CA2237336C - Use of nuclear magnetic resonance to design ligands to target biomolecules - Google Patents

Abstract

The present invention provides a process of designing compounds which bind to a specific target molecule. The process includes the steps of a) identifying a first ligand to the target molecule using two-dimensional 15N/1H NMR
correlation spectroscopy; b) identifying a second ligand to the target molecule using two-dimensional 15N/1H NMR correlation spectroscopy; c) forming a ternary complex by binding the first and second ligands to the target molecule; d) determining the three-dimensional structure of the ternary complex and thus the spatial orientation of the first and second ligands on the target molecule; and e) linking the first and second ligands to form the drug, wherein the spatial orientation of step (d) is maintained.

Description

- = -USE OF NUCLEAR MAGNETIC RESONANCE
TO DF',SIGN l,IGAI~DS TO TARG~T RIOMO-,h'CUT,li.S

Technical Field of the Invention The present invention pertains to a method for the use of two--lim~n.cional lSN/lH
NMR correlation spectral analysis to design ligands that bind to a target bi--molecule.

Back~Fround of the Invention One of the most powerful tools for discovering new drug leads is random SCl~.,i,lg of synthetic chemi~l and natural product ~l~t~b~es to discover compounds that bind to a particular target molecule (i.e., the i~ntific~tion of ligands of that target). Using this method, ligands may be irl~tifi~d by their ability to form a physical association with a target molecule or by their ability to alter a function of a target molecule.
When physical binding is sought, a targe~ molecule is typically exposed to one or more compounds ~.u~e~,lt;d of being ligands and assays are ~,.Çol,lled to r~termine if complexes between the target molecule and one or more of those compounds are formed.
Such assays, as is well known in the art, test for gross changes in the target molecule (e.g., changes in size, charge, mobility) that in~ t~ complex formation.
Where functional changes are measured, assay conditions are est~hli~h~rt that allow for measurement of a biological or chen ic~l event related to the target molecule (e.g., enzyme catalyzed reaction, lcc~l~lo~ e(li~ enzyme activation). To identify an alteration, the function of the target molecule is d~,Le~ ihled before and after exposure to the test compounds.
Existing physical and functional assays have been used successfully to identify new drug leads for use in designing therapeutic compounds. There are, however, limitations inherent to those assays that con~p~{".ise their accuracy, reliability and efficiency.
A major shortcoming of exi~ting assays relates to the problem of "false positives". In a typical functional assay, a "false positive" is a compound that triggers the assay but which compound is not effective in eliciting the desired physiological response. In a typical phys*al assay~ a "false positive" is a compound that, for example, attaches itself to the target but in a non-specific manner (e.g., non-specific binding). False positives are particularly prevalent and problematic when screening higher concentrations of putative ligands because many compounds have non-specific affects at those concentrations.
In a sirnilar fashion, existing assays are plagued by the problem of "false negatives", which result when a compound gives a negative response in the assay but which compound is actually a ligand for the target. False negatives typically occur in assays that use concentrations of test compounds that are either too high (resulting in toxicity) or too low relative to the binding or dissociation constant of the compound to the target.
Another major shortcoming of existing assays is the limited amount of information provided by the assay itself. While the assay may correctly identify compounds that attach to or elicit a response from the target m-)lecl~ ., those assays typicaUy do not provide any i.lÇ ~ f;on about either specific binding sites on the target mnlecn1e~ or ~Llu~iluie activity relationships I~L~ l the compound being tested and the target molecule. The inability to provide any such inro~ Lion is particuIarly prohlem~tic where the scr~,~ g assay is being 5 used to identify leads for further study.
It has l ~,C~ y been suggested that X-ray crystallography can be used to identify the bin-1ing sites of organic solvents on macrom- l~cllles However, this method cannot clet~rmine the relative binding ~ffiniti~s at dirrclcl - sites on the target. It is only applicable to very stable target proteins that do not dGualulc in the presence of high a ncentr~tions of 10 organic solvents. Moreover, this approach is not a sc~ g method for rapidly testing many compounds that are ~hemic~lly diverse, but is limited to mapping the binding sites of only a few organic solvents due to the long time needed to ~Ic~.. ;l-~. the individual crystal structures.
Clompounds are screened to identify leads that can be used in tne design of new drugs that alter the function of the target biomolecule. Those new drugs can be ~llu~iLuli~ analogs of i~entified leads or can be conjugates of one or more such lead cc,lllyOullds. Because of the problems inherent to eYi~tin~ scr~ellulg m~,thotl.~, those methods are often of little help in ~lPsignin~ new drugs.
There cnntinlles to be a need to provide new, rapid, effic~i~nt accurate and reliable 20 means of screening compounds to identify and design ligands that specifically bind to a particular target.

Brief Sull~ of the lnvention In its principal aspect, the present invention provides a process for the design and 25 i~len ~; fic,~t;on of compounds which bind to a given target biomolecule. That process compri~es the steps of: a) identifying a first ligand to the target molecule using two-dimensional l~N/~H NMR correlation spectroscopy; b) identifying a second ligand to the target molecule using two--limt-n~ional 15N/lH NMR correlation spectroscopy; c) forrning a ternary complex by binding the first and second ligands to the target molecule; d) ~1~l~....i.1i.~g 30 the three tlimpncinnal structure of the ternary complex and thus the spatial ~rient~tion of the first and second ligands on the target molecule; e) linking the first and second ligands to form the drug, wherein the spatial orientation of step (d) is m~int~ine~l This aspect of the present invention uses the two-~limen~ional l5N/lH NMR
correlation spectroscopic screening process as set forth below to identify a first and 35 subsequent ligands that bind to the target molecule. A complex of the target molecule and two or more ligands is formed and the three-dimensional structure of that complex is determined preferably using NM~ spectroscopy or X-ray crystallography. That three-dimensional ~ ~ = - .

structure is use~ to ~1et~o.rmine the spatial u. ;~ ;on of the ligands re}ative to each other and to the target molecule.
Based on the spatial nri~nS~tion, the ligands are linked together to form the drug. The selection of an applupriate linking group is made by ~ g the spat;al ori~nt~tion of the ligands to one another and to the target molecule based upon principles of bond angle and bond length u~çulmaLion well known in the organic cl~ l art.
Thus, the molecular design aspect of the present invention comrri~es identifying a first ligand moiety to the target molecule using two--limencional lSN/lH NMR correlation spectroscopy; identifying subsequent ligand moieties to the target mn1~cl11e using two-~limt~ncional lSN/lH NMR correlation spectroscopy; forming a complex of the first and subsequent ligand moieties to the target molecule; ~~t~ g the three ~lim.on~ional structure of the complex and, thus, the spatial ori~ntAti~ n of the first and ~.ubse~luenL ligand moieties on the target molecule; and linking the first and subse~uGI~ ligand moieties to form the drug to ;lll~ the spatial nrit~nt~tinn of the ligand moi~o*~5 The i~ ntificAtion of subse~luçnt ligand moieties can be ywro~ ed in the ~hsçnce or presence of the first ligand (e.g., the target molecllle can be bound to the first ligand before being exposed to the test compounds for i~lentifir~tion of the second ligand).
The present invention further col~ pl~tes a drug ~lçeigne~ by the design process of this invention.
(~hemi~sll compounds can be screened for binding to a given target biomolecule by a process involving the steps of a) first g~,nG~ g a first two-dimensional lSN/lH NMR
correlation spectrum of a lSN-labeled target molecule; b) exposing the labeled target molecuLe to one or a nli~lul~, of chemi~l compounds; c) next, generating a second two-flim~on~ional lSN/lH NMR correlation spectrum of the labeled target molecule that has been exposed to 2~ one or a ll~l ue of compounds in step (b); and d) compAring said first and second two-~lim~ n~ional lSN/lH NMR correlation spectra to determine differences be~ween said first and said second spectra, the dirr~ ces identifying the presence of one or more compounds that are ligands which have bound to the target molecule.
Where the process screens more than one compound in step (b), that is, a ~ ~G ofcompounds, and where a diLrGlGIlce between the first spectrum generated from the target molecule alone and that generated from the target molecule in the presence of the mixture, additional steps are performed to identify which specific compound or compounds contained ~ in the ll-L~Lulc iS binding to the target molecule. Those additional steps comprise the steps of e) exposing the 15N-labeled target molecule individually to each compound of the mixture, f ) ~5 generating a two-dimensional 1 5N/lH NMR correlation ~e1LI Ulll of the labeled target molecule that has been individually exposed to each compound; and g) comparing each spectrum generated in step f) to the first spectrum generated from the target molecule alone to WO 97/18469 PCT~US96/18312 ~le/c~ . . .i ne ~;liLrc;r~llces in any of those co~ >a,~;l spectra, the diLrt;~ ces idcntiry"lg the ~r,;,cnce of a compound that is a iigand which has bound to the target molecule.Because the chemicAl shift values of the particular lSN/lH signals in the two-dimensional correlation ~e~;l ulll correspond to known specific locations of atomic groupings 5 in the target molecule (e.g., the N-H atoms of the arnide or peptide link of a particular amino acid residue in a poly-peptide), the s(;~ g process allows not only for the for irl~-ntifir~ation of which compound(s) bind to a particular target molecule, but also permit the dcti . .. ~i~.AIion of the particular binding site of the ligand on the target molecule.
The dissociation constant, KD, for a given ligand and its target molecule can bedet~rrnined by this process, if desired, by performing the steps of a) gent~r~ting a first two-dirnensional lSN/lH NMR correlation ~ ulll of a lSN-labeled target molecule; b) exposing the labeled target molecule to various cnn~en~T~ti~nS of a ligand; c) g I~-A~ a two-. li . . .~ ional IsN/lH NMR correlation ~e~ ulll at each conf~entratiQn of ligand in step (b); d) CO~ rAI ing each ~e;LIum f~om step (c) to the first spectrum from step (a); and e) cAl-~ulAting the dissociation constant bc~ ~,cn the target molecule and the ligand from those ~lirr~,~,nces according to the equation:

KD = ([P~O - X)([I~O - X3 x An advantageous cArability of the screening method is its ability to ~l~terrnine the dissociation constant of one ligand of the target molecule in the presence of a second molecule already bound to the ligand. This is generally not possible with prior art methods which employ "wet chemical" analytical methods of det~ormining binding of a ligand to a target molecule ~ub~LLate.
The process of det~rminin~ the dissociation constant of a ligand can be pulru~ ed in the presence of a second bound ligand. Accordingly, the 15N-labeled target molecule is bound to that second ligand before exposing that target to the test compounds.
The ability of the scree~ g method to de~.lll,lle not only the existence of binding between one ligand and the target molecule, but also the particular site of binding in the presence of a second bound ligand, permits the capability to design a drug that comprises two or more linked moieties made up of the lig~n(ls In a ~c~re~lc;d embodiment of the present invention, the target molecule used in the molecular design process is a polypeptide. The polypeptide target is preferably produced in recombinant form from a host cell tran~r~,l,ned with an ~ ,ssion vector that contains a polynucleotide that encodes the polypeptide, by culturing the transformed host cell in a medium that contains an ~ssimilAhle source of lSN such that the recombinantly produced polypeptide is labeled with lSN.

CA 02237336 l998-05-ll Brief ]~)eswl~,Lion of the Drawings In the drawings which form a portion of the specific~tion FIG. 1 shows a 15N/lH correlation ~e.iLlu,n of the DNA binding domain of ulliru~ ly 15N-labeled human papillomavirus E2. The s~e.iL, u-n (80 complex points, 4 scans/fid) was acquired on a 0.5 mM sample of E2 in 20 mM phosphate (pH 6.5), 10 mM
dithiothreitol (D~I) and 10% d~lt~..;lll.l oxide (D20).
FIG. 2 shows 15N/lH correlation spectra of the DNA binding domain of uniformly 5N-labeled human papillomavirus E2 before (thin multiple contours) and after (thick single o contours) addition of a final test compound. The final concentr~tion of compound was 1.0 mM. All other conditions are as stated in FIG. 1. Selected residues that show signific~nt changes upon binding are inAi~fltecl FIG. 3 shows lSN/lH-correlation spectra of the DNA binding domain of unirollllly15N-labeled human papillomavirus E2 before (thin multiple co~ u,~,) and after (thick single 15 contours) addition of a second test compound. The final concentration of compound was 1.0 mM. All other conditions are as stated in FIG. 1. Selected residues that show c; ~.... ric~
changes upon binding are indicated.
FIG. 4 shows 15N/lH correlation spectra of the catalytic domain of ulliLollllly ISN-labeled stromelysin before (thin multiple contours) and after (thick single ~;ollLoL-l~,) addition of a test compound. The final concentration of compound was 1.0 mM. The spectra (80 complex points, 8 scans/fid) were acquired on a 0.3 mM sample of SCD in 20 mM TRIS (pEI
7.0), 20 mM CaCl2 and 10% D2O. Selected residues that show ~ignific~nt changes upon binding are indicated.
FIG. 5 shows 1SN/lH correlation spectra of the Ras-binding domain of uniforrnly 15N-labeled RAF peptide (residues 55-132) before (thin multiple contours) and after (thick single contours) addition of a test co.,.~oulld. The final concentration of compound was 1.0 mM. The spectra (80 complex points, 8 scans/fid) were acquired on a 0.3 mM sample of the RAF fragment in 20 nM phosphate (pH 7.0), lO mM DTT and 10% D20. Selected residues that show ~ignifi~nt changes upon binding are inAic~teA
FIG. 6 shows lSN/lH correlation spectra of uniformly 15N-labeled FKBP before {thin multiple contours) and after (thick single contours) addition of a test compound. The final concentration of compound was l.0 mM. The spectra (80 complex points, 4 scans/fid) was acquired on a 0.3 mM sample of FKBP in 50 mM phosphate (pH 6.5), lO0 mM NaCland 10% D20. Selected residues that show significant changes upon binding are in~ te~.

3~ FIG. 7 shows a first depiction of the NMR-derived structure of the DNA-binding domain of E2. The two monomers of the symmetric dimer are oriented in a top-bottom fashion, and the N- and C-termini of each monomer are inAi(~.ateA (N and C for one monomer, N* and C* for the other). Shown in ribbons are the residues which exhibit .~ignifi{~nt chPmical shift changes (ao(1H)>0.04 ppm; ao(l5N) >0.1 ppm) upon binding to a first test compound. These residues co~ ond to the DNA-recognition helix of E2. Selected residues are numbered for aid in vic~ 7~tinn.
FIG. 8 shows a second depiction of the NMR-derived structure of the DNA-binding domain of E2. The two monomers of the ~y.~ . ;c dimer are ." ~~.~1~;1 in a top-bottom fashion, and the N- and C~-termini of each monomer are in~lic~te~l (N and C for one mnnt~m~r, N* and C* for the other). Shown in ribbons are the residues which exhibit eignific~nf chemical shift changes (a~(lH)~0.04 ppm; ao(l5N) >0.1 ppm) upon binding to a second test compound. These residues are located prirnarily in the dimer int~ ce region. Scle~ ed residues are numbered for aid in vieu~li7~tinn.
FIG. 9 shows a depiction of the NMR-derived structure of the catalytic domain ofstromelysin. The N- and C-termini are intli~te-l Shown in ribbons are the residues which exhibit cignif~ nt ch-qmic~l shift changes (ao(lH)>0.04 ppm; ~o(15N) ~0.1 ppm) upon binding to a test compound. These either form part of the S 1 ' binding site or are spatially ts ~ ~unal to this site. Selected residues are numbered for aid in vicll~li7~on.
FIG. 10 shows a ribbon plot of a ternary complex of first and second ligands bound to the catalytic domain of stromelysin.
FIG. 11 shows the correlation between the NMR hin-lin~ data and a view of the NMR-derived three-rlim~-ncinn~l structure of l?KBP.
FIG. 12 shows a ribbon plot of a ternary complex involving FKBP, a fragment analog of ascomycin, and a b~n7~nili~le compound.

Detailed Description of the Invention The present invention provides a rapid and ef~lcient method for ~l~ci~ninp ligands that 2~ bind to ~ d~eu~c targetmolecules.
T i~:~n~l~ are i(lentifitA by testing the binding of molecules to a target molecule (e.g., protein, nucleic acid, etc.) by following, with nuclear m~gnetic resonance (NMR)spectroscopy, the changes in chemic~l shifts of the target molecule upon the ~drlitinn of the ligand compounds in the ~l~t~k~e From an analysis of the cherni~zll shift changes of the target molecule as a function of ligand concentration, the binding affinities of ligands for biomolecules are also detf.rrnin~-A.
The location of the binding site for each ligand is ~leterrnin~ from an analysis of the chemical shifts of the biomolecule that change upon the addition of the ligand and from nuclear Ov~rh~ r effects (NOEs) bt;lw~en the ligand and biomolecule.
Infc" IllaLion about the structure/activity relationships between ligands i~.ontifi~ by such a process can then be used to design new drugs that serve as ligands to the target molecule. By way of exarnple, where two or more ligands to a given target molecule are nhfi~-l, a complex of those ligands and the target molecule is formed. The spatial ~" ;f~ ;nn of the ligands to each other as well as to the target molecule is derived from the three-~im~n~ional ~L~u~ e. That spatial nriPnt~tion defines the ~ t~nf~e l~lwee~l the binding sites of the two ligands and the ~ritont~ti~n of each ligand to those sites.
Using that spatial orientation data, the two or more ligands are then linked togeeher to 5 form a new ligand. T inking is accomplished in a manner that m~int~in~ the spatial orient~tion of the ligands to one another and to the target molecule.
There are numerous advantages to the NMR-based discovery and design processes ofthe present invention. First, because a process of the present invention id~ntifi~ gands by directly measuring binding to the target molecule, the problem of false positives is 10 ~ignifi~ntly recluced Because the presene process identifies specific binding sites to the target molecule, the problem of false positives reslllting from the non-specific binding of compounds to the target mc l~cllle at high co~ r, . ~ ;onc is kl i ~ t~ A
Second, the problem of false negatives is ~i~nific~ntly reduced because the present process can identify compounds that specifically bind to the target molecule with a wide range 15 of dissociation constants. The dissociation or binding constant for compounds can actually be ~leterrnin~d with the present process.
Other advantages of the present invention result from the variety and ~let~ fl data provided about each ligand from the discovery and design processes.
Because the location of the bound ligand can be ~ir~ fl from an analysis of the 20 chemical shifts of the target molecule that change upon the addition of the ligand and from nuclear Overh~ e.r effects (NOEs) between the ligand and biomolçcule, the binding of a second ligand can be measured in the presence of a first ligand that is already bound to the target~ The ability to simultaneously identify binding sites of dirr~ ligands allows a skilled artisan to 1) define negative and positive cooperative binding between ligands and 2) design 25 new drugs by linking two or more ligands into a single compound while m~in~ining a proper orientation of the ligands to one another and to their binding sites.
Further, if multiple binding sites exist, the relative affinity of individual binding moieties for the dirÇeie"t Wnding sites can be measured from an analysis of the ch~mic~l shift changes of the target molecule as a function of the added concentration of the ligand. By 30 simultaneously screening numerous structural analogs of a given compound, clet~ileA
structure/activity relationships about ligands is provided.
In part, the present invention provides a process of screening compounds to identify ligands that bind to a specific target molecule. That process comprises the steps of: a) generating a first two-~imen~ional lSN/lH NMR correlation spectrum of a 15N-labeled target - 3~ molecule; b) exposing the labeled target molecule to one or more compounds; c) generating a second two-dimensional 15N/lH NMR correlation spectrum of the labeled target molecule that has been exposed to the compounds of step (b); and d) comparing the first and second spectra to d~ e whether dirLr,.~ces in those two spectra exist, which differences in~1ins~t~
the presence of one or more ligands that have bound to the target molecule.
Where a process of the present invention screens more than one compound in step (b) and where a dirr~.~,.lce l,~ ~n spectra is observed, additional steps are performed to identify 6 which specific compound is binding to the target ninlec~ s Those ~ isinll~l steps comprice gen.or~ting a two-llim~ncional lSN/lH NMR correlation spectrum for each individual compound and c~"..p,..;.-g each ~.~e~ um to the first spe~l-ul,. to determin~. whether dirr~l~nces in any of those co~ ,d spectra exist, which ~liLrGlGIlces intiif~tr the presence of a ligand that has bound to the target molecule.
Any 15N-labeled target molecule can be used in a process of the present invention.
Because of the .~ olL~Ice of proteins in medicinal çh~mictry, a ~lGrGll~ target molecule is a polypeptide. The target m~lçcnl~. can be labeled with 15N using any means well known in the art In a ~ ,r~ ,d çmho~ nt the target molecule is prepared in l~,COlllbill~l~ form using transrc,l"~cd host cells. In an especially ~lGr~Gd embodiment, the target molecule is a polypeptide. Any polypeptide that gives a high resolultion NMR ~e~ ulll and can be partially or ullirollllly labeled with lSN can be used. The ~lG~dld~ion of ulliru~ ly 15N-labeled exemplary polypeptide target molecules is set forth hGlcillarlGl in the Examples.
A L~ rGllGd means of plG~illg adequate q~l~ntiti~s of uni~ullllly lSN-labeled polypeptides is to transform a host cell with an G~ ssion vector that contains apolynucleotide that encodes that polypeptide and culture the tran~roll--ed cell in a cultore medium that contains ~ccimil~hle sources of lSN. Accimil~hle sources of 15N are well known in the art. A pLGrGll~,d such source is 15NH4Cl.
Means for p~GJJdlillg G~ ion vectors that contain polynucleotides encoding specific polypeptides are well l~nown in the art. In a similar manner, means for transforming host cells with those vectors and means for culturing those transformed cells so that the polypeptide is expressed are also well known in the art.
The screening process begins with the generation or acquisition of a two-~lim~c-onal lSN/lH correlation spectrum of the labeled target molecul~. Means for generating two-dimensional 1SN/lH correlation spectra are well known in the art [See, e.g., D. A. Egan~ et al., Biochemistry, 32:8, pgs. 1920-1927 (1993); Bax. A.~ (~rzesiek~ S., Acc. Chem. Res., 26:4, pgs. 131-138 (1993)].
The NMR spectra that are typically recorded in the screening procedure of the present invention are two-riim~ncinnal 15N/1H heteronuclear single ~luanlul-l correlation (HSQC) spectra. Because the lSN/~H signals corresponding to the backbone amides of the proteins are usually well-resolved, the chernical shift changes for the individual arnides are readily monitored.
In generating such spectra, the large water signal is suppressed by spoiling gradients.
To facilitate the acquisition of NMR data on a large number of compounds (e.g., a ~l~t~b~ce WO 97/18469 PCTtUS96/18312 of synthetic or naturally occurring small organic compounds), a sample changer is employed.
Using the sample changer, a total of 60 samples can be run lm~t~nde-l Thus, using the typical acquisition p~ f t''--~. (4 scans per free induction decay (fid), 100-120 HSQC spectra can be acquired in a 24 hour period.
To facilitate procescing of the NMR data, con-l,ut~,l programs are used to ~ .r~,. and o~ lic~lly process the multiple two--lim~nsion~l NMR data sets, including aroutine to ~ntom~ti~lly phase the two-.l;."e.-sional NMR data. The analysis of the data can be facilitated by rO" ,~ .g the data so that the individual HSQC spectra are rapidly viewed and compared to the HSQC ..~e~ w~ of the control sample c~ i.-g only the vehicle for the added compound (DMSO), but no added compound. Detailed descriptions of means of generating such two--lim~o-ncional 15N/lH correlation spectra are set forth hele"lar~ in the Examples.
A r~ ,se~ /e two-~iim~.nsional lSNIlH NMR correlation ~7~JeCIlUlll of an lSN-labeled target molecule (polypeptide) is shown in FIG. 1 (the DNA-binding domain of the E2 1 5 protein).
Following acquisition of the first ~ye~ ulll~ the labeled target molecule is exposed to one or more test compounds. Where more than one test conlpowld is to be tested ~cimlllt~n~ously, it is ~,~GÇellcd to use a d~t~k~ce of compounds such as a plurality of small molecules. Such molecules are typically dissolved in perdc;uL~ d dimethylsulfoxide . The compounds in the ~t~kace can be purchased from vendors or created according to desired needs.
Individual compounds can be select~cl inter alia on the basis of size (molecular weigh~
= 100-300) and molecular diversity. Compounds in the collection can have dirreiellt shapes (e.g., flat aromatic rings(s), puckered aliphatic rings(s), straight and branched chain aliphatics with single, double, or triple bonds) and diverse functional groups (e.g., carboxylic acids, esters, ethers, ~mine,s, aldehydes, ketones, and various heterocyclic rings) for n~ ing the possibility of discovering compounds that interact with widely diverse binding sites.
The NMR screening process utilizes ligand concentrations ranging from about 0.1 to about 10.0 mM. At these concentrations, compounds which are acidic or basic can cignific~ntly change the pH of buffered protein solutions. Chemucal shifts are sensitive to pH
changes as well as direct binding interactions, and "false positive" ch~,mic~l shift changes, which are not the result of ligand binding but of changes in pH, can therefore be observed. It ~ is thus necessary to ensure that the pH of the buffered solution does not change upon addition of the ligand. One means of controlling pH is set forth below.
Compounds are stored at 263~K as 1.0 and 0.1 M stock solutions in dimethylsulfoxide (DMSO). This is nececc~ry because of the limited solubility of the ligands in aqueous solution. It is not possible to directly adjust the pH of the DMSO solution. In .

addition, HCl and NaOH form insoluble salts in DMSO, so ~lt~rn~tive acids and bases must be used. The following approach has been found to result in stable pH.
The 1.0 M stock solutions in DMSO are diluted 1:10 in 50 mM phosphate, pH 7Ø
The pH of that diluted aliquot solution is measured. If the pH of the aliquot is Im~h:~n~ed 5 (i.e., re",ail,s at 7.0), a working solution is made by ~ tin~ the DMSO stock solution 1:10 to make a 0.1 M solution and that solution is stored.
If the pH of the diluted aliquot is less than 7.0, ethanolamine is added to the 1.0 M
stock DMSO sntlltion, that stock solution is then diluted 1:10 with phosphate buffer to make another aliquot, and the pH of the aliquot rechecked.
If the pH of the diluted aliquot is greater than 7.0, acetic acid is added to the 1.0 M
stock DMSO solution, that stock solution is then diluted 1:10 with phosphate buffer to make anotner aliquot, and the pH of the aliquot ~ L rA
Ethanolarnine and acetic acid are soluble in DMSO, and the proper equivalents are added to ensure that upon transfer to aqueous buffer, the pH is unchanged. Adjusting the pH
is an interactive procecs, repeated until the desired result is obL~ined.
Note that this procedure is performed on 1:10 dilutions of 1.0 M stock solutions (100 mM ligand) to ensure that no pH changes are observed at the lower cormçntr~tionc used in the ,filllGnLs (0.1 to 10 rnM) or in dirrGlGIlt/~,dker buffer ~,y ,L~n~s.
Following exposure of the 15N-labeled target molecule to one or more test compounds, a second two--lirnencional 15N/lH NMR correlation spectrum is gen~r~t~
That second spectrum is generated in the same manner as set forth above. The first and second spectra are then col,.~ ed to determine whether there are any differences beL~n the two spectra. Differences in the two-flimPncion~l 15N/1H NMR correlation spectra that in~ te the presence of a ligand correspond to 15N-labeled sites in the target molecule.
26 Those di~GrG-Ices are de~Gl"~ned using standard procedures well known in the art.
By way of example, FIGs. 2, 3, 4, 5 and 6 show comparisons of correlation spectra before and after exposure of various target molecules to various test compounds. A ~let~
descnption of how these studies were performed can be found hereinafter in Examples 2 and 3.
Particular signals in a two-~imencional 15N/lH correlation spect-rum correspond to specific nitrogen and proton atoms in the target molecule (e.g., particular amides of the amino acid residues in the protein). By way of example, it can be seen from FIG. 2 that chemical shifts in a two-dimensional 15N/lH correlation of the DNA-binding domain of E2 exposed to a test compound occurred at residue positions 15 (I15), 21 (Y21), 22 (R22) and 23 (L23).
It can be seen from FIG. 2 that the binding of the ligand involved the isoleucine (Ile) residue at position 15, the tyrosine (Tyr) residue at position 21, the arginine (Arg) residue at position 22 and the leucine (Leu) residue at position 23. Thus, a process of the present WO 97/18469 PCTtUS96tl8312 invention can also be used to identify the specific binding site bcL~ c~ll a ligand and target molecule.
The region of the protein that is responsible for binding to the individual compounds is ic~en*fie~ from the particular amide signals that change upon the addition of the 5 compounds. These signals are ~cign~d to the individual amide groups of the protein by standard procedures using a variety of well-established heteron~ e~r multi-.li...~ nal NMR

To discover molecules that bind more tightly to the protein, molecnl~5 are selected for testing based on the ~L~u~;Lul~/activity rel~tic)n~hiI s from the initial screen and/or structural 10 information on the initial leads when bound to the protein. By way of example, the initial screening may result in the i~lentifi~tion of lig~ncls> all of which contain an ~ laLiC ring.
The second round of scr~ni-lg would then use other aromatic molecules as the test compounds.
As set forth helci~larLc~ in Example 2, an initial s~ elullg assay for binding to the 15 catalytic domain of stromelysin i~l~ntifieA two biaryl compounds as lig~nrl.~ The second round of screening thus used a series of biaryl derivatives as the test colllyou,lds.
The second set of test compounds are initially screened at a c~ CÇn I ~ ~11 ;on of 1 mM, and binding constants are measured for those that show affinity. Best leads that bind to the protein are then col}l~alcd to the results obtained in a functional assay. Those compounds 20 that are suitable leads are ~hemi~lly modified to produce analogs with the goal of discovering a new ph~ eutical agent.
The present method also provides a process for de~Glll~li~lg the dissociation constant between a target molecule and a ligand that binds to that target molecule. That process comprises the steps of: a) generating a first two-~imt-n~innal lSN/lH NMR correlation 25 spectrum of a 15N-labeled target molecule; b) titrating the labeled target molecule with various concentrations of a ligand; c) gGnGl~Ling a two--lim--n~iona} lSN/lH NMR correlation spectrum at each concentration of ligand from step (b); d) comparing each spectrum from step (c) both to the first spectrum from step (a) and to all other spectra from step (c) to 4uanLiry difr~,~cnces in those spectra as a function of changes in ligand concentration; and e) calculating 30 the dissociation constant (KD) beLween the target molecule and the ligand from those dirr~lellces.
Because of their importance in medicinal chemistry, a ~lGrGlred target molecule for ~ use in such a process is a polypeptide. In one L~IGrellGd embodiment, a process of de~.lllilling the dissociation constant of a ligand can be ~Glfolllled in the presence of a second 35 ligand. In accordance with this embodiment, the 15N-labeled target mo}ecule is bound to that second ligand before exposing that target to the test compounds.
Binding or dissociation constants are measured by following the lSN/lH chf mic~lshifts of the protein as a function of ligand concentration A known concentration ([P]O) of the target mo~ is mixed with a known cqnc en~tion ([L]o) of a previously iclçntif ligand and the two--iim~ncional 15N/lH correlation spe~;L-u,ll was acquired. From this spectrum, observed chemical shift values (~obs) are obtained. The process is r~eaL,d for varying concentrations of the ligand to the point of s~tl-r~tiQn of the target molecule, when 5 possible, in which case the limiting chçmic~1 shift value for saturation ~~sa~ is measured.
In those ~itn~-on~ where saturation of the target molecule is achieved, the dissociation collsL~L for the binding of a particular ligand to the target molçcnle is c~ ted using the formula KD = ([P]O - X) ([~]0 - X) where [P]o is the totaI molar concentration of target mnl~nle; [L]o is the total molar 10 co,~e"L,~lion of ligand; and x is the molar concentration of the bound species. The value of x is ri~ ~A from the equation:

X = ~obs ~ ~free where of ree is the chernil~l shift of the free species; ~obs is the observed chemi~l shift; and is the dirr~ ce belw~en the limitin~ ch~mi~:~l shift value for saturation (~sat) and the chemic~l shift value of the target molecule free of ligand (~free)-The ~ oçi~tir~n constant is then ~leterrnine~ by varying its value until a best fit to theobserved data is obtained using standard curve-fitting statistical methods. In the case whe re ~sat is not directly known, both KD and ~sat are varied and subjected to the same curve-fifflng procedure.
The use of the process described above to ~let~rmine the dissociation or binding20 affinity of various ligands to various target mflle~ules is set forth hereinafter in Examples 2 and 3.
Preferred target molecules, means for generating spectra, and means for comparing spectra are the same as set forth above.
In its principal aspect, the present invention provides a process of ~ie~ignin~ new 2~ ligands that bind to a specific target molecule by linl~ng together two or more molecules that bind to the target molecule.
The initial step in the design process is the j(lentific~tic-n of two or more ligands that bind to the specific target molecule. The identification of such ligands is done using two-~limt n~ional 15N/1H NMR correlation spectroscopy as set forth above.

Once two or more ligands are iclentifi~d as binding to the target molecule at diL
sites, a complex bGtv~,~;n the target molecule and ligands is formed. Where there are two n~lc, that complex is a ternary complex. Q..~ . y and other complexes are formedwhere there are three or more lig~n(1s Complexes are formed by mixing the target molecule simlllt~ne,ously or sequentially with the various ligands under ~ çs that allow those ligands to bind the target.
Means for d- t~ lhlg those conditions are well known in the art.
Once that complex is formed, its three-rlim~nsion~l structure is ~let~ h~ Any means of deLG~ g three-din e.ncinn~l structure can be used. Such mfthorls are well 1 ç known in the art. Exemplary and p,~rG,l~d methods are NMR and X-ray crystallography.
The use of three-~lim~nsion~l double- and triple resonance NMR to ~l~l .. ;nP the three-ion~ Llu~;Lulc of two ligands bound to the catalytic domain of stromelysin is set forth in detail h~ ft~,~ in F.xampl~ 4.
An analysis of the three-(l; --~ )~ional SLluc~ulc reveals the spatial nrient~tion of the ligands relative to each other as well as to the conrol,llaLion of the target molecule. First, the spatial oriP.nt~tion of each ligand to the target molecule allows for illentific~tion of those portions of the ligand directly involved in binding (i.e., those portions interacting with the target binding site) and those portions of each ligand that project away from the binding site and which portions can be used in subsequent linking procedures.
Second, the spatial rtrient~tion data is used to map the positions of each ligand relative to each other. In other words, discrete ~list~n~e~s between the spatially oriented ligands can be calculated.
Third, the spatial orientation data also defines the three-~iimPnsional relationships amongst the ligands and the target. Thus, in addition to calculating the absolute ~iist~nç~s between lig~n-l~, the angular orient~ti~ns of those ligands can also be det~ ~--i--e~
Knowledge of the spatial orientations of the ligands and target is then used to select linkers to link two or more ligands together into a single entity that contains all of the lip~n~is The design of the l~nkers is based on the distances and angular nriçnt~tion needed to ~ "~h~
each of the ligand portions of the single entity in proper orie~nt~t;l~n to the target.
The three--lim.on~ional conformation of suitable linkers is well known or readily ascertainable by one of ordinary skill in the art. While it is theoretically possible to link two or more ligands together over any range of ~list~nce and three--lim~-n~ional projection, in practice certain limit~tinns of distance and projection are ~lGÇc;lled. In a L~l~r~l~d embodiment, ligands are sepa,~Led by a distance of less than about 15 Angstroms (A), more preferably less than about 10 ~ and, even more preferably less than about 5 A.
Once a suitable linker group is identified, the ligands are linked with that linker.
Means for linking ligands are well known in the art and depend upon the chemiçSIl structure of thc ligand and the linking group itself. T ig~nrls are linked to one another using those portions of the ligand not directly involved in binding to the target molecule.
A deta~led descli~Lion of the design of a drug that inhibits the proteolytic activity of stromelysin, which drug was cle~ignçtl using a process of the present invention is set forth 5 hc,.~,~7~ in Example 4.
The following Examples illustrate ~ Gd embotli~ of the present invention and are not limiting of the ~perific~tion and claims in any way.

Example 1 ]~ aL~.Lion Of Uniformly 15N-Labeled TargetMolecules A. Stromelysin Human stromelysin is a 447-arnino acid protein believed to be involved in proteolytic degradation of cartilage. Cartilage proteolysis is believed to result in tlçgr,7r7.~tive loss of joint cartilage and the resnlting ;~ 1 of joint function observed in both osteoarthritis and rh~l-m~toirl arthritis. The protein possesses a series of r,70~inc incl7lrlin~ N-t~rn~in~l latent and ~G~Lide domains, a C-terminal domain homologous with homopexin, and an internal catalytic clom~in Studies have shown that removal of the N-tr-rmin~l prosequence of approximately eighty amino acids occurs to convert the proenzyme to the 45 kDa mature enzyme.
Furthr,rmore, studies have shown that the C-terminal homopexin homologous domain is not required for proper folding of the catalytic domain or for interaction with an inhihitor (See, e.g., A. I. Marcy, Biochemistrv~ 30: 6476-6483 (1991). Thus, the 81-256 amino acid residue intt-rn~l segment of stromelysin was selected as the protein fragment for use in identifying compounds which bind to and have the potential as acting as inhihitnr~ of stromelysin.
To employ the method of the present invention, it was neces~. y to prepare the 81-256 fragment (SEQ lD N0: 1) of stromelysin in which the peptide backbone was isotopically enriched with and 15N. This was done by inserting a plasmid which coded for the production of the protein fragment into an E. coli strain and ~rowing the g~nr-t~ ly-modified bacterial strain in a limiting culture medium enriched with NH4Cl and C-glucose.The isotopically t-.nri(~h~1 protein fr~gm~nt was isolated from the culture medium, purified, and subsequently used as the basis for ev~ ting the binding of test compounds.
The procedures for these processes are described below.
Human skin fibroblasts (ATCC No. CRL 1507) were grown and int11lcefl using the procedure described by Clark et al., Archiv. Biochem. and Biophys., 241: 36-45 (1985).
Total RNA was isolated from 1 g of cells using a Promega RNAgentst~ Total RNA Isolation System Kit (Cat.# Z5 1 10, Promega C~orp., 2800 Woods Hollow Road, Madison, WI 5371 1-5399) following the manufacturer's instructions. A 1 ,ug por~on of the RNA was heat-den~Lul~,d at 80~C for five ~ uL~S and then ~,ubje~;lGd to reverse transcriptase PCR using a GeneAmp~ RNA PCR kit (Cat.~ N808-0017, Applied Biosystems/Perkin-Elmer, 761 MainAvenue, Norwalk, CT 06859-0156) following the manufacturer's instructions.
Nested PCR was pe,ru~ ed using first primers (A) GAAATGAAGAGTC TTCAA
(SEQ ID NO:3) and (B) GCGTCCCAGGTTCTGGAG (SEQ ID NO:4) and thirty-five cycles of 94~C, two minutes; 45 C, two mi~luLGs, and 72~C three min~ltes This was followed by reamplification with internal ~lilllGl~7 (C) ATACCATGGCCTATCCAT TGGATGGAGC
(SEQ ID NO:5) and (D) ATAGGATCCTTAGGTCTCAGGGGA GTCAGG (SEQ ID NO:6) using thirty cycles under the same conditions described ul~ ly above to generate a DNA
coding for amino acid residues 1 -256 of human stromelysin.
The PCR fragment was then cloned into PCR cloning vector pT7Blue~R) (Novagen, Inc., 597 Science Drive, Madison, WI 53711) according to the m~mlf~ctllrer's instrnction~
The resllltin~ plasmid was cut with NcoI and BamHI and the stromelysin fr~m~nt was subcloned into the Novagen expression vector pET3d (Novagen, Inc., 597 Science Drive, Madison, WI 53711), again using the m~nnf~r,turer's instructions.
A mature stromelysin expression construct coding for amino acid residues 81-256 plus an inhi~*ng methionine was generated from the 1-256 expression construct by PCR
~mplifir~tion The resulting PCR fragment was first cloned into the Novagen pT7Blue(R) vector and then subcloned into the Novagen pET3d vector, using the m~nnf~rtnrer's instructions in the manner described above, to produce plasmid (pETST-83-256). This final plasmid is if lentic~l to that described by Qi-Zhuang et al., Biochemistry. 31: 11231 - 11235 (1992) with the exception that the present codes for a peptide sequence beginning two amino acids earlier, at position 81 in the sequence of human stromelysin.
Plasmid pETST-83-256 was transformed into E. coli strain BL21(DE3)/pLysS
2~; (Novagen, Inc., 597 Science Drive, Madison, VVI 53711) in accordance with the manufacturer's instructions to generate an expression strain, BL21(DE3)/pLysS/pETST-255-1.
A preculture l..e~ . was p~c~al~;d by dissolving 1.698 g of Na2HP4-7H2O~ 0.45 g of KH2P04, 0.075 g NaCl, 0.150 g 1SNH4Cl, 0.300 13C-glucose, 300 ~lL of lM aqueous MgSO4 solution and 15 ,uL of aqueous CaC12 solution in 150 mL of deionized water.
The resulting solution of preculture medium was sterilized and transferred to a sterile 500 mL baffle flask. Tmmt~ t--ly prior to inoculation of the preculture medium with the bacterial strain, 150 ~L of a solution con~ining 34 mg/mL of chloramphenicol in 100%
ethanol and 1.5 mL of a solution cont~inin~ 20 mg/mL of ampicillin were added to the flask contents.
The flask contents were then inoculated with 1 mL of glycerol stock of genetically-modified E. Coli, strain BL21(DE3)/pLysS/pETST-255-1. The flask contents were shaken (225 ~pm) at 37~C until an optical density of 0.65 was observed.

WO 97/18469 PCT/US96tl8312 Af. .~ l;nnnntri~nt"~r~ waspreparedbydissolving 113.28gofNa2~IP4-7H20, 30 g of KH2PO4, 5 g NaCl and 10 mL of 1% DF-60 antifoam agent in 9604 mL of deionized water. This solution was placed in a New Brunswick Scientific Micros r~,L~ ,nter (E~dison, NJ) and st~rili7r~l at 121~C for 40 minutes.
Tmm~ t-~.1ypriortoinoculationoftheÇ~.. -Il~.l-lnl;on~-,P.Ii-"--, thefollowingpre-sterilized components were added to the f~.rm~nt~tion vessel contents: 100 mL of a 10%
aqueous solution of 15NH4Cl, 100 mL of a 10% aqueous solution of 13c-glllrose~ 20 rnL of an aqueous lM solution of MgSO4, 1 mL of an aqueous lM CaC12 solution, 5 mL of an aqueous solution of thiamin hydrochloride (10 mg/mL), 10 mL of a solution co~ ;.l;llp 34 mg/mL of chlor~mph- ni~ol in 100% ethanol and 1.9 g of ampicillin dissolved in the chloramphenicol solution. The pH of the r~snlting solution was adjusted to pH 7.00 by the addition of an aqueous solution of 4N H2SO4.
The preculture of E. Coli, strain BL21(DE3)/pLysS/pETST-255-1, from the shake-flask scale procedure described above was added to the r~ .., cnnl ~ .. " ~i and cell growth 5 was ailowed to proceed until an optical density of 0.48 was achieved. During this process, the f~rmr~ntpr contents were ~uLu~ llc~lly mslint~inr~cl at pH 7.0 by the ~d~ition of 4N H2S04 or 4N KOH as needed. The dissolved oxygen content of the r~..~ contents was m~.int,.inr d above 55% air s?~hlr~.hion through a c~cad~ 1 loop which increased agitation speed when the dissolved oxygen content ~ko~ed below 55%. Air was fed to the f~
conLGllLs at 7 standard liters per minute (SLPM) and the culture te. l lp~ ., was .~ 1 at 37'C throughout the process.
The cells were harvested by centrifugation at 17,000 x g for 10 minutes at 4''C and the resulting cell pellets were collected and stored at -85~C. ~he wet cell yield was 3.5 g/L.
Analysis of the soluble and insoluble fractions of cell lysates by sodium dodecyl sulfate 26 polyacrylamide gel electrophoresis (SDS-PAGE) revealed that a~lu~-aL~ly 50% of the 5N-stromelysin was found in the soluble phase.
The isotopica71y-labeled stromelysin fragment prepared as ~leserihed above was purified employing a modification of the technique described by Ye, et al., E3iochemistry. 31:
1 123~-1 1235 (1992).
The harvested cells were suspended in 20 mM Tris-HCI buffer (pH 8.0) sodium azide solution containing 1 mM MgC12, 0.5 mM ZnC12, 25 units/mL of Benzonase~ enzyme, and an inhibitor ~ Lul~; made up of ~(2-aminoethyl)-benzenesulfonyl flllon~le ("AEBSF"), Leupeptin(~, Aprotinin(~9, and Pepstatin(~ (all at concentrations of 1 ~lglmL. AEBSF, Leupeptin(~. A~ Linil~ , and Pe~:jL~ are available from ~Am~riczln Tn~-rn~ nal Chemical, 17 SLl~Ll-l~ore Road, Natick, MA 01760.) The resulting Il~ixLulc was gently stirred for one hour and then cooled to 4~C. The cells were then sonically disrupted using a 50% duty cycle. The resulting lysate was WO 97/18469 PCT/US96/183i2 centrifuged at 14,000 rpm for 30 .~ es and the pellet of in~o}llble fraction frozen at -80~C
for subsequent processing (see below).
Solid ~mmonillm sulfate was added to the supernatant to the point of 20% of saturation and the reslllting solution loaded onto a 700 mL phenyl sepharose fast flow ("Q-Sepharose FF") column (Pharmacia Biotech., 800 (~çntennizll Ave., P. O. Box 1327, Piscataway, NJ 08855). Prior to loading, the sepharose column was equilibrated with 50 mM Tris-HCl buffer (pH 7.6 at 4~C), 5 mM CaC12, and 1 M (NH4)2SO4. The loaded column was eluted with a linear gradient of decreasing co~ ns of aqueous (NH4)2S04 (from 1 down to 0 M) and increasing concentr~tic)ns of aqueous CaC12 (from 5 o to 20 mM) in Tris-HCl buffer at pH 7.6.
The active fractions of eluate were collected and colue~ at~d in an Amicon stirred cell (Amicon, Inc., 72 Cherry Hill Drive, Beverly, MA 01915). The conce~ a~d sample was dialyzed overnight in the starting buffer used with the Q-Sepharose FF column, 50 mM Tris-HCl (pH 8.2 at 4~C) with 10 mM CaCk.
The dialy_ed sample was then loaded on the Q-Sepharose FF column and eluted witha linear ~r~ nt conlplising the starting buffer and 200 mM NaCl. The purified soluble fraction of the isotopically-labeled stromelysin fragment was concentrated and stored at 4~C.
The pellet was s~ hili7~A in 8M g~ n~ ne-Hcl. The solution was centrifuged for 20 ...i..~l~esat20,000rpmandthe:~ul)e...~t~ntwasaddeddropwisetoafoldingbuffer comprising 50 mM Tris-HCl (pH 7.6), 10 mM CaC12 0.5 mM ZnC12 and the inhibitor coc~ctail of AEBSF, Leupeptin~, A~lo~ , and Pepstatin(~ (all at concentrations of 1 ,ug/mL). The volume of folding buffer was ten times ehat of the sup~rn~t~nt The mixture of supernatant and folding buffer was centrifuged at 20,000 rpm for 30 min~ltes The supem~t~nt from this centrifugation was stored at 4~C and the pellet was subjected twice to the steps described above of sol~lhili7~tion in g~ niflin~-HCl, refolding in buffer, and centrifugation. The final :~Up~Lllaklllt~ from each of the three centrifugations were combined and solid ammonium sulfate was added to the point of 20% sZ~tnr~tion The resulting solution thus derived from the insoluble fraction was subjected to pnrific~tinn on phenyl Sepharose and Q-Sepharose as described above for the soluble fraction.
The purffled soluble and insoluble fractions were combined to produce about 1.8 mg of purified isotopically-labeled stromelysin 81-256 fragment per gram of original cell paste.

B. Human papillomavirus ~HPV~ E2 Inhibitors The papillomaviruses are a family of small DNA viruses that cause genital warts and cervical carcinomas. The E2 protein of HPV regulates viral transcription and is required for viral replication. Thus, molecules that block the binding of E2 to DNA may be useful therapeutic agents against HPV. The protein rather than the DNA was chosen as a target, WO 97/18469 PCT/US96/183i2 bccau~.e it is expected that agents with greater selectivity would be found that bind to the protein rather than the DNA.
The DNA-binding domain of human papillomav~rus E2 was cloned from the full length DNA that codes for E2 using PCR and ovc~ lcssed in bacteria using the T7 5 c;~ ion system. Uniforrnly 15N-labeled protein was isolated from b?~et~.ri~ grown on a minim~l n,eJiu~-l c~ i.lg lSN-labeled protein was isolated from b~f~t~ori~7 grown on a minim~l m-qAinm co~ ;rlg lSN-labeled arnmonium chlc ririe The protein was purified from the bacterial cell lysate using an S-sepharose FastFlow column ~l~-e~l..;lihr~te~i with buffer (50 mM Tris, 100 mM NaCl, 1 mM EDTA, pH = 8.3).
The protein was eluted with a linear gradient of 100-500 mM NaCl in buffer, pooled, and applied to a Mono-S column at a pH = 7Ø The protein was eluted with a salt gradient (100-500 mM), conc~ LIaLed to 0.3 mM, and exchanged into a TRIS (50 mM, pH = 7.0buffered H20/D20 (9/1) solution conL~...ir-g sodium azide (0.5%).

C. RAF
Uniformly lSN-labeled Ras-binding domain of the RAF protein was ~.c~,d as described in Emerson et al., Biochemistry. 34 (21): 6911-6918 (1995).

D. FKBP
Uniformly lSN-labeled recombinant human 3:~K binding protein (FKBP) was prepared as described in Logan, et al., J. Mol. Biol., 236: 637-648 (1994).

Example 2 Screening Compounds Using Two-Dirnensional 15N/IH
NMR Correlation Spectral ~n~lysis Tne catalytic domain of stromelysin was ~ al~i in accordance with the proceduresof Example 1. The protein solutions used in the screening assay contained the unilol~-ly lSN-labeled catalytic domain of stromelysin (0.3 mM), aceLohy~ xarnic acid (S00 mM), CaC12 (20 mM), and sodium azide (0.5%3 in a H20/D20 (9/1~ TRIS buffered solution (50 mM, pH=7.0).
Two--limen~i-)nal 15N/IH NMR spectra were generated at 29~C on a Bruker AMX500 NMR ~.~e~Ll~nl;;L~l equipped with a triple resonance probe and Bruker sample changer. The 15N/lH HSQC spectra were acquired as 80 x 1024 complex points usingsweep widths of 2000 Hz (15N, tl) and 8333 E~z (lH, t2). A delay of 1 second l)e~wt;en scans and 8 scans per free induction decay(fid) were employed in the data collection. All NMR spectra were processed and analyzed on Silicon Graphics COIllpU~ . using in-house-written software.

A first two-~ cion~l l5N/lH NMR correlation S~c il~unl was ac~luucd for the l5N-labeled stromelysin target molecule as described above. The stromelysin target was then exposed to a d~t~h~e of test compounds. Stock solutions of the compounds were made at lO0 mM and l M. In addition, a combin~t;on library was ~lG~d that contained 8-lO5 compounds per sample at a concentration of lO0 mM for each compound.
The pH of the l M stock solution was adjusted with acetic acid and eth~nol~mine so that no pH change was observed upon a l/lO dilution with a lO0 mM phosphate burr~ ,d solution (pH = 7.0). It is important to adjust the pH, because small ch~n~s in pH can alter the ch~mi~l shifts of the biomolecules and complicate the i~ ylckllion of the NMR data.
The compounds in the ~l~t~bz-~e were selected on the basis of size (mnl~cnlslr weight =
100-300) and molecular diversity. The molecules in the collection had (li~r,.c~-t shapes (e.g., flat a~ulll~Lic rings(s), ~u~;h~,.cd aliphatic rings(s), straight and branched chain ~iph~ti~s with single, double, or triple bonds) and diverse functional groups (e.g., carboxylic acids, esters, ethers, amines, aldehydes, ketones, and various heterocyclic rings) for m~ximi7.ing the possibility of discovering compound that interact with widely diverse binding sites.
The NMR samples were ~l'c~al~d by adding 4 ~l of the DMSO stock solution of the compound n~i~Lulcs that contained each compound at a cul~clllla~ion of lO0 mM to 0.4 rnl H20/D20 (9/1) burr~l~ solution of the uniformly 15N-labeled protein. The final concentration of each of the compounds ;n the NMR sample was about l mM.
In an initial screen, two compounds were found that bind to the catalytic domain of stromelysin. Both of these compounds contain a biaryl moiety. Based on these initial hits, structurally simila-r compounds were tested against stromelysin. The s~ucture of those biaryl compounds is represented by the structure I, below. (See Table I for definitions of Rl-R3 and A l-A3).

~ A~A3- R3 In the second round of screening, binding was assayed both in the absence and in the presence of sa~ulaLi~lg amounts of acetohydroxamic acid (500 mM).
Many of the biaryl compounds were found to bind the catalytic domain of stromelysin. FIG. 4 shows a r~l~s~ntative two--iimton~ional l5N/lH NMR correlation spectrum before and after exposure of stromelysin to a biaryl test compound. It can be seen 30 from FIG. 4 that the compound caused chemical shifts of lSN-sites such as those ~lesi~n;-t.od Wl24, Tl87, Al99 and G204.
These sites correspond to a tryptophan (Trp) residue at position 124, a threonine (Thr) at position l 87, an alanine (Ala) at position l99, and a glycine (Gly) at position 204 of SEQ
ID NO. l. FIG. 9 shows the correlation between the NMR binding data and a view of the NMR-derived three~lim~n~ional ~u~ uc of the catalytic domain of stromelysin. The ability to locate the specific binding site of a particular ligand is an advantage of the present invention.
Some compounds only bound to stromelysin in the presence of hyd~ alllic acid.
Thus, the bin~ling affinity of some compounds was enh~nccd in the presence of the hydroxarnic acid (i. e. cooperative). These results ~Y~mrlify another important c~r~bility of the present S~;l~.lillg assay: the ability to identify compounds that bind to the protein in the presence of other molecules.
Various biaryl compounds of ~ u~ c I were tested for binding to stromelysin at 10 differing concentrations. The 15N/lH spectra ~ d at each concer~ tinn were evaluated to quantify ~liLr~nces in the spectra as a function of compound conf~ntr~ti~n A binding or dissociation constant (KD)was calcnl~tçA, using standard procedulcs well known in the art, from those dirrc.~nces. The rçsults of this study are shown in Table 1. The values for Rl-R3 and Al-A3 in Table 1 refer to the corresponding positions in the structure I, above.

Table 1 Compound No. Rl R2 R3 Al A2 A3KD(mM) H OH H C C C 1.1 2 CH2OH H H C C C 3.2 3 Br H OH C C C 1.3 4 H H H N N C 1.6 CHO H H C C C 1.7 6 OCH3 NH2 H C C C 0.4 7 H H H N C C 0.2 Table 1 (Continued) Compound No. Rl R2 R3 Al A2 A3 KD(mM) 8 OCOCH3 H H C C C~ 0.3 9 OH H OH C C C0.01 H H H N C N0.4 11 OH H H C C C0.3 12 OH H CN C C C0.01 The data in Table 1 show the utility of a process of the present invention in .lCt~ ~g dissociation or binding constants between a ligand and a target molecule.
Another advantage of an NMR sl;lce~ g assay of the present invention is the ability to correlate observed chP.mic~,~l shifts from the two-~ n~ n~l 15N/lH NMR
correlation spectra with other spectra or pro~ections of target molecule configuration. The results of a representative such correlation are shown in FIG. 9, which depicts regions within 10 the polypeptide at which binding ~,vith the substrate molecule is most likely occurring. In ~his Figure, the ~pd~ Wnding regions in stromelysin are shown for Compound 1 (from Table 1).
Compounds from the ~l~t~b~e were screened in a similar manner for binding to theDNA-binding ~lon~z~in of the E2 protein. Those compounds had the structure II below, where Rl-R4 and A are defined in Table 2.
R~
~A~ R3 II

NMR exp~nmt~nf~ were performed at 29~C on a Bruker AMX500 NMR specllollle~
equipped with a triple resonance probe and Bruker sample changer. The 15N-/lH HSQC
- spectra were acquired as 80 x 1024 complex points using sweep widths of 2000 Hz (15N,t~ ) and 8333 Hz (lH, t2). A delay of 1 second between scans and 4 scans per free induction 20 decay were employed in the data collection. All NMR spectra were processed and analyzed on Silicon Graphics computers.

CA 02237336 1998-0~-wo 97/18469 PCT/US96/1831 PIGs. 2 and 3 show represent~tive two-Aimen~io~ SN/IH NMR correlation spectra before and after exposure of the DNA-binding domain of E2 to a f~st and second test compound, respectively.
It can be seen from FIG. 2 that the first test compound caused chemic~1 shifts of l~N-sites such as those desi~n~t~d I15, Y21, R22 and L23. Those sites col.~s~und to an isoleucine (~e) residue at position 15, a tyrosine residue (Tyr) at position 21, an arginine (Arg residue at position 22 and a leucine (Leu) residue at position 23 of SEQ ~ NO. 6.
It can be seen from FIG. 3 that the second test compound caused ch.o.mi~ ~l shifts in the particular 15N-sites de~ign~t.-d I6, G1 l, H38, and T52. Those sites correspond to an 1 Q isoleucine (Ile) residue at position 6, a glycine (Gly) residue at position 11, a hictiAine (EI;s3 residue at position 38 and a threonine (Thr) at position 52 of SEQ ID NO. 6.
FIGs. 7 and 8 show the correlation b~,L~ ,n those NMR binding data and a view ofthe NMR-derived three-Aim~,n~ional ~LLUC~U1C of the DNA-binding domain of E2.
Several structurally similar compounds caused ch~mic~l shift changes of the protein signals when screened at a c~-n~ . "*-ln of 1 mM. Two distinct sets of amide resonances were found to change upon the addi*ion of the compounds: one set of signals corresponding to amides located in the ~-barrel formed between the two monomers and a second set cc,-l~,syollding to amides located near the DNA-binding site.
For ex~mrlP,, compounds cOI .li. h~ two phenyl rings with a carboxylic acid ~tt~Chf!A
to the carbon linking the two rings only caused chemi~l shift changes to the amides in the DNA-binding site. In contrast, benzophenones and phenoxyphenyl-cont~ining compounds only bound to the ~3-barrel. Other compounds caused chemic~l shift changes of both sets of signals but shifted the signals in each set by different amounts, suggesting the presence of two distinct binding sites.
By monitoring the chemical shift changes as a function of ligand conce~ ion, binding constants for the two binding sites were also measured. The results of those studies are summarized below in Table 2.

CA 02237336 1998-05-ll WO 97/18469 PCT/US96/183i2 Table 2 Comp. A ~1 R2R3 R4 DNA13-baIrelFilter No. KD(m~KD(tT~ binding a~.say - 13 CO H H H OH >50 û.6 14 O H H H CH2OH >50 2.0 a H H COO H 2.0 >50 +
16 a Cl ClCOO H 0.1 >50 +
17 a H HCH2COO H 4.2 4.9 +
18 a H HCH=CHCOO H 1.2 6.2 +
19 0 H HCH2cH2cH(cH3) H 0.5 0.2 +

O H HCOCH2CH2COO H 2.7 4.8 +
a a dash (-) for A in~lic~t~s no atom (i.e., byphenyl linkage) Uniforrnly 1 5N-labeled Ras-binding domain of the RAF protein was prepared as 5 rlesçrikerl in Example 1 and screened using two--~im~n~i-nal lSN/lH NMR correlation spectral analysis in accordance with the NMR procedures described above. The results of a represçm~tive study are shown in FIG. 5, which depicts two-~limen~ional 15N/lH NMR
correlation spectra both before and after exposure to a test compound.
Uniformly 15N-labeled FKBP was plc~alcd as described in Example 1 and screerEe dlo using two--1imen~ional lSN/lH NMR correlation spectral analysis in accordance with the NMR procedures described above. The results of a repres~nt~tive study are shown in FIG.

6, which depicts two-tlim-oncional 15N/lH NMR correlation spectra both before and after exposure to a test compound.

Exarnple 3 Comparison of NMR~ Enzymatic, Filter Bindin~ and Gel Shift Screening Assays Studies were performed to compare binding constants of ligands to various biomolecules, determined by the NMR method of the present invention, to similar results 20 obtained from prior art methods.
In a first study, binding constants were determined, both by the NMR method of the present invention, and by a prior art enzymatic assay. The target molecule was the catalytic domain of stromelysin prepared in accordance with the procedures of Example 1. The NMR
binding constants, KD, were derived using two-dimensional 15N/lH NMR correlation -spectroscopy as described in Example 2. The KD values so obtained were co~ d to an inhibition COI~ KI as det~ ....i..eA in an ~,n~y~latic assay.
The en~ylnatic assay measured the rate of cleavage of a fluorogenic ~,ulJ~ by following the ffuorescence increase upon peptide cleavage which causes a s~ ;nn be~n 5 the fluorophore and quencher. Enzymatic activity was measured using a matrix of different concentrations of acetohyd,~;~l,ic acid and biaryl compounds. The assay is a motlifi~tion of the method described by H. Weingarten, et al. in Anal. Biochem.. 147: 437-440 (1985) employing the fluorogenic substrate properties ~l~sçriberl by E. Matayoshi, et al. in Science:
247: 954-958 (1990).
Eight acetohy~ amic acid concentrations were used ranging from 0.0 to 1.0 M, andsix compound concentrations were used, rçsnlting in a total of 48 points. Individual compound concenl.a~ion varied due to solubility and potency.
All NMR mea,ul~n~en~s were ~ Çull"ed in the presence of 500 mM acetohy~u~a",ic acid, except for the titration of acetohydroxamic acid itself. Dissociation con.~t~nt~ were 5 obtained from the dependence of the observed t~h~mi~l shift changes upon added ligand.
Inhibition const~nt.~ were then obtained from the inhibition data using standard procedures.
The results of these studies are ~7. ~ ed below in Table 3, which shows the comparison of NMR-derived dissociation cl-n~ s (E~D) with inhibiti~n con~t~nt~ measured in the enzyme assay (KI)~ using a fluorogenic substrate.
2~
Table 3 Compound No. NMR KD (rnM3 Assay KI (mM3 4 1.6 7.4 7 0. 17 0.32 9 0.16 0-70 0.40 1.8 12 0.02 0. 1 1 Acetohycllu~nic acid 17.0 21.1 The data in Table 3 show that a NMR process of the present invention provides a rapid, efficient and accurate way of fl~ t~ l;ll;ng dissociation or binding con~Lanls of ligands to target biomolecules. Comparison of the binding c~ n~t~ntS clet~ ;..~1 by the two methods result in the same ranking of potencies of the compounds tested. That is, while the values for a given substrate as clc t~ ed by the two methods are not equal, they are proportional to one another.
In a second study, the results for binding of the DNA-b;n-ling domain of E2 to its target DNA were obtained by prior art methods and con~altd with results obtained by the methQ~1 of the present invention. The target was the DNA-binding domain of E2, ~c~,d in accordance with the procedures of Fxample 1. NMR S~ illg assays and NMR processes for ~1G~ ;ng ligand dissociation con~t~nt~ were ~ rol.ned as set forth above in F.Y~mple 2.
The binding constant from the NMR process was com~d to the results of a physical, filter binding assay that measured binding of DNA to the target. The high-throughput filter binding assay was performed using E2, ~l~e;l according to Example 2 above. The 33P-labeled DNA construct comprised a 10,329 base pair plasmid formed by inserting the HPV-11 genome, con~i~ g three high affinity and one low aff~ity E2 binding sites, into the PSP-65 plasmid (Promega, Madison, WI).
The binding ~ffinhi~s at the dirr~ sites as ~let~-rminecl by NMR were conll,~Gd for a subset of the compounds to the inhibition of E2 binding to DNA as measured in the filter binding assay. As shown in Table 2 above, the activities deL~ ed in the filter binding assay correlated closely wi~ the binding ~ffinit;çs calculated from the amides of the DNA-binding site but noe to the ~ffinities measured for the 13-barrel site. This is con~ictent with the relative locations of each site.
In an alternative study, a comparison of the NMR-~let~rmine~l binding results was made with similar results obtained by a prior art gel-shift assay using techniques well known in the art. The gel-shift assay was pGlr~llned using a GST fusion protein which contained full length E2 and a 33P-labeled 62 base pair DNA fragment cont~ining two E2 binding sites.
The method identified numerous compounds which gave positive results in the gel-shift assay. Some of these positive results, however, were believed to be due to binding to the DNA, since in these cases, no binding to the E2 protein was observed using the NMR
method of this invention. These compounds were shown to indeed bind to DNA rather than to E2, as evi~çnçe-l by changes in the cl~n~cal shifts of the DNA rather than the protein upon the addition of the compounds. These data show that yet another advantage of the present 3~ invention is the ability to minimi7e the occurrence of false positives.

FY~mr1e 4 DP~;~n of a potent. non-peptide inhibitor of stromelysin Studies were pG~Çu l~-ed to design new ligands that bound to the catalytic domain of stromelysin. Because stromelysin undergoes autolysis, an inhibitor was sought to block the 5 ~legr~ tion of stromelysin. That inhibitor would f~ it~te the screening of other potential ligands that bind to other sites on the enzyme.
The criteria used in selecting con~L,ou,lds in the scr~nil~g for other binding sites was based pnmarily on the size of the ligand. The sm~llPst ligand was sought that had enough solubility to Sd~uldlG (>98% occupancy of enzyme) and inhibit the enzyme.
The cloning, G~.Gssion, and pnrif;~ticln of the catalytic domain of stromelysin was accompli~hP~I using the procedures set forth in FY~mple 1. An initial step in the design of the new ligand was the i-lentific~ti~n of a first ligand that bound to the stromelysin target. Such identification was carried out ul accordance with a two-~iim~-n~ n~ 15N/lH NMR correlation screening process as disclosed above.
A variety of llydlu~al~c acids of the general formula R-~CO)NHOH were s-, ~,~d for binding to stromelysin using the procedures set forth in Fx~mple 2. Of the compounds tested, acetohydroxamic acid [CH3(CO)NHOHl best s~ticfipA the selection criteria: it had a binding affinity for strûmelysin of 17 mM and had good water solubility. At a concentration of 500 mM, acetûhydroxamic acid inhihited the degradation of the enzyme, allowing the 20 screening of other potential ~ n~l~
The second step in the design process was the iclentific~tlnn of a second ligand that bound to the target stromelysin at a site dirr~ from the binding site of acetohyd~ ~ic acid. This was accompli~hed by screening compounds for their ability to bind stromelysin in the presence of saturating amounts of acetohydroxamic acid. Details of procedures and 25 results of this second i(lPntific~ti~n step are set forth above in Example 2.The compound idçntifi~ as a second ligand from these studies and used in subsequent design steps was the compound de~ip;n~ted as Compound #4 in Table 1 (See Example 2).
The next step in the design process was to construct a ternary complex of the target 30 stromelysin, the first ligand and the second ligand. This was accomplished by exposing the stromelysin target to the two ligands under conditions that resulted in complex formation.
The three-rlim~n~innal structure of the ternary complex was then ~let- rrninPd using NMR
spectroscopy as described below.
The 1H, 13c~, and 15N backbone resonances of stromelysin in the ternary complex 35 were assigned from an analysis of several 3D double- and tIiple-resonance NMR spectra (A.
Bax, et al.. Acc. Chem. Res.. 26: 131-138 (1993)). The Ca resonances of adjacent spin systems were illentifi~d from an analysis of three-dimensional (3~) HNCA (L. Kay, et al., L
Ma~n. Reson.. 89: 496-514 (1990)) and HN(CO)CA (A. Bax, et al., J. Bio. NMR. 1: 99 WO g7/18469 - 27 - PCT/US96/18312 (1991)) spectra recorded with i(~en*~l spectral widths of 1773 Hz (35.0 ppm), 3788 Hz (30.1 ppm), and 8333 Hz (16.67 ppm) in the Fl(l5N), F2(13C) and F3( H) dimen.cion.c, lc~G..~ rely.
The data matrix was 38(tl) x 48(t2) x 1024(t3) complex points for the HNCA
7~1G1LIUIII, and 32(tl) x 40(t2) x 1024(t3) complex points for the HN(CO)CA spectrum. Both spectra were acquired with 16 scans per i.._rc~--ent~ A 3D QCA(CO)NH ~L)G~;LIUIII (S.
.r7~Si~, et al., J. Am. Chem. Soc.. 114: 6261-6293 (1992)) was collected with 32(tl, 15N) x 48(t2, 13C) x 1024(t3, lH) complex points and 32 scans per in~ cn~ Spectral widths were 1773 Hz (35.0 ppm), 7575.8 Hz (60.2 ppm), and 8333 Hz (16.67 ppm) in the 0 N, C and H dirnensions, respectively.
For all three spectra, the lH carrier frequency was set on the water iGsona"ce and the 15N carrier frequency was at 119.1 ppm. The 13C carrier ~ u~ cy was set to 55.0 ppm in HNCA and HN(CO)CA c"~ .nl.~, and 46.0 ppm in the QCA(CO)NH G~)~ t The backbone ~ were colLL"...ed from an analysis of the crosspeaks observed in an lsN-sG~ Gd 3D NOESY-HSQC spectrum and a 3D HNHA-J ~e~ u.~.
The 15N-separated 3D NOESY-HSQC spectrum (S. Fesik, et al., J. Magn. Reson.~ 87:588-593 (1988)); D. Marion, etal., J. Am. Chem. Soc.. 111: 1515-1517 (1989)) wascollected with a mixing time of 80 ms. A total of 68(tl, 15N) x 96(t2, lH) x 1024(t3, lH) complex points with 16 scans per increment were collected, and the spectral widths were 1773 Hz (35.0 ppm) for tne 1 N dimension, 6666.6 Hz (t2, H, 13.3 ppm), and 8333 Hz (16.7 ppm) for the lH dimension.
The 3D HNHA-J spectrum (G. Vuister, et al., J. Am. Chem. Soc.~ 115: 7772-7777 (1993)), which was also used to obtain 3JHNHa coupling con~t~nt~ was acquired with 35(tl, 15N) x 64(t2, lH) x 1024(t3, lH) complex points and 32 scans per increment.
Spectral widths and carrier frequencies were identical to those of the lSN-sr~ 1 NOESY-HSQC spectrum. Several of the H~ signals were assigned using the HNHB wc~ t.
The sweep widths were the same as in the 15N-separated NOESY-HSQC spectrum that was acquired with 32(tl, N) x 96(t2, H) x 1024(t3, H) complexpoints.
The lH and 13C chemical shifts were assigned for nearly all sidechain resonances. A
3D HCCH-TOCSY spectrum (L. Kay, et al., J. Ma~n. Reson.. 101b: 333-337 (1993)) was acquired with a mixing time of 13 ms using the DIPSI-2 sequence (S. Rucker, et al., Mol.
Phvs., 68: 509 (1989)) for 13C isotropic mixing. A total of 96 (tl, 13C) x 96(t2, lH) x 1024(t3, lH) complex data points were collected with 16 scans per increment using a spectral width of 10638 Hz (70.8 ppm, wl), 4000 Hz (6.67 ppm, w2), and 4844 (8.07 ppm, w3).
3~ Carrier positions were 40 ppm, 2.5 ppm, and at the water frequency for the 13C, indirectly detected lH, and observed lH (iimen~ions7 respectively.
Another 3D HCCH-TOCSY study was performed with the 13C carrier at 122.5 ppm to assign the aromatic r~sid~les The spectra were collected with 36(tl, C) x 48(t2,1H) x = = = == = . ===.= = = = = =
CA 02237336 l998-05-ll WO 97/18469 PCT/US96/183i2 1024 ~t3, H) complex points with spectral widths of 5263 Hz (35.0 ppm, wl), 3180 Hz (5.30 ppm, w2), and 10,000 (16.7 ppm, w3). Carrier positions were 122.5 ppm, 7.5 ppm, and at the water frequency for the 13C, indirectly d~tectefl lH, and observed lH ~limen~ ns, respectively.
A 13C-s~alat~d 3D NOESY-HMQC ~e:CllUln (S. Fesik, etal., J. Magn. Reson 87: 588-593 (1988)); D. Marion, etal., J. Am. Chem. Soc.. 111: 1515-1517 (1989)) was recorded using a rnixing time of 75 ms. A total of 80 (tl, 13C) x 72 (t2, lH) x 1024 (t3, lH) complex data points with 16 scans per ill~ en~ were coll~cteA over ,epec~l widtns of 10638 Hz (70.49 ppm, wl), 6666.6 Hz (13.3 ppm, w2), and 8333.3 Hz (16.67 ppm, w3). The10 lH carrier frequencies were set to the water resonance, and the 13C carrier frequency was placed at 40.0 ppm.
Stereospecific ~ ignm~nt~ of methyl groups of the valine and leucine residues were obtained by using a biosynthe~ic approach (Neri etal., Biochem.. 28: 7510-7516 (1989)) on the basis of the 13C-13C one-bond coupling pattern observed in a high-resolution IH, 13C-HSQC spectrum (G. Bo-lçnh~lsen, etal., J. Chem. Phys. Lett.~ 69: 185-189 (1980)) of a fractionally C-labeled protein sample. The ~e~ u~ll was acquired with 200( 3C, tl) x 2048( H, t2) complex points over spectral widths of 5000 Hz (39.8 ppm, 13C) and 8333 Hz (16.7 ppm, lH). Carrier positions were 20.0 ppm for the 13C dimrn~ion~ and at the water frequency for the lH ~iim~n~ion To detect NOEs between the two ligands and the protein, a 3D 12C-filtered, 13C-edited NOESY spectrum was collected. The pulse scheme consisted of a double 13C-~llter sequence (A. Gçmmt ~r, et al., J. Ma~n. Reson.. 96: 199-204 (1992)) concatenated with a NOESY-E~MQC sequence (S. Fesik, et al., J. Ma~n. Reson.. 87: 588-593 (1988)); D.Marion, et al., J. Am. Chem. Soc.. 111: 1515-1517 (1989)) . The spectrum was recorded with a mixing time of 80 ms, and a total of 80 (tl, 13C) x 80 (t2, lH~ x 1024 (t3, IH) complex points with 16 scans per increment. Spectral widths were 8865 Hz (17.73 ppm, wl), 6667 Hz (13.33 ppm, w2), and 8333 Hz (16.67 ppm, w3), and the carrier positions were 40.0 ppm for the carbon ~lim~ncion and at the water frequency for both proton dimensions.
To identify amide groups that exchanged slowly with the solvent, a series of lH,15N-HSQC spectra (G. Bodenhausen, et al., J. Chem. Phys. Lett.. 69: 185-189 (1980)) were recorded at 25~C at 2 hr intervals after the protein was exchanged into D2O. The acquisition of the first HSQC spectrum was started 2 hrs. after the ~ n of D20.
All NMR spectra were recorded at 25~C on a Bruker AMX500 or AMX600 NMR
35 spectrometer. The NMR data were processed and analyzed on Silicon Graphics computers.
~n all NMR experiments, pulsed field gradients were applied where a~p.ol,~iate as described (A. Bax, et al., J. Ma~n. Reson.. 99: 638 (1992)) to afford the suppression of the solvent signal and spec~al artifacts. Quadrature detection in indirectly detected t~imencions was CA 02237336 1998-05-ll accomplished by using the States-TPPI method (D. Marion, et al., J, Am. Chem. Soc.. 111:
1515- 1517 (1989)). Linear prediction was employed as described (E. Olejnic7~k, et al., J.
Magn. Reson.. 87: 628-632 (1990)).
The derived three-~1im~n~ional structure of the ternary complex was then used to5 define the spatial ori~nt~tinn of the first and second ligands to each other as well as to the target stromelysin molecule.
Distance ~ derived from the NOE data were cl~ifiçd into six categories based on the NOE cross peak intensity and given a lower bound of 1.8 A and upper bounds of 2.5 A, 3.0 A, 3.5 A, 4.0 A, 4.5 A, and 5.0 A, respectively. Restraints for q, torsional angles 10 were derived from JHNHa coupling constants measured from the 3D HNHA-J spectrum (G. Vuister, et al., J. Am. Chem. Soc.. 115: 7772-7777 (1993)). The q3 angle wased to 120%+40% for 3JHNHa > 8.5 Hz, and 60%+40% for 3JHNHa < 5 Hz.
Hydrogen bonds, iflentifi~l for slowly ex~h~nginp amides based on initial structures, were defined by two restr~int~: 1.8-2.5 A for the H-O distance and 1.8-3.3 A for the N-O
~ t~n~e- Structures were calculated with the X-PLOR 3.1 program (A. Brunger, "XPLOR
3.1 Manual," Yale University Press, New Haven, 1992) on Silicon Graphics co~ u~
using a hybrid distance geometry-~im~ teA ~nne~ling approach (M. Nilges, et al., FEBS
L~., 229: 317-324 (1988)).
A total of 1032 ~l~lo~"aL~ in~l~lu~o~ t~nce restraints were derived from the NOE data. In addition, 21 lln~mhiguous int~rmnl~:cular distance le~L-ahlts were derived from a 3D 12C-filtered, 13C-edited NOESY spectrurn. Of the 1032 NOE l~ involving the protein, 341 were intra-residue, 410 were sequential or short-range between residues separated in the primary sequence by less than five amino acids, and 281 were long-range involving residues s~ted by at least five re~i-1nes.
In addition to the NOE distance restraints, 14 ~ dihedral angle r~ ints were included in the structure c~ tic)ns that were derived from three-bond coupling con~nt~ (3JHNH0c) det~l",ined from an HNHA-J spectrum (G. Viioster, et al., J. Am. Chem. Soc.. llS: 7772-7777 (1993)). The c;~ e~ ldill~ also included 120 distance restraints corresponding to 60 hydrogen bonds. The amides involved in hydrogen bonds were identified based on their characteristically slow exchange rate, and the hydrogen bond partners from initial NMR structures calculated without the hydrogen bond restraints. The total number of non-reAIln-l~nt exp~rirrl~nt~lly-derived restraints was 1166.
- The ~L~u;~ ;S were in excellent agreement with the NMR experiment~l restraints.
There were no distance violations greater than 0.4 A, and no dihedral angle violations greater - 35 than 5 degrees. In addition, the cimll~ ed energy for the van der Waals repulsion term was small, in~lic~ting that the structures were devoid of bad inter-atomic contacts.The NMR structures also exhibited good covalent bond geometry, as indicated by small bond-length and bond-angle deviations from the corresponding i-le~li7eA parameters.

The average atomic root mean square deviation of the 8 ~LlucLul~S for residues 93-247 from the mean coordinates was 0.93 A for backbone atoms (Ca, N, and C'), and 1.43 A for all non-hydrogen atoms.
A ribbon plot of the ternary complex involving stromelysin, acetohydlu,~amic acid (the first ligand), and the second ligand is shown in Fig 10. The structure is very sirnilar to the global fold of other matrix metallo~ Leillases and consists of a five-s~n~led ~-sheet and three a-helices.
The catalytic zinc was located in the binding pocket. It was cooldil~a~d to three hi~fi~in~s and the two oxygen atom of acetohy~ a,l ic acid. A biaryl group of the second 10 ligand was located in the S 1 ' pocket ~Lwcen the second helix and the loop formed from residues 218-223. This deep and narrow pocket is lined with hydrophobic residues which make favorable contacts with the ligand.
Based on the three~ n~ LIUl;Lu~c of the ternary complex as ~let~ yl above and the structure/activity relationships observed for ~e binding to stromelysin of ~LIuCluLal 15 analogs of the second ligand (i.e., other biaryl compounds), new molecules were cle~igned that linked together the acetohydroxamic acid to biaryls.
As shown in Table 4 below, the initial biaryls chosen contained an oxygen linker and the ~bs~n~e or presence of CN para to the biaryl link~g~ Initial lilLkers collL~.ed varying lengths of methylene units. Means for linking compounds with linkers having varying 20 lengths of methylene units are well known in the ar~
Table 4 H

HO' N~X~

R

CA 02237336 l998-05-ll X R Stromelysin Compound Tnhibiti~n 21 (cH2)2 H 0.31 ~M
22 ~CH2)3 H 110 ,uM
23 (CH2)4 H 38%@ 100 ~I
24 (CH2)5 H43%@ 100 ,uM
(cH2)2 CN 0.025 ~LM
26 (CH2)3 CN 3.4 ~LM
27 (cH2)4 CN 3.5,uM
28 (CH2)5 CN 1.7 IlM

As ~;~e ;~d based on the better binding of the CN substi~uted biaryls to stromelysin, the CN derivatives exhibited better stromelysin inhibition. The compound that exhibited the 5 best inhibition of stromelysin con~ l a linker with two methylene units.
The present invention has been described with .c~r~,nce to ~l~r~ ,d embof1imf~nt.c Those emboflimPntc are not limiting of the claims and specification in any way. One of ur~ ~y skill in the art can readily envision changes, modifications and alterations to those embo-1imt-ntc that do not depart from the scope and spirit of the present invention.

F.x~mple S
Desi~n of potent. novel inhibitors of FKBP
Studies were ~3t;l~llned to design novel ligands that bound to FK-binding protein (FKBP) .
The cloning, expression and plmfication of FKBP was accomplished as set forth inExample 1. An initial step in the design of the new ligand was the i~entific:~tion of a first ligand that bound to the FKBP target. Such ir~ntific~tion was carried out in accordance with a two-dirnensional 15N/lH NMR correlation s~ g process as disclosed above.
A variety of low -molecular weight fragments and analogs of several known potent20 immnnnsuppressants (i.e. ascomycin, rapamycin) were screened for binding to FKBP using the procedures as set forth in example 2. Of the compounds tested~ compound 29, below WO 97118469 PCT/US96/t8312 o~\ /
ro /--~ ,p OCI~3 ~_~N~ OCH3 best s:~ti~fi~l the selection criteria: it had a binding affinity for FKBP of 2 ~I (rrlca~u~ by fluoresence by the methods known in the art) and s~tllr~t~ the protein (~ 98% occupancy of the binding site) at ligand conc~nfrations of l mM.
The second step in the design process was the i~ nhfil~atif~n of a second ligand that 5 bound to the target FKBP at a site di~r~ from the binding site of compound 29. This was accomplished by s~ g compounds for their ability to bind to FKBP in the ~,~,sellce of .c~hlr~ting amounts of the ascomy(;ill fr~gm~nt analog (compound 29). Details of procedures for this second i~l~ntific~tion step are as set forth in ex~mr1e 2.
In an initial screen, a compound was found that co.lt~ a ben7~nili~1e moiety. Based 10 on this initial hit, structurally similar compounds were obtained and tested against FKBP. The structure of these ben7~nili~e. compounds is r~l~,sell~ed by the structure m, below (see Table 5 for definitions of Rl-R4).

Rl~ H Fi2 R4 m In the second round of screening, binding was assayed both in the presence and in the absence of saturating amount of compound 29 (l mM).
A structure-activity relationship was developed for these diphenyl amide compounds as set forth in Table A. Fig. 6 shows a reprçsçnt~tive two--lim--n~ional 15N/lH correlation spectrum before and after exposure of FE~BP to a diphenyl amide test compound. It can be seen from Fig. 6 that the compound caused ch~omic~l shifts of 15N sites such as those ~lesi~n~te~l I50, Q53, E54, and V55. These sites correspond to an isoleucine (Ile) at position 50, a gl~ min~ (Gln) at position 53, a glutamate (Glu) at position 54, and a valine (Val) at position 55 of SEQ ID NO # 7. Figure l l shows the correlation between the NMR binding data and a view of the NMR-derived three-dimensional structure of FKBP. The ability to locate the spec~fic binding site of a particular ligand is an advantage of the present invention.

WO 97/18469 PCT/US96/183i2 Some compounds only bound to FKBP in the presence of compound 29. Thus ~e binding affinity of some compounds was ~nh~ncel in the presence of compound 29. These results exemplify yet another i.npo~ t c~p~hility of the present screening assay which is the ability to identify compounds that bind to the protein in the ~lcsw-ce of other molecules.
Various bon7~nili~1e compounds were tested for binding to FKBP at mul*ple ligandconcentra*ons. The 15N/IH correla*ion spectra ge~ cl at each c~n~entr~*on were evaluated to quan*fy dirr~,rences in the spectra as a function of compound concentration. A binding or dissociation constant (Kd) was calculated, using standard procedures well known in the art, from those dirr r~,nces. The results of this study are shown in Table 5. The values for Rl-R4 in Table 5 refer to the corresponding positions in the structure m, above.

Table 5 Compound RI R2 R3 R4 Kd (mM) No.
OH OH H H 0.8 31 H H OH H 1.4 32 H H H OH 0.5 33 OH H H OH 0.1 34 OH H H H 0.6 OH H CH3 OH 0.5 36 H H H H >5.0 37 H OH H H >5.0 The data in Table S show the utility of a process of the present invention in determining dissociation or binding con~ between a ligand and a target molecule.The next step in the design process was to construct a ternary complex of the target FKBP, the first ligand and the second ligand. This was accompliched by exposing the FKBP
target to the two ligands under conditions that resulted in complex formation. The location and orientation of the ligands were then determined as described below.
The lH, 13C and 15N resonances of FKBP in the ternary complex were ~igne~l from an analysis of several 3D double- and triple-resonance NMR spectra. The ~ nm~ont process was aided by known ~si~nments of FKBP when complexed to ascomycin (R. Xu, et al., Biopolymers. 33: 535-550, 1993). lH sidechain and lSN/lH backbone resonances were i(içn~ifie-l from an analysis of three-dimensional (3D) HC(CO)NH spectra recorded with spectral widths of 200û Hz (39.5 ppm), 6250 Hz (12.5 ppm) and 8333 Hz (16.7 ppm) in the Fl(l5N~, F2(1H) and F3(1H) ~imencions, respectively, and with a data matrix of 46(tl) x CA 02237336 l998-0~-ll WO 97/18469 PCT/US96/183i2 80(t2) x 1024(t3) complex points and 16 scans per increment. lH andl3C ~ ,ch~in and Ca resonances were i~lçntifie-1 from an analysis of 3D HCCH-TOCSY spectra (L. Kay, et al. J.
M~n. Reson., 101b:333-337, 1993) recorded with spectral widths of 7541.5 Hz (60.0 ppm), 6250 Hz (12.5 ppm) and 8333 Hz (16.7 ppm) in the Fl(l3C), F2(1H) and F3(1H) ~im~,n~ion~, respectively, and with a data matrix of 48(tl) x 64(t2) x 1024(t3) complex points and 16 scans per in~ ~nl. Intermolecular NOEs be~n the ligand and FKBP were obtained from an analysis of a 3D 12C-filtered, 13C edited NOESY ~e.iL,unl. The pulse scheme con.~i~tçfl of a double 13C filter sequence (A. Ge,mmç~ .r, et al., J. Ma~n. Reson., 96:199-204, 1992) conc~tto.n~ted with a NOESY-HMQC sequence (S. Fesik, et al-, I. A~M.
Chem. Soc., 111:1515-1517, 1989). The ~L)e~L.um was recorded with a mixing tirne of 350 ms and a total of 46(tl, 13C) x 64(t2, llI) x 1024(t3, lH) complex points and 16 scans per ine,clllent. Spectral widths of 7541.5 Hz (60.0 ppm), 6250 Hz (12.5 ppm) and 8333 Hz (16.7 ppm) were used in the ~1(13C), F2(1H) and F3(~ im~n~ions, ~cspc.,Lively.
In all spectra, the 15N carrier frequency was set at 117.4 ppm, the 13C carrier frequency was set at 40.0 ppm, and the lH carrier frequency was set on the water resonance.
All spectra were recorded at 303K on a Bruker AMX500 NMR s~e~ mc,~l. The NMR data were processed and analyzed on Silicon Graphics COIII~)U~ ;. In all NMR exp~
pulsed field gradients were applied where applopiiate as described (A. Bax, et al., J. M~n ~eson, 99:638, 1992) to afford the ~u~plcssion of the solvent signal and spectral artifacts.
Q~ lr?.~lre detection in the indirecdy detected ~lim~,n~ion was accompli~h~1 by using the States-TPPI method (D. Marion, etal. ~. Am. Chem. Soc. 87: 1515-1517, 1989). ~inear prediction was employed as described (E. Olejniczak, et al., J. Magn. Reson., 87: 628-632, 1 990).
Distance restraints derived from the NOE data were cl~if-iPd into ~ree categories based on the NOE crosspeak intensity and were given a lower bound of 1.8 A and upper bounds of 3.0, 4.0 and 5.0 A. A total of 17 intermolecular distance restraints between the protein and cc,~ o~ d 33 and 10 intermol~cular distance ~ beL~ the protein and compound 29 were used to define the location and orientation of the compounds when bound to FKBP using the known three-dimensional coordinates for the FKBP protein stlucture. A
ribbon plot of the ternary complex involving FKBP, a fragment analog of ascomycin ~compound 29), and a bçn7~nilicle compound (compound 33) is shown in Figure 12.
Based on the three--lim~n~ional structure of the ternary complex as determined above and the structure activity relationships observed for the binding to FKBP of structural analogs of the second compound, a new molecule was ~le~ignt A that linked the ascomycin ~r~gn~ent analog to the ben7~nilic~e compound. This compound, shown below, Ho~N9~ '~

C~OCH3 IV

has a 19 nM affinity for FKBP as dclf . ~ d by fluolcsc~ ce titrations. This is a 100-fold 5 increase in potency over the as~ ycil~ fragment analog (compound 29) alone (Kd = 2 ~).

As shown by the above non-limitin~ examples, the present invention relates to a process for clesi,~nin~ a high-affinity ligand to a given target molecule, cnmI ri~ing a) iclentirying by the sclce~ g processes dçsrribef1 herein at least two ligands10 which bind to disting binding sites on the target molecu}e using mnlti-Tim~,n~innal NMR
spectroscopy;
b) forming at least a ternary complex by exposing the at least two ligands to the target molecule;
c) clel~,. ~--i--;--~ the three flimen~iQn~l structure of the complex and thus the spatial 15 orientation of the at least two ligands on the target mnlec~ ; and d) using the spatial orien~tinn ~let~,rminerl in step c) to design the high affinity ligand which ~LI u~;Luldlly resembles a combination of the at least two ligands which bind to distinct sites on the target molecule. Preferrably, the high-affinity ligand designe~l in the above process serves as or is the basis for a drug which binds to a given target molecule and 20 pclrcJlllls, in vitro and in vivo, a targeted theld~t;u~ic effect in m~mm~l~ incln~1ing hnm~n,~ in need of LlcaLnlellt thereof.
The process also relates to de~igning a high-affinity ligand to a given target molecule comprising:
a) identifying a first ligand to the target molecule using mnltirlimt-,n~innal NMR
25 spectroscopy;
b) idellLiry..lg a second ligand to the target molecule using mlll*~lim~-,n~innal NMR spectroscopy wherein the second ligand may be the same or different than the first ligand and wherein the second ligand binds to a different site on the target molecule than the first ligand;

c) forming a ternary complex by binding the first and second ligands to ~~e target molecule;
d) cle~ . . . i ..; . .g the three ~lim~n~iona'~ structure of the complex and thus the spatial ori~.nt~tion of the first ligand and the second ligand on the target molecule; and e) ~le~igning the ligand wh~ ,;n the spatial oritont~tinn of step d) is "~ u~l In the process ~lescrihe l above, the first and second ligands may have the identicle mnlecnl~r structure or formula wherein the moiety binds to at least two hinding sites on the target molecule. The ligand that is based upon the structural comhin~tion of the first and second ligands then serves as a drug lead or drug upon actual synthesis of that comhin~A compound and evaluation in the ay~~uL,liate biological assays. The synthesis of the comhinç~ ligand, high-affinity ligand, drug lead or drug is achieved through synthetic or biological means.
CollcG~Lually, as in~lie~tf~ ~~u~ lloUt the spe~ifin~tion, the first and second ligands are linked (joined) together by carbon atoms, he~loa~ll~s, or a comhin~tion thereof to form the ligand or drug lead. The processes described herein, of course, include ~yll~heses of the high-affinity ligand by linear or non-linear (convergent) means which nltim~t~ly produce the linked (combined) first, second or more ligands.
The first and second ligands may also have dirr~,lGnt molecular SLIU~;~UIGS and either of the ligands may or may not bind to the other (&stinct~ bin&ng site on the target molecule.
In more detail, the process of the invention also relates to ~le~igning a high affinity ligand to given target mnlecllle, compri~ing a) plel,~ing an isotopically-labeled target mnleclllF wherein said molecule is enriched with an NMR cletect~ble isotope;
b) generating a mlllti~1im~n~inn~ NMR spectra of the isotopically-~ l~1 target molecule;
c) screening ~e isotopically-labeled target molecule by exposing the target molecule to a plurality of compounds to identify by mnlti~1im~n~iQnal NMR spectroscopy at least a first and second ligand which bind to &stinct sites on the target molecule;
d) forming at least a ternary complex by exposing at least the first and second ligand to the isotopically-labeled target molecule;
e) ~lelf ~ g the spatial nri~n~hnn of the at least first and second ligand on the isotpically-labeled target molecule;
f) using t'ne spatial ori~nt~tion clet.-rmine~l in step e) to design the high affinity ligand based upon the comkin~tion of the at least first and second ligands. Of course, a plurality of ligands (1 + n) can be combined to form a high affinity ligand which has the spatial orientation of the I + n (n = 1 -~) combined lig~n(1.~. After the high-affinity ]igand has been llesigned~ the process may further include the step f) of making the high affinity ligand by synthetic or biological means. The at least two ligands (first and second ligand) may be linlced by carbon atoms (e.g. by methylene or alkylene units) or by heleluatollls (e.g. by CA 02237336 1998-05-ll nitrogen, oxygen, sulphur) or by other atoms which ,.,~ ;.. or a~plu~ e the spatial orientation of the 1 + n ligands to the target molecule. Depending upon the lig~n~e, the molecules may also be colllbi lled or joined (linked) d-- e~;lly to each other without intervening alkylene or het~,.ualulll linker units. The high affinity ligand produced from the 1 + n 5 combined ligands pç~ dbly shows an increase in binding potency to the target molecule in relation to any one of the 1 + n lig~n~l~, The present invention, therefore, incl~l~es hign-affinity ligands designed by the processes shown herein wherein said high-affinity ligand has an in.;l~,ase in binding potency (Kd) to the given target molecule over the at least two ligands which bind to distinct sites on the given target molecule.
The present invention also relates to a method for discovering high affinity ligands using ~, U~;IUl~ ,-activity relationships obtained from nuclear m~gntotic resonance wherein said method cc..~-p. ;~es constructing a high-affinity ligand from ligands which bind to a subsite of a target molecule by;
i) S~ g low molecular weight ( < 450 MW) compounds which bind to a~5 subsite 1 of the target molecule;
iu) screening analogs prepared from the initial results obtained in step i) to optimize binding to subsite 1;
iii) s~ ,nillg for low molecular weight (< 450 mw) compounds and cull~,sponding analogs which bind to a nearby binding site, subsite 2, of the target molecule 20 using mnltiAim.oneil~nal NMR spectroscopy to measure binding affinity; wh~,leill, after steps (i) - (iii), lead fragmentc are generated;
iv) combining lead fr~gment~ generated from steps i) - iii) to design a high affinit~y ligand. ~ombining can be achieved by synthetic or biological means. Synthetic means includes organic synthesis of the combined ligand. Biological means inchlcies ~r.l I l lr.11~ n or 25 generation of the combined ligand through a cellular vehicle or system. P~cr~ldbly, the target molecule is a polypeptide. The present invention also relates to the method as recited above wherein the combination of fr~gment~ produces a ligand with a higher binding potency (Kd) than the individual fragments to the target molecule.

W 0 97/18469 PCT~US96/18312 u~:~ LISTING

~l) GENERAL INFORMATION:
(i) APPLICANT: Fesik, Stephen W.
Hajduk, Philip J.
Olejniczak, Edward T.
(ii) TITLE OF INVENTION: Use o~ Nuclear Magnetic Resonance to Design Ligands to Target Biomolecules (iii) NUMBER OF SEQUENCES: 7 (iv) CORRESPONDENCE ADDRESS:
(A) ADDRESSEE: Steven F. Weinstock, Dept. 377 AP6D, Abbott Laboratories (B) STREET: l00 Abbott Park Road (C) CITY: Abbott Park (D) STATE: Illinois (E) COUNTRY: USA
(F) ZIP: 60064 (vi) COMPUTER READABLE FORM:
(A) MEDIUM TYPE: Floppy disk (B) COMPUTER: IBM PC compatible (C) OPERATING SYSTEM: PC-DOS/MS-DOS
(D) SOFTWARE: PatentIn Release #l.0, Version #l.30 (vi) CURRENT APPLICATION DATA:
(A) APPLICATION NUMBER:
(B) FILING DATE:
(C) CLASSIFICATION:
(viii) ATTORNEY/AGENT INFORMATION:
(A) NAME: Anand, Mona (B) REGISTRATION NUMBER: 34537 (ix) TELECOMMUNICATION INFORMATION:
(A) TELEPHONE: (847) 937-4559 (B) TELEFAX: (847) 938-2623 CA 02237336 l998-0~-ll W O 97/18469 PCT~US96/18312 (2) INFORMATION FOR SEQ ID NO:l:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 174 amino acids (B) TYPE: amino acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (xi) SEQUENCE DESCRIPTION: SEQ ID NO:l:
Phe Arg Thr Phe Pro Gly Ile Pro Lys Trp Arg Lys Thr His Leu Thr Tyr Arg Ile Val Asn Tyr Thr Pro Asp Leu Pro Lys Asp Ala Val Asp Ser Ala Val Glu Lys Ala Leu Lys Val Trp Glu Glu Val Thr Pro Leu Thr Phe Ser Arg Leu ~yr Glu Gly Glu Ala Asp Ile Met Ile Ser Phe Ala Val Arg Glu His Gly Asp Phe Tyr Pro Phe Asp Gly Pro Gly Asn Val Leu Ala His Ala Tyr Ala Pro Gly Pro Gly Ile Asn Gly Asp Ala His Phe Asp Asp Asp Glu Gln Trp Thr Lys Asp Thr Thr Gly Thr Asn Leu Phe Leu Val A$a Ala His Glu Ile Gly His Ser Leu Gly Leu Phe Hi~ Ser Ala Asn Thr Glu Ala Leu Met Tyr Pro Leu Tyr ~is Ser Leu Thr Asp Leu Thr Arg Phe Arg Leu Ser Gln Asp Asp Ile Asn Gly Ile Gln Ser Leu Tyr Gly Pro Pro Pro Asp Ser Pro Glu Thr Pro WO 97/18469 PCTnUS96/18312 (2) INFORMATION FOR SEQ ID NO:2:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 83 amino acids (B) TYPE: amino acid (C) STRAWDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (xi) SEQUENCE DESCRIPTION: SEQ ID NO:2:
Met Ala Thr Thr Pro Ile Ile His Leu Lys Gly Asp Ala Asn Ile Leu l 5 l0 15 Leu Cy5 Leu Arg Tyr Arg Leu Ser Lys Tyr Lys Gln Leu Tyr Glu Gln Val Ser Ser Thr Trp His Trp Thr Cys Thr Asp Gly Lys His Lys Asn Ala Ile Val Thr Leu Thr Tyr Ile Ser Thr Ser Gln Arg Asp Asp Phe Leu Asn Thr Val Lys Ile Pro Asn Thr Val Ser Val Ser Thr Gly Tyr Met Thr Ile (2) INFORMATION FOR SEQ ID NO:3:
(i) SEQUEN OE CHARACTERISTICS:
(A) LENGTH: 18 base pairs (B~ TYPE: nu~leic acid (C) STRAWDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:3:
GAAATGAAGA ~ CAA l8 CA 02237336 1998-0~-11 (2) INFORMATION FOR SEQ ID No:4:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 18 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:4:
GCGTCCCAGG ~ l~GAG 18 (2) INFORMATION FOR SEQ ID NO:5:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 28 ba~e pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID NO:5:

(2) INFORMATION FOR SEQ ID NO:6:
(i) SEQUENCE CHARACTERISTICS:
(A) LENGTH: 30 base pairs (B) TYPE: nucleic acid (C) STRANDEDNESS: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: DNA (genomic) (xi) SEQUENCE DESCRIPTION: SEQ ID No:6:

CA 02237336 l998-05-ll WO 97/18469 PCT~US96/183i2 (2) INFORMATION FOR SEQ ID NO:7:
(i) SEQUENCE CHARACTERISTICS:
~A) LENGTH: 107 amino acids (B) TYPE: amino acid (C) STRA~n~n~cs: single (D) TOPOLOGY: linear (ii) MOLECULE TYPE: peptide (xi) SEQUENCE DESCRIPTION: SEQ ID NO:7:
Gly Val Gln Val Glu Thr Ile Ser Pro Gly Asp Gly Arg Thr Phe Pro Lys Arg Gly Gln Thr Cys Val Val His Tyr Thr Gly Met Leu Glu Aqp Gly Lys Lys Phe Asp Ser Ser Arg Asp Arg Asn Lyq Pro Phe Lys Phe Met Leu Gly Lys Gln Glu Val Ile Arg Gly Trp Glu Glu Gly Val Ala Gln Met Ser Val Gly Gln Arg Ala Lys Leu Thr Ile Ser Pro Asp Tyr Ala Tyr Gly Ala Thr Gly His Pro Gly Ile Ile Pro Pro His Ala Thr Leu Val Phe Asp Val Glu Leu Leu Lys Leu Glu

Claims (18)

WHAT IS CLAIMED IS:

1. A process for designing a high affinity ligand to a given target molecule, comprising:
a) identifying at least two ligands to the target molecule which bind to distinct binding sites on the target molecule using multidimensional NMR spectroscopy;
b) forming at least a ternary complex by exposing the at least two ligands to the target molecule;
c) determining the three dimensional structure of the complex and the spatial orientation of the at least two ligands on the target molecule; and d) using the spatial orientation determined in step c) to design the affinity ligand.

2. A process for designing a high-affinity ligand to a given target molecule comprising:
a) identifying a first ligand to the target molecule using multidimensional NMR
spectroscopy;
b) identifying a second ligand to the target molecule using multidimensional NMR spectroscopy wherein the second ligand may be the same or different than the first ligand and wherein the second ligand binds to a different site on the target molecule than the first ligand;
c) forming a ternary complex by binding the first and second ligands to the target molecule;
d) determining the three dimensional structure of the complex and thus the spatial orientation of the first ligand and the second ligand on the target molecule; and e) designing the high-affinity ligand wherein the spatial orientation of step d) is maintained.

3. The process according to claim 2 wherein the first ligand is different than the second ligand.

4. A process for designing a high affinity ligand to a given target molecule, comprising:
a) preparing an isotopically-labeled target molecule wherein said molecule is enriched with an NMR detectable isotope;
b) generating multidimensional NMR spectra of the isotopically-labeled target molecule;
c) screening the isotopically-labeled target molecule by exposing the target molecule to a plurality of compounds to identify by mu1tidimensional NMR spectroscopy at least a first and second ligand which bind to distinct sites on the target molecule;
d) forming at least a ternary complex by exposing at least the first and second ligand to the isotopically-labeled target molecule;
e) determining the spatial orientation of the at least first and second ligand on the isotopically-labeled target molecule;
f) using the spatial orientation determined in step e) to design the high affinity ligand based upon the combination of the at least first and second ligands.

5. A process according to step 3 further comprising, following step f), g) making the high affinty ligand by synthetic or biological means.

6. A high-affinity ligand designed by the process of claim 1 wherein said high-affinity ligand has an increase in binding potency to the given target molecule over the at least two ligands which bind to distinct sites on the given target molecule.

7. A process of designing a drug that serves as a ligand to a given target molecule comprising the steps of:
a) identifying a first ligand to the target molecule using two-dimensional 15N/1H NMR correlation spectroscopy;
b) identifying a second ligand to the target molecule using two-dimensional 15N/1H NMR correlation spectroscopy;
c) forming a ternary complex by binding the first and second ligands to the target molecule;
d) determining the three dimensional structure of the ternary complex and thus the spatial orientation of the first and second ligands on the target molecule; and e) linking the first and second ligands to form the drug, wherein the spatial orientation of step (d) is maintained.

8. The process of Claim 7 wherein the identification of the first ligand is accomplished by generating a first two-dimensional 15N/1H NMR correlation spectrum of a uniformly 15N-labeled target molecule, exposing the labeled target molecule to one or more chemical compounds, generating a separate two-dimensional 15N/1HNMR correlation spectrum for each of the compounds, and comparing each spectrum to the first spectrum to determine whether differences in those spectra exist, which differences would indicate the presence of a first ligand that has bound to the target molecule.

9. The process of Claim 7 wherein the identification of the second ligand is accomplished by generating a first two-dimensional 15N/1H NMR correlation spectrum of a uniformly 15N-labeled target molecule, exposing the labeled target molecule to one or more chemical compounds, generating a separate two-dimensional 15N/1H NMR correlationspectrum for each of the compounds, and comparing each spectrum to the first spectrum to determine whether difference in those spectra exist, which differences would indicate the presence of a second ligand that has bound to the target molecule.

10. The process of Claim 9 wherein the target molecule is bound to the first ligand before being exposed to the compounds.

11. The process of Claim 8 wherein the differences in the two-dimensional 15N/1H NMR correlation spectra are chemical shifts at particular 15N-labeled sites in the target molecule and chemical shifts in protons attached to those 15N-labeled sites.

12. The process of Claim 9 wherein the differences in the two-dimensional 15N/1H NMR correlation spectra are chemical shifts at particular 15N-labeled sites in the target molecule and chemical shifts in protons attached to those 15N-labeled sites.

13. The process of Claim 7 wherein the three dimensional structure of the ternary complex is determined using NMR spectroscopy or X-ray crystallography.

14. The process of Claim 7 wherein the target molecule is a polypeptide.

15. A drug designed by the process of Claim 1.

16. A method for discovering high-affinity ligands to target molecules using structure-activity relationships obtianed from nuclear magnetic resonance, comprising:
i) screening low molecular weight (< 450 mw) compounds which bind to a subsite 1 of a given target molecule using multidimensional NMR to measure binding affinity;
ii) screening analogs prepared from binding results obtained in step i) to optimize binding of a first fragment to the target molecule;
iii) screening for compounds and corresponding analogs which bind to a nearby binding site, subsite 2, of the target molecule using multidimensional NMR to measure binding affinity to optimize binding of a second fragment to the target molecule; and iv) combining the first and second fragments to design a high-affinity ligand.

17. A method according to claim 16 wherein the target molecule is a protein.

18. A method according to claim 16 wherein the high-affinity ligand has a higherbinding potency to the target molecule than the fragments thereof.