectively, inter-
`act with the leucines?” In other structures, a methyl
`group is within van der Waals distance to these resi-
`dues, as with 9 and 1133 (Figure 3C).
`GRID and X-SITE predicted this hydrophobic binding
`site (Table 4