Leukemia (2020) 34:416–426
`https://doi.org/10.1038/s41375-019-0582-x
`
`ARTICLE
`
`Acute myeloid leukemia
`In vivo efficacy of mutant IDH1 inhibitor HMS-101 and structural
`resolution of distinct binding site
`
`● Thomas Klünemann2
`● Julia Weder2
`● Charu Gupta1
`● Ramya Goparaju1
`Anuhar Chaturvedi1
`Michelle Maria Araujo Cruz1
`● Arnold Kloos1
`● Kerstin Goerlich1
`● Renate Schottmann1
`● Basem Othman1
`Eduard A. Struys3
`● Heike Bähre4
`● Denis Grote-Koska5
`● Korbinian Brand5
`● Arnold Ganser1
`● Matthias Preller2
`Michael Heuser1
`
`●
`
`●
`
`●
`
`Received: 11 December 2018 / Revised: 1 July 2019 / Accepted: 23 July 2019 / Published online: 4 October 2019
`© The Author(s), under exclusive licence to Springer Nature Limited 2019
`
`Abstract
`IDH produces
`Mutations in isocitrate dehydrogenase 1 (IDH1) are found in 6% of AML patients. Mutant
`R-2-hydroxyglutarate (R-2HG), which induces histone- and DNA-hypermethylation through the inhibition of epigenetic
`regulators, thus linking metabolism to tumorigenesis. Here we report the biochemical characterization, in vivo antileukemic
`effects, structural binding, and molecular mechanism of the inhibitor HMS-101, which inhibits the enzymatic activity of
`mutant IDH1 (IDH1mut). Treatment of IDH1mut primary AML cells reduced 2-hydroxyglutarate levels (2HG) and induced
`myeloid differentiation in vitro. Co-crystallization of HMS-101 and mutant IDH1 revealed that HMS-101 binds to the active
`site of IDH1mut in close proximity to the regulatory segment of the enzyme in contrast to other IDH1 inhibitors. HMS-101
`also suppressed 2HG production, induced cellular differentiation and prolonged survival in a syngeneic mutant IDH1 mouse
`model and a patient-derived human AML xenograft model in vivo. Cells treated with HMS-101 showed a marked
`upregulation of the differentiation-associated transcription factors CEBPA and PU.1, and a decrease in cell cycle regulator
`cyclin A2. In addition, the compound attenuated histone hypermethylation. Together, HMS-101 is a unique inhibitor that
`binds to the active site of IDH1mut directly and is active in IDH1mut preclinical models.
`
`1234567890();,:
`1234567890();,:
`
`These authors contributed equally: Anuhar Chaturvedi, Ramya
`Goparaju
`
`These authors contributed equally: Matthias Preller and
`Michael Heuser
`
`Supplementary information The online version of this article (https://
`doi.org/10.1038/s41375-019-0582-x) contains supplementary
`material, which is available to authorized users.
`
`* Matthias Preller
`preller.matthias@mh-hannover.de
`* Michael Heuser
`heuser.michael@mh-hannover.de
`
`1 Department of Hematology, Hemostasis, Oncology and Stem Cell
`Transplantation, Hannover Medical School, Hannover, Germany
`
`2
`
`Institute for Biophysical Chemistry, Hannover Medical School,
`
`Introduction
`
`Mutations in the active site arginine residue (R132) of
`isocitrate dehydrogenase 1 (IDH1) have been found in
`about 6–10% of acute myeloid leukemia (AML) patients
`[1, 2], which confer a neomorphic function to the mutant
`enzyme
`and
`result
`in
`elevated
`levels
`of R-2-
`hydroxyglutarate (R-2HG) [3, 4]. IDH1 mutations lead to
`a block in cellular differentiation and promote tumorigen-
`esis partly due to global DNA and histone hypermethylation
`by disruption of α-ketoglutarate dependent
`enzymes
`through R-2HG [5–7]. Thus, the inhibition of oncogenic
`
`Hannover and Centre for Structural Systems Biology,
`Hamburg, Germany
`3 Department of Clinical Chemistry, VU University Medical Center,
`Amsterdam, The Netherlands
`4 Research Core Unit Metabolomics, Institute of Pharmacology,
`Hannover Medical School, Hannover, Germany
`
`5
`
`Institute of Clinical Chemistry, Hannover Medical School,
`Hannover, Germany
`
`Rigel Exhibit 1040
`Page 1 of 11
`
`

`

`In vivo efficacy of mutant IDH1 inhibitor HMS-101 and structural resolution of distinct binding site
`
`417
`
`mutant IDH1 represents an opportunity for therapeutic
`intervention. The crystal structures reported from X-ray
`crystallographic studies are currently being used as protein
`models of choice to screen inhibitors which specifically
`target mutant IDH1, while sparing wild-type IDH1 [3].
`Human cytosolic IDH1 consists of asymmetric subunits
`forming a dimer. The crystal structure of wild-type IDH1
`consists of a large domain, a small domain and a clasp
`domain [8]. A deep cleft is formed by the large domain, the
`small domain and a second small domain from a different
`subunit, which together form the active site of the enzyme
`[8]. The hydrophilic active site has pockets for isocitrate, a
`metal ion, and NADP binding sites [8]. An amino acid
`change at arginine 132 in the active site to histidine or
`cysteine induces a conformational change leading to
`increased affinity to α-ketoglutarate, which is then catalyzed
`to R-2HG [9]. Thus, altered substrate-binding affinity con-
`fers a neomorphic function to the mutant IDH1 enzyme.
`Several IDH1 inhibitors have been designed that can
`target
`the conformational change induced by a single-
`mutated amino acid, and some of the IDH1 inhibitors like
`AG-120,
`IDH-305, FT-2102, and BAY1436032 have
`entered clinical trials or even have been approved (AG120/
`ivosidenib) [10–16]. Crystal structures of several IDH1
`inhibitors
`(chemically related BAY1436032, AG-881,
`IDH1-305, IDH1-125, and IDH1-146) in complex with
`mutant IDH1 have shown that
`they bind to allosteric
`pockets,
`inducing a conformational change and conse-
`quently inhibit the catalytic activity [12, 17–19]. Until now
`there is no approved or clinically active inhibitor, which
`binds to the active site of mutant IDH1. By computational
`screening of ~500,000 compounds, we have previously
`identified the IDH1 mutant-specific inhibitor HMS-101,
`which has shown efficacy towards IDH1 mutant cells
`in vitro [20].
`In this study, we evaluated the antileukemic effect of
`HMS-101 in a syngeneic mutant IDH1 mouse model and
`patient-derived human AML xenograft models as well as
`performed biochemical and structural studies to evaluate the
`molecular mechanism of this compound. From this study,
`we conclude that HMS-101 binds to the active site of
`mutant IDH1, which is different from other IDH1 inhibitors,
`inhibits cellular proliferation and induces differentiation in
`IDH1 mutant leukemia cells.
`
`Materials and methods
`
`IDH activity assay
`
`The enzymatic activity of pure IDH proteins was assessed
`by measuring the rate of consumption or the production of
`NADPH spectrophotometrically at 340 nm wavelength. A
`
`reaction mixture was prepared with 100 mM Tris, pH 7.4,
`1% BSA, and 5 mM of MnCl2 along with 250 μM NADPH
`as the co-factor for IDH1 and IDH2 mutants and 50 mM
`HEPES, pH 7.5, 150 mM NaCl, and 20 mM of MgCl2
`along with 1 mM NADP for IDH1 and IDH2 wt enzyme
`catalysis. Reactions containing 500 nM of mutant IDH1 or
`mutant IDH2 proteins were initiated by adding 250 μM
`α-ketoglutarate as a substrate, whereas reactions containing
`3 nM of wt IDH1 or wt IDH2 proteins were initiated upon
`addition of 4 mM isocitrate. All chemicals for the activity
`assays were purchased from Sigma-Aldrich (Munich,
`Germany). Efficacy of the drug was tested by incubating
`IDH1 proteins with increasing concentrations of HMS-101
`or PBS (control) for 30 min prior to the experiment. Assays
`were performed with Multiskan FC (Life Technologies, CA,
`USA), and the specific activity was calculated from the
`slope value obtained from kinetic reactions.
`
`Crystallization, data collection, and structure
`determination
`
`Crystals of the IDH1mut-HMS-101 complex were obtained
`by co-crystallization of the purified IDH1mut protein with a
`concentration of 10 mg/ml in the presence of 2 mM HMS-
`101 by the sitting drop vapor diffusion method at 4 °C. The
`mixture was incubated for 30 min on ice and subsequently
`mixed with an equal volume of the reservoir solution (0.2 M
`di-ammonium citrate, 20% (w/v) PEG 3350) without the
`addition of NADP or α-ketoglutarate. The crystals were
`transferred to a cryoprotection solution containing reservoir
`solution mixed with 2 mM HMS-101 and additional 25%
`ethylene glycol, and subsequently flash-cooled in liquid
`nitrogen. Diffraction data were collected at beamline
`PROXIMA-1,
`Synchrotron
`SOLEIL, Gif-sur-Yvette,
`France. The dataset was processed using XDS [21], and the
`structure was solved by molecular replacement using Phaser
`[22] of Phenix [23] and the structure of IDH1 wt (pdb:
`1T09) [8] as the search model. Structure refinement was
`carried out with Phenix and a random 5% of the data was
`excluded for cross-validation. Model building and valida-
`tion was done using Coot [24]. The final coordinates were
`stored in the protein database [25] with the identification
`code 6Q6F.
`In silico screening of IDH1mut inhibitors, compound
`preparation, retroviral vectors, and infection of primary
`bone marrow cells, cloning, expression, and purification of
`recombinant IDH1 mutant proteins, microscale thermo-
`phoresis, 2-hydroxyglutarate quantification, patient sam-
`ples,
`clonogenic
`progenitor
`assay,
`antibodies
`for
`immunophenotyping and morphologic analysis, cell culture
`and treatment, cell viability and cell counts, pharmacoki-
`netic and toxicity analysis of HMS-101, bone marrow
`transplantation,
`treatment
`and monitoring
`of mice,
`
`Rigel Exhibit 1040
`Page 2 of 11
`
`

`

`A. Chaturvedi et al.
`
`418
`
`Fig. 1 Validation of candidate
`inhibitors of mutant IDH1.
`a Concentration of R-2HG
`per million cells in mouse
`HoxA9 + IDH1mut-transduced
`mouse bone marrow cells
`incubated with 10 μM of HMS-
`101, HMS-102, HMS-103, and
`100 μM of HMS-104 or DMSO
`for 72 h (mean ± SEM, n = 3).
`b IC50 of HMS-101 in
`HoxA9 + IDH1 wt, HoxA9 +
`IDHmut R132C and HoxA9 +
`IDH1mut R132H cells treated
`for 72 h (mean ± SEM, n = 3).
`c Concentration of R-2HG per
`million cells in mouse
`HoxA9 + IDH2mut transduced
`mouse bone marrow cells
`incubated with 10 μM and 100
`μM of HMS-101 or DMSO for
`72 h (mean ± SEM, n = 3).
`d IC50 of HMS-101 in HoxA9
`+ IDH2 wt, HoxA9 + IDH2
`R140Q, and HoxA9 + IDH2
`R172K cells treated for 72 h
`(mean ± SEM, n = 3).
`***P < 0.001
`
`immunoblotting and statistical
`quantitative RT-PCR,
`methods are described in detail in Supplementary Materials.
`
`Results
`
`Validation of candidate inhibitors of mutant IDH1
`
`A parallelized, high-throughput computational screening
`approach of ∼0.5 million compounds from the Zinc data-
`base [26] against mutant IDH1, allowed us to identify four
`promising candidates for the specific inhibition of IDH1mut
`(Supplementary Table S1). These compounds showed high-
`binding energies against
`the high-resolution IDH1mut
`crystal structure (pdb: 3INM) [3] and a marked preference
`towards mutant IDH1 over wild type, when re-docked
`against an IDH1 wt crystal structure (pdb: 1T0L) [8]. We
`used a subset of the lead-like library of the Zinc database to
`ensure preferable pharmacokinetic properties of the com-
`pounds. Both, limiting the search area during computational
`docking to the isocitrate binding site, as well as covering
`both the isocitrate and NADP binding sites, suggested
`favorable binding of these ligands to the isocitrate binding
`pocket on IDH1mut with predicted binding energies in the
`range of −12 kcal/mol, giving computed ligand efficiencies
`(LE) of 0.5–0.8. In contrast, a second population was found
`for the larger search area with smaller parts of the ligands
`overlapping with the NADPH binding site,
`featuring
`slightly lower predicted overall binding energies. For both
`
`identified binding sites and all four compounds, the com-
`puted affinities towards IDH1mut were markedly higher
`than those obtained with IDH1 wt. The on-target efficacy of
`these candidate inhibitors was evaluated by their effect on
`R-2HG levels, which is specifically produced by mutant
`IDH1. Murine hematopoietic bone marrow cells were
`transformed with HoxA9 and the IDH1R132C mutant as
`described previously [20], and were treated with the four
`candidate IDH1 inhibitors for 72 h. The concentration of R-
`2HG was significantly reduced by HMS-101 and only
`slightly decreased by HMS-102, HMS-103, and HMS-104
`(Fig. 1a and Supplementary Fig. S1A). We further eval-
`uated their effect on cell viability in murine HoxA9 +
`IDH1mut cells expressing either the IDH1R132C or the
`IDH1R132H mutation. HMS-101 reduced the IC50 for both
`mutation
`types
`compared with
`IDH1 wild
`type-
`overexpressing cells, suggesting broad activity against dif-
`ferent types of IDH1 mutations (Fig. 1b). However, HMS-
`101 did not inhibit R-2HG production in HoxA9 immor-
`talized mouse bone marrow cells expressing IDH2 R140Q
`and R172K mutants (Fig. 1c and Supplementary Fig. S1B)
`and the cellular IC50 of HMS-101 was 300–1000-fold
`higher in IDH2 R140Q (300 μM), IDH2 R172K (1000 μM),
`and IDH2 wt (400 μM) cells compared with IDH1mut cells,
`indicating the specificity of HMS-101 towards mutant IDH1
`(Fig. 1d). In order to determine the binding affinity of HMS-
`101 to IDH1mut, we used microscale thermophoresis
`measurements (Supplementary Fig. S1A). The binding
`affinity of HMS-101 toward IDH1mut in the absence of
`
`Rigel Exhibit 1040
`Page 3 of 11
`
`

`

`In vivo efficacy of mutant IDH1 inhibitor HMS-101 and structural resolution of distinct binding site
`
`419
`
`Fig. 2 HMS-101 inhibits the enzymatic activity of mutant IDH1.
`a Inhibition of IDH1mut R132C or IDH1mut R132H enzymatic
`activity measured by the rate of consumption of NADPH and IDH1 wt
`activity by the rate of production of NADPH in the presence of
`increasing concentrations of HMS-101 compared with the PBS treated
`control (mean ± SEM, n = 3). b Inhibition of R-2HG production by
`IDH1mut R132C, IDH1mut R132H, IDH2mut-R140Q, and IDH2mut-
`
`R172K proteins in the presence of increasing concentrations of HMS-
`(mean ± SEM, n = 3).
`101 compared with PBS-treated control
`c Concentration of R-2HG/mg protein in primary human AML cells
`harboring different IDH1 and IDH2 mutations 24 h after HMS-101
`treatment, calculated as percentage of PBS control-treated cells.
`IC50 values for mutant IDH1 are provided in the graph (mean ± SEM,
`n = 3)
`
`substrate and co-factor is in the low micromolar range (Kd:
`IDH1R132H 14.1 μM). In the presence of NADPH and
`αKG, the Kd of HMS-101 is shifted to higher values (Kd:
`IDH1R132H 445.9 μM and IDH1R132C 191.9 μM), indi-
`cating that HMS-101 competes with either NADPH or α-
`KG for binding to mutant IDH1. The other candidate
`inhibitors showed no differential cytotoxic effect in IDH1-
`mut compared with IDH1 wt cells (Supplementary fig.
`S2A–C). We,
`therefore, selected HMS-101 for
`further
`characterization.
`
`patients (IC50: IDH1mut R132C 0.7 μM, IDH1mut R132H
`0.8 μM, Fig. 2c). While R-2HG production was efficiently
`inhibited within 24 h, viability and cell counts of primary
`AML cells were hardly affected within this short time frame
`(Supplementary Fig. S6A, B). Thus, HMS-101 pre-
`ferentially interferes with the enzymatic activity of mutant
`IDH1.
`
`X-ray diffraction revealed the binding of HMS-101 in
`the active site of mutant IDH1
`
`HMS-101 inhibits the enzymatic activity of mutant
`IDH1
`
`We purified His-tagged wild type and mutant IDH1 proteins
`(IDH1mut R132C and IDH1mut R132H)
`from the
`BL21DE3 bacterial strain (Supplementary Fig. S3). The
`preferential reactivity of the wild-type IDH1 protein with
`isocitrate and of the mutant IDH1 protein with αKG was
`confirmed in activity assays
`in vitro (Supplementary
`Fig. S4). While IDH1 wt and IDH2 wt enzymatic activities
`were only affected at higher compound concentrations
`(IC50: IDH1 wt 95 μM, Fig. 2a, and IDH2 wt 110 μM,
`Supplementary Fig. S5), HMS-101 selectively inhibited the
`enzymatic activity of both IDH1 mutation types in a dose-
`dependent manner with an IC50 of 4 μM (IDH1mut R132C)
`and 5 μM (IDH1mut R132H), accounting for a 19- and 24-
`fold selectivity over IDH1 wt, respectively. The conversion
`from αKG to R-2HG by the IDH1mut pure protein in a cell
`free solution was also efficiently inhibited by HMS-101
`(IC50: IDH1mut R132C 6 μM, IDH1mut R132H 7 μM),
`however, the IC50 of HMS-101 was not reached until a
`concentration of 1 mM for
`the IDH2 mutant protein
`(Fig. 2b). We validated the specificity towards mutant IDH1
`by treating primary AML cells from IDH1 and IDH2
`mutated patients. R-2HG was reduced in a dose-dependent
`manner in IDH1 mutant but not in IDH2 mutant AML
`
`To verify the predicted binding of HMS-101 to the active
`site and to determine the detailed binding mode of the
`inhibitor, we co-crystallized IDH1mut R132H with HMS-
`101 in the absence of substrate or co-factor, and solved the
`structure by molecular replacement to a resolution of 3.3 Å
`(Fig. 3a). Statistics for diffraction data and structure
`refinement are shown in Supplementary Table S2. The
`overall fold of the IDH1mut-HMS-101 complex resembles
`the reported IDH1mut crystal structures, featuring a mutated
`homodimeric structure with an open/quasi-open active site
`conformation in the subunits and a width of the active site
`entrance of 19.7 and 17.0 Å (measured between residues 76
`and 250 of the second subunit), respectively. Accordingly,
`the back clefts are closed with widths of 11.1 and 11.3 Å
`(measured between residues 199 and 342). The large
`domains of the subunits show the typical Rossmann fold,
`while the small domains fold in an α/β sandwich, and the
`clasp domains of the two subunits form together two four-
`stranded β-sheets. As has been seen before in crystal
`structures of IDH1mut in the open/quasi-open conformation
`(pdb: 3MAP and 3MAR) [9], the regulatory segment (seg-2,
`residues 271 to 286) in the active site is disordered and was
`not observable in the obtained HMS-101 complex crystal
`structure. This important structural segment, together with
`the hinge segment (seg-1, residues 134–141) undergoes
`substantial
`conformational
`rearrangements during the
`
`Rigel Exhibit 1040
`Page 4 of 11
`
`

`

`420
`
`A. Chaturvedi et al.
`
`Fig. 3 Crystal structure of HMS-101 bound to the active site of mutant
`IDH1. a Overview of the co-crystallized structure of mutated IDH1 in
`complex with HMS-101 in a cartoon representation. The subunits of
`mutated IDH1 are depicted in cyan and red, and the inhibitor was
`found in both subunits. b Close-up view of segment-1 (seg-1, purple),
`featuring a slightly different conformation with rearrangements of the
`catalytically important residue Tyr139, as compared with the IDH1 wt
`crystal structure (pdb: 1T09, transparent gray structure). c Close-up
`view on the active site showing rearrangements of the active site
`residues that are involved in the enzymatic activity. Note the slightly
`more open active site due to the conformational changes of the large
`domain (cyan). The subdomains are colored cyan and red, seg-1, and
`
`seg-2 (of IDH1 wt crystal structure, pdb: 1T09) are shown in purple
`and yellow, respectively. For comparison, the IDH1 wt crystal struc-
`ture is depicted as transparent gray structure. d The 2Fo-Fc density map
`of HMS-101 and surrounding residues of the binding site. The map
`was contoured at 1.0 σ. e Interaction diagram, showing the HMS-101
`binding site and contact residues. Green dashed lines indicate hydro-
`gen bonds and purple dashed lines show halogen bonds. Color code:
`green: unpolar residues, red: negatively charged residues, purple:
`positively charged residues, cyan: polar residues, beige: glycine resi-
`dues. f Close-up view of the HMS-101 binding site with contact
`residues shown as sticks. Dashed lines indicate favorable interactions
`with the protein
`
`transition from the open conformation to the catalytically
`competent closed conformation [8]. Seg-1 could be resolved
`in our structure and adopts a slightly different conformation
`as compared with IDH1 wt in the open or closed state, with
`changes in the Tyr139 conformation (pdb: 1T09 and 1T0L,
`Fig. 3b) [8].
`Compared with the known IDH1mut crystal structure in
`the presence of α-KG and NADPH (pdb: 3MAP), the large
`domains in the HMS-101-bound structure of both subunits
`
`are moved slightly outwards with root mean square devia-
`tions of 1.94 and 2.19 Å to adopt a slightly more open
`active site. Several rearrangements of active site residues
`were observed upon ligand binding,
`including residues
`Ser94, Arg100, Arg109, and Lys212 (Fig. 3c). The ligand
`HMS-101 binds to both subunits in a compact, bent con-
`formation, and is nestled to a site that partially overlaps with
`the NADPH binding site in other known IDH1 crystal
`structures (Fig. 3a, d). Favorable interactions with protein
`
`Rigel Exhibit 1040
`Page 5 of 11
`
`

`

`In vivo efficacy of mutant IDH1 inhibitor HMS-101 and structural resolution of distinct binding site
`
`421
`
`Fig. 4 HMS-101 inhibits
`proliferation and induces
`differentiation in primary human
`AML. a Representative FACS
`plots from IDH1/2 wt and
`IDH1mut AML patient cells
`showing proportions of CD14+
`and CD15+ cells after treatment
`with either 1 μM HMS-101 or
`DMSO for 14 days ex vivo.
`b Percentages of CD14+ or
`CD15+ cells in four IDH1 wt
`human AML patients cells (left)
`and three IDH1mut human
`AML patients cells (right) after
`treatment with either 1 μM
`HMS-101 or DMSO for 14 days
`ex vivo. Two IDH1mut patients
`carry R132C and one an R132H
`mutation. c Morphology of bone
`marrow cells from IDH1 wt and
`IDH1mut AML patients treated
`with either 1 μM HMS-101 or
`DMSO for 14 days ex vivo
`(×1000 magnification)
`
`residues are formed between the piperazine ring of HMS-
`101 and the main chain nitrogen of Val312, as well as the
`pyrrolidin ring and the side chain of Arg314 (Fig. 3e, f). In
`addition, a halogen bond between the fluorophenyl and
`Asn328 is found, and a number of hydrophobic interactions
`with residues Leu288, Gly310, and Val312 stabilize the
`binding. The total protein surface interaction area in contact
`with HMS-101 comprises ~419 Å. Thus, our data support
`binding of HMS-101 to the active site of mutant IDH1.
`
`HMS-101 inhibits proliferation and induces
`differentiation in primary human AML
`
`We determined the effects of treatment with HMS-101 in
`primary human AML cells with wild-type or mutant IDH1
`on cellular differentiation. IDH1mut human AML cells that
`were cultured in suspension medium ex vivo showed
`marked upregulation of the myeloid differentiation markers
`
`CD15 and/or CD14 (Fig. 4a, b). Viability and cell counts of
`human AML cells were more efficiently inhibited by HMS-
`101 than by control treatment in IDH1mut cells (Supple-
`mentary Fig. S7A–B). In HMS-101 treated IDH1mut AML
`we observed morphologic changes consistent with mono-
`cytic differentiation (Fig. 4c), demonstrating that HMS-101
`induces myeloid differentiation in AML cells. No mor-
`phological changes suggestive of granulocytic or monocytic
`differentiation were observed in wild-type IDH1 AML cells
`treated with DMSO or HMS-101 (Fig. 4c).
`
`HMS-101 inhibits proliferation, induces myeloid
`differentiation, and prolongs survival in leukemic
`mice in vivo
`
`The maximum tolerated dose of HMS-101 was identified by
`treating C57BLJ/6 mice with varying doses of HMS-101. A
`dose of 40 mg/kg HMS-101 intraperitoneally once daily
`
`Rigel Exhibit 1040
`Page 6 of 11
`
`

`

`422
`
`A. Chaturvedi et al.
`
`Fig. 5 HMS-101 inhibits proliferation, induces myeloid differentiation
`and prolongs survival in leukemic mice in vivo. a Unbound HMS-101
`plasma concentrations in C57BL6/J mice treated with a daily dose of
`16, 40, and 160 mg/kg HMS-101 for 9 days. Plasma was collected
`before the next injection on days 1, 2, 7, and 8 (mean ± SEM of five
`animals/dose). The dashed line indicates the in vitro IC50 in HoxA9
`IDH1mut cells. b Absolute concentration of R-2HG in the serum of
`mice transplanted with HoxA9 IDH1mut cells and treated with HMS-
`101 at a dose of 40 mg/kg for 8 weeks (mean ± SEM). c Engraftment
`of HoxA9 IDH1mut cells in peripheral blood of mice treated with
`either vehicle (left) or HMS-101 at a dose of 40 mg/kg at the indicated
`time points (mean ± SEM). d White blood cell count, e hemoglobin
`
`level, and f platelet count in peripheral blood at different time points
`after the start of treatment with vehicle or HMS-101 at a dose of
`40 mg/kg (mean ± SEM). g Morphology and fluorescence of peripheral
`blood cells from HoxA9 + IDH1mut transplanted mice treated with
`vehicle (left) or HMS-101 (right) at 15 weeks after treatment (×400
`original magnification). Mutant IDH1 was expressed from a retroviral
`vector that co-expresses GFP. Thus, GFP-positive cells indicate IDH1
`mutant leukemic cells. h Survival of HoxA9 + IDH1mut transplanted
`mice treated with either vehicle or HMS-101. *P < 0.05, **P < 0.01,
`***P < 0.001. Hash (#) indicates week 15 after transplantation or at
`death if the mouse died before week 15 due to leukemia
`
`produced plasma levels equivalent to the in vitro IC50 in
`HoxA9 IDH1mut cells [20] (Fig. 5a) and was well tolerated
`in mice with no visible signs of toxicity on body weight,
`complete blood counts, spleen weight, and serum chemistry
`(Supplementary Fig. S8A–I and Supplementary Table S3).
`HMS-101 was further evaluated for its antileukemic effects
`in vivo in a previously described mouse model induced by
`co-expression of HoxA9 and IDH1mut R132C [20]. HMS-
`101 was applied intraperitoneally once daily starting on day
`5 after transplantation and continued until death at a dose of
`40 mg/kg. After 8 weeks of
`treatment, R-2HG was
`
`significantly reduced in the serum of HMS-101 treated
`compared with solvent-treated mice (Fig. 5b). While leu-
`kemic
`engraftment
`in peripheral blood continuously
`increased in solvent treated mice, it decreased in five of
`seven mice after week 6 of transplantation (Fig. 5c). At
`15 weeks or at death of the mice that had died before week
`15 WBC count was significantly lower, and hemoglobin
`and platelet count were significantly higher in HMS-101
`treated mice compared with control mice (Fig. 5d–f). Fif-
`teen weeks after transplantation HMS-101-treated mice
`showed mostly
`differentiated GFP-positive/IDH1mut-
`
`Rigel Exhibit 1040
`Page 7 of 11
`
`

`

`In vivo efficacy of mutant IDH1 inhibitor HMS-101 and structural resolution of distinct binding site
`
`423
`
`Fig. 6 HMS-101 induces
`differentiation in primary IDH1
`mutant AML cells. a Absolute
`concentration of R-2HG in the
`serum of PDX-IDH1R132C
`mice treated with HMS-101 at a
`dose of 40 mg/kg or vehicle for
`18 weeks (mean ± SEM).
`b Percentage of human CD14+
`cells in peripheral blood of
`PDX-IDH1R132C mice at
`different time points with either
`vehicle or HMS-101 at dose of
`40 mg/kg (mean ± SEM).
`c Survival of PDX-IDH1R132C-
`transplanted mice treated with
`either vehicle or HMS-101.
`d Colony forming cell assay of
`IDH1/2 wt, IDH1mut/NRASwt,
`IDH1mut/NRASmut, and
`IDH2mut primary cells from
`AML patients treated with
`HMS-101 relative to DMSO-
`treated cells (mean ± SEM).
`*P < 0.05, ***P < 0.001
`
`transduced cells (neutrophils) in peripheral blood, while
`control-treated mice had mostly undifferentiated GFP-
`positive/IDH1mut-transduced myeloid progenitor cells in
`peripheral blood (Fig. 5g). Immunophenotyping of periph-
`eral blood cells at 15 weeks after transplantation confirmed
`that HMS-101 shifted the myeloid cells from the more
`immature phenotype CD11b+ Gr1− to the more mature
`phenotype CD11b+ Gr1+ (Supplementary Fig. S9).
`Importantly, these effects of HMS-101 attributed to differ-
`entiation induced a significant survival benefit with a
`median survival of 66 days for control
`treated and of
`84 days in HMS-101-treated mice (Fig. 5h and Supple-
`mentary Fig. S10).
`HMS-101 was also evaluated in a PDX model of an
`IDH1-mutated AML patient. This patient had IDH1
`p.R132C, NRAS p.Q61R, ETV6 p.R103SfsTer9, PTPN11
`p.S502P, ASXL1 p.E947Ter, RUNX1 p.R169KfsTer44, and
`EZH2 p.P527S mutations. Treatment was started with
`HMS-101 or solvent intraperitoneally once daily starting on
`day 45 after transplantation and continued until death at a
`
`dose of 40 mg/kg body weight. At 18 weeks, the R-2HG
`concentration in serum declined by 2.9-fold in HMS-101
`treated mice (Fig. 6a) and at 22 and 26 weeks after trans-
`plantation, the proportion of CD14, a marker of monocytic
`differentiation, on human cells was significantly higher in
`HMS-101 treated mice compared with controls (Fig. 6b).
`Median survival was significantly prolonged by 20 days in
`HMS-101 treated mice (median survival 210 vs 230 days,
`Fig. 6c). In an independent, second PDX model, which
`harbored IDH1 p.R132H, DNMT3A p.R882H, PTPN11
`p.A72T, and NPM1 p.T288CfsTer12 mutations, the per-
`centage of human CD45+ cells in the peripheral blood of
`mice increased in vehicle-treated animals but was essen-
`tially absent
`in HMS-101 treated mice (Supplementary
`Fig. S11A). In an independent third PDX model, NSG mice
`were transplanted with primary IDH1/IDH2 wt, NPM1
`p.W288CfsTer12 and TET2 p.G1931D mutant AML cells.
`Both HMS-101 and vehicle-treated mice had similar per-
`centages of human CD45+ cells in peripheral blood of mice
`(Supplementary Fig. S11B). There was no significant
`
`Rigel Exhibit 1040
`Page 8 of 11
`
`

`

`424
`
`A. Chaturvedi et al.
`
`Fig. 7 HMS-101 induces the expression of differentiation genes,
`inhibits cell cycle associated genes and restores histone methylation in
`HoxA9 + IDH1mut cells. a Gene expression levels of differentiation-
`inducing genes CEBPA and PU.1 in in vitro cultured HoxA9 cells
`transduced with IDH1 wt or IDH1mut and treated with either vehicle
`(CTL) or 1 μM HMS-101 for 9 days. Gene expression was determined
`by quantitative RT-PCR relative to the housekeeping gene Abl1 and
`
`was normalized to gene expression in IDH1 wt vehicle-treated cells
`(mean ± SEM). b Gene expression levels of cell cycle genes genes
`CCNB1 (Cyclin B1) and CCNA2 (Cyclin A2) as described above
`(mean ± SEM). c Western blot of histone H3 trimethylation levels at
`residues H3K4, H3K9, H3K27, and H3K36 in HoxA9 IDH1 wt and
`HoxA9 IDH1mut cells treated with 1 μM HMS-101 for 9 days. *P <
`0.05, **P < 0.01, ***P < 0.001, ns, non significant
`
`difference between the number of colonies formed by
`IDH1mut/NRASwt and IDH1mut/NRASmut primary AML
`cells in the presence of HMS-101 compared with control-
`treated cells, suggesting that NRASmut is not predictive of
`response to HMS-101. Further, HMS-101 did not inhibit the
`colony formation of IDH2 mutant AML patient cells indi-
`cating specificity towards mutant IDH1 (Fig. 6d).
`
`HMS-101 induces the expression of differentiation
`genes, inhibits cell cycle associated genes, and
`restores histone methylation in HoxA9 + IDH1mut
`cells
`
`Murine HoxA9 + IDH1mut cells were treated with HMS-
`101 or solvent for 9 days to evaluate the impact of HMS-
`101 on gene expression. CEBPA and PU.1 are critical
`transcription factors for granulocytic and monocytic dif-
`ferentiation, respectively. Both genes were strongly upre-
`gulated upon HMS-101 treatment in IDH1mut but not IDH1
`wt cells (Fig. 7a). The cell cycle regulator CCNA2, but not
`CCNB1 was significantly downregulated in HMS-101-
`treated IDH1mut cells (Fig. 7b), suggesting that the inhi-
`bitor induces differentiation and inhibits cell cycle, as
`reported for other IDH1 inhibitors before [12, 13, 27]. As
`R-2HG inhibits several histone demethylases like the
`H3K4me3, H3K9me3, and H3K27me3 histone demethy-
`lases KDM4A, KDM4C [5, 27–29], we evaluated the effect
`of HMS-101 on histone methylation. H3K4me3 and
`H3K27me3 were increased in IDH1mut cells compared
`with IDH1 wt cells. This increase in H3K4me3 and
`H3K27me3 methylation was efficiently reduced by HMS-
`
`101 in IDH1mut cells with no effect in IDH1 wt cells
`(Fig. 7c). In addition, H3K36me3 methylation, which is
`reported to be increased in IDH1mut knock-in mice [28],
`also declined after treatment with HMS-101 in IDH1mut
`cells but not in IDH1 wt cells (Fig. 7c). These changes in
`gene expression and histone methylation were accompanied
`by a marked reduction in cell viability and absolute cell
`counts in IDH1mut cells treated with HMS-101 (Supple-
`mentary Fig. S12A, B). In summary, HMS-101 induces
`differentiation,
`inhibits cell cycle, and reverses aberrant
`histone methylation in IDH1mut leukemias.
`
`Discussion
`
`We characterized the unique properties of the IDH1 inhi-
`bitor HMS-101, which competitively inhibits mutant IDH1
`and binds in the proximity of the NADPH binding site in
`the active site of the enzyme, thereby inhibiting the con-
`version of α-ketoglutarate (KG) to R-2HG. HMS-101 is
`active in vivo, supresses R-2HG production, induces mye-
`loid differentiation and prolonges survival in syngeneic and
`IDH1mut xenograft mouse models.
`IDH1 is one of the few enzymes, which is capable of a
`novel enzymatic catalysis as a consequence of a single
`amino acid mutation in the active center, resulting in
`incomplete catalytic product formation of R-2HG [3]. Ele-
`vated levels of R-2HG in the cells result in the disruption of
`α-ketoglutarate dependent enzymatic functions because of
`structural similarity resulting in epigenetic changes leading
`to IDH1-induced tumors [5, 10, 30]. Several allosteric IDH1
`
`Rigel Exhibit 1040
`Page 9 of 11
`
`

`

`In vivo efficacy of mutant IDH1 inhibitor HMS-101 and structural resolution of distinct binding site
`
`425
`
`inhibitors are in the pipeline to deliver a targeted therapy
`including a pan IDH1/IDH2 mutant
`inhibitor (AG-881)
`