3928
`
`J. Aifcd. Chern. 1998, 41, 3928-3939
`
`Articles
`
`Investigations of Neurotrophic Inhibitors of FK506 Binding Protein via Monte
`Carlo Simulations
`
`Nlichclle L. La1nb and Willia1n L. Jorgensen*
`
`Dcpartrnent of Chcn1ist1y, }"ale Unil'ersity, 1'le1v Haven, Connecticut 06520-8107
`
`Rccci1'cd JantJa1y 27, 1998
`
`The binding and solution-phase properties of six inhibitors of FK506 binding protein (FKBP 12)
`\Vere investigated using free energy perturbation techniques in Monte Carlo statistical
`n1echanics si1nulations. These noninununosuppressivc n1olccules are of current interest for
`their ncurotrophic activity \Vhen bound to FKBP 12 as \Vell as for their potential as building
`blocks for che1nical inducers of protein din1erization. H.clativc binding affinities \Vere con1puted
`and analyzed for ligands differing by a phenyl ring, an external phenyl or pyridyl substituent,
`and a pipecolyl or prolyl ring. Such results are, in general, valuable for inhibitor optiinization
`and, in the present case, bring into question sonic of the previously reported binding data.
`,.,
`
`Meo
`
`Introduction
`1'he a-ketoan1ide functionality of the innnunosup(cid:173)
`pressant natural product FI<S06 (Figure 1) is retained
`in 1nany of the highest affinity ligands that have been
`developed to inhibit the rota1nase (cis-trans peptidyl(cid:173)
`prolyl ison1erase, or PPlase) activity 1 of the FI<S06
`binding protein (FKBP12, MW= 12 kDa).2 Originally,
`interpretation of the crystal structure of Fl(506-
`FI<BP12 led to the belief that the a-ketoan1ide n1imics
`a t\visted-an1ide transition state of peptide bond iso1ner(cid:173)
`ization, although an endogenous substrate for FI<BP12
`had not been discovered. It \vas thought that blockage
`of the iso1nerase active site prevented n1odification of
`do\vnstrean1 proteins necessary for T~cell activation, and
`this \Vas the source of the observed in11nunosuppression.
`A sirnilar n1echanis1n had been proposed for the activity
`of the undecapeptide cyclosporin A (CsA). \Vhich inhibits
`the PP!ase cyclophilin, although neither the natural
`products nor the proteins are ho1nologous. Ho\vever,
`evidence that rota1nase inhibition \Vas not sufficient for
`itnn1unosuppression soon began to niount.3 Rapan1ycin
`(Figure I), another fungal 1nolecule structurally shnilar
`to Fl(506, inhibited FI<BP12 but appeared to influence
`a later stage of the '!'-cell cycle. Schreiber and co(cid:173)
`\vorkers4 1nade a significant contribution \Vith the syn(cid:173)
`thesis of a n1olecule \Vhich retained the Fl<BP 12 binding
`do1nain of FI<S06 and rapan1ycin (pyranose ring, a-ke(cid:173)
`toan1ide, pipccolate ester, and cyclohexyleth(en)yl groups).
`but in \Vhich the 1nacrocycle \Vas contracted. This
`n1olecule \Vas a rota1nase inhibitor but did not prevent
`1'-cell proliferation.
`It later becan1e clear that the fonnation of an in1n1u(cid:173)
`nosuppressant-hn1nunophilin co111plex results in a gain
`of function for the protein. 'fhe CsA-cyclophilin and
`FI<S06-Fl<BP12 pairs each present a recognition sur(cid:173)
`face to the calcium-dependent, serine/threonine phos(cid:173)
`phatase, calcincurin (CN).5 The FK50G-FKBP12 com(cid:173)
`plex binds at least 10 A fro1n the active site of CN and
`
`10
`
`0
`
`OH r
`" !
`
`w~
`
`FK-506
`
`0
`
`,,,
`
`10
`
`0
`
`0
`
`H
`
`f>
`
`' " I
`
`~
`r
`
`0
`
`0
`
`QH
`'¥
`
`"'
`
`0
`
`Rapamycin
`Figure 1. Structures and atotn nutnbering for the ilnn1uno(cid:173)
`suppressants FK506 and rapamycin.
`
`$0022-2623(98)00062-4 CCC: SIS.OD © 1998 Atnerican Cheinical Society
`Published on Web 09/19/1998
`
`Roxane Labs., Inc.
`Exhibit 1019
`Page 001
`
`

`

`f\1eurotrophic Inhibitors of FK506 Binding Protein
`
`Journal of A'1edicinal Che1nistiy, I 998, Vol. 4 I, No. 2 I 3929
`
`Table 1. Exprrl!nental Activities for Selected FKBP 12 Ligands
`
`con1pd
`
`stnicture
`
`rotarnase
`Ki. n?-.1
`
`neurite outgrowth
`EDs 0, nM"
`
`0.3
`
`250', 11 O'
`
`300
`
`130'. J 65'
`
`8.5
`
`52'
`
`53
`
`0.05
`
`2
`
`3
`
`4
`
`5
`
`6
`
`•1 Data on ncurltc outgrowth fro1n chick dorsal root ganglia reported In ref 2. Ii Data frmn Guilford Pharmaceuticals, refs 2 and 59.
`'"Data frmn S1nithKllnc Beecha1n, refs 13 and 16.
`
`1nust block binding of subsequent phosphorylated pro(cid:173)
`teins and thus the T-cell signaling path\\'ay.6·7 Reports
`of the association of ca\ciun1 channels containing -Leu(cid:173)
`Pro- sequences \Vi th both FI<BP 12 and CN are filling
`in another long-standing piece of the FI<BP12 puzzle,
`as these may represent endogenous "ligands" for FI<BP12
`1ni1nicked by FI<506.8 In contrast, rapa1nycin-FI<BP12
`interrupts a distinct signaling cascade through its
`interaction \Vith another protein, generally tenned
`FRAP (FI<BP-rapa1nycin-associated protein).9- 11 A
`crystallographic structure of this ternary co1nplcx con(cid:173)
`finns the recognition require1nents for rapainycin. 12 In
`both Fl< BP 12 ligands. it is the portion of the 1nacrocycle
`opposite the a-ketoan1idc-pipecolic acid 1noicty, the
`~effector" region, \vhich contacts calcineurin.
`As part of an effort to design lo\V 1nolecular \Veight
`PPiase inhibitors as scaffolds for the iln1nunosuppres·
`sive effector con1ponents, the crystal structure of
`1-FI<BP12 (Table l) \Vas solved at S1nithl(line Bcechan1
`in 1993.13 Figure 2 sho\vs the binding 1node revealed
`for the a-ketoa1nide and pipecolyl portion of 1. The keto
`carbonyl (04) contacts aro111atic hydrogens of Tyr26,
`Phe36, and Phe99 , and the pipecoline ring sits over 1'rp59,
`The 3-phcnylpropyl 1noiety binds in the solvent-exposed
`FK506-cyclohexyl groove of FKBP12 between lle 56 and
`
`Figure 2. Position of co1npound 1 (yello\v) in the aro1natic
`binding pocket of FKBP12 (green). 13 Molecular graphics hn(cid:173)
`ages \Vere produced using the fvlidasPlus soft\vare systen1 froin
`the Cosnputer Graphics Laboratory, University of California,
`San Francisco.r.o
`Tyr-82, and these residues for1n hydrogen bonds \Vith the
`ester (02) and a1nide (03) carbonyl oxygens of the
`ligand. 'fhe I-phenyl substituent interacts \Vith Phe46
`and the tertiary pentyl g1·oup of the inhibitor. A
`
`Roxane Labs., Inc.
`Exhibit 1019
`Page 002
`
`

`

`3930 Journal ofAiedicinal Che111ist1y, 1998, Vol. 41, f•lo. 21
`
`La1nb and Jorgensen
`
`Schc1ne 1
`
`FKBP·A
`
`FKBP + B
`
`FKBP · B
`
`--;
`ii Ge
`
`Figure 3. FKBP 12-bound confonnation of 1 (yello\v) overlaid
`\Vith that of FKSOG (red). 13·15
`
`con1parison of the bound conforn1ation of 1 and FI<SOG
`presented in Figure 3 detnonstrates that this n1ode is
`consistent \Vith that found in crystallographic structures
`ofFK506-FKBP12 and rapamycin-FKBPlZ.14.lS How(cid:173)
`ever, the ability to fonn hydrogen bonds to the Glu54
`carbonyl observed in co1nplexcs \Vith FI<S06 (C24-0I-l)
`and rapan1ycin (C28-0H) is not present in this ligand.
`Binding patterns sitnilar to those for 1 tnay be expected
`for compounds 2 and 3 (Table 1) as \Vell. 13·16 An ex(cid:173)
`cellent analysis of FI<BP12-ligand interactions, includ(cid:173)
`ing discussion of previously unpublished ato1nic struc(cid:173)
`tures, is included in a revie\v of protein-ligand recog(cid:173)
`nition 1notifs by Babine and Bender. 11
`An additional activity for rotarnase inhibitors of this
`class has expanded interest in these con1pounds beyond
`their potential in itn1nunosuppressant drug design. As
`revie\ved recently by I-Ia1nilton and Steiner, 2 FI<S06 has
`been sho\vn to induce the regeneration of damaged
`nerves in anhnal 1nodels of Parkinson's and Alzheiiner's
`diseases. Furthennore, the enriched concentration of
`Fl<BP12 in neurons has been associated \Vith nitric
`oxide synthesis, ncurotrans1nitter release, and neurite
`extension. Potent, nonhnn1unosuppressive FI<BP12
`ligands, such as V-10,367 18- 20 and GPl-1046 (6, Table
`
`o Ro
`MeO~o:
`
`Meo
`
`OMe
`
`I ...:
`N
`
`V-10,367
`
`1).21 - 23 are able to pro1note neuronal gro\vth in vitro and
`in vivo \Vithout the addition of exogenous gro\vth factors.
`They have a better therapeutic potential than gro\vth
`factors in that they arc orally bioavailable and able to
`cross the blood~brain barrier. The requireinent of
`binding to FI<BP12 for neuronal activity has also been
`de1nonstrated, but there is no linear relationship bc(cid:173)
`t\veen rota111ase inhibition and activity in neuronal
`cclls.2 FI<BP12 binding is apparently necessary but not
`sufficient for stirnulation of nerve gro\vth, suggesting
`
`that, as in T-cclls, the complex may 1nodify the function
`of an additional target.
`Another use of this class of Fl< BPI 2 ligands has also
`e111erged. The ability of the inununosuppressants 'to
`induce protein heterodin1erization and the kno\vledge
`of ligand n1odifications that prevent this association has
`been exploited for control of cellular signaling patlnvays,
`protein translocation, and gene activation.24·25 Target
`proteins are first artificially attached to the in1n1uno(cid:173)
`philins (FKBP12 or cyclophilin), CN, or FRAP. The
`ligands then1selves or synthetic ho1no- or hctcrodin1crs
`of FI<S06, CsA, or rapa1nycin then bring their protein
`partners together, resulting in the proxhnity of the tar(cid:173)
`get proteins and trans111ission of signa1.24 - 3o Recently,
`di111crs of 7 have been used to effect cellular apoptosis
`and to induce transcription, again \Vithout the hn1nu(cid:173)
`nosuppressive effecls of further binding to calcineurin. 31
`This technique of ~che1nically induced di1nerization",
`used \Vith s1nall, cell-per111eable n1oleculcs such as 7, is
`
`n
`0 t"'y0~oMo
`,,Co oQ-
`
`OMo
`
`"- I
`
`Linker
`
`7
`
`designed to have application in cellular gene therapy.
`Given the diverse biological applications of these
`a-ketoainidc ligands and that only slight differences in
`structure can have profound effects on activity, \Ve have
`used theoretical techniques to probe the binding of
`cornpounds 1-6 (Table l) at the aton1ic level, in both
`structural and energetic tcrn1s. Previous sitnulations
`of FI<BP12 have addressed the rotan1ase n1echanisn1
`applied to peptide substrates32 ·33 and the hnportance of
`Tyr82 in binding FI<506.3'1 Our current approach has
`focused on free energy perturbation (FEP) calculations,
`using Monte Carlo (MC) n1ethods rather than n1olecular
`dynan1ics (MD) for sa1npling. Con1puted relative free
`energies of binding, \Vhich are obtained fron1 shnula(cid:173)
`tions of the ligands in solution and bound to the protein,
`n1ay be con1pared \Vi th those obtained fro1n experitnen(cid:173)
`tal binding constants (Sche1ne 1), A veragcs of the
`co1nputed structures 1nay then be used to analyze the
`origin of the differences in binding affinities.
`'fhe MC n1cthod used here has been validated \Vi th a
`study of benzamidine inhibitors of trypsin35 and \Vas
`further applied to the analysis of orthogonal CsA(cid:173)
`cyclophilin pairs as co1nponents of a syste1n for chetni(cid:173)
`cally induced dhnerization,36
`'I'he present study is
`ain1cd at understanding factors that influence the
`binding of 1 and its analogues. In particular, the effects
`of ren1oval of the I-phenyl group, conversion of the
`3-phenyl to 3-(3-pyridyl), and ring contraction of the
`
`Roxane Labs., Inc.
`Exhibit 1019
`Page 003
`
`

`

`f1lcurotrophic Inhibitors of FK506 Binding Protein
`
`.Journal ofJ\-fcdicinal Chen1istry, 1998. Vol. 41, No. 21 3931
`
`pipecolyl ring to prolyl are exan1ined. There are dis~
`crcpancles in the binding data rron1 the t\VO experiinen(cid:173)
`tal sources, as indicated by the results for 2 and 3 in
`·rable I. Fron1 the crystal structure for 1 bound (Figure
`2), the pyridine nitrogen of 3 is anticipated to be solvent
`exposed. Thus, it \vould nonnally not be expected to
`favor the lo\ver dielectric cnviron1nent of a protein (E ~
`4) over that of bulk \Yater (E ~ 80),37 in contrast to the
`binding results fron1 Guilford Pharn1aceuticals. This
`\Vas pursued through co1nputations for the 2, 3 and 5,
`6 pairs.
`l-Ia1nilton and Steiner have also pointed out
`that 5 and 6 are the first exatnples ofprolyl cotnpounds
`that bind better than their pipecolyl analogues, but the
`high affinity is attributed only to "itnproved design" .2
`To investigate further, differences in free energies of
`binding \Vere co1nputed for t\vo pairs of pipecolyl and
`prolyl ligands. Co1npounds 2 and 5 represent the
`unusual case \Vith the prolyl ligand (5) as the better
`inhibitor. Co1npounds 1 and 4 represent the n1ore
`co1nn1on situation in \Vhich the presu1nably tnorc hy(cid:173)
`drophobic pipccolyl ligand (1) has higher affinity for
`FKBP12.
`
`Con1putational Details
`
`Paran1ctrization and Initial MC Sitnulations.
`'!'he crystal structure of l-FI<BP12 at 2.0 A resolution 13
`fron1 the Brookhaven Protein Data Bank38 (entry l fkg)
`\Va.s used as the starting point for the sin1ulations. The
`co1nputational protocol for the MC shnulations \Vas the
`san1e as in previous applications.35·36 The good precision
`that is obtainable for free energy changes \Vith this
`n1ethodology \Vas addressed extensively in ref 35. 'fhe
`MC sa1npling included variation of all bond angles and
`dihedrals of the ligand and protein side chains as \Vell
`as overall rotation and translation of the ligand and
`\Vater n1olecules. The protein backbone atoms \Vere held
`fixed in their crystallographic positions. 1'his 1nakes the
`i'v1C simulations 111ore rapid, and the approxhnation is
`justified for FI<BP 12. f{estricted backbone 1notion on
`the picosccond tiine scale has been noted for native
`FKBP 12, 39 and ligand binding further rigidifies the
`protein structure, as de1nonstratcd by the close rese111-
`blance an1ong ato111ic structures of FI<BPl 2 in nuinerous
`FI<BP12-ligand coniplexes. 17 To be consistent \Vith
`prior MD calculations on the FI(506-FKBPI2 syste111,40
`all 79 residues \Vithin 12 A of FI<506 in its cocrystal
`structure \Vith FI<BP12 14 \Vere sa1npled. This provided
`a greater nun1bcr of 1noving side chains than \vould be
`found in a 12 A region around 1.
`The OPLS united-ato111 force field 41 \Vith all-aton1
`aron1atic groups42 provided n1ost paran1eters for the
`protein: paran1eters for the inhibitors also ca1ne fro111
`this source and fron1 a previous MD study of FI<S06.43
`1\ listing of para1neters for the inhibitors is provided in
`the Supporting Inforn1ation. 'I'he torsional paran1eters
`for the an1ino acid residues \Vere derived fro1n fitting to
`torsional energy profiles obtained fro1n ab initio calcula(cid:173)
`tions \Vith the 6-31G" basis sct. 44 Any n1issing paran1-
`eters \Vere derived by fitting to i\1M2·15 energy profiles,
`\Vhich \Vere generated using Macromodel. 46 A scale
`factor of 112 \vas applied to all 1-4 nonbondcd interac(cid:173)
`tions. l·Iistidlnes 25, 87, and 94 arc kno\vn to be un(cid:173)
`protonated,47 and they \Vere designated as O-tauton1ers
`based on visual inspection. '1'his tauto1neric state has
`
`also been chosen in lvfD shnulations of Fl<BP 12-ligand
`con1plcxcs in solution.32·3·l.
`The unbound ligands and protcin--ligand cornplexes
`\Vere solvated \Vith 22 A spheres containing 1477 and
`939 TIP4P \\later 1nolecules, respectively. A half(cid:173)
`hannonic potential \Vi th a 1.5 kcal/11101 A2 force constant
`\Vas employed to prevent \Vaters fron1 1nigrating a\vay
`fron1 the cluster. A 9 A residue-based cutoff \Vas used
`for all nonbondcd interactions; if any pair or ato111s fro1n
`t\VO residues \Vas \Vithin this distance, all nonbonded
`interactions bct\veen the residues \Vere included in the
`energy evaluation. The list of nonbonded interactions
`\Vas updated every 2 x 105 configurations during the
`sin1ulations.
`All f\1.onte Carlo shnulations \Vere perforn1ed \Vith the
`MCPRO progra111.'18 An advantage of using internal
`coordinate MC n1ethods is the ability to focus sa1npling
`on specific regions and degrees of freedon1 of interest.
`Consequently, bond lengths \Vere fixed to their crystal
`structure values, and aro1natic rings \Vere treated as
`rigid units. To prevent inversion at sp3 centers such
`as a-carbons and to enforce planarity of sp 2 centers for
`n1ore efficient san1pllng, hnproper dihedral angles \Vere
`not varied except as noted belo\v. Othcr\visc, all bond
`angles and dihedrals in the 111oving portion of the syste1n
`\Vere san1pled.
`The l\1C siinulations \Vere carried out for 25 °C on
`Silicon Graphics \Vorkstations and on a cluster of
`personal con1putcrs using Pentiun1 p1·ocessors. It 111ay
`be noted that the cxperhncntal results con1e fro111 an
`assay for rotan1ase inhibition. 49 This \videly used
`procedure for 1neasuring FI<BP 12 binding affinities is
`usually perforn1ed son1e\vhat bclo\v roo1n ten1perature,
`e.g., near 10 °C. 13 The solvent \Vas first sa111pled for 1
`111illion (A1J configurations to re1nove any highly repul(cid:173)
`sive initial contacts \Vith the solutes. Then, 81\1! con(cid:173)
`figurations \Vere perfonned to equilibrate the 1-FI<BP 12
`con1plex.
`'I'he sa1ne protocol \Vas follo\ved for 1 in
`solution, beginning \Vith the bound confor1nation taken
`fro1n the 1-FI<BP 12 structure. During equilibration,
`the conforn1ation of the bound ligand ren1aincd sim(cid:173)
`ilar to the crystal conforn1ation; ho\vcvcr, partial in(cid:173)
`version of the pipecolyl ring occurred in solution to
`S\vltch it fron1 a chair to a half-chair conforrnation
`(Figure 4). In gas-phase opthnizations of ligand 1 \Vi th
`the present force field, the adopted ring confortnation
`is favored by 1.3 kcal/Jnol. 1'his is likely an artifact of
`using the AMBER C2-N-CH bending para1neters \Vi th
`00 = 118°, \Vhich 'vas not designed for a piperidine
`ring.50 The difference is expected to have little effect
`on the con1putcd free energy changes since the niutated
`phenyl rings are not in contact \Vith the pipecolyl ring
`or, in the case of the ring contraction, the chair con(cid:173)
`fonnation \Vas enforced (vi de infra).
`Free energy changes \Vere calculated during the MC
`sinnilations according to standard procedures of statis(cid:173)
`tical perturbation thcory.5!-53 The difference in free
`energy of binding (6.6.Gti) for 111olccule B relative to
`nlolecule A (Schen1e 1) 1nay be obtained fro1n transfor(cid:173)
`n1ations of the ligands in solution and bound to the
`protein according to eq 1:
`
`FEP Situulation Protocol for 1-2. This perturba-
`
`Roxane Labs., Inc.
`Exhibit 1019
`Page 004
`
`

`

`3032 Journal ofi\-fedicinal Chernistry, 1998, Vol. 41, /\'o. 21
`
`Larnb and Jorgensen
`
`Figure 4. Stereovie\vs of unbound ligand I. The initial geoinetry fro1n the 1-FKBP 12 crystal structure has the pipecolic ester
`substituent in an axial confonn~tion (top). Subsequent equilibration resulted in partial inversion of the ring {botto111).
`
`tion involved the rernoval of the !-phenyl ring of 1 to
`obtain 2. The aton1s of the phenyl group \Vere converted
`to "du1n1ny" ato1ns \Vithout charge or Lennard-Jones
`paran1eters, and the length of the bond connecting the
`substituent to the re1nainder of the ligand \Vas reduced
`to 0.65 A \Vith all other phenyl ring bonds reduced to
`0.35 A. The transforn1ation of 1-2 \Vas carried out in
`13 \Vindo\vs \Vith double-\vide san1pling, \Vhich yield 26
`free energy incre111ents. 51 A coupling paran1eter, J., \Vas
`e1nployed such that ), = 0 corresponds to the initial
`state, 1, and A= 1 corresponds to the final state, 2. 1'he
`first six \vindo\VS used ~). = ±0.025, \Vhile the ren1ain(cid:173)
`ing \Vinclo\VS used AA = ±0.050. All \Vere equilibrated
`\Vi th 2-4M configurations of sa1npling; the last config(cid:173)
`uration of the previous \Vindo\v \Vas used to start the
`next one. Averaging \Vas done in batches of 2 x 105
`configurations, \Vith data collected over a total of 4-7 lvf
`configurations in each \Vindo\v. For subsequent analy(cid:173)
`ses of hydrogen bonding, an additional 1 Iv! configura(cid:173)
`tions \Vere generated at the endpoints of the shnula(cid:173)
`tions.
`FEP Shnulation Protocol for 2----..3, 5-6. The next
`transforrnation addressed \Vas the conversion of a phen(cid:173)
`yl 1noiety to a 3-pyridyl ring.
`'fhis perturbation is
`straightfor\vard: the analogous perturbation of benzene
`to pyridine had been perforrned in the develop1nent of
`OPLS all-ato1n (OPLS-AA) paran1eters for pyridinc. 54
`As before, the standard phenyl ring structure \Vas
`transforrned to a pyridine geon1etry dctennined froin
`1nicro\vave expcrhnents. 51 A rnodel of 5 \Vas required
`prior to the conversion of fr-----6 and \Vas obtained by
`1napping a prolyl ring onto the final structure of 2 fro1n
`the 1-2 FEP calculation. In each shnulation, the prolyl
`or pipecolyl ring \Vas flexible. 'fhe perturbation protocol
`for these calculations \Vas slightly 111odified fro1n that
`
`used for 1-2 to take advantage of the acquisition of a
`nc\V parallel cotnputing syste1n \Vithin our laboratory.
`Seven double-\vide \Vindo\vs \Vere run in parallel, \Vi th
`4-BA1 configurations sa1nplcd during the equilibration
`phase and \Vith data collected over 4- l2M configura(cid:173)
`tions. A gas-phase FEP calculation \Vas also perfonned
`for 5-6 to allo\v estin1ation of the relative free energies
`of hydration of the t\VO ligands.
`FEP Sitnulation Protocol for 2-5, 1-4. In our
`experience, perturbations bet\veen different cyclic sys(cid:173)
`terns require 1nuch care to ilnplen1ent and can be
`particularly s\o\v to converge. The necessity of account(cid:173)
`ing for both changes in bonded and nonbonded interac(cid:173)
`tions \Vithin the ring as one aton1 disappears makes this
`a technically difficult perturbation. One \Vay to sin1plify
`the present calculations is to drive the ring fro1n one
`fixed six-n1en1bered ring confonnation to a fixed five-
`1nen1bered ring confonnation, "disappearing" the re(cid:173)
`n1aining atoin and sin1ultaneously reeling it in to\vard
`the others. For this rigid perturbation, changes in
`energy \Vi thin the ring need not be n1onitored, as these
`intraligand differences should be very sitnilar in each
`environn1ent (bound and unbound). Ho\vever, other
`possible confor1nations for the rings \vould not be taken
`into account, and the results could be sensitive to the
`path chosen.
`The silnulations for the unbound and bound trans(cid:173)
`fonnations \Vere started fron1 the final bound confonna(cid:173)
`tions of 2-FI<BP12 or 1-FI<BP 12 above \Vi th the chair
`conforn1ation for the pipecolyl ring. The final prolyl ring
`geo1netry \Vas obtained fron1 a gas-phase optitnization
`of the bound confonnation of 2 \Vith one ring ato1n
`converted to a du1n1ny atoin, as illustrated in Figure 5.
`Other than for the internal structures of the pipecolyl
`and prolyl rings, the sa1npling for the ligands included
`
`Roxane Labs., Inc.
`Exhibit 1019
`Page 005
`
`

`

`/\leurotrophic Inhibitors of FK506 Binding Protein
`
`Journal of Aledicinal Che1nist1y, 1998, Vol, 41. !\fa. 21 3933
`
`Figure 5. The initial and final ring confonnatlons of the 2-5 perturbation. The dununy at01n in 5 is noted Du.
`lS,0 r·-- • - 1-- ·•
`t "
`
`'.:I
`
`,------•----;--
`
`--i-
`
`. --- :::> - --
`
`_:: =~ -'---~~~-
`
`5.0
`
`,------~r-
`
`b
`
`all bond angles and dihedral angles, as before. Prior to
`the FEP calculation, the unbound ligands, 2 and 1, \Vere
`each resolvated and relaxed \Vi th 4.4M configurations
`of solvent-only san1pling, follo\ved by BM configurations
`of full equilibration. T\venty-one \Vindo\vs \Vere used
`to perfor1n the ring contraction in s1nall steps. Fortu(cid:173)
`nately. convergence \Vas rapidly achieved \Vithin the 4M
`configurations perfonned for both equilibration and
`averaging.
`
`Results and [)iscussion
`Effect of the 1-Phenyl Group, 1-2. Re1noval of
`the 1-phcnyl rnoiety fro1n I is a large perturbation but
`a co1nputationally attractive choice given the available
`struclural infonnation and the sizable difference in
`binding affinity (Table 1). The free energy change as a
`function of A fron1 the FEP calculations for the unbound
`and bound ligands proceeded sn1oothly (Figure 6a), \Vi th
`1'.G,q ~ 10.25 ± 0.31 kcal/mo! and 1'.GFKBP ~ 11.69 ±
`0.31 kcal/1110! (Table 2). 1'he resultant relative free
`energy of binding (t\i\Gb) of 1.4 kcal/mol obtained
`according to Sche1ne 1 is then in excellent agree1nent
`\Vith the experilnental observations of 1.4-1.6 kcal/Inol
`(Table 2).
`A con1parison of the averaged structures fro111 the
`sin1ulations \Vith the crystal structure yields several
`interesting observations. First, the average root-mean(cid:173)
`squarcd (rrns) deviations for non-hydrogen alo1ns be(cid:173)
`t\veen the structures sa1npled for the co1nplex of 1 and
`the crystal structure 13 \Ve1·e co1nputed. The average nns
`for the aton1s in the side chains that \Vere varied is 0.7
`A, and the average nns for the ligand 1 is 1.1 A. The
`corresponding 1naxhnu1n nns values for individual
`structures did not exceed 0.8 and 1.4 A, and the values
`at the end of the MC run \Vere 0.7 and 0.9 A. Thus, the
`MC san1pled .structures did not drift far fro1n the crystal
`structure, though sonic differences emerged. In the
`n-ketoan1ide region of the ligand, the crystallographic
`interaction of Oil \Vith an <:-hydrogen of Phe99 is not
`1naintained, but there is a frequent interaction of the
`S-hydrogcn \\1ith the atnide carbonyl oxygen (03). 'I'he
`hydrogen bonds to llc56 and Tyr82 are unaffected by the
`perturbation {Figure 2). Within con1pound 1, the I-phen(cid:173)
`yl and isopentyl groups ren1ain in contact. Ho\vevcr,
`the I-phenyl group n1oves a\vay so1ne fro1n Phe46 ; the
`shortest contact bet\veen aro1natic carbons increases
`fron1 4.5 A in the crystal structure to 5.6 A in the
`silnulation. In both the crystal structure and fro1n the
`
`-5.0 ~-~~--~-~-~--~-~-~-~~
`
`:::r
`
`'°
`
`-100
`
`:: r-~·
`
`~~~~=
`,.,; ''"'"''''" ~ "-,;
`;o;..
`--~"'
`&~-
`.,,"'-, .. ,,.
`w·
`,,lf·
`• .. ""
`
`··;;· .. ;,-,,,
`
`-15.0
`M
`
`·-
`
`OJ
`
`U ~ U
`
`U M ~ U
`
`U
`
`I•
`
`Figure 6. (a) Free energy profile for the transfonnation of
`1-2, \Vith error bars for each \Vindo\v sho\vn. (b) Profiles for
`2-3 and 5-6 (squares). (c) Profiles for 2-5 and 1-4
`(squares). Solid lines represent the unbound sinnliations, and
`the dashed lines result fron1 shnulations of FKBP12-ligand
`cotnplexes.
`
`'
`
`MC siinulations, the I-phenyl and isopcntyl groups pack
`\Vell into the hydrophobic pockel lhat is outlined by
`Phe46, Phe36, Jlc9o, Ile9 1, Tyr82 , and His87 (Figure 2). Loss
`of hyrdrophobic contacts upon ren1oval of the I-phenyl
`group is unfavorable for binding.
`'!'he sin1ulations also suggest that son1e specific
`contacts \Vith the I-phenyl group 1nay be relevant. As
`highlighted in Figure 7, the I-phenyl substituent 1nakes
`aryl CI-1· .. Q contacts \Vi th the backbone oxygen of Glu54
`and the Tyr82 hydroxyl oxygen, and it has an a1nino(cid:173)
`aro1natic interaction \Vith the c-nitrogen of I-lis87 . While
`the interactions of the I-phenyl group in 1 \Vith Tyr82
`and Glu 54 are found in n1ost of the structures analyzed,
`the interaction \Vith I-Iis87 occurs in only 29% of the
`analyzed structures. Average distances and frequencies
`
`Roxane Labs., Inc.
`Exhibit 1019
`Page 006
`
`

`

`3934 Journal ollvledicinal Chen1is1ry. 1998. Vol. 41, /\'o. 21
`
`La1nb and Jorgensen
`
`Table 2. Experin1enta\ Binding Free Energies, Calculated Free Energy Changes, and a Con1parison of Experimental and Calculated
`Relative Binding Free Energiesa
`
`exptJb
`
`ca led
`
`exptlc
`. - -
`"" c,,
`1.6, 1.4
`-0.4, 0.3
`-1.0
`-1.1
`0.7
`
`ca led
`MC,,
`1.4
`0.8
`0.9
`-1.7
`-1.9
`
`A-B
`1-2
`2-3
`5-6
`2-5
`1-4
`
`6.GA
`-10.6, -10.9
`-9.0, -9.5
`-10.1
`-9.0, -9.5
`-10.6, -10.9
`
`"'~
`-9.0, -9.5
`-9.4, -9.2
`-11.1
`-10.l
`-9.9
`
`f..,,GFKBP
`f..,,Gaq
`11.69 ± 0.31
`10.25 ± 0.31
`2.32 ± 0.08
`1.52 ± 0.09
`1.48 ± 0.07
`0.59 ± 0.08
`-10.07 ± 0.10
`-8.36 ± 0.09
`-8.76 ± 0.11
`-6.86 ± 0.11
`.~All free energies in kcal/11101. b Absolute binding free energies are derived front rota1nase inhibition data given In Table I, using 11C
`=/?Tin K; and T= 25 °C (298 K). c Relative binding free energies are only listed \Vhen experi1nenta\ data frmn the sa1ne source 1nay be
`con1pared.
`
`tion, 2-<~ and s--..6. Experhnental data fro1n Guilford
`Phannaceuticals suggests that 111odification to a pyridyl
`substituent i111proves binding of 3-phenylpropyl con1-
`pounds, although in one case this is contradicted by data
`obtained at Stnithl(linc Bcccharn (Table 2). Considering
`the ca. 4 kcal/n1ol n1orc favorable free energy of hydra(cid:173)
`tion of pyridine than of benzene54 and assun1ing that
`the pyridyl co111pounds bind in a n1anncr shnilar to 1
`in the solvent-exposed region of the binding pocket, the
`phenyl to pyridyl conversion \vould be expected to
`decrease binding affinity. Sin1ply put, a dipole is better
`solvated in a n1ediun1 \Vith a higher average dielectric
`constant.
`First, the salvation of the ligands \Vas addressed. 'fhe
`aqueous transforn1ation of z-..3 resulted in a free energy
`difference of 1.52 ± 0.09 kcal/n1ol, son1e\vhat larger than
`that for 5~6 (0.59 ± 0.08 kcal/mol). The gas-phase
`transfonnation of 5 to 6 yielded a free energy change of
`3.15 ± 0.04 kcalhnol, \Vhich cornbines for a net relative
`free energy of hydration of -2.6 kcal/1110! favoring 6.
`This is reasonable given the previous benzene to pyri(cid:173)
`dine results,5 4 and the larger, flexible ligand. Hydrogen
`bonds to \vater \Vere shnilar for both sets of ligands: as
`expected, the pyridyl nitrogen provides an additional
`acceptor site in 3 and 6 (Table 4).
`The free energy profiles for both pairs of unbound and
`bound perturbations are displayed in Figure 6b. They
`are again notably sn1ooth. As seen fro1n 1'able 2, the
`transfonnations in the protein are ca. 1 kcalhnol less
`favorable than those in solution, and the net result is a
`consistent preference for the phenylpropyl ligands. No
`direct protein contacts arc 1nade by the pyridyl nitrogen
`aton1; its interaction \Vith \vatcr is n1aintained ('fable
`4). For this hydrogen bond, the opthnal interaction
`energy is -6.2 kcal/Jnol, 51 n1uch stronger than an aryl
`Cl-1···0- or N- interaction.
`I·lo\vever, the pyridyl
`nitrogen does participate in a hydrogen-bonding net(cid:173)
`\vork \Vi th \Vater n1oleculcs and carbonyl oxygens in the
`protein backbone. As reflected in Figure 9, a t\vo-\vater
`bridge bet\veen the oxygen of Val55 and the nitrogen is
`found consistently in 6-FKBP12. In 3-FKBP12 the
`interaction is often 1nediated by three \Vater n1olccules,
`\Vith a fc\v structures in \Vhich one 1nolecule also bridges
`to the Gln53 carbonyl oxygen (Figure 9). The VaJ55
`oxygen has been noted previously as a consensus
`hydration site in FI<BP12-ligand con1plexes.57 Crystal(cid:173)
`lographic \Vatcrs from the 1-FI<BP12 structure 13 \Vere
`not explicitly included in the calculations, so it is
`gratifying that this interaction is established during the
`MC shnulations. Water surrounding the ligand in the
`binding pocket can also bridge longer distances: for
`exa1nple, in 6-Fl<BP12, four 111oleculcs link the hy-
`
`Figure 7. Intennolecular aryl CH·"N, 0 contacts \vith His87
`Tyr'2 , and Glu 54 1nade in 1-FKBP12 that are lost on trans(cid:173)
`fonnation to 2-FKBP 12. One representative configuration
`frmn the Monte Carlo shnulation is illustrated.
`
`,
`
`of occurrence of these and other key interactions be(cid:173)
`t\vccn the ligands and FI<BP12 are sun1n1arized in
`1'able 3. Such aryl interactions are con1111only observed
`in protein crystal structures, and their orientational
`distributions have been analyzed.55,56 To provide a
`sense of the strength of the aryl CH···X interactions
`\Vith the OPLS-AA force field, gas-phase opthnizations
`for co1nplexes of benzene \Vith phenol. hnidazole, N-
`111cthylacctan1ide, and \Vater yielded the interaction
`energies sho\vn in Figure 8. With the hydrogen bonds
`constrained to be linear, the intrinsic interaction ener(cid:173)
`gies arc in the 1-2 kcalhnol range.
`In both cornplexes, the 3-phenylpropyl n1oicty rcn1ains
`in the FJ(506-cyclohexyl region of FI<BP 12, and there
`is usually a \Vater 1110\ccule \Vell-positioned on the face
`of one or both phenyl rings in 1 (Figure 7). For
`reference, in the gas phase, the opthnal JT hydrogen bond
`bet\veen benzene and a \Vater n10\ecule has an interac(cid:173)
`tion energy of ca. -3.5 kcalhnol.42 An arornatic hydro(cid:173)
`gen of the 3-phenylpropyl group often interacts \Vi th the
`carbonyl oxygen of Val55 in 1-FI<BP12; this contact is
`shorter and 1nore frequent in 2-FI<BP12 (Table 3). In
`addition, \Vhile the 1-phcnyl ring no longer interacts
`\Vith the Glu5'1 backbone carbonyl oxygen \Vhen the
`transfonnation to 2 is con1plete, the 3-phenyl ring shifts
`to pick up this contact. Considering these observations,
`it is likely that the inajor contribution to the \Veaker
`binding for 2 than 1 is the overall reduction in hydro(cid:173)
`phobic interactions and, possibly, the specific loss of the
`aryl CJ-I•"O, N contacts bet\veen the 1-phenyl group and
`'fyr82 and His