Evaluation of ERK as a therapeutic target in acute myelogenous leukemia A novel ERK inhibitor for AML

Introduction

The MAPK signaling cascade, comprised of key kinases including Raf, MEK1 and MEK2 (MAP2K1 and MAP2K2), and ERK1 and ERK2 (p44 MAPK and p42 MAPK), contributes to the growth, survival and transformation of hematopoietic cells and is a promising target for acute myeloid leukemia (AML) therapy (refs ^1–2). RAS is a major target of activating mutations in AML, and ERK is a key regulator of multiple signaling pathways.

Despite its prevalence in AML, targeting of mutant RAS directly has proven to be difficult because of its molecular configuration (ref. ³). Attempts have been made to target Ras signaling through other, more indirect, approaches, such as inhibition of MEK, mTOR, ATK, PI3K, and Raf, however clinical testing of these alternative strategies have been disappointing because of compensatory activation of other signaling molecules or negative feedback loops (refs. ^4–6).

ERK is the main downstream effector in MAPK signaling; it is anticipated that ERK inhibition may be better able to circumvent feedback and oncogenic bypass associated with molecules further upstream, such as MEK (ref. ²). Only a few ERK inhibitors have been clinically investigated in the context of solid tumors (refs. ^7–8), and there has been little investigation of ERK inhibitors as a therapeutic for blood cancers, leukemia specifically. One novel ERK1/2 inhibitor, LY3214996 (Eli Lilly), was first reported to be efficacious in preclinical models, including BRAF or NRAS mutant melanoma, BRAF or KRAS mutant colorectal, lung and pancreatic cancer xenografts or PDX models (ref. ⁹), and is currently in Phase I trials for patients with advanced cancer (NCT02857270).

As little is known about the effectiveness of ERK inhibitors in the context of heme malignancies, we were interested in investigating the anti-leukemic potential of LY3214996, as both a single agent and in combination with chemotherapeutic drugs and other targeted inhibitors. We tested LY3214996 against AML cell lines and primary AML specimens driven by and expressing a variety of oncogenes, including mutant RAS, both in vitro and in vivo.

Materials and Methods

Cell proliferation studies, apoptosis assays, cell cycle analysis, immunoblotting, and drug combination studies were carried out as previously described (ref. ¹⁰). Bioluminescence imaging/total body bioluminescence was carried out as previously described (refs. ^11–12). Detailed information regarding these assays and in vivo studies, as well as sources of cell lines and chemical compounds and biologic reagents, is provided in the Supplementary Data section.

Results

LY3214996 inhibits growth of oncogene-driven Ba/F3 cell lines and human AML cell lines

We first tested LY3214996 efficacy in isogenic Ba/F3 cells engineered to over-express oncogenes, including KRAS-G12D, JAK2-V617F, FLT3-ITD, SYK-TEL, FLT3-ITD+SYK-TEL, and BCR-ABL. LY3214996 more potently inhibited proliferation of the oncogene-driven lines as compared to parental cells (Supplementary Figure 1A,B and Table 1A). Mutant RAS-positive cells were among the most sensitive, displaying a potency around 10-fold lower than that of the MEK inhibitor, AZD6244. Growth inhibition of cells correlated with G1 arrest with minimal apoptosis (Supplementary Figures 2–3).

Table 1. (A).

Treatment of Ba/F3-derived cell lines with LY3214996: Estimated IC50s.

Cell lines	LY3214996
	2-day (IC50, nM)	3-day (IC50, nM)
Ba/F3	4382+/−445	4625+/−570
Ba/F3-KRAS-G12D	402+/−121	347+/−141
Ba/F3-EPOR+JAK2-V617F	>4000	1268+/−74
Ba/F3-FLT3-ITD	1947	2445+/−658
Ba/F3-SYK-TEL	2086	848+/−98
Ba/F3-FLT3-ITD+SYK-TEL	575	268+/−19
Ba/F3-BCR-ABL	>4000	2647+/2925

Data are presented as mean+/−SD.

Consistent with these results, LY3214996 inhibited proliferation of human AML cell lines expressing and/or driven by oncogenes, including FLT3-ITD, AML1/ETO, mutant RAS and mutant JAK2, with IC₅₀s ranging from around 300 nM (several mutant RAS- and AML1-ETO-expressing cells were among the most sensitive) to around 1000 nM (Table 1B and Supplementary Figure 1C, G–J). In comparison, AZD6244 killed mutant RAS-expressing cells with higher potency (IC₅₀s of around 20–50 nM) than wt RAS-expressing cell lines (IC₅₀s of >100 nM) (Supplementary Figure 1D and Table 1B).

Table 1. (B).

Treatment of human AML cell lines with LY3214996: Estimated IC50s.

Cell lines	LY3214996		AZD6244 Torin 2		crenol-anib	mido-staurin	quizar-tinib
	2-day (IC50, nM)	3-day (IC50, nM)	2-day (IC50, nM)	2-day (IC50, nM)	2-day (IC50, nM)	2-day (IC50, nM)	2-day (IC50, nM)
MOLM14	1218+/−408	647+/−172	202+/−36	ND	2.1+/−0.5	11+/−5	0.4+/−0.03
Kasumi-1-luc+	674+/−234	688+/−83	757+/−739	4+/−2
SKNO-1-luc+ *	345+/−106	379+/−81	119+/−81	37+/−16
HL60	603+/−188	873+/−284	47+/−6	ND
OCI-AML3	367+/−146	377+/−77	38+/−10	25+/−18
K052	283+/−100	326+/−72	24+/−14	11+/−7
NB4-luc+	682+/−237	515+/−40	31+/−8	6+/−0.3
SKM-1	509+/−17	525+/−82	83+/−53	7+/−3
NOMO-1	1422+/−320	1030+/−294	84.8+/−20	ND
HEL	>10,000	5295+/−232	ND	ND

Data are presented as mean+/−SD. FLT3 inhibitors, crenolanib, midostaurin, and quizartinib, the MEK inhibitor, AZD6244, and the mTOR inhibitor, Torin 2, were used to treat AML cell lines as controls.

^*SKNO-1-luc+ cells were treated with LY3214996 in the presence of GM-CSF (IC50=491+/−244), SKNO-1 cells (not expressing luciferase) were treated with LY3214996 in the presence of GM-CSF (IC50=454.2+/−124.3), and Kasumi-1 cells (not expressing luciferase) were treated with LY3214996 (IC50=441.4+/−109.9). Oncoprotein expression in human AML lines: MOLM14 (FLT3-ITD, MLL-AF9) Kasumi-1-luc+ (KIT-N822K, AML1-ETO, p53 mutation) SKNO-1-luc+ (KIT-N822K, AML1-ETO, p53 mutation) HL60 (NRAS Q61L, myc+) OCI-AML3 (NPM1 (type A), DNMT3A R882C, NRAS Q61L) K052 (NRAS G13R, p53 mutation) NB4-luc+ (KRAS A18D, p53 mutation, PML-RARalpha) SKM-1 (KRAS K117N, EZH2 mutation (Y641C), p53 mutation) NOMO-1 (KRAS G13D, p53 mutation, MLL-AF9) HEL (JAK2-V617F, p53 mutation)

Anti-proliferative effects of LY3214996 correlated with G1 arrest in all drug-responsive lines with minimal apoptosis (Figure 1, panel A (left side figure) and Supplementary Figures 2–3). Growth inhibition also correlated with inhibition of phosphorylation of S6 and a decrease in c-myc (Figure 1, panel A (middle and right side figures) and Supplementary Figure 4). Taken together, these data suggest that LY3214996 is a potentially effective AML therapeutic, with targeted inhibition of MAPK-mediated signaling in AML cells, and cells expressing mutant RAS and AML1/ETO being particularly responsive.

Figure 1. Anti-leukemic effects of LY3214996 in vitro and in vivo against human AML cell lines, primary AML cells and AML primagraft cells, and potentiation of anti-proliferative effects of LY3214996 by chemotherapy agents and targeted small molecule inhibitors.

(Panel A) LY3214996 inhibits cell cycle progression and inhibits c-myc and phospho-S6. (Left side figure) Cell cycle analysis of K052 cells following 2 days of treatment with LY3214996. (Middle and right side figures) Immunoblotting of protein lysate prepared from K052 cells or Kasumi-1-luc+ cells treated respectively for 2 hr with LY3214996 and analyzed for c-myc or phospho-S6 expression. (Panel B) LY3214996 inhibits growth of primary AML cells and in vivo growth of AML cells. (First figure on left) Proliferation study: 3-day treatment of a mutant RAS-positive AML primagraft sample. (Second figure from left) Comparison of effects of 1000 nM LY3214996 on proliferation of primary AML patient samples treated for 3 and 6 days, respectively. (Third figure from left) Comparison of survival of vehicle-treated versus LY3214996 (100 mg/kg 1X daily)-treated mice harboring OCI-AML3-luc+ cells. Log-rank (Mantel-Cox) test; P=0.0001.Gehan-Breslow-Wilcoxon test; P=0.0001. (Fourth figure from left) Comparison of survival of vehicle-treated versus LY3214996 (50 mg/kg 2X daily)-treated mice harboring HL60-luc+ cells. Log-rank (Mantel-Cox) test; P<0.0001.Gehan-Breslow-Wilcoxon test; P=0.0001. (Panel C) Combination indices for combinations of LY3214996 with chemotherapy agents or targeted agents, generated by Calcusyn software for drug combinations tested. Proliferation assays/combination studies were carried out for 3 days. Calcusyn combination indices can be interpreted as follows: CI<0.1 indicate very strong synergism; values 0.10–0.30 indicate strong synergism; values 0.30–0.70 indicate synergism; values 0.70–0.85 indicate moderate synergism; values 0.85–0.90 indicate slight synergism; values 0.90–1.10 indicate nearly additive effects; values 1.10–1.20 indicate slight antagonism; values 1.20–1.45 indicate moderate antagonism; values 1.45–3.30 indicate antagonism; values 3.30–10.0 indicate strong antagonism; values >10.0 indicate very strong antagonism.

LY3214996 overrides drug resistance

As drug resistance remains a significant problem limiting use of standard of care therapies for AML, we explored the ability of LY3214996 to override resistance in a cell line-based model. Ara-c (cytarabine)-resistant MV4–11 cells were generated by outgrowth of cells following selection with 300 nM cytarabine as previously described (ref. ¹³). Anti-proliferative effects of LY3214996 were equipotent with respect to parental MV4–11 cells and Ara-c-resistant MV4–11 cells (Supplementary Figure 1E,F), suggesting that LY3214996 can override resistance to standard chemotherapy.

LY3214996 inhibits growth of primary AML cells

To investigate the clinical potential of LY3214996, we tested it against human primary AML samples. A mutant RAS-positive AML primagraft sample (Supplementary Table 1) was potently growth-inhibited following 3 days of LY3214996 treatment (IC₅₀ between 10 and 100 nM) (Figure 1, panel B (first figure on left)). LY3214996 treatment for 3–6 days led to inhibition of proliferation of several AML patient samples (Figure 1, panel B (second figure from left), Supplementary Figure 5A,B, E–G and Supplementary Table 2), with 50% or more growth inhibition observed at 1000 nM as compared to 10–20% inhibition of normal PBMC growth. Growth inhibition at 1000 nM LY3214996 for several patient samples was equivalent to or exceeded growth inhibition observed for three human AML cell lines tested as controls (Figure 1, panel B (second figure from left) and Supplementary Figure 5C,D).

Inhibition of growth of mutant RAS-positive leukemia in vivo

We utilized two mouse models of mutant NRAS-positive leukemia to examine the in vivo effects of LY3214996. Treatment of mice harboring OCI-AML3-luc+ cells for 28 days via oral gavage with 100 mg/kg LY3214996 led to a significant decrease in leukemia burden (P=0.011, day 15, and P=0.0002, day 22) and circulating leukemic cells (P=0.0286, day 22) as compared to vehicle-treated control mice and significantly prolonged survival (P=0.0001) (Figure 1, panel B (third figure from left) and Supplementary Figure 6A,B,D,E,F). Treatment of mice harboring HL60-luc+ cells for 28 days via oral gavage with 50 mg/kg LY3214996 twice daily also led to a significant decrease in leukemia burden (P=0.0280, day 22) and a significant increase in survival (P=0.0001), as compared to vehicle control-treated mice (Figure 1, panel B (fourth figure from left) and Supplementary Figure 7A). There was no significant change in body weight for LY3214996-treated mice in either study, suggesting toleration of treatment (Supplementary Figures 6C and 7B).

LY3214996 synergizes with standard chemotherapy agents and targeted therapy

Given the demonstrated ability of LY3214996 to override resistance to standard chemotherapy, we hypothesized that LY3214996 might be potentially useful as a chemosensitizing agent. LY3214996 synergistically combined with 5-azacytidine or decitabine and against many of the human AML cell lines tested, and generally displayed additive to synergistic effects with Ara-c (Figure 1, panel C). LY3214996 also potentiated effects of targeted inhibitors such as the FLT3 inhibitor, midostaurin, the MEK inhibitor, AZD6244 (selumetinib; ARRY-142886), and the cdk4/6 inhibitor, palbociclib (Ibrance), against mutant RAS- and AML1/ETO-expressing cell lines (Figure 1, panel C and Supplementary Figure 8).

Discussion

LY3214996 is a novel competitive ATP inhibitor targeting ERK1 and ERK2 and is under clinical investigation for solid tumors. We wished to explore the effectiveness of ERK inhibition as a putative anti-AML therapy, given the importance of MAPK signaling in leukemic cell transformation and hyperactivation of this pathway across all AML subtypes. Here, we demonstrate that LY3214996 potently and selectively inhibits growth of human AML cell lines and isogenic murine cell lines driven by and/or expressing different oncogenes in vitro. In addition, primary AML cells were more sensitive to LY3214996 treatment than normal donor PBMCs, suggesting preferential killing of transformed cells over untransformed cells. As expected, drug effects were predominantly due to G1 arrest. This is not surprising given that ERK activation is necessary for cell proliferation to proceed, and ERK plays a significant role in the growth factor-independent G2/M phase of cell cycle progression (ref. ¹⁴). LY3214996 inhibition of cell growth correlated with downregulation of c-myc and inhibition of phospho-S6, two molecules associated with MAPK signaling. LY3214996 overcame drug resistance in a cell line-based model of Ara-c resistance, and its effects were potentiated by chemotherapeutic agents used as a standard treatment regimen for AML, including Ara-c, 5-azacytidine, and decitabine, as well as several targeted therapies. Importantly, LY3214996, as a single agent, significantly delayed mutant N-RAS-positive leukemia growth in two distinct orthotopic xenograft leukemia models and significantly prolonged survival as compared to vehicle-treated control mice.

Findings presented here support development of LY3214996 as a therapeutic for mutant RAS-positive AML, as well as AML expressing or driven by other oncogenes that signal through this pathway. As ERK and the MAPK pathway are generally activated in AML, targeting common downstream effectors of various oncogenes represents a strategy that can be utilized broadly across AML. Thus, ERK1/2 inhibitors such as LY3214996 would be expected to be a viable therapeutic strategy for AML with the mutant RAS patient population being of particular interest given the slightly higher sensitivity to ERK inhibition and the limited treatment options. Further clinical investigation of LY3214996 as a potential treatment strategy for AML driven by RAS and other oncogenes is thus warranted.