Advan. Enzyme Regul., Vol. 38, pp. 135–152, 1998
`# 1998 Elsevier Science Ltd. All rights reserved
`Printed in Great Britain
`0065-2571/98/$19.00 + 0.00
`
`PII: S0065-2571(97)00017-4
`
`M U L T I P L E F O L A T E E N Z Y M E
`I N H I B I T I O N : M E C H A N I S M O F A
`N O V E L P Y R R O L O P Y R I M I D I N E -
`B A S E D A N T I F O L A T E L Y 2 3 1 5 1 4
`( M T A )
`
`CHUAN SHIH, LILLIAN L. HABECK, LAURANE G. MENDELSOHN,
`VICTOR J. CHEN and RICHARD M. SCHULTZ
`
`Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN 46285,
`USA
`
`I N T R O D U C T I O N
`Since the late 1950 s, extensive research e€orts have been devoted to the
`discovery and development of antifolate antimetabolites as chemothera-
`peutic agents for the management of neoplastic diseases. However, it was
`in the last 10 to 15 years, due to the rapid advances of medicinal chem-
`istry, x-ray protein crystallography, molecular biology, pharmacology and
`clinical medicine, that a significant number of new generation antifolates
`were brought forward for clinical development. Several folate-based anti-
`metabolites are currently being investigated in clinical trials. These include
`Lometrexol
`(6R-5,10-dideazatetrahydrofolic
`acid, DDATHF)
`(1-3),
`LY309887 (4) and AG2034 (5), which exhibit potent and selective inhi-
`bition of glycinamide ribonucleotide formyltransferase (GARFT, EC
`2.1.2.2) of the purine de novo biosynthetic pathway; trimetrexate (6), eda-
`trexate (7, 8) and PT523 (9) which act on dihydrofolate reductase (DHFR,
`
`Abbreviations: r, recombinant; h, human; m, murine; TS, thymidylate synthase; DHFR,
`dihydrofolate reductase; GARFT, glycinamide ribonucleotide formyltransferase; AICARFT,
`aminoimidazole carboxamide ribonucleotide formyltransferase (EC 2.1.2.3); C1-S, C1 tetra-
`hydrofolate synthase; D/C, the protein domain of C1-S containing the 5; 10-methylenetetra-
`hydrofolate dehydrogenase (EC 1.5.1.5) and 5,10-methenyltetrahydrofolate cyclohydrolase
`activities; D/C/S, the full length enzyme of C1-S containing 5; 10-methylenetetrahydrofolate
`dehydrogenase, 5; 10-methenyltetrahydrofolate cyclohydrolase and 10-formyltetrahydrofolate
`synthetase activities (EC 6.3.4.3); FPGS, folylpolyglutamate synthetase; HEPES, N-[2-hydro-
`xyethyl]piperazine-N’-[2-ethanesulfonic acid]; MTT,3-[4; 5-dimethylthiazol-2yl]-2,5-diphenylte-
`trazolium bromide; DDATHF, 5; 10-dideazatetrahydrofolic acid; Lometrexol, 6R-DDATHF;
`ME, mercaptoethanol; NADPH, b-nicotinamide adenine dinucleotide phosphate; reduced
`form; ATP, adenosine 5’-triphosphate; 6R-MTHF, 6[R]-5; 10-methylene-5,6; 7,8-tetrahydro-
`folate;
`LY231514, N-[4-[2-(2-amino-3;
`4-dihydro-4-oxo-7H-pyrrolo[2,3-d]pyrimidin-5-
`yl)ethyl]-benzoyl]-L-glutamic acid; MTA, multitargeted antifolate.
`
`135
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`136
`
`C. SHIH et al.
`
`FIG. 1. The structure of N-[4-[2-(2-amino-3,4-dihydro-4-oxo-7H-pyrrolo[2,3-d]pyrimidin-5-
`yl)ethyl]-benzoyl]-L-glutamic acid, LY231514.
`
`EC 1.5.1.3); Tomudex1 (10, 11), AG337 (12), BW1843U89 (13) and
`ZD9331 (14) which target specifically at the enzyme thymidylate synthase
`(TS, EC 2.1.1.45) of the pyrimidine biosynthesis.
`N-[4-[2-(2-amino-3,4-dihydro-4-oxo-7H-pyrrolo[2,3-d]pyrimidin-5-yl)-
`ethyl]-benzoyl]-L-glutamic acid, LY231514 (Multitargeted Antifolate,
`MTA), is a structurally novel antifolate antimetabolite that possesses the
`unique 6-5 fused pyrrolo[2,3-d]pyrimidine nucleus instead of the more
`common 6-6 fused pteridine or quinazoline ring structure (15) (Fig. 1). As
`a ‘‘classical’’ antifolate, LY231514 was found to be one of the best sub-
`strates for the mammalian folylpolyglutamate synthetase (FPGS, EC
`6.3.2.17) and it is believed that polyglutamation and the polyglutamated
`metabolites of LY231514 play profound roles in determining both the
`selectivity and the antitumor activity of
`this novel agent
`(15, 16).
`Preliminary cell culture end-product reversal studies in human CCRF-
`CEM and murine L1210 leukemia cells have demonstrated that thymidine
`(5 mM) alone was not able to fully reverse the cytotoxic action of
`LY231514. Both thymidine (5 mM) and hypoxanthine (100 mM) were
`required to fully protect cells from the growth inhibitory activity exerted
`by LY231514 (15). This reversal pattern is significantly di€erent from
`such as methotrexate, Tomudex1 and the
`other known antifolates
`GARFT inhibitor DDATHF, and suggests that TS is only partially re-
`sponsible
`for
`the antiproliferative action of
`this novel antifolate.
`LY231514 and its polyglutamates may inhibit other folate-requiring
`enzymes, such as DHFR, or enzymes along the de novo purine biosyn-
`thetic pathway. This report summarizes our findings on the polyglutama-
`tion profile of LY231514, the activity of LY231514-polyglutamates (glu3
`and glu5) against various folate-requiring enzymes, the cell culture rever-
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`MULTIPLE ENZYME INHIBITION BY LY231514
`
`137
`
`sal and cross-resistance pattern of LY231514 and, finally, the e€ects of
`LY231514 on cellular folate and nucleoside triphosphate pools.
`
`M A T E R I A L S A N D M E T H O D S
`
`Materials
`Methotrexate, hypoxanthine and thymidine were obtained from Sigma
`Chemical Company, St. Louis, MO. Methotrexate polyglutamates were
`purchased from Dr. Schirks Laboratories, Jona, Switzerland. LY231514
`and Lometrexol were prepared at Lilly Research Laboratories,
`the g-glutamyl derivatives of
`Indianapolis,
`IN. The
`syntheses of
`LY231514 were by the method of Pawelczak et al. (17). L-[14C(U)]-glu-
`tamate (NEC-290E) was acquired from Dupont NEN, Boston, MA. The
`Enzfitter microcomputer package was purchased from Biosoft, Ferguson,
`MO. Human CCRF-CEM lymphoblastic leukemia cells were obtained
`from St. Jude Children’s Hospital, Memphis, TN. Human GC3/C1 cells
`were obtained from Dr.
`Janet Houghton of St.
`Jude Children’s
`Hospital, Memphis, TN. MCFTDX, H630, H630TDX, H630R10 cells were
`obtained from Dr. P. J. Johnston, Department of Oncology, Queen’s
`University, Belfast, North Ireland. HCT-8 cells were purchased from the
`American Type Culture Collection, Rockville, MD. The recombinant
`enzymes used were all obtained in purified form from the following
`sources: rhTS from Dr. D. V. Santi, University of California at San
`Francisco, San Francisco, CA; trifunctional mGARFT from Dr. R. G.
`Moran of Medical College of Virginia, Richmond, VA; rhDHFR from
`Dr. M. Ratnam of Medical College of Ohio, Toledo, Ohio and
`Anatrace Co.
`(Maumee, OH). Two forms of rhC1 tetrahydrofolate
`synthase were obtained from Dr. R. E. Mackenzie of McGill University,
`Montreal, Canada: these are (A) the 101kD D/C/S full length trifunc-
`tional enzyme, containing the activities of the dehydrogenase, cyclohy-
`drolase, and synthetase activities and (B) the D/C domain, the 35kD
`truncated version of C1-S which contained only the dehydrogenase and
`cyclohydrolase activities.
`
`Determination of Kinetic Constants for Hog Liver FPGS and
`Substrate
`Vmax and Michaelis constants for the conversion of folate analogues to
`their corresponding diglutamate forms were determined using FPGS puri-
`fied from mouse (18) and hog liver. Purification of the hog liver FPGS
`was carried out through chromatographing column step as previously
`described by Cichowicz and Shane (19). The pooled activity peak was dia-
`lyzed against 100 mM Tris-HC1, 50 mM b-ME, 20% glycerol, pH 8.4, ali-
`quoted, and stored at (cid:255)708C. An approximate 8000-fold purification was
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`138
`
`C. SHIH et al.
`
`achieved with a specific activity of 110 nmol/hr/mg when assayed with
`200 mM MTX and 250 mM L-[14C]-glutamate. Km and Vmax values were
`determined using 5 to 6 folate analogue concentrations in duplicate per
`experiment. The assay conditions were 100 mM Tris, 10 mM MgCl2,
`5 mM ATP, 20 mM KCl, 100 mg/ml BSA, 100 mM b-ME, 1 mM L-[14C]-
`glutamate (4 mCi/mmol), pH 8.9, in a final vol of 0.25 or 1.0 ml at 378C.
`The amount of protein added (1–6 mg) and incubation time (1–3 hr) varied
`to minimize the formation of higher polyglutamates at low substrate con-
`centration, and were determined to be in the linear range. Reactions were
`halted by addition of ice-cold 10 mM L-glutamate, pH 7.5. Labeled and
`unlabeled substrate was separated from unincorporated [14C]-glutamate by
`binding and elution from Waters Sep-Pak Plus C18 cartridges as pre-
`viously described by Jansen et al. (20). Km and Vmax values were deter-
`mined by non-linear fitting the data to a rectangular hyperbola using the
`Enzfitter microcomputer package.
`
`In Vitro Polyglutamation and Separation of Polyglutamates by HPLC
`Using the assay conditions listed above, the in vitro polyglutamation of
`selected folate analogues was examined. Reactions consisted of 1 or
`20 mM folate substrate, 2 mg of protein, 0.25 ml final vol, and were incu-
`bated for 1, 8, or 24 hr aseptically. Reactions were halted by boiling for
`3 min, and were centrifuged at 14,000(cid:2) g for 10 min to remove particu-
`lates prior to injection of 100 ml of sample. Folylpolyglutamates were sep-
`arated and quantitated using a modification of the reversed-phase HPLC
`method described by Montero and Llorente (21). Our protocol utilized a
`4.6 mm(cid:2) 25 cm Beckman Ultrasphere IP 5 mM column, 1.5 ml/min flow
`rate, and the following gradient elution: (1) 100% A (0.1 M ammonium
`acetate pH 5.5 + 1% acetonitrile) for 5 min, (2) linear gradient to 9% B
`(acetonitrile) over 20 min (3) 5 min at 9% B, (4) linear gradient to 100%
`B over 5 min, (5) linear gradient to 100% A over 5 min, (6) 2 min at
`100% A (total run time 42 min). The amount of polyglutamate product
`formed was calculated from the [14C]-glutamate incorporated in the radio-
`active peaks, and was corrected for the increasing specific activity of
`increasing glutamate chain length.
`
`Enzyme Assays and Methods
`The Ki values for TS, DHFR, GARFT, AICARFT and C1 tetrahydro-
`folate synthase and the tight binding kinetic analysis for LY231514-glu3
`for TS and DHFR were determined as described previously (22).
`
`In Vitro Cell Culture Studies
`Dose response curves were generated to determine the concentration
`required for 50% inhibition of cell growth (IC50). Test compounds were
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`MULTIPLE ENZYME INHIBITION BY LY231514
`
`139
`
`dissolved initially in DMSO at a concentration of 4 mg/ml and further
`diluted with cell culture medium to the desired concentration. CCRF-
`CEM leukemia cells in complete medium were added to 24 well Cluster
`plates at a final concentration of 4.8(cid:2) 104 cells/well
`in a total vol of
`2.0 ml. Test compounds at various concentrations were added to dupli-
`cate wells so that the final vol of DMSO was 0.5%. The plates were
`incubated for 72 hr at 378C in a 5% CO2-in-air atmosphere. At the end
`of the incubation, cell numbers were determined on a ZBI Coulter coun-
`ter. Control wells usually contained 4–6(cid:2) 105 cells at the end of the in-
`cubation. For several studies, IC50s were determined for each compound
`in the presence of either AICA (300 mM), thymidine (5 mM), hypox-
`anthine (100 mM), or combination of
`thymidine (5 mM) plus hypox-
`anthine (100 mM).
`For adherent tumor cells, we used a modification of the original MTT
`colorimetric assay described by Mosmann (23) to measure cell cytotox-
`icity. The human tumor cells were seeded at 1(cid:2) 104 cells in 100 ml of
`assay medium/well in 96-well flat-bottom tissue culture plates (Corstar,
`Cambridge, MA). Assay medium contained folic acid-free RPMI-1640
`medium supplemented with 10% fetal calf serum and either 2 nM folinic
`acid or 2.3 mM folic acid as the sole folate source. Well 1A was left blank
`(100 ml of growth medium without cells). Stock solutions of antifolates
`were prepared in Dulbecco’s phosphate-bu€ered saline (PBS) at 1 mg/ml,
`and a series of 2-fold dilutions were subsequently made in PBS. Ten
`microliter aliquots of each concentration were added to triplicate wells.
`Plates were incubated for 72 hr at 378C in a humidified atmosphere of 5%
`CO2-in-air. MTT was dissolved in PBS at 5 mg/ml, 10 ml of stock MTT
`solution was added to all wells of an assay, and the plates were incubated
`at 378C for two additional hours. Following incubation, 100 ml DMSO
`was added to each well. After thorough formazan solubilization, the
`plates were read on a Dynatech MR600 reader, using a test wavelength of
`570 nm and a reference wavelength of 630 nm. The IC50 was determined
`as the concentration of drug required to inhibit cell growth by 50% com-
`pared to an untreated control.
`
`Folate and Nucleotide Pool Studies
`For the folate pools study, the CCRF-CEM cells (2–5(cid:2) 105 per ml)
`were labeled with 100 nM of 3H-folinic acid (20 Ci/mmol) for 16 hr, then
`the drugs were added and incubation was allowed for the indicated dur-
`ation. The extraction and analyses of folate pools were conducted accord-
`ing to the methods published by Wilson and Horne (24). For the
`nucleotide pools study, CCRF-CEM cells were seeded in fresh complete
`medium at 3(cid:2) 105 cells per ml and cultured for 12 to 16 hr before drug
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`140
`
`C. SHIH et al.
`
`TABLE 1. SUBSTRATE ACTIVITY OF LY231514 AND OTHER ANTIFOLATES FOR
`MOUSE AND HOG LIVER FPGS
`
`Compound
`
`Km (mM)*
`
`rel Vmax*,$
`
`rel Vmax/Km%
`
`9.32 1.6
`0.802 0.11
`166.02 14
`
`1.0
`0.6320.18
`0.5020.09
`
`Mouse Liver FPGS
`DDATHF (Lometrexol)
`LY231514
`Methotrexate
`Hog liver FPGS
`16.42 1.0
`DDATHF (Lometrexol)
`1.0
`1.0
`1.92 0.5
`0.7420.10
`LY231514
`6.40
`116.02 14
`0.5120.08
`Methotrexate
`0.07
`*Values listed are mean2standard error for nr3 or 21/2 range for n = 2 replicate exper-
`iments.
`$The ratio of Vmax for a substrate to the Vmax of DDATHF with either mouse or hog
`liver FPGS.
`%Vmax/Km of a substrate relative to that of DDATHF.
`
`1.0
`13.7
`0.031
`
`treatment. The nucleotide analyses were performed as those described by
`Garrett and Santi (25).
`
`R E S U L T S A N D D I S C U S S I O N
`Polyglutamation Profile of LY231514
`Polyglutamation plays an essential role in determining the overall bio-
`chemical and pharmacological properties of the ‘‘classical’’ antifolates.
`The formation of polyglutamated metabolites of folates and antifolates
`results in the intracellular accumulation of polyglutamated metabolites to
`levels that are significantly higher than could otherwise be achieved at
`steady state by the parent compounds, and thus serves as an important
`cellular retention mechanism for folates and antifolates (26, 27). The
`unique 6-5 fused pyrrolo[2,3-d]pyrimidine nucleus structure of LY231514
`makes it an excellent substrate for the mammalian FPGS. The substrate
`activities of LY231514 and several other antifolates for mouse and hog
`liver FPGS are listed in Table 1 (16). Compared to other antifolates,
`LY231514 had Km values significantly lower than that of DDATHF (8 to
`11-fold) and MTX (60 to 207-fold) both for mouse and hog liver FPGS.
`The first order rate constant (k’, Vmax/Km) of LY231514 was also signifi-
`cantly greater than that of DDATHF and MTX, indicating again the su-
`perior capacity of LY231514 to be processed by FPGS to the diglutamate
`level. Recent studies indicated that LY231514 is an even better substrate
`for the recombinant human FPGS (R. G. Moran, personal communi-
`cation) and this makes LY231514 one of most e(cid:129)cient substrates for the
`enzyme FPGS studied to date.
`It has been shown that the e(cid:129)ciency of folate conversion to diglutamate
`by mammalian FPGS does not predict the e(cid:129)ciency of glutamate addition
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`MULTIPLE ENZYME INHIBITION BY LY231514
`
`141
`
`to longer chain length polyglutamates, and hence the distribution of poly-
`glutamates (28). This concern was addressed by the in vitro incubation of
`various antifolates with hog liver FPGS. After incubation of various com-
`pounds (MTX, Lometrexol and LY231514) with FPGS for di€erent
`
`FIG. 2. In vitro polyglutamation of MTX, Lometrexol and LY231514. Reaction mixes
`(0.25 ml) contained 1 mM (250 pmol) or 20 mM (5000 pmol) analogue, 5 mM ATP, 1 mM
`[14C]-L-glutamate, and were incubated for 1, 8, or 24 hr at 378C. The polyglutamated pro-
`ducts contained in 100 ml of reaction mix were separated by reversed-phase HPLC, and quan-
`titated by correcting for the increasing specific activity of the products. Total amount of
`polyglutamated product formed (pmol) is indicated for each analogue. Data represent an
`average of 3 separate experiments.
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`142
`
`C. SHIH et al.
`
`length of time (1, 8 and 24 hr), the resulting polyglutamates were separ-
`ated by reversed-phase HPLC and the extent of polyglutamation esti-
`mated (Fig. 2). The data clearly indicated that LY231514 can be rapidly
`and e(cid:129)ciently converted into the higher chain length polyglutamates (tri-,
`tetra- and pentaglutamates) either under low (1 mM) or high (20 mM) sub-
`strate concentrations. In comparison, methotrexate, which had the lowest
`relative first order rate constant (0.07 vs 6.40 for LY231514) was not con-
`verted beyond the diglutamate and yielded the least amount of total poly-
`glutamated product at all
`time points and with both substrate
`concentrations. The di€erence between the GARFT inhibitor Lometrexol
`and LY231514 was most apparent at 1 mM and 1 hr. While Lometrexol
`produced almost exclusively the triglutamate, LY231514 was converted
`mostly to the triglutamate (50%) and tetraglutamate (48%), with some
`small amount (2%) of pentaglutamate. After 8 hr (1 mM), the distribution
`of both compounds tended to shift to higher polyglutamates, and after
`24 hr the pentaglutamate became the predominate (76%) form of poly-
`glutamates for LY231514. At higher substrates concentrations, di€erent
`distributions of polyglutamates were observed. Under 20 mM substrate
`concentrations,
`it was found that the polyglutamate products of both
`Lometrexol and LY231514 were shifted to shorter chain length relative to
`the 1 mM reactions. This observation was consistent with reports in the lit-
`erature which indicate that substrate inhibition of FPGS activity may
`have occurred at higher concentrations. These data suggested that this
`pyrrolo[2,3-d]pyrimidine-based antifolate is an extremely e(cid:129)cient substrate
`for the enzyme FPGS. The polyglutamation reaction occurred rapidly and
`e(cid:129)ciently, and LY231514 was converted to long chain length polygluta-
`mates (tri-, tetra- and pentaglutamates) by FPGS and did not stop at the
`diglutamate stage.
`
`Folate Enzyme Inhibition Studies
`The
`inhibition of
`recombinant hTS, hDHFR, mGARFT and
`hAICARFT by LY231514 and its synthetic polyglutamates (glu3 and
`glu5) is summarized in Table 2. The parent monoglutamate LY231514
`inhibited rhTS with a Ki of 1092 9 nM. It has been well documented
`that mammalian TS showed a strong preference for polyglutamated folate
`substrates (29, 30). The longer chain g-glutamyl derivatives of LY231514
`demonstrated significantly enhanced a(cid:129)nity toward rhTS. The addition of
`two extra g-glutamyl residues (glu3) to LY231514 resulted in a 68-fold re-
`duction of the Ki value (Ki = 1.6 nM). Further extension of the gluta-
`mate tail (LY231514-glu5) only slightly increased the a(cid:129)nity toward rhTS
`(Ki = 1.3 nM). LY231514 was also found to be a very potent inhibitor
`for human DHFR (Ki = 7.0 nM). In contrast to rhTS, attachment of ad-
`ditional g-glutamyl residues to LY231514 had little e€ect on the inhibition
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`MULTIPLE ENZYME INHIBITION BY LY231514
`
`143
`
`TABLE 2. INHIBITORY ACTIVITY OF LY231514, METHOTREXATE AND THEIR
`POLYGLUTAMATES AGAINST RHTS, RHDHFR, RMGARFT AND RHAICARFT
`
`Ki (mean2s.e, nM)
`
`Compound
`
`rhTS
`
`rhDHFR
`
`rmGARFT
`
`rhAICARFT
`
`LY231514
`LY231514-glu3
`LY231514-glu5
`MTX
`MTX-glu5
`
`10929
`1.620.1
`1.320.3
`13,000
`47
`
`7.02 1.9
`7.12 1.6
`7.22 0.4
`0.004
`0.004
`
`9,3002690
`380292
`65216
`80,000
`2,500
`
`3,580
`480
`265
`143,000
`56
`
`of DHFR; the glu3 and glu5 of LY231514 exhibited identical Ki values
`against rhDHFR (7.1 nM and 7.1 nM, respectively). Tight binding analy-
`sis showed that LY231514-glu3 inhibited both TS (Fig. 3) and DHFR
`(Fig. 4) in a competitive manner.
`When LY231514 was tested against the enzymes along the purine de
`novo biosynthetic pathway,
`it only demonstrated moderate inhibition
`
`FIG. 3. Morrison analysis of tight-binding inhibition of rhTS by LY231514-glu3. A velocity
`versus inhibitor concentration curve is shown from a representative experiment illustrating
`the concentration-dependent inhibition of rhTS (29 nM) in the presence of [6R]-5,10-methyl-
`ene-5,6,7,8-tetrahydrofolate (6R-MTHF; 15 mM) and 100 mM deoxyuridylate monophos-
`phate. Inset, Ki app values were determined by the non-linear fitting of data collected at
`three concentrations of 6R-MTHF to the Morrison Equation using the ENZFITTER micro-
`computer package. The Ki value (1.3 nM) was determined from the slope of the graph
`Kiapp versus [6R-MTHF] using a Km for 6R-MTHF of 3.0 mM.
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`144
`
`C. SHIH et al.
`
`FIG. 4. Morrison analysis of tight-binding inhibition of rhDHFR by LY231514-glu3. A vel-
`ocity versus inhibitor concentration curve is shown from a representative experiment illustrat-
`ing the concentration-dependent inhibition of rhDHFR (16.7 nM) in the presence of [6R,S]-
`7,8-dihydrofolate (12 mM) and 10 mM NAFPH. Inset, Ki app values were determined by the
`non-linear fitting of data collected at three concentrations of [6R,S]-7,8-dihydrofolate to the
`Morrison Equation using the ENZFITTER microcomputer package. The Ki value (7.1 nM)
`was determined from the slope of the graph Kiapp versus [(6R,S)-7,8-dihydrofolate] using a
`Km for [6R,S]-7,8-dihydrofolate of 0.15 mM.
`
`toward GARFT (recombinant mouse, Ki = 9.3 mM). However, the triglu
`and pentaglu of LY231514 had significantly enhanced inhibitory activity
`against GARFT, with Ki values of 380 nM (24-fold) and 65 nM (144-
`fold), respectively. The pentaglu of LY231514 also inhibited human
`AICARFT (31) with a Ki of 265 nM. Kinetic analysis also demonstrated
`competitive inhibition pattern of LY231514-glun against both GARFT
`and AICARFT. Finally, LY231514 and its polyglutamates were competi-
`tive inhibitors of both the dehydrogenase and synthetase domains of C1
`tetrahydrofolate synthase. The Ki values for the mono-, tri- and pentaglu-
`tamyl derivatives of LY231514 were 9.9, 3.9 and 4.7 mM, respectively, for
`the dehydrogenase and 329, 25.4 and 1.6 mM for the synthetase. The in-
`hibitory activity of LY231514 on C1 tetrahydrofolate synthase was rela-
`tively less potent when compared with the activity against other targets
`such as TS, DHFR and GARFT. However, cell culture experiments have
`estimated that the intracellular drug concentrations of LY231514 can
`reach to 50 mM (R. M. Schultz, unpublished observation); at these con-
`centrations the activity of C1 tetrahydrofolate synthase can also be greatly
`suppressed through the inhibition by LY231514-glun. LY231514 thus has
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`MULTIPLE ENZYME INHIBITION BY LY231514
`
`145
`
`demonstrated inhibitory activity against multiple folate enzymes including
`TS, DHFR, GARFT, AICARFT, C1 tetrahydrofolate synthase and can
`be considered as a novel multitargeted antifolate (MTA). The important
`role of TS in serving as a rate-limiting enzyme in folate metabolism, as
`well as the relative order of inhibitory potency toward TS by LY231514-
`indicate that TS is a major site of action for LY231514.
`glun, all
`Inhibition of DHFR and other enzymes along the de novo purine biosyn-
`thetic pathway may also contribute significantly to the overall antiproli-
`ferative e€ect of LY231514. This unique mode of action was further
`supported by additional cell-based studies (vide infra).
`For comparison, it was reported that the polyglutamates of methotrex-
`ate (MTX) also inhibited multiple folate-dependent enzymes (32). It was
`reported by Chabner et al. (33, 34) that the pentaglutamate of MTX
`(MTX-glu5) demonstrated a significant increase in a(cid:129)nity toward rhTS
`(Ki = 47 nM) and AICARFT (Ki = 56 nM) when compared with the
`parent monoglutamate. However, the a(cid:129)nity of MTX (and its polygluta-
`mates)
`for DHFR (Ki = 4 pM) was
`several orders of magnitude
`(>12,000-fold) higher than its a(cid:129)nity for TS and AICARFT, suggesting
`that the primary intracellular target of MTX may still be at DHFR.
`
`Cell Culture End-product Reversal and Cross-resistance Studies
`LY231514 is a very cytotoxic agent for CCRF-CEM leukemia cells in
`culture. It was found that the potent antiproliferative e€ect of LY231514
`can be completely prevented by leucovorin whereas only partial protection
`was observed with thymidine. With the presence of 5 mM thymidine, the
`IC50 of LY231514 increased only by ca. 6–10 fold in contrast to complete
`protection observed for pure TS inhibitors (22). This reversal pattern of
`LY231514 was further characterized in various human tumor cell lines
`such as GC3/C1 colon carcinoma and HCT-8 ileocecal carcinoma. It was
`observed that 5 mM thymidine treatment of these cells only increased the
`IC50 of LY231514 by 18.7-fold (GC3/C1), and by 15-fold (HCT-8)
`(Table 3). Hypoxanthine (100 mM) alone did not markedly influence cyto-
`toxicity of LY231514. Similarly, AICA (300 mM) did not modulate cyto-
`toxicity. However, the combination of thymidine plus hypoxanthine can
`completely reverse
`the
`cytotoxicity of LY231514 in all
`cell
`lines
`(IC50s>20 mM). The reversal pattern of LY231514 was also significantly
`di€erent from that of MTX. Neither thymidine nor hypoxanthine could
`protect the cells from the cytotoxic actions of MTX at all drug concen-
`trations.
`The unusual reversal pattern observed for LY231514 also suggests that
`in addition to TS other important inhibitory sites may exist for this agent.
`The higher degree of protection by thymidine at low drug concentrations
`indicated that TS is a major target for LY231514. Addition of hypox-
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`146
`
`C. SHIH et al.
`
`TABLE 3. END-PRODUCT REVERSAL STUDIES WITH LY231514*
`
`Cell line
`
`CCRF-CEM
`GC3/C1
`HCT
`
`alone
`
`25
`34
`220
`
`LY231514 IC50 (nM)
`
`dThd$
`
`Hypoxanthine%
`
`dThd and
`Hypoxanthine
`
`138
`637
`3104
`
`32
`34
`1077
`
`>40,000
`>40,000
`>40,000
`
`*Cytotoxicity determined by MTT analysis after 72 hr exposure to drug, mean of triplicate
`determinations did not exceed 10% of mean.
`$With the addition of 5 mM thymidine (dThd).
`%With the addition of 100 mM hypoxanthine.
`
`anthine together with thymidine fully reversed the cytotoxicity of
`LY231514, suggesting that, at higher concentrations, inhibition of DHFR
`and/or purine de novo biosynthetic enzymes was responsible for other sec-
`ondary cytotoxic actions of the drug, a conclusion that is very much con-
`sistent with the enzymatic results.
`This hypothesis was further supported by examination of LY231514
`against cell lines that were developed and resistant to TS-specific inhibi-
`tors. For example, it was found that H630-R10 cells (35), which was re-
`sistant
`to the pyrimidine-based TS inhibitor 5-FU (with a 39-fold
`amplification of TS protein), demonstrated significantly less resistance
`toward the multitargeted antifolate LY231514 (Table 4). Similar results
`were obtained for a MCFTDX subline (36) (resistant to Tomudex1 with a
`
`TABLE 4. CYTOTOXICITY* OF LY231514 AGAINST RESISTANT CELL LINES
`DERIVED FROM MCF-7 HUMAN BREAST CARCINOMA AND H630 HUMAN
`COLON CARCINOMA CELLS
`
`Cell line
`
`MCF-7 (WT)
`MCFTDX$
`H630 (WT)
`x
`H630TDX
`H630R10**
`
`LY231514 IC50 (mM)
`
`Resistance factor
`
`0.0081
`0.7242
`0.1412
`0.630
`0.760
`
`89.4 (15,000)%
`
`4.5 (50,000)%
`5.4 (10,000)***
`
`*Cytotoxicity determined by MTT assay following 72 hr exposure to LY231514. Data rep-
`resent the mean of two separate experiments, each involving triplicate determinations. The re-
`sistant cell lines were obtained from Dr. P. J. Johnston, Department of Oncology, Queen’s
`University, Belfast, North Ireland.
`$Tomudex-resistant MCF-7 cell
`line with 20-fold TS gene amplification and 40-fold
`increase in intracellular TS protein levels.
`%Data in parenthesis: resistance factor of ZD-1694 in resistant lines, reported by P. J.
`Johnston et al. (ref. 35).
`xTomudex-resistant H630 cell line with 48-fold decrease in FPGS and 50-fold decrease in
`Tomudex transport.
`**5-FU-resistant H630 cell line with 39-fold increase in TS protein levels.
`***Data in parenthesis: resistance factor of ZD-1694 in resistant lines (P. J. Johnston, per-
`sonal communication).
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`MULTIPLE ENZYME INHIBITION BY LY231514
`
`147
`
`20-fold TS gene amplification and 40-fold increase in intracellular TS pro-
`tein levels) which also exhibited much less resistance to LY231514. These
`cross resistance patterns observed for LY231514 further support the ear-
`lier conclusion that TS is not the sole molecular target for this novel pyr-
`rolopyrimidine-based antifolate antimetabolite. It was also interesting to
`note that significant lower level of cross-resistance was observed for
`LY231514 toward H630TDX cells. H630TDX cells were highly resistant to
`quinazoline TS inhibitor Tomudex1 through impaired polyglutamation
`and transport, two important mechanisms for the cytotoxicity of ‘‘classi-
`cal’’ antifolates.
`
`Preliminary Results of the E€ects of LY231514 on Folate and
`Nucleoside Triphosphate Pools
`The e€ects of LY231514 and several other antifolates (methotrexate,
`LY309887) on cellular folate and nucleotide metabolism were examined in
`CCRF-CEM cells. It was discovered that the e€ects of methotrexate and
`the GARFT inhibitor LY309887 on cellular folate pools were fully con-
`sistent with their known mechanism of action. By labeling the CEM cells
`with 100 nM 5-formylTHF, it was observed that exposing the cells to
`0.1 mM MTX caused the loss of 10-formylTHF, THF (including 5,10-
`methyleneTHF) and 5-methylTHF and with concomitant accumulation of
`dihydrofolate (DHF), a metabolic response that is consistent with the
`blockade of DHFR by methotrexate. The GARFT inhibitor LY309887
`on the other hand caused accumulation of 10-formylTHF, a direct conse-
`quence from the inhibition of GARFT, which utilizes 10-formylTHF as
`the one carbon donor for its enzymatic reactions.
`Due to inability of the assay to distinguish THF and methyleneTHF,
`the e€ects of the multitargeted antifolate LY231514 on cells could not be
`readily studied. However, exposure of cells to LY231514 (300 nM, also 10
`times the IC50) seemed to have triggered a slight decrease in THF and a
`compensatory increase in the level of 10-formylTHF. In light of the
`observed accumulation of 10-formylTHF by the specific GARFT inhibitor
`LY309887, it is tempting to attribute this small, yet significant, increase of
`10-formylTHF to the inhibition of GARFT and AICARFT (antipurine
`e€ect) by LY231514-glun, a hypothesis which can be verified by direct
`measurement of the e€ect of LY231514 on the metabolic flux from glycine
`to inosinic acid.
`In the ribonucleotide pool studies, it was discovered that both MTX
`and LY309887 caused rapid depletion of both purines, ATP and GTP,
`and had moderate e€ects on the pyrimidines, UTP and CTP. However,
`LY231514 produced no significant e€ects on any of the ribonucleotide tri-
`phosphates at concentrations 10 times their IC50 in CEM cells. In con-
`trast, all
`three compounds demonstrated more dramatic e€ects on
`
`Lilly Ex. 2078
`Sandoz v. Lilly IPR2016-00318
`
`

`
`148
`
`C. SHIH et al.
`
`deoxyribonucleotide pools. In response to LY309887, dATP declined
`rapidly, followed closely by dCTP, and then later by dGTP and dTTP at
`a slower rate. Methotrexate rapidly depleted all four deoxyribonucleotide
`levels. The e€ect of LY231514 on deoxyribonucleotide levels is consistent
`with reports of other TS inhibitors (37). It was found that LY231514 was
`able to induce rapid losses in dTTP, dCTP, and dGTP. However, an
`increase of dATP level was observed for cells treated with LY231514. It
`will be interesting to examine the di€erence of the rate of accumulation of
`dATP induced by LY231514 (MTA) or by other specific TS inhibitors,
`since Chong and Tattersall (37) reported that the combination of a
`GARFT inhibitor and TS inhibitor prevented the rise in the dATP pool
`seen with the TS inhibitor alone. The mechanism for the changes in dATP
`levels induced by TS inhibitors has not been well understood. However, it
`is noteworthy that for methotrexate, a drug with both antipyrimidine and
`antipurine e€ects, depletion of dTTP occurred without a concomitant
`increase in dATP. In summary, these studies showed that LY231514
`exhibited unique metabolic e€ects that were quite distinct from those of
`MTX and LY309887. The folate pools (accumulation of 10-formylTHF)
`data suggest that in addition to the primary e€ect on thymidylate syn-
`thesis, LY231514 may produce an antipurine e€ect by interfering with the
`enzymes along the de novo purine biosynthetic pathway.
`
`S U M M A R Y
`