US 20090181415Al (19) United States (12) Patent Application Publication (10) Pub. No.: US 2009/0181415 A1 Bitter et al. (43) Pub. Date: Jul. 16, 2009

(54) PREDICTION OF GENOTOXICITY (22) Filed: Dec. 19, 2008

(76) Inventors: Hans Marcus Ludwig Bitter, San Related US. Application Data Francisco, CA (US); David Michael Goldstein, San Jose, CA (60) Provisional application No. 61/107,161, ?led on Oct. (US); Nina Gonzaludo, Palo Alto, 21, 2008, provisional application No. 61/015,291, CA (U S); Stephan Kirchner, ?led on Dec. 20, 2007. Loerrach (DE); Kyle Louis Kolaja, San Mateo, CA (US); Andrew Publication Classi?cation James Olaharski, Sunnyvale, CA (51) Int. Cl. (Us) C12Q 1/48 (2006.01) Correspondence Address: (52) U.S. Cl...... 435/15 ROCHE PALO ALTO LLC PATENT LAW DEPT. M/S A2-250 (57) ABSTRACT 3431 HILLVIEW AVENUE PALO ALTO, CA 94304 (US) The likelihood that a compound Will exhibit genotoxicity in a micronucleus test is predicted by the ability of the compound (21) Appl. No.: 12/339,499 to inhibit a plurality of from a selected group. US 2009/0181415 A1 Jul. 16, 2009

PREDICTION OF GENOTOXICITY PFTKl, PCTK1, PCTK3, CDK2, GSK3A, CDK3, CLK2, MELK, BRSK2, CAMK1, STK3, MYLK, CDK5, FLT3, FLT3.ITD, PRKR, and AMPKOt2, Wherein inhibition of at [0001] This application claims priority from Us. Ser. No. least tWelve of the 22 kinases by at least 50% indicates a 61/107,161, ?led Oct. 21, 2008 andU.S. Ser. No. 61/015,291, likelihood that said test compound Will demonstrate genotox ?led Dec. 20, 2007, both incorporated herein by reference in icity. If at least 12 of the 22 primary kinases are inhibited by full. 100%, this strongly and reliably indicates that the test com pound Would test as toxic in the MNT assay. Another aspect FIELD OF THE INVENTION of the invention comprises the method Wherein the group of [0002] This invention relates generally to the ?eld of toxi kinases further comprises one or more kinases selected from cology. More particularly, the invention relates to methods for the group of secondary kinases consisting of SLK, NUAK1, predicting genotoxicity, and methods for screening com CAMKK2, BRSK1, GSK3B, TTK, CAMK2G, ALK, AAK1, pounds for potential genotoxicity. ACVR2A, CLK1, BIKE, SNARK, LIMK2, PIP5K1A, STK16, LIMK1, DAPK1, PTK2B, CDK9, RPS6KA1.Kin. BACKGROUND OF THE INVENTION Dom.1, and CLK4. [0008] Another aspect of the invention is the method for [0003] The micronucleus test (“MNT”) is a common assay screening candidate compounds for potential genotoxicity, in the pharmaceutical industry routinely used to detect chro comprising providing a plurality of compounds; and deter mosome damage. A micronucleus forms When Whole chro mining the ability of each compound to inhibit the mosomes or fragments do not incorporate into activity of a number of kinases selected from the group con the daughter nuclei folloWing the completion of mitosis. sisting of CAMK2A, CAMK2D, DYRK1B, MAPK15, Aneugens and clastogens, chemicals Which cause chromo PCTK2, PFTKl, PCTK1, PCTK3, CDK2, GSK3A, CDK3, somal loss/ gain and breakage, respectively, Will cause signi? CLK2, MELK, BRSK2, CAMK1, STK3, MYLK, CDK5, cant increases in micronuclei formation and can be detected FLT3, FLT3.ITD, PRKR, and AMPKOt2, Wherein inhibition using the assay. Thus, micronuclei are biomarkers of chro of at least ?ve of said kinases by 100% indicates a likelihood mosome damage and the micronucleus assay is a sensitive that said test compound Will demonstrate genotoxicity. method to detect chemicals Which are aneugens and/or clas Another aspect of the invention comprises the method togens. The micronucleus assay is Widely used in the phar Wherein the group of kinases further comprises the group maceutical industry as evidence of genotoxicity (or lack consisting of SLK, NUAK1, CAMKK2, BRSK1, GSK3B, thereof). TTK, CAMK2G, ALK, AAK1, ACVR2A, CLK1, BIKE, [0004] HoWever, performing the micronucleus assay is SNARK, LIMK2, PIP5K1A, STK16, LIMK1, DAPK1, laborious and time consuming, false positive results can PTK2B, CDK9, RPS6KA1.Kin.Dom.1, and CLK4. occur When testing at cytotoxic doses, and large amounts of [0009] One aspect of the invention is a method for predict supplies (cells, reagents for cell-line maintenance, and com ing the genotoxicity of a compound, the method comprising pound) are required to perform the assay. providing a test compound; and determining the ability of the [0005] Kinases are responsible for phosphorylat compound to inhibit the kinase activity of a number of kinases ing substrates and disseminating inter- and intracellular sig selected from the group consisting of CDK2, CLK1, nals, including the initiation, propagation, and termination of DYRK1B, ERK8, GSK3A, GSK3B, PCTK1, PCTK2, chromosome replication during mitosis. Kinases are often STK16, TTK, CLK2, ERK3, and PRKR, or the group con targeted for inhibition by pharmaceutical companies because sisting of CDK2, CLK1, DYRK1B, ERK8 (MAPK15), many signaling cascades have knoWn roles in a variety of GSK3A, GSK3B, PCTK1, PCTK2, STK16, TTK, CDK17, diseases. Small molecule kinase inhibitors (SMKIs) often are CLK4, and PCTK3, Wherein inhibition of at least ?ve of said developed to competitively bind to the kinase ATP binding kinases by 100% indicates a likelihood that said test com pocket, blocking the ability of the to phosphorylate pound Will demonstrate genotoxicity. substrates. SMKIs often inhibit many kinases in addition to [0010] Another aspect of the invention is the method for the desired target due to the highly conserved nature of the screening candidate compounds for potential genotoxicity, ATP binding pocket Within the kinome, thus toxicities asso comprising providing a plurality of compounds; and deter ciated With off-target kinase inhibition is a concern for this mining the ability of each compound to inhibit the kinase pharmaceutical class of compounds. In particular, post activity of a number of kinases selected from the group con metaphase genetic toxicity, manifested as positive micro sisting of CDK2, CLK1, DYRK1B, ERK8, GSK3A, GSK3B, nucleus results, is a common toxicological liability for PCTK1, PCTK2, STK16, TTK, CLK2, ERK3, and PRKR, or SMKIs. the alternate group consisting of CDK2, CLK1, DYRK1B, ERK8 (MAPK15), GSK3A, GSK3B, PCTK1, PCTK2, SUMMARY OF THE INVENTION STK16, TTK, CDK7, CLK4, and PCTK3, Wherein inhibition [0006] We have noW invented a method for predicting or speci?c binding of at least ?ve of said kinases by 100% Which compounds Will demonstrate positive (i.e., genotoxic) indicates a likelihood that said test compound Will demon results in a micronucleus assay, using a method that is faster, strate genotoxicity. uses smaller quantities of reagents, and is easily automated. [0007] One aspect of the invention is a method for predict DETAILED DESCRIPTION OF THE INVENTION ing the genotoxicity of a compound, the method comprising De?nitions providing a test compound; and determining the ability of the compound to inhibit the kinase activity of a number of kinases [0011] Unless otherWise stated, the folloWing terms used in selected from the group of primary kinases consisting of this application, including the speci?cation and claims, have CAMK2A, CAMK2D, DYRK1B, MAPK15, PCTK2, the de?nitions given beloW. It must be noted that, as used in US 2009/0181415 Al Jul. 16, 2009

the speci?cation and the appended claims, the singular forms [0016] All patents and publications identi?ed herein are “a”, “an,” and “the” include plural referents unless the context incorporated herein by reference in their entirety. clearly dictates otherwise. [0012] The term “genotoxicity” as used herein refers to General Method compounds that produce chromosomal aberrations, including [0017] The invention provides a method for quickly deter breakage (clastogens) or abnormal copy number (aneugens). mining the likelihood that a given compound Will exhibit In this context, “genotoxicity” refers to a positive result in a genotoxicity in an MNT assay by examining the interaction micronucleus test. A “likelihood of genotoxicity” means spe betWeen the compound and a number of kinases (kinase bind ci?cally that the compound in question is predicted to dem ing and/ or inhibition) . As kinase inhibition and/ or binding can onstrate genotoxicity in an MNT With at least 75% con? be determined quickly, and by using automated methods, the dence. method of the invention enables high-throughput screening of [0013] The term “test compound” refers to a substance compounds for genotoxicity (or lack thereof). Which is to be tested for genotoxicity. The test compound can [0018] Thus, one aspect of the invention is a method for be a candidate drug or lead compound, a chemical interme predicting the genotoxicity of a compound, said method com diate, environmental pollutant, a mixture of compounds, and prising providing a test compound; determining the ability of the like. the compound to inhibit the kinase activity of at least ten [0014] The term “kinase” refers to an enZyme capable of kinases selected from the group consisting of CDK2, CLK1, attaching and/ or removing a phosphate group from a DYRKlB, ERK8, GSK3A, GSK3B, PCTK1, PCTK2, or molecule. “Inhibition of kinase activity” refers to the abil STK16, TTK, CLK2, ERK3, and PRKR, Wherein inhibition ity of a compound to reduce or interfere With such phos of at least ?ve of said kinases by 100% indicates a likelihood phatase activity. As binding af?nity of a small molecule for a that said test compound Will demonstrate genotoxicity. given kinase correlates Well With the ability of said molecule [0019] Another aspect of the invention is the method to inhibit the kinase activity, binding a?inity is considered described above, Wherein the second step further comprises synonymous With kinase activity herein, and high binding determining the ability of the compound to inhibit the kinase a?inity is considered equivalent to high kinase inhibitory activity of at least one kinase selected from the group con activity. The correlation betWeen binding a?inity and kinase sisting of MKNK2, SgK085, PIM2, TNNI3K, KIT, MELK, inhibition is described by M. A. Fabian et al., Nature Biotech AURKA, CLK3, AAK1, DCAMKL3, LIMK1, FLTl, nol (2005) 23:329-36, incorporated herein by reference in MAP2K4, PIM3, AURKB, ERK2, CSNK1A1L, DAPK3, full. MLCK, CLK3, PFTKl, PRKD3, AURKC, ERK5, STK17A, [0015] The term “primary kinases” refers to the folloWing MST4, CDK3, MYLK, CDC2L1, QIK, CDK11, , set of kinases (also identi?ed by accession number in paren PDGFRB, PRKCM, MAPK4, PIP5K2B, CSNKlD, theses): CAMK2A (N Pi741 960. 1), CAMK2D RPS6KA1.Kin.Dom.1, CDK5, PLK3, BIKE, PLK4, (AAD20442.1),DYRK1B (NPi004705.1), MAPK15 (NPi CAMK2A, STK3, CSNK2A1, STK17B, CDK8, MAP2K6, 620590.2),PCTK2 (CAA47004.1),PFTK1 (NPi036527.1), PIM1, MAP2K3, CDK7, IKKe, TGFBR2, CDK9, CLK4, PCTK1 (NPi006192.1), PCTK3 (NPi002587.2), CDK2 and PCTK3. ( dependent kinase 2, NPi001789.2), GSK3A (NPi [0020] Another aspect of the invention is the method 063937.2), CDK3 (NPi001249.1), CLK2 (NPi003984.2), Wherein the test compound is tested at a concentration of MELK (NPi055606.1), BRSK2 (NPi003948.2), CAMK1 about 10 [1M. Another aspect of the invention is the method (NPi003647.1), STK3 (NPi006272.1), MYLK (NPi Wherein the second step comprises determining the ability of 444254.3), CDK5 (NPi004926.1), FLT3 (NPi004110.2), the compound to inhibit the kinase activity of at least tWelve FLT3.ITD (NPi004110.2), PRKR (NPi002750.1), and kinases selected from the identi?ed group. Another aspect of AMPKOt2 (NPi006243.2). The term “secondary kinases” the invention is the method Wherein the second step com refers to the folloWing set: SLK (NPi055535.2), NUAK1 prises determining the ability of the compound to inhibit the (NPi055655.1), CAMKK2 (NPi006540.3), BRSK1 (NPi kinase activity of all kinases in the group. 1158061), GSK3B (NPi002084.2), TTK (NPi003309.2), [0021] Another aspect of the invention is a method for CAMK2G (NPi751913.1), ALK (NPi004295.2), AAK1 predicting the genotoxicity of a compound, by providing a (NPi055726.3), ACVR2A (NPi001607.1), CLK1 test compound; and determining the ability of the compound (AAA61480.1), BIKE (NPi060063.2), SNARK (NPi to inhibit the kinase activity of at least ten kinases selected 112214.1), LIMK2 (NPi005560.1), PIP5K1A (AAC50911. from the group consisting of CDK2, CLK1, DYRKl B, ERK8 1), STK16 (CAA06700.1), LIMK1 (NPi002305.1), DAPK1 (MAPK15), GSK3A, GSK3B, PCTK1, PCTK2, STK16, (NPi004929.2), PTK2B (NPi775267.1), CDK9 (NPi TTK, CDK7, CLK4, and PCTK3, Wherein inhibition of at 001252.1), RPS6KA1.Kin.Dom.1 (NPi002944.2), and least ?ve of said kinases by 100% indicates a likelihood that CLK4 (NPi065717.1). The term “identi?ed kinases” refers the test compound Will demonstrate genotoxicity. to the folloWing set of kinases (also identi?ed by accession [0022] Another aspect of the invention is the method number): CDK2 (NMi001798.2), CLK1 (NMi004071.1), Wherein the second step further comprises determining the DYRKlB (NPi004705.1), ERK8 (aka MAPK15, ability of the compound to inhibit the kinase activity of at NPi620590.2), GSK3A (D63424.1), GSK3B (NPi002084. least one kinase selected from the group consisting of 2), PCTK1 (NMi006201.2), PCTK2 (CAA47004.1), MKNK2, SgK085, PIM2, TNNI3K, KIT, MELK, AURKA, STK16 ( )NMi003691.1, TTK (NMi003318.2), CLK2 CLK3,AAK1, DCAMKL3, LIMK1, FLTl, MAP2K4, PIM3, (NMi003993.1), ERK3 (N Pi002739.1), and PRKR (NMi AURKB, ERK2, CSNK1A1L, DAPK3, MLCK, CLK3, 002759.1). “Alternate identi?ed kinases” refers to the set of PFTKl, PRKD3, AURKC, ERK5, STK17A, MST4, CDK3, kinases consisting of CDK2, CLK1, DYRKlB, ERK8 MYLK, CDC2L1, QIK, CDK11, PLK1, PDGFRB, PRKCM, (MAPK15), GSK3A, GSK3B, PCTK1, PCTK2, STK16, MAPK4, PIP5K2B, CSNKlD, RPS6KA1.Kin.Dom.1, TTK, CDK7, CLK4, and PCTK3. CDK5, PLK3, BIKE, PLK4, CAMK2A, STK3, CSNK2A1, US 2009/0181415 A1 Jul. 16, 2009

STK17B, CDK8, MAP2K6, PIMl, MAP2K3, CDK7, IKKe, [0030] Thus, one can prepare a test substrate having, for TGFBR2, CDK9, CLK4, and PCTK3. example, a plurality of immobilized kinases, preferably com [0023] Another aspect of the invention is the method prising a plurality the primary kinases or identi?ed kinases. In Wherein the test compound is tested at a concentration of one embodiment, the substrate comprises all of the primary about 10 [1M. kinases. In another embodiment, the substrate further com [0024] Another aspect of the invention is the method prises a plurality of the secondary kinases. In another embodi Wherein the second step comprises determining the ability of ment, the substrate comprises all of the primary and second the compound to inhibit the kinase activity of at least tWelve ary kinases. In another embodiment, the substrate comprises kinases selected from the group. all of the identi?ed kinases. In another embodiment, the sub [0025] Another aspect of the invention is the method strate further comprises a plurality of the alternate identi?ed Wherein the second step comprises determining the ability of kinases. In another embodiment, the substrate comprises all the compound to inhibit the kinase activity of all kinases in of the identi?ed kinases and the secondary kinases. The the group. Another aspect of the invention is a method for kinases can be immobilized directly (i.e., by adsorption, screening compounds for potential genotoxicity, comprising: covalent bond, or biotin-avidin binding or the like) to the providing a plurality of test compounds; and determining the surface, or indirectly (for example by binding to a ligand that ability of each compound to inhibit the kinase activity of at is tethered to the surface by adsorption, covalent bond, biotin least ten kinases selected from the group consisting of CDK2, avidin or other linkage). The kinases are then contacted With CLKI, DYRKIB, ERK8, GSK3A, GSK3B, PCTKI, the test compound(s), and the a?inity (or enzyme inhibition) PCTK2, STK16, TTK, CLK2, ERK3, and PRKR, or the determined, for example by measuring the binding of labeled alternate group consisting of CDK2, CLKl, DYRKIB, compound or loss of labeled competitor. ERK8 (MAPKIS), GSK3A, GSK3B, PCTKI, PCTK2, [0031] The kinase a?inity of each compound is measured STK16, TTK, CDK7, CLK4, and PCTK3; Where inhibition against at least ten of the 22 primary or 13 identi?ed kinases. of at least ?ve of said kinases by 100% indicates a likelihood Use of a larger number of kinases selected from these sets that said test compound Will demonstrate genotoxicity. results in a prediction of genotoxicity With higher con?dence. [0026] Another aspect of the invention is the method fur A compound With high total activity (for example, demon ther comprising rejecting compounds that demonstrate a like strating high af?nity for at least ?ve of the primary or identi lihood of genotoxicity. ?ed kinases, preferably eight or more) has a high likelihood of [0027] Another aspect of the invention is the method genotoxicity: this compound is predicted to test positive for Wherein the ability of the compound to inhibit the kinase genotoxicity in the MNT. A compound having loW total activ activity is determined by measuring the binding a?inity of the ity (for example, shoWing only loW a?inity for the selected compound for said kinases. kinases, or shoWing high a?inity to only l-4 selected kinases) [0028] Another aspect of the invention is a test substrate, is predicted to test negative in the MNT. comprising: a solid support; and immobilized on said solid [0032] Candidate drugs that test positive in the assay of the support, the kinases CDK2, CLKl, DYRKIB, ERK8, invention (i.e., that are predicted to demonstrate genotoxicity GSK3A, GSK3B, PCTKI, PCTK2, STK16, TTK, CLK2, in the MNT) are generally identi?ed as “genotoxic” or ERK3, and PRKR or the kinases CDK2, CLKl, DYRKIB, “potentially genotoxic”, and rejected or otherWise dropped ERK8 (MAPKIS), GSK3A, GSK3B, PCTKI, PCTK2, from further development. In the case of high-throughput STK16, TTK, CDK7, CLK4, and PCTK3. Another aspect of screening applications, such compounds can be ?agged as the invention is the test substrate of claim 14, further com toxic (for example, by the softWare managing the system in prising immobilized on said solid support, a kinase selected the case of an automated high-throughput system), thus from the group consisting of MKNK2, SgK085, PIM2, enabling earlier decision making. TNNI3K, KIT, MELK, AURKA, CLK3, AAKl, [0033] Thus, one can use the method of the invention to DCAMKL3, LIMKI, FLTl, MAP2K4, PIM3, AURKB, prioritize and select candidate compounds for pharmaceutical ERK2, CSNKIAIL, DAPK3, MLCK, CLK3, PFTKI, development based in part on the potential of the compound PRKD3, AURKC, ERKS, STK17A, MST4, CDK3, MYLK, for genotoxicity. For example, if one has prepared a plurality CDC2L1, QIK, CDKI, PLKI, PDGFRB, PRKCM, MAPK4, of compounds (e.g., 50 or more), having similar activity PIP5K2B, CSNKID, RPS6KA1.KD1, CDKS, PLK3, BIKE, against a selected target, and desires to prioritize or select a PLK4, CAMK2A, STK3, CSNKZAI, STK17B, CDK8, subset of said compounds for further development, one can MAP2K6,PIM1, MAP2K3, CDK7, IKKe, TGFBR2, CDK9, test the entire group of compounds in the method of the CLK4, and PCTK3. invention and discard or reject all those compounds that [0029] In practice, binding and inhibition can be deter exhibit positive signs of genotoxicity. This reduces the cost of mined using methods knoWn in the art. See, for example, M. pharmaceutical development, and the amount invested in any A. Fabian et al., Nature Biotechnol (2005) 23:329-36, incor compound selected for development by identifying an impor porated herein by reference in full. In general, the binding tant source of toxicity early on. Because the method of the a?inity of a compound for a given kinase correlates Well With invention is fast and easily automated, it enables the bulk the ability of the compound to inhibit the activity of that screening of compounds that Would otherWise not be possible kinase, so that binding af?nity is a reliable substitute for or practical. inhibitory activity. Binding a?inity may be determined by a [0034] Environmental pollutants and the like can also be variety of methods knoWn in the art; for example by competi identi?ed using the method of the invention, in Which case tive assay using an immobilized kinase (or an immobilized such compounds are typically identi?ed for further study into test compound, or an immobilized competing ligand, any of their toxic properties. In this application of the method of the Which may be labeled). Compounds and kinases can be invention, one can fractionate an environmental sample (for immobilized by standard methods, for example by biotinyla example, soil, Water, or air, suspected of contamination) by tion and capture on a streptavidin-coated substrate. knoWn methods (for example chromatography), and subject US 2009/0181415 A1 Jul. 16, 2009

said fractions to the method of the invention. Fractions that selected to observe acceptable toxicity (decrease of the rela display signs of genotoxicity can then be further fractionated, tive cell count (RCC) beloW 50%) or clear signs of precipi and (using the method of the invention), the responsible toxic tation in the aqueous medium. If the compound Was soluble agents identi?ed. Alternatively, one can perform the method and non-toxic, a maximal dose level of 5000 ug/mL Was set. of the invention using pure or puri?ed compounds that are For assessment of cytotoxicity, relative cell counts (RCC, as suspected of being environmental pollutants to determine % negative control) Were calculated. Slides Were prepared by their potential for genotoxicity. Because the method of the setting the cell density to approximately 1><106 cells/mL and invention is fast and easily automated, it enables the bulk centrifuging onto clean glass slides using a cytospin (1000 screening of samples that Would otherWise not be possible or rpm, 5 min). Fixation of cells and storage Was performed in practical. ice cold methanol (—200 C., at least 4 h). Slides Were incu [0035] The folloWing additional kinases can also be tested: bated for 5 min With H 33258 (1 ug/mL PBS/CMF) and high a?inity of a compound for one or more of these addi mounted With 10 [LL antifade for ?uorescence microscopy. A tional kinases (in addition to a majority of the primary or minimum of 3 concentration levels Were analysed for the identi?ed kinases) correlates With a higher likelihood of presence of micro-nucleated cells With the aid of an epi?uo genotoxicity. The additional kinases (and accession numbers) rescence microscope equipped With appropriate ?lter sets. A are: MKNK2 (NMi017572.1), SgK085 (NPi001012418. compound is considered to possess clastogenic/aneugenic 1), PIM2 (NMi006875.1), TNNI3K (NMi015978.1), KIT activity if one or more concentrations shoW at least a 2-fold (NMi000222), MELK (N Mi014791 .1), AURKA (NMi increase in the number of micronucleated cells in comparison 003600.1), CLK3 (NMi003992.1), AAK1 (NMi014911. to the concurrent negative control. 1), DCAMKL3 Q(Pi047355.6), LIMK1 (NMi002314.2), [0038] Fifty-four compounds Were selected for inclusion in FLT1 (NMi002019.2), MAP2K4 (NPi003001.1), PIM3 the training set, based on a number of criteria including selec (NPi001001852.1), AURKB (NMi004217.1), ERK2 tive kinase inhibition pro?les, minimization of redundancy, (NMi138957.1), CSNK1A1L (NMi145203.1), DAPK3 and chemical diversity. From an internal database of SMKIs, (NMi001348.1), MLCK (NPi872299.1), CLK3 (NMi only compounds that had selective kinase inhibition pro?les 003992.1), PFTK1 (NPi036527.1), PRKD3 (NPi005804. Were considered, Where a selective compound Was consid 1), AURKC (NMi003160.1), ERK5 (NPi002740.2), ered to be one that inhibited six or feWer kinases at single STK17A (NMi004760.1), MST4 (NMi016542.2), CDK3 point inhibition values greater than 95%, and eleven or feWer (NPi001249.1), MYLK (NPi444254.3), CDC2L11(NPi kinases at values greater than 85%. Kinase inhibition Was 277023.1), QIK Q(Mi041314.4), CDK11 (NPi055891.1), determined using the method set forth in M. A. Fabian et al., PLK1 (N Mi005030.2), PDGFRB (NMi002609.2), Nature Biotechnol (2005) 23:329-36. In cases Where a num PRKCM (N Mi002742.1), MAPK4 (N Pi002738.2), ber of compounds Were selective against many of the same PIP5K2B (NPi003550.1), CSNKlD (NMi001893.3), kinases, only one of the compounds Was selected, to minimiZe RPS6KA1 (KD1) (NMi002953.3), CDK5 (NPi031694.1), redundancy or over-representation of those kinases. After PLK3 (NMi004073.1), BIKE (NMi017593.2), PLK4 these ?ltering steps, a chemically diverse set Was selected (NMi014264.2), CAMK2A (NMi015981.1), STK3 (NPi based on physical properties, including A Log P, molecular 006272.2), CSNK2A1 (NMi001895.1), STK17B (NMi Weight, number of hydro gen donors and acceptors, number of 004226.1), CDK8 (NPi001251.1), MAP2K6 (NMi rotatable bonds, number of atoms, number of rings, number 002758.3), PIM1 (NMi002648.1), MAP2K3 (NPi002747. of aromatic rings, and number of fragments. Diversity Was 2), CDK7 (NPi001790.1), IKKe (NPi054721.1), TGFBR2 de?ned using the “Diverse Molecules” ?lter, based on a maxi (NMi003242.4), CDK9 (N Pi001252. 1), CLK4 (NMi mum dissimilarity method, in SciTegic’s Pipeline Pilot 6.0.2. 0206661), and PCTK3 (NPi002587.2). [0039] Inhibition pro?les against 290 kinases and in vitro MNT results Were acquired for each compound in the training EXAMPLES set (N:54). Three different readouts Were obtained for the MNT results: negative (N:22), positive (N:26), and Weakly Example 1 positive (N:6). The six Weakly positive Were assigned to [0036] To identify the set of kinases that Would indicate a either negative or positive labels based on the % MN cells at likelihood that a test compound Would demonstrate genotox the concentration at Which the inhibition pro?les Were per icity, the folloWing analysis Was carried out. First, 54 suitable formed. This led to ?ve of the six compounds being re small molecule kinase inhibitors (“SMKIs”) Were selected to assigned as negative, giving a total of 27 negative and 27 form a training set. Second, for each compound in the training positive compounds. set, an in vitro MNT result and single point inhibition pro?les [0040] Pre-processing Was ?rst performed across the set of against 290 kinases Were acquired. A statistical analysis Was all inhibition pro?les to remove uninformative or biased then performed to (1) build a model using said single point kinases. Kinases With no variance across the set of 54 com kinase inhibition pro?les to predict said MNT result and (2) pounds Were removed, as they Were not informative. JNK and identify the kinases correlated With MNT results. Finally, the p38 isoforms Were removed to reduce the bias of the large model Was validated against an additional set of 33 SMKIs number of compounds in the training set that Were developed not used for training. to target those kinases. To ensure that the removal of IN K and [0037] The in vitro micronucleus assay has been described p38 isoforms did not introduce a different form of bias, We in detail previously (M. Fenech, Mutation Res (2000) 455(1 performed an additional analysis Whereby We considered 2):81-95). The established permanent mouse lymphoma cell only those training set compounds not developed for these line L5178Y tk+/_ (ATCC CRL 9518), groWing in suspen kinase targets, and found that none of the IN K and p38 iso sion, Was used for this experiment. In general, compounds forms Were correlated With MNT results. Were tested up to 500 ug/mL, and at least 12 concentration [0041] Feature selection (FS) and pattern recognition (PR) levels Were tested. The top dose for evaluation Was generally Were performed in several phases in order to build the model. US 2009/0181415 A1 Jul. 16, 2009

For all analyses, cross validation Was used to assess the model CLK4, and PCTK3. The kinases selected for the tWo models performance over several trials. Each trial randomly split the are highly similar, demonstrating the robustness of the single initial data into a training set and a test set; the training set Was point kinase inhibition model. used to build the temporary model, and the test set Was used [0047] To assess the utility of the ?nal model, an additional to predict results and then verify performance. Feature selec set of 33 compounds Were used as a validation set. These 33 tion methods Were used to determine Which kinases, or “fea compounds Were not included in the initial set of 54, but each tures”, Were likely to correlate most With MNT result. In each compound included a single point inhibition value against the trial, the inhibition values against the features chosen Were thirteen model kinases, plus an in vitro MNT result. Given the used as input for a pattern recognition method, Which then validation data, the model Was able to accurately predict the predicted the positive or negative result. MNT result of all compounds, and thus performed With an [0042] In the ?rst phase, feature selection methods Were accuracy of 76%, Which lies Within our estimated accuracy of divided into tWo groups: methods that could handle a large the model based on cross-validation. input data set (PS 1 ), and methods that performed better With [0048] While the present invention has been described With less data (PS2). Different combinations of PS1, PS2, and PR reference to the speci?c embodiments thereof, it should be Were tested over several trials using 10 ?ve-fold cross-vali understood by those skilled in the art that various changes dations. The combination of methods With the loWest mean may be made and equivalents may be substituted Without error rate Was chosen for the next phase of the analysis. This departing from the true spirit and scope of the invention. In combination includes a Kolmogorov-Smimov/T-test hybrid addition, many modi?cations may be made to adapt a par algorithm for PS1, Random Forests for PS2, and Support ticular situation, material, composition of matter, process, Vector Machines for PR (T. Hastie et al., “The Elements of process step or steps, to the objective spirit and scope of the Statistical Learning” (2001, Springer-Verlag); R. O. Duda et present invention. All such modi?cations are intended to be al., “Pattem Classi?cation, 2'” Ed.” (2000, Wiley-Inter Within the scope of the claims appended hereto. science); and “Feature ExtractioniPoundations and Appli cations” (2006, Springer-Verlag, I. Guyon et al. Eds.)). Example 2 [0043] The chosen combination of methods from the ?rst phase Were tuned for optimal performance. Several param [0049] Proceeding as described in Example 1 above, but eters Were optimiZed, including the number of kinases to be employing an expanded set of training compounds (1 13 com used in the model. The tuning process shoWed that Within pounds instead of 54), the primary and secondary kinases several trials, the mean error rate Was loWest When the number Were identi?ed as more accurately predicting a positive of kinases chosen as signi?cant after PS1 and PS2 Was 13. (toxic) result in the MNT assay. The primary kinases identi Thus the model Was adjusted With the optimal parameters, ?ed are (accession number in parentheses): CAMK2A (N Pi then speci?ed to choose the 13 most signi?cant features as 7419601), CAMK2D (AAD20442.1), DYRKlB (NPi input for PR. 004705.1), MAPK15 (NPi620590.2), PCTK2 (CAA47004. [0044] The accuracy of the model using this combination of 1), PPTKl (NPi036527.1), PCTKl (NPi006192.1), feature selection and pattern recognition methods, number of PCTK3 (NPi002587.2), CDK2 (cyclin dependent kinase 2, features, and optimal tuning parameters Was then assessed by NPi001789.2), GSK3A (NPi063937.2), CDK3 (NPi performing 50 ?ve-fold cross-validations. Importantly, the 001249.1), CLK2 (NPi003984.2), MELK (NPi055606.1), feature selection and pattern recognition Was performed BRSK2 (NPi003948.2), CAMKl (NPi003647.1), STK3 Within each cross-validation fold. The resulting model had an (NPi006272.1), MYLK (NPi444254.3), CDKS (NPi accuracy of 80%:4%: that is, the model on average correctly 004926.1), PLT3 (NPi004110.2), PLT3.ITD (NPi004110. predicted MNT results 80% of the time. 2), PRKR (NPi002750.1), and AMPK0t2 (NPi006243.2). [0045] The 50 ?ve-fold cross-validations Were also used to The secondary kinases identi?ed are: SLK (NPi055535.2), determine the kinases correlated With MNT result. The selec NUAKl (NPi055655.1), CAMKK2 (NPi006540.3), tion of kinases Was based on the number of times a kinase Was BRSKl (NPi115806.1), GSK3B (NPi002084.2), TTK chosen as signi?cant amongst the 250 trials (50 ?ve-fold (NPi003309.2), CAMK2G (NPi751913.1), ALK (NPi cross-validations). 55 out of the original 290 kinases Were 004295.2),AAK1 (NPi055726.3),ACVR2A (NPi001607. chosen at least once as signi?cant. Those kinases that Were 1), CLKl (AAA61480.1), BIKE (NPi060063.2), SNARK chosen With a of greater than 50% (N:13) Were (NPi112214.1), LIMK2 (NPi005560.1), PIP5K1A selected to be included in the ?nal model. Over multiple runs (AAC50911.1), STK16 (CAA06700.1), LIMKl (NPi of testing, the kinase inhibition pro?les against these 13 002305.1), DAPKl (NPi004929.2), PTK2B (NPi775267. kinases Were found to be signi?cant in predicting actual MNT 1), CDK9 (NPi001252.1), RPS6KA1.Kin.Dom.1 (NPi result at least 50% of the time. That is, SMKIs With a positive 0029442), and CLK4 (NPi065717.1). in vitro MNT result tended to have high levels of inhibition [0050] Proceeding as described in Example 1 above, 113 against the thirteen kinases. small molecule kinase inhibitors Were screened for their abil [0046] For each SMKI, the model consists of single point ity to inhibit 290 kinases. The model Was developed as set kinase inhibition pro?les against the folloWing 13 kinases: forth in Example 1 above, except that micronucleus results CDK2, CLKl, DYRKlB, ERK8 (MAPK15), GSK3A, Were based upon concentration, such that positive micro GSK3B, PCTKl, PCTK2, STK16, TTK, CLK2, ERK3, and nucleus results occurring at concentrations above 10 [1M Were PRKR. Additionally, an in vitro MNT assay result at the reclassi?ed as negative, While results that Were positive beloW concentration in Which the kinase screen Was performed is that threshold Were classi?ed as positive. Thirty of the 113 included. A second model based upon quantitative binding small molecule kinase inhibitors Were classi?ed as positive, constants consisted a second (overlapping) set of thirteen Whereas 83 Were negative. All negative classi?cations Were kinases: CDK2, CLKl, DYRKlB, ERK8 (MAPK15), independent of concentration. Instead of using 250 trials (50 GSK3A, GSK3B, PCTKl, PCTK2, STK16, TTK, CDK7, ?ve-fold cross-validations), 500 trials Were used. US 2009/0181415 A1 Jul. 16, 2009

[0051] This resulted in identi?cation of 22 primary kinases, 5. The method of claim 3, Wherein step b) comprises deter inhibition of Which correlated strongly With positive (toxic) mining the ability of the compound to inhibit the kinase MNT results. The primary kinases identi?ed Were CAMK2A, activity of all kinases in said group. CAMK2D, DYRKlB, MAPK15, PCTK2, PFTKl, PCTKl, 6. A method for predicting the genotoxicity of a compound, PCTK3, CDK2, GSK3A, CDK3, CLK2, MELK, BRSK2, said method comprising: CAMK1, STK3, MYLK, CDK5, FLT3, FLT3.ITD, PRKR, andAMPKOt2. If a test compound exhibits inhibition of about a) providing a test compound; 100% against at least 12 of the 22 primary kinases, this model b) determining the ability of the compound to inhibit the predicts that it Will exhibit a positive (toxic) response in the kinase activity of at least ten kinases selected from the MNT assay. The likelihood of a positive MNT response cor group consisting of CAMK2A (N Pi741 960. 1), relates With the number of kinases inhibited, and the degree to CAMK2D (AAD20442.1), DYRKlB (NPi004705.1), Which they are inhibited. MAPK15 (NPi620590.2), PCTK2 (CAA47004.1), [0052] In addition, a further group of 22 secondary kinases PFTKl (NPi036527.1), PCTKl (NPi006192.1), Was identi?ed, inhibition of Which (in conjunction With one or PCTK3 (N Pi002587.2), CDK2 (cyclin dependent more primary kinases) correlates strongly With positive MNT kinase 2, NPi001789.2), GSK3A (NPi063937.2), results. The secondary kinases identi?ed Were SLK, NUAKl, CDK3 (NPi001249.1), CLK2 (N Pi003984.2), MELK CAMKK2, BRSKl, GSK3B, TTK, CAMK2G, ALK, AAK1, (NPi055606.1), BRSK2 (NPi0039482), CAMK1 ACVR2A, CLK1, BIKE, SNARK, LIMK2, PIP5K1A, (NPi003647.1), STK3 (NPi006272.1), MYLK (NPi STK16, LIMK1, DAPKl, PTK2B, CDK9, RPS6KA1.Kin. 444254.3), CDK5 (NPi004926.1), FLT3 (NPi Dom.l, and CLK4. Where a test compound exhibits inhibi 004110.2), FLT3.ITD (NPi004110.2), PRKR (NPi tion of the primary kinases, inhibition of several secondary 0027501), and AMPK0t2 (NPi006243.2), Wherein kinases further increases the probability of a positive MNT inhibition of at least ?ve of said kinases by 100% indi result. cates a likelihood that said test compound Will demon strate genotoxicity. What is claimed: 7. The method of claim 6, Wherein step b) further com 1. A method for predicting the genotoxicity of a compound, prises determining the ability of the compound to inhibit the said method comprising: kinase activity of at least one kinase selected from the group consisting of SLK (NPi055535.2), NUAKl (NPi055655. a) providing a test compound; 1), CAMKK2 (NPi0065403), BRSKl (NPi1158061), b) determining the ability of the compound to inhibit the GSK3B (NPi0020842), TTK (NPi003309.2), CAMK2G kinase activity of at least ten kinases selected from the (NPi751913.1), ALK (NPi004295.2), AAK1 (NPi group consisting of CAMK2A (N Pi741960. 1), 055726.3), ACVR2A (NPi001607.1), CLK1 (AAA61480. CAMK2D (AAD20442.1), DYRKlB (NPi004705.1), 1), BIKE (NPi060063.2), SNARK (NPi12214.1), LIMK2 MAPK15 (NPi620590.2), PCTK2 (CAA47004.1), (NPi005560.1), PIP5K1A (AAC50911.1), STK16 PFTKl (NPi036527.1), PCTKl (NPi006192.1), (CAA06700.1), LIMK1 (NPi002305.1), DAPKl (NPi PCTK3 (NPi0025872), CDK2 (N Pi001789.2), 004929.2), PTK2B (NPi775267.1), CDK9 (NPi001252. GSK3A (NPi063937.2), CDK3 (NPi001249.1), 1), RPS6KA1.Kin.Dom.1 (NPi002944.2), and CLK4 (NPi CLK2 (NPi0039842), MELK (NPi055606.1), 065717. 1). BRSK2 (NPi0039482), CAMK1 (NPi003647.1), 8. The method of claim 6, Wherein said test compound is STK3 (NPi006272.1), MYLK (NPi444254.3), CDK5 tested at a concentration of about 10 uM. (NPi004926.1), FLT3 (NPi004110.2), FLT3.ITD (NPi004110.2), PRKR (NPi002750.1), and 9. The method of claim 6, Wherein step b) comprises deter AMPKOt2 (N Pi006243.2), Wherein inhibition of at mining the ability of the compound to inhibit the kinase least ?ve of said kinases by 100% indicates a likelihood activity of at least tWelve kinases selected from said group. that said test compound Will demonstrate genotoxicity. 10. The method of claim 9, Wherein step b) comprises 2. The method of claim 1, Wherein step b) further com determining the ability of the compound to inhibit the kinase prises determining the ability of the compound to inhibit the activity of all primary kinases in said group. kinase activity of at least one kinase selected from the group 11. A method for screening compounds for potential geno consisting of SLK (NPi055535.2), NUAKl (NPi055655. toxicity, said method comprising: 1), CAMKK2 (NPi006540.3), BRSKl (NPi1158061), a) providing a plurality of test compounds; GSK3B (NPi0020842), TTK (NPi003309.2), CAMK2G b) determining the ability of each compound to inhibit the (NPi751913.1), ALK (NPi004295 .2), AAK1 (NPi kinase activity of at least ten kinases selected from the 055726.3), ACVR2A (NPi001607.1), CLK1 (AAA61480. group consisting of CAMK2A (N Pi741 960. 1), 1), BIKE (NPi060063.2), SNARK (NPi112214.1), CAMK2D (AAD20442.1), DYRKlB (NPi004705.1), LIMK2 (NPi005560.1), PIP5K1A (AAC50911.1), STK16 MAPK15 (NPi620590.2), PCTK2 (CAA47004.1), (CAA06700.1), LIMK1 (NPi002305.1), DAPKl (NPi PFTKl (NPi036527.1), PCTKl (NPi006192.1), 004929.2), PTK2B (NPi775267.1), CDK9 (NPi001252. PCTK3 (N Pi002587.2), CDK2 (cyclin dependent 1), RPS6KA1.Kin.Dom.1 (NPi002944.2), and CLK4 (NPi kinase 2, NPi001789.2), GSK3A (NPi063937.2), 065717. 1). CDK3 (NPi001249.1), CLK2 (N Pi003984.2), MELK 3. The method of claim 1, Wherein said test compound is (NPi055606.1), BRSK2 (NPi0039482), CAMK1 tested at a concentration of about 10 uM. (NPi003647.1), STK3 (NPi006272.1), MYLK (NPi 4. The method of claim 1, Wherein step b) comprises deter 444254.3), CDK5 (NPi004926.1), FLT3 (NPE mining the ability of the compound to inhibit the kinase 004110.2), FLT3.ITD (NPi004110.2), PRKR (NPi activity of at least tWelve kinases selected from said group. 0027501), and AMPK0t2 (NPi006243 .2); US 2009/0181415 A1 Jul. 16, 2009

wherein inhibition of at least ?ve of said kinases by 100% 0039482), CAMKl (NPi003647.1), STK3 (NPi indicates a likelihood that said test compound Will dem 0062721), MYLK (NPi444254.3), CDKS (NPi onstrate genotoxicity. 004926.1), FLT3 (NPi0041102), FLT3.ITD (NPi 12. The method of claim 11, further comprising: 0041102), PRKR (NPi002750.1), and AMPKotZ c) rejecting compounds that demonstrate a likelihood of (NPi0062432). genotoxicity. 15. The test substrate of claim 14, further comprising: Immobilized on said solid support, a kinase selected from 13. The method of claim 1, Wherein the ability of the compound to inhibit the kinase activity is determined by the group consisting of SLK (NPi0555352), NUAKl measuring the binding a?inity of the compound for said (NPi055655.1), CAMKK2 (NPi0065403), BRSKl kinases. (NPi115806.1), GSK3B (NPi0020842), TTK (NPi 0033092), CAMKZG (NPi751913.1), ALK (NPi 14. A test substrate, comprising: 0042952), AAKl (NPi0557263), ACVRZA (NPi A solid support; and 001607.1),CLK1 (AAA61480.1), BIKE (NPi060063. Immobilized on said solid support, the kinases CAMKZA 2), SNARK (NPi112214.1), LIMK2 (NPi005560.1), (NPi741960.1), CAMKZD (AAD20442.1), DYRKlB PIP5K1A (AAC50911.1), STK16 (CAA06700.1), (NPi004705.1), MAPK15 (NPi6205902), PCTK2 LIMKl (NPi002305.1), DAPKl (NPi0049292), (CAA47004.1),PFTK1 (NPi036527.1),PCTK1 (NPi PTKZB (NPi775267.1), CDK9 (NPi001252.1), 006192.1), PCTK3 (NPi0025872), CDK2 (cyclin RPS6KA1.Kin.Dom.1 (NPi0029442), and CLK4 dependent kinase 2, NPi0017892), GSK3A (NPi (NPi065717.1). 0639372), CDK3 (NPi001249.1), CLK2 (NPi 0039842), MELK (NPi0556061), BRSK2 (NPi