13
`
`
`
`. Tyrosine Kinase Inhibitors Against
`EGF Receptor-Positive Malignancies
`
`Elise A. Sudbeck, Sutapa Ghosh, Xing-Ping Liu,
`Yaguo Zheng, Dorothea E. Myers, and Fatih M. Uckun
`
`1. Introduction
`The role of protein tyrosine kinases (PTKs)in the survival of cancer cells
`and their potential use in anticancer therapy has led to their selection as anti-
`cancer drugtargets. Tyrosine kinases which are being studied for this purpose
`include epidermal growth factor receptor (EGFR) (—-6), Janus kinases (JAKs)
`(7-13), Bruton’s tyrosine kinase (BTK) (14-16), platelet-derived growth fac-
`tor (PDGF) (17), protein kinase C (PKC) (18-24), Lek (25,26), Trk (27-30),
`and others. The strategies used to attenuate or disable kinases implicated in
`cancer includethe use of antibodies, immunoconjugates, ligand-binding cyto-
`toxic agents, and small-molecule inhibitors. Each of these strategies has shown -
`some promise for the treatment of cancer. Herceptin (31-35), for example,is
`an immunotherapeutic agent that binds to the extracellular domain of HER2
`(also referred to as ErbB-2, a tyrosine kinase belonging to the same family as
`EGFR) at nanomolar levels. EGF-genistein (BGF-gen)is an EGFR-binding
`cytotoxic agentthat also showspotency in the nanomolar range (2,36) and will
`be discussed in this chapter. The search for new small molecules that inhibit
`kinases has involvedtraditional approaches, including the testing of natural
`products, random screening of chemical libraries, the use of classical struc-
`ture—activity relationship studies, and the incorporation of structure-based drug
`design approaches and combinatorial chemistry techniques. As a result, sev-
`eral promising small-molecule inhibitors have also been identified in recent
`years that may prove useful as potent new anticancer drugs.
`Small-molecule inhibitors of kinases that show. promise as anticancer agents
`include inhibitors of EGFR. EGF exerts pleiotropic biologic effects by binding
`
`From: Methods in Molecular Biology, vol. 166: Immunotoxin Methods and Protocols
`Edited by: W. A. Hall © Humana Press Inc., Totowa, NJ
`
`193
`
`j
`
`|
`
`j |
`
`
`
`OSI EXHIBIT 2003
`APOTEX V. OSI
`IPR2016-01284
`
`OSI EXHIBIT 2003
`APOTEX V. OSI
`IPR2016-01284
`
`

`

`194
`
`Sudbecketal.
`
`to ErbB-L (37-40). In breast cancer, the expression of EGFR is a significant
`and independent indicator for recurrence and poorsurvival (41-43). Recent
`studies provided evidence that EGFR serves as an endogenousnegative regu-
`lator of apoptosis in breast cancer cells (2). Many classes of small molecules
`have been reported in recent years that inbibit EGFR kinase (1,3-6,44~-47).
`Genistein (Fig. 1), a naturally occurring isoflavone found in soybeans,is an
`inhibitor of EGFR (6). When genistein is linked to EGF, the potency against
`EGFR increases from an ICsy value of >10 1 to a value in the nanomolar
`range (2).
`Pyrazolopyrimidines(see Fig. 1) were foundto inhibit EGFR with ICsp val-
`ues ranging around 1-8 nM (3). Two pyrazolopyrimidines with reported ICs
`values below 10 nM (3) also showedhigh selectivity towards some nonreceptor
`tyrosine kinases (c-Src, v-Abl, and serine/threonine kinases such as PKC-a
`and CDK1). The quinazoline derivative CP-358,774 (45) inhibits EGFR with
`an ICsg of 2 nM and reduces EGFR autophosphorylation in intact tumorcells
`with an ICsq of 20 nM. This inhibition is selective for EGFR relative to other
`tyrosine kinases examined as determined by assays of isolated kinases and
`whole cells. Despite the reported profound in vitro potency (K; = 5 pM) and
`selectivity of the ATP-competitive brominated quinazoline derivative PD153035
`(ig. 1; 4,5), the compoundfailed to show significant in vitro-orin vivoeffi-
`cacy against cancer cells. Other quinazolines reported include PD168393 and
`PD160678, which selectively target and irreversibly inactivate EGFR through
`covalent modification of a cysteine (Cys?”*) residue presentin the ATP-bind-
`ing pocket (44). These compoundsalsointeract in an analogous fashion with
`ErbB2 (which has a conserved Cys residue at the same position) but have no
`activity against IR, PDGFreceptor, FGFR, and PKC. The compoundshave not
`been tested against BTK and JAK3, which also contain conserved cysteine
`residues at the correspondingposition.
`A series of new quinazoline compoundstargeting EGFR have been designed
`more recently using structure-based methods.In this study, a three-dimensional
`modelof the kinase domain of EGFR wasconstructed (I) using known coordi-
`nates of homologous kinase domains as reference coordinates (Hematopoietic
`cell putative protein tyrosine kinase [HCK; 48], fibroblast growth factor recep-
`tor [FGFR; 49,50], and insulin receptor kinase [IRK; 57]). The EGFR model
`was used along with an inhibitor docking procedureforthe rational design of
`compoundspredicted to bind favorably to EGFR. The EGFR modelindicated
`that inhibition may be significantly improved by increasing the sizeof the func-
`tional groups attached to the 4-anilinoqninazoline molecular scaffold. Chemi-
`cally relevant substitutions at the 3', 4’ and 5' positions on the anilino ring lead
`to the successful design of a dibromo quinazoline derivative, WHI-P97, with
`an ICs9 value of 2.5 1M in EGFR kinase inhibition assays. WHI-P97 effec-
`
`

`

`EGFRInhibitors for Treatment of Cancer
`
`195
`
`.
`
`,
`
`N
`
`om
`:
`ahhBr
`a
`7
`Oo.
`HO
`Cy
`YN NH
`. C }
`W?.
`a He
`CO on
`OHO
`Pyrazolopyrimidine
`PD153035
`Genistein
`.
`derivative
`(,7,4'-trihydroxyisoflavone)
`HOURx =:52pM@
`
`
`
`4-6 nM (3), 101M (46)
`
`EGFR IC,)> 102M so =25pMEGFR IC,, = 1 nM (5),
`
`
`
`OQ. oeoseoO
`
`WHI-P97
`EGFR IC, = 2.5 pM
`
`LFM (A77 1726)
`EGER IC,= 5.4 1M
`
`PD168393
`EGER IC,, = 0.45 nM
`
`WHI-P154
`EGER IC, = 5.6 1M
`
`H3
`
`a
`
`ay
`
`LFM-A12
`EGER IC,, = 1.71M
`
`PD160678
`EGFR IC,, = 0.70 nM
`oO
`
`WHI-P131
`JAK IC, = 9.1 1M
`EGFR IC,, = 4.2 pM
`Br
`
`Hs!
`
`ng
`
`N Ni
`
`it
`
`LEM-A13
`BTK IC,, = 2.5 1M
`
`Fig. 1. Examples of tyrosine kinase inhibitors.
`
`tively inhibited the in vitro invasiveness of EGFR-positive human cancer cells
`in a concentration-dependent manner. The quinazoline derivatives WHI-P97,
`WHI-P131, WHI-P154, WHI-P180, and WHI-P197 with 3’ or 4'OH substitu-
`tion on the anilino moiety were predicted to form an additional hydrogen bond
`with Asp®3! in the ATP-binding region of EGFR that may enhancebinding.
`The EGFR inhibition values for WHI-P97, WHI-P131, WHI-P154, WHI-P180,
`and WHI-P197 ranged from 2.5 to 5.6 [LM in kinase assays. However, the
`
`

`

`196
`
`Sudbecket al.
`
`quinazolines tested were not specific for EGFR. For example, the EGFR inhib-
`itor WHI-P154 (EGFR ICsp = 5.6 LM) also inhibited other tyrosine kinases
`such as HCK (ICsp = 12 WM), JAK3 (Cs = 130 11.) andspleen tyrosine kinase
`(SYK) (C59 = 150 1) (46).
`‘
`°
`In terms of selectivity for EGFR, leflunomide metabolites may show the
`most promise. Earlier studies reported that the immunosuppressive activity of
`leflunomide is due to its metabolite A77 1726 (a-cyano-B-hydroxy-B-methyl-
`N-[4-(trifluromethyl)phenyl]-propenamide, or LFM), which is rapidly formed
`in vivo, functions as a pyrimidine synthesis inhibitor (52) and also inhibits the
`tyrosine kinase activity of EGFR (53). The leflunomide metabolite analog
`LEM-A12 (Fig. 1) showed inhibition of EGFR with an ICs9 value of 1.7 1M
`and killed >99% of human breast cancercells in vitro bytriggering apoptosis
`(1). Both LFM-A12 and WHI-P97 inhibited the in vitro invasiveness ofEGFR-
`positive human breast cancer cells at micromolar concentrations and induced
`apoptotic cell death. In addition, LFM-A12 inhibited the proliferation (ICs) =
`26.3 uM) andin vitro invasiveness (ICs5y = 28.4 1.4) of EGFR-positive human
`breast-cancercells in a concentration-dependentfashion.
`Like the quinazolines WHI-P97 and WHI-P154, the design of LFM-A12
`wasaided by a model of the EGFR kinase domain.In kinase assays, LFM-A12
`was found to be specific for EGFR and did not inhibit other-PTKs such-as
`BTK, HCK, JAKI, JAK3, IRK, and SYK at concentrations ranging from 175
`to 350 jtM. The observedselectivity of LFM-A12 for EGFR likely results from
`its molecular shape and from favorable interactions with unique EGFR resi-
`dues that are not present in the kinase domains of other PTKs. Likewise, unfa-
`vorable interactions with unique residues of other PTKsthat are not found in
`the EGFR kinase domain may also contribute to this selectivity. This observa-
`tion is in contrast to the observed inhibition of several kinases (EGFR, HCK,
`JAK3 and SYK) by WHI-P154. Thefirst contributing factor for the nonselec-
`tivity of WHI-P154 maybe the inhibitor’s complementary shape with the hinge
`region of the binding cavity of all seven kinases, which in turn leadsto favor-
`able hydrophobic contact between the compoundand the residuesin this cav-
`ity. Additionally, predicted hydrogen bonding interactions with all seven
`kinases may enhanceits binding with each of them.
`The structure-based method used to design leflunomide inhibitors of EGFR
`was also successfulfor the identification of small-molecule inhibitors of JAK3
`(7). JAK3 is expressed abundantly in primary leukemic cells from children with
`acute lymphoblastic leukemia (ALL). The constniction of a three-dimensional
`model ofJAK3 (7) was used to design a quinazoline inhibitor, WHI-P131 (ig. 1,
`shownto have specificity for JAK3. WHI-P131 inhibited JAK3 (Cs) = 9.1 pM)
`but not JAK1 or JAK2 anddid notinhibit the ZAP/SYK-family tyrosine kinase
`SYK, TEC-family tyrosine kinase BTK, Sre-family tyrosine kidase LYN, or
`
`

`

`EGFRInhibitors for Treatment of Cancer
`
`197
`
`the receptor-family tyrosine kinase IRK, even at concentrationsas high as 350 1M
`(7). WHI-P131 induced apoptosis in JAK3-expressing human leukemia cell
`_lines but not in JAK3-negative melanoma or squamouscarcinomacells.
`Anotherstudy to identify kinase inhibitors focused on inhibitors of BTK as
`antileukemic agents with apoptosis-promoting properties (14). A three-dimen-
`sional homology model of the BTK kinase domain was constructed(14) and
`- inhibitor docking proceduresled to the identification of an LFM analog, LFM-
`.A13 (Fig. 1), which was found to be a potent and specific inhibitor ofBTK. LFM-
`AJ3 inhibited recombinant BTK with an IC.value of 2.5 [LM,butit did notaffect
`the enzymaticactivity of other protein tyrosine kinases including JAK1 and JAK2,
`Sre-family kinase HCK, and receptor-family tyrosine kinases EGFR and IRK,at
`concentrations as high as 278 iM. LFM-A13 also enhaiiced the chemosensitivity
`of BTK-positive B-lineage leukemiacells to vincristine atidCeramide.
`Although several agents have been identified in recentyears that inhibit tyro-
`sine kinases such as EGFR, JAK3, and BTK,a future challengeis to ensure the
`specificity of inhibitors for one targeted tyrosine kinase. A successful strategy
`to accomplish this involves conjugating small-molecule inhibitors to ligand-
`binding entities; this enables the inhibitor to be delivered to a specific tyrosine
`kinase. An example of this strategy is to link a kinase inhibitor (soybean-
`derived genistein, 5,7,4'-trihydroxyisoflavone) to a protein (recombinantliuman
`EGF)that binds to a receptor kinase (EGFR). The resulting protein-inhibitor
`conjugate is an EGFR-directed cytotoxic agent (EGF-gen) with PTK inhibi-
`tory activity (2,36), which will be describedin further detail in this chapter. (A
`similar method was successfully appliedto the targeted delivery of genistein to
`CD19-receptor—associated vital PTK and shows considerable promise for more
`effective treatment of human leukemias and lymphomas/25,54))
`
`2. Materials and Methods
`
`2.1. Structure-Based Design of Small-Molecule Inhibitors of EGFR
`
`The three-dimensional coordinates ofthe EGFR kinase domain used in protein—
`inhibitor modeling studies (Fig. 2) were constructed based on a structural align-
`mentof the sequence of EGFR with the sequences of knowncrystal structures of
`other protein kinases (kinase domains ofHCK[48], FGFR [50], IR[55], and cAPK
`[56)) as described previously (1). The procedure was also usedto construct homol-
`ogy models for JAK1, JAK3 (7), BTK (14), and SYK (Mao, C., unpublished data).
`Molecular docking and scoring procedures were used to estimate binding of
`inhibitors in the catalytic site of EGFR (Tables 1 and 2, Figs. 3 and 4)(1,47).
`Leflunomide metabolite analogs such as LFM-A12 (Scheme1) and quinazoline
`compounds such as WHI-P97 (Schemes 2&3) were synthesized, and their ability
`to inhibit EGFR in breast cancercells was tested as previously described (1,57).
`
`

`

`198
`
`.
`
`Sudbecketal.
`
`2.2. EGF-Genistein Immunoconjugates
`2.2.1. Preparation ofEGF-Genistein Immunoconjugates
`Recombinant human EGF (chEGF) was produced in £. coli harboring a
`_ genetically engineered plasmid that.contains a synthetic gene for haman EGF
`fused at the N-terminus to a hexapeptide leader sequence for optimal protein
`expression and folding. The rhEGF fusion protein precipitated in the form of
`inclusion bodies, and the mature protein was recovered by trypsin cleavage
`followed by purification using ion-exchange chromatography and HPLC (2,36).
`The recently published photochemical conjugation method using the hetero
`bifunctional, photoreactive crosslinking agent sulfosuccinimidyl 6-(4’azido-2'-
`nitrophenylamino)hexanoate (Sulfo-SANPAH) (25)-wasused to prepare the
`EGF-genistein (EGF-gen) conjugate. Photolytic generationOf areactive sing-
`let nitrene on the other terminus of EGF-SANPAH in the presenceof a 10-fold
`molar excess ofdifferentially hydroxyl-protected genistein (gen) resulted in
`the attachmentofgenviaits available C7-hydroxyl group to the Lys?® or Lys*®
`residues of EGF. The resulting sample was purified using size-exclusion
`HPLC, and reverse-phase HPLC (2). Electrospray-ionization mass spectrom-
`etry (62,63) was used to determine the stoichiometry of gen and EGF in EGF-
`gen. }*J-gen was also used to confirm the stoichiometry of gen and EGFin
`EGF-gen and to verify the removal of free gen and gen-labeled EGF-EGF
`homoconjugates by the described purification procedure. The purity of EGF—
`125]_sen was assessed by SDS-PAGEandautoradiography.
`2.2.2. Binding of EGF-'*5|-Gen to Breast Cancer Cells
`Ligand-binding assays using EGF—!5I-gen (2.0 x 108 cpm/pimol), !*I-gen
`(3.8 x 108 cpm/umol) and !4I-EGF (2.2 x 10'* cpm/imol; Amersham) were
`performed using standard procedutes as previously described (25,64). Thecell
`lines in ligand-binding assays included the EGFR-positive breast cancer cell lines
`MDA-MB-231 and BT-20, as well as the EGFR-negative human leukemiacells
`lines NALM-6(pre-B leukemia) and HL-60 (promyelocytic leukemia).
`
`Fig. 2. (opposite page) Themolecular-surface representation of the homology
`model of the EGFR kinase domain. A small-molecule inhibitor (multicolor) is shown
`docked into the ATP-bindingsite (active site, yellow) ofEGFR. Taken fromn Ghosh et
`al., (1998) Clin. Cancer Res. 4, 2657-2668 (1).
`
`Fig. 3. (opposite page) Docked position ofLFM-A12 (yellow)in the catalytic site of
`the EGFR kinase domain model. The EGFR catalytic site residues are shown as space-
`filling atoms. Taken from Ghoshetal. (1999) Clin. Cancer Res. 5, 4264-4272 (46).
`
`
`emeeeee
`
`Nemaueearned:Petetenenets
`
`vere
`
`

`

`
`ee —T
`
`Fig.2.
`
`

`

`
`
`200
`
`.
`
`Sudbecketal.
`
`eat
`
`Raatg
`
`
`Fig. 4. The superimposed docked positions of several 4-anilinoquinazolines-in the
`catalytic site of the EGFR kinase domain model. Taken from Ghoshet al. (1999)
`Anticancer Drug Des., 14, 403-410 (47).
`
`Fig. 5. The effects of EGF-gen on tumor progression in SCID mice xenografted
`with MDA-MB-231 human breast cancer cells. White arrows indicate the approxi-
`mate size of the tumors in (A) a PBS-treated mouse (control) and (B) an EGF-gen—
`treated mouse after 60 d. Ruler units shown are in centimeters.
`
`—
`
`asmeeAeeteiaee
`
`

`

`“S—
`
`al,
`
`the
`19)
`
`[Japss00e7Zwos
`
`
`
`
`ISSEVCVCIN§«oTet-ag-vaN«(AM)DTTi)Ey,oDg:SsBION(%)(y)xpunoduop
`
`
`
`
`
`
`
`
`
`
`
`0}yuveur)azoydssqozdAueunpimouinjoajoArepunogsyjsepouoq*(69)urexdoidsysAffouuoDBusnpaye]nayedwaresoRpINsJeINI2;0Wy
`S0616861vs880S(145S0O®—s«OSZ*yo-dWaT
`£97vesLtel68osS6Ioe=DOF=ZLV-WT
`ry;Lsige.HeNO
`O0E<o0E<oor<oot<OvE9€POLIOZ=JGIP-S‘TSsELV-INST
`00€<o0€<SbL6991PO€pOLI0%ro|LY-WATvoniqiyuyYADs‘sone,'ypayzinayeg‘saioaguopoeIa}U]
`
`
`
`aoe<O0E<-O00T;<oat<vee16€Sol«EZIo“=O1V-WAT00E<00E<oor<oor<SpESBELSI,Le1ID- OV-WATo0E<o0E<-001<2%Sov26b681orefySY-WATaoe<O0E<oot'<ool<pSLOE8919€Zig-u6V-WHT
`
`
`00E<00€<oot<O01<LOEL6E.8STplzeUuON=PLYgoe<00E<ooli<ool<L8€aorE9181ZauTIV-WaT
`o0E<o0E<ool<001<Z9€L6E9112td€V-WIT
`
`Q0E<o0E<001<oor<SEtrp1816%"19-9PVT
`
`00g<ode<ool<001<E8Eber891OE1g-0SY-WAT
`
`O0E<00E<ool<LEery9p0818@Zlod=@V-WTone<-Q0E<oor<ze6¢P2pLLI9%jg=1V-WAT
`Gant)5]Anorxojorh-yuOnIqIyuT=IPN'TpnydeygsiSW
`
`
`
`
`Sy(1)sBojeuyWAT]0Ayo}xo}O}AQ[199J89UeDeAnISOg-YADTpue‘(°%D))
`
`
`
`
`
`
`
`
`(a2)(WD‘O8sICues‘sHOTENWISIENIS|OWE“TLSIsu])wresTordipuryaryuyuorourysso0syeownduraay)uopaseLytpn],
`
`
`
`
`
`
`
`‘suonisodpayoopuopaseq‘IpmryAqpoyepnoreourajosdtrayoeIMOOUrgoRTINSTe[Nds[OUrJoadeUsoIedety‘soRJINSPILING,
`
`‘gnoajoursipdnoyeuryeuysworeazoydspreyIIAOuINIOAOUBuLTEYsSNIPEwaATsJo(o[Naa[oUT3ayeMvJUasaidar
`
`
`
`
`
`
`
`
`
`
`
`
`(I)8997-LSOT“pSAN420UDD"UND(8661)Te39YsoUDWayuoyEL
`
`
`
`201
`
`
`
`

`

`202
`
`.
`
`Sudbecketal.
`
`Table 2
`EGFRInteraction Scores, Estimated K;, Values,
`
`and Measured!Cz, Data for 4-Anilinoquinazolines
`
`EGFR
`MS? BS? Lipo Ludi?—InhibitionLudi
`
`
`Compound R3 R4
`RS (A2
`(A?) Score HB‘ Score ‘Kj(uM)
`ICs (UM)
`WHI-P79
`Br H
`H 302
`208
`610
`1
`570
`2.0
`10.0
`(PD153035)
`2.5
`0.09
`701
`2
`655
`224
`329
`Br OH Br
`WHI-P97
`4.0
`0.60
`622
`1
`661
`226
`WHI-PI11 Br CH; H 305
`4.2
`0.40
`639
`2
`593
`202
`WHI-P131
`H OH H 290
`5.6
`0.14
`685
`2
`639
`218
`WHI-P154 Br OH H 307
`645 036 __ 40
`2
`599
`204
`WHLPI80 OH H
`H 292
`
`
`
`
`
`
`215 629 2 675 0.18WHLP197 Cl OH H 298 3.5aavaOC
`
`“Molecular surface area calculated using Connolly’s MS program (69). Defined as the bound-
`ary of volume within any probe sphere (meant to represent a water molecule) of given radius
`sharing no volumewith hard sphere atoms that make up the molecule.
`’8uried surface, the molecular surface in contact with protein calculated by Ludi based on
`docked positions.
`°The numberof hydrogen bonds between the protein andthe inhibitor.
`“Ludi K; values were calculated based on the empirical score function in the Ludi program
`(71). Ideal hydrogen bond distances and angles between compoundsandprotein are assumedin
`all cases for Ludi score and KX; calculation.
`Taken from Ghosh et al. (1999) Anticancer Drug Des. 14, 403-410 (47).
`
`2.2.3. Immunocytochemistry of EGF-Gen
`Immunocytochemistry was used to examine the surface expression of EGFR
`on breast cancer cells, to ‘evaluate the uptake of EGF-gen by breast cancer
`cells, and to examine the morphologic features of EGF-gen-treated cancercells.
`To detect the EGFR-EGF-gen complexes, cells were incubated with a mixture
`of a monoclonal antibody directed at the extracellular domain of haman EGFR
`and a polyclonal rabbit anti-gen antibody (2). After rinsing with PBS,cells were
`incubated with a mixtureof a goat antimouse IgG antibody conjugated to FITC,
`and donkeyantirabbit IgG conjugated to Texas Red, Cells were washed in PBS,
`counterstained with toto-3, and viewed using a confocal microscope.
`
`

`

`EGFRInhibitors for Treatment of Cancer
`
`203
`
`noJL, +nn\cits OG J,
`ciisoprepy
`.
`2-
`
`1
`
`HeFLT
`oO”
`iLFM-AI2
`
`CHscoct
`
`Scheme1. Synthesis of LFM-A12 (59,60).
`
`sO R ——r CH30
`
`CH30
`
`CHO
`
`R=3'5-diBr, 4-0H
`or 3°-Br,4-OH
`
`3
`4
`i ‘ 5
`=p
`
`HN
`
`SN
`
`2 7
`WHI-P97: A = 3'5'-diBr, 4'-OH °
`WHI-P154: R = 3'-Br,4'-OH
`
`Scheme 2. Synthesis of quinazoline derivatives WH1-P97 and WH1-P154 (58).
`
`9°
`°
`1)s
`CH,;0 OH1)80, CH30 NH, _£uSO4
`
`
`CH30
`2) NH,OH
`CH30:
`NaBH:
`wee
`NOz
`NO,
`
`(1)
`
`(@)
`
`1
`9
`
`
`
`
`CHy
`CH30
`CHO
`2
`NH2
`N
`
`CH; NH, _HCO,H=CHs0 POC CH30 “N
`
`(3)
`
`‘
`
`(5)
`
`Scheme 3, Synthesis of a quinazoline derivative precursor (61,62).
`
`2.2.4. In Vitro Treatment of Cells with EGF-Gen
`
`In order to determine the cytotoxic activity of EGF-gen against breast
`cancercells, cells were treated with various concentrations of EGF-gen and
`then used in either apoptosis assays or clonogenic assays (2). Controls
`included cells treated with G-CSF-gen (an irrelevant cytokine—gen conjugate
`that does not react with EGFR), cells treated with unconjugated EGF plus
`unconjugated gen, cells treated with unconjugated gen or unconjugated EGF,
`and cells treated with PBS, pH 7.4. In some experiments, excess G-CSF or EGF
`’ was added to the EGF-gen—containing treatment medium to showthat the cyto-
`toxicity of EGF-gen can beselectively blocked by excess EGF but not G-CSF.
`
`
`
`[men
`
`

`

`204
`
`.
`
`Sudbecket al.
`
`2.2.5. Immune-Complex Kinase Assays and Antiphosphotyrosine
`Immunoblotting
`
`-
`
`After treatment with EGF-Gen, cells were stimulated with EGF and cell
`lysates were immunoprecipitated with an anti-EGFR antibody reactive with
`the Ala35!_Asp*64 sequence of human EGFR. EGFR immune complexes were
`examined for tyrosine phosphorylation by Western blot analysis as previously
`described (66). All antiphosphotyrosine Western blots were subjected to den-
`sitometric scanning and for each time point a percent inhibition value was
`determined.
`
`2.2.6. Apoptosis Assays Using EGF-Gen
`
`MC540 binding and propidium iodide (PI) permeability (indicators of
`apoptosis) were simultaneously measuredin breast cancercells after exposure to
`EGF-gen(either without any cytokine preincubation or following preincubation
`with excess unconjugated EGF or G-CSF), unconjugated gen, unconjugated EGF
`plus unconjugated gen, or G-CSF-gen,as previously described (66). To detect
`the DNA fragmentation in apoptotic cells, cells were harvested after treatment
`with EGF-gen andDNAwas prepared from Triton-X-100detergentlysates for
`analysis of fragmentation as previously described (66).
`-
`van
`
`2.2.7, Clonogenic Assays
`
`After treatment with EGF-gen, G-CSF—gen, unconjugated EGF, unconju-
`gated gen, or PBS, cells were resuspended in clonogenic medium. Cells were
`cultured and cancer-cell colonies were enumerated on a grid using an inverted-
`phase microscope of high optical resolution. Results were expressed as the
`percent inhibition of clonogenic cells at a particular concentration of thetest
`agent. Dose-survival curves were constructed using the percent control sur-
`vival results for each drug concentration as the data points, and the IC.) values
`were calculated.
`
`2.2.8. Crossreactivity of Human EGF and Antihuman EGFR
`Antibodies with MouseEGFR
`Livers and thymus of BALB/c mice were frozen in liquid nitrogen and 5 pm—
`thick tissue sections were prepared using a cryostat. The sections were pro-
`cessed for standard indirect immunofluorescence using a monoclonal antibody
`directed at the extracellular domain of human EGFR as the primary antibody
`and a goat antimouse IgG conjugated to FITC as the secondary antibody (2). In
`parallel, sections were also stained by direct immunofluorescencestaining tech-
`niques with FITC-conjugated EGF.
`
`

`

`peneeeeumnaetnaornRennerCoNPAA
`
`pe
`
`EGFRinhibitors for Treatment of Cancer
`
`205
`
`2.2.9, Mouse Toxicity Studies
`The toxicity profile of EGF-gen in BALB/c mice was examinedas previ-
`ously reported for other biotherapeutic agents (25,67). In single-dosetoxicity
`studies, female BALB/c mice were administered an intraperitonealbolus injec-
`tion of EGF-gen in 0.2 mL PBS, or 0.2 mL PBSalone (control mice). In cumu-
`lative toxicity studies, mice received a total of 2800 lg (140 mg/kg) EGF-gen
`intraperitoneally over 28 consecutive days. Mice were monitored daily for
`mortality to determine the 30-d LDsg values (36).
`
`2.2.10. Treatment of SCID Mouse Xenograft Model of Human
`Breast Cancer
`The left hind legs of CB.17 SCID mice were inoculated.subcutaneously with
`MDA-MB-231 breast cancercells in 0.2 mL PBS. SCID miceinoculated with
`human breast cancer cells were treated with EGF-gen (36). Mice were moni-
`tored daily for health status and tumor growth. Primary endpoints of interest
`were tumor growth and tumor-free survival outcome. Estimationoflife table
`outcome and comparisons of outcome between groups were doneas previ-
`ously reported (6,25,67). The efficacy of EGF-gen against established tumors
`was examinedby treating SCID mice with subcutaneous MDA-MB-231 xeno-
`grafts with EGF-gen on 10 consecutive days and determining the tumor diam-
`eter daily for 20 d from the start of therapy. Control mice weretreated with 0.2
`mL PBSfor 10 consecutive days.
`
`2.2.11. Pharmacokinetic Studies of SCID Mice Treated
`with EGF-Genistein
`Tissue distribution studies in SCID mice were performed using EGF—!*I-
`gen and !?5]-gen as described previously (25). A flow-limited physiologic phar-
`macokinetic model was used to characterize the tissue disposition of EGF-gen
`in non—tumor-bearing as well as tumor bearing SCID mice (25,36,68).
`
`3. Results
`
`3.1. Structure-BasedDesign of Small-Molecule Inhibitors of EGFR
`3.1.1. Predicted Binding of LFM-A12 with the EGFR Kinase Domain
`Based on modeling studies of the kinase domain of EGFR, a binding mode
`was proposed for LFM analogs (Z). The predicted binding mode allows LFM
`analogs to maintain close contact with the hinge region of EGFR.Theinhibitor
`can fit into a space in the EGFR catalytic site defined by Leu®™ and Val’ on
`one side, and Leu®?° and Thr®*? on the other. Thenitrile nitrogen of LFM-A12
`waspredicted to interact with the amide of Met’®? via hydrogen bonding. In
`
`

`

`206
`
`.
`
`Sudbecketal.
`
`addition, the para-substituted OCF; group on LFM-A12 can form close con-
`tacts with residues Thr’®and Asp®#!, Interaction scores, calculated K, values,
`and measured IC.9data for several LFM analogsarelisted in Table 1. Of the
`compoundspredicted to interact most favorably with EGFR, LFM and LFM-
`. Al2 showedthe mostpotent activity against EGFR in kinaseassays, and LFM-
`A12 was the most cytotoxic against breast cancer cells.
`
`3.1.2, Modeling Studies of 4-Anilinoquinazoline Derivatives
`with the EGFR Kinase Domain
`
`Kinaseinhibition properties have also been evaluated for derivatives of 4-ani-
`linoquinazoline(4,5,47). In modeling studies aimed at identifying quinazoline
`derivatives with a high likelihood to bind favorably to theCatalytic site of the
`EGFR, K, values were estimated based on predicted binding interactions
`- between the inhibitor and catalytic-site residues of the EGFR (47). The model
`of the EGER binding pocketwas used in combination with docking procedures
`to predict the favorable placement of chemical groups with defined sizes on a
`molecular template. These studies led to the design of six quinazoline deriva-
`tives, WHI-P97, WHI-P111, WHI-P131, WHI-P154, WHI-P180, and WHI-
`P197. The various docked positions of each quinazoline dérivative were
`qualitatively evaluated in terms of an estimated K; value and consequently com-
`pared with the ICs values of the compounds in EGFR kinaseinhibition assays.
`Table 2 lists the interaction scores and estimated K;, valuesfor the quinazoline
`derivatives. This most favorable docked position allowedthe quinazoline inhib-
`itor to maintain close contacts with the hinge region of EGFR. The quinazoline
`moiety in the molecule can align itself along the hinge region ofEGFR,andthe
`N1nitrogen ofthe quinazoline group can form a hydrogen bondwith the back-
`bone carbonyl atom of the EGFR Met’?residue.
`In the final docking mode, the 6,7-OCH; groups ofthe inhibitor faced the
`solvent accessible region, and the anilino ring was surrounded by residues
`Thr76, Asp®3!, Thr®3°, and Val”, This modelofthe anilinoquinazoline bound
`to the EGFR kinase domain is consistent with that reported by others (71).
`WHI-P97, WHI-P131, WHI-P154, WHI-P180, and WHI-P197 with 3’ or 4'-OH
`substitutions on the anilino moiety can form an additional hydrogen bond with
`Asp®3!, WHI-P79 was the least potent inhibitor in the series evaluated,fol-
`lowed by WHI-P111. The activity of these two compounds maybe affected by
`the absenceof 3' or 4'-OH substitutions that can form a second hydrogen bond
`with EGFR residues and enhance binding.
`The crystal structures of the HCK—quercetin complex (48) and two FGFR—
`inhibitor complexes (50) revealed that PTK inhibitors can bind to the PTK
`catalytic (ATP-binding) site. When the catalytic sites of these PTK crystal
`
`

`

`ereaeneaaceetdAoeaNeeeeecereeeeARnAN
`
`structures were superimposed, all atoms of the three PTK inhibitors present
`fell within the plane of a triangle defining the ATP-binding region, and each
`molecule was in close contact with the hinge region and a conserved Aspresi-
`due (Asp*?! in EGFR). Theinhibitors characteristically occupied only half of
`the bindingsite, closest to the hinge region. The fact that the inhibitors reside
`close to the hinge region seemsto correlate with tighter binding and may be an
`important determinant for inhibitor binding. In the case of EGFR,the size and
`relatively planar shapeofthe catalytic site within the constructed EGFR kinase
`domain may contribute to its ability to form favorable interactions with mol-
`ecules such as quinazoline derivatives and LFM analogs. This observation was
`in good agreement with conclusions derived from structure~activity relation- .
`ships for pyrrolo- and pyrazoloquinazoline compounds (71) and wasincorpo-
`rated into the described modeling strategy forEGFR. ~~
`While most of the catalytic site residues of the EGFR kinase domain are
`conserved relative to other tyrosine kinases, a few specific variations are
`observed. In the EGFR modeling studies, the docked inhibitors were located
`between tworegions of mostly hydrophobic residues. EGFR residues on one
`side of the dockedinhibitor included Leu®*, Val’, Lys?2!, and Ala”!9, which
`are conserved in EGFR, HCK, FGFR and IRK. EGFR residues on the other
`side of the docked inhibitor included Leu®2° and Thr®°, whith varyint FGFR
`(Leu, Ala) and IRK (Met, Gly). EGFR residues Asn®!8 and Asp*?! (opposite
`the hinge) are conservedin all four PTKs. Residue Thr’in the EGFR hinge
`region changesto Val in FGFR,and to Metin IRK. Residue Thr®*° in EGFR
`changes to Ala in FGPR,and to Gly in IRK. Oneside of the binding pocket
`contains Cys’73 in EGFR andis therefore considerably more hydrophobic than
`the corresponding residue ofplatelet derived growth factor (PDGF)R (Asp),
`FGFR (Asn), and IRK (Asn). These residue identity differences may provide
`the basis for designing selective inhibitors of the EGFR tyrosine kinase.
`Thecatalytic site of EGFR may have specific features that can be advanta-
`geous for the design of inhibitors. Molecules that can fit into the triangular
`binding region of the EGFR catalytic site that can also form favorable contacts
`with the hinge region are likely to bind more strongly and inhibit EGFR more
`effectively. The docked position of LFM-A12 in the catalytic site of EGFR is
`shownin Fig. 3. It indicates that the molecule can maintain close contact with
`the hinge region of EGFR and can form a hydrogen bond betweenthenitrile
`nitrogen and the amide of Met’®,
`The docked positions of several quinazoline derivatives in the EGFR cata-
`lytic site are shownin Fig. 4. Like LFM-A12, quinazolines can maintain close
`contacts with the hinge region of EGFR,and the N1 nitrogen of the quinazoline
`group can hydrogen bond with the backbone carbonyl atom ofMet”, WHI-P97,
`which can form the most favorable interactions with EGFR (two hydrogen
`
`
` 4
`
`is
`ieni
`in
`eett
`iitie
`it
`ifot
`if
`
`pjae
`
`EGFRInhibitors for Treatment of Cancer
`
`207
`
`ngeaSenettngnnerapnpeteot
`
`
`TesteleottetgSeheSUreSieittWarWoeecerpameneroomet
`
`
`
`fo
`
`

`

`208
`
`:
`
`Sudbecketal.
`
`bonds and hydrophobic contacts), was the most potent of the quinazolines
`tested (EGFR ICs = 2.5 1M). However, in the EGFR model, the binding vol-
`ume of the EGFR-catalytic site is much larger than the volume occupied by
`WHI-P97 and LFM-A12.Increasing thesize of the ligand by using larger ring
`systems might increase the contact area between the receptor and ligand and
`thus enhancebinding. Interactions ofthe inhibitor with nonconserved residues
`such as Cys?>! and Thr®° in the catalytic site of EGFR mayalso beutilized for
`the design of more potentand selective inhibitors of EGFR.
`,
`
`3.2. EGF-Gen Immunoconjugates
`
`3.2.1. Biologic Activity of EGF-Gen Immunoconjugates
`Treatment of EGFR with EGF-genresulted in decreasédtyrosine phospho-
`rylation of the EGFR in a concentration-dependent fashion (2). Whereas EGF-
`gen exhibited marked PTK-inhibitory activity in MDR-MB-231 cells at
`concentrations as low as 0.1 1M in the treatment medium, unconjugated gen
`_ did not significantly affect the EGFR tyrosine phosphorylation even at a 10 uM
`concentration. The inhibitory effect of EGF-gen was blocked by preincubation
`of cells with excess EGF but not by excess G-CSF. Immune-complex kinase
`assays were usedto assessthe effects of EGF-gen on-the enzymatic activities
`of EGFR-associated Src PTK in MBA-MB-231 cells. EGF-gen treatmentinhib-
`ited the Src kinase. Unlike EGF-gen, a mixture of unconjugated gen and EGF
`or G-CSF-—gen did not inhibit the Src kinase activity in MDA-MB-231 cells.
`Thus, EGF-gen is a potent inhibitor of both the EGFR tyrosine