LSD1 Inhibitor T-3775440 Inhibits SCLC Cell Proliferation by Disrupting LSD1 Interactions with SNAG Domain Proteins INSM1 and GFI1B

Small-cell lung cancer (SCLC) accounts for approximately 15% of all lung cancers and represents the most aggressive subtype. The 5-year survival rate of SCLC is reportedly less than 7%. Although an initial response to standard chemotherapy is observed in most SCLC cases, nearly all patients relapse within 6–12 months. Many clinical trials investigating SCLC treatments have been conducted; however, there has been little improvement in treatment outcomes over the past few decades (1). In contrast to non–small cell lung cancer (NSCLC), effective molecularly targeted therapeutics for SCLC have not been developed due to an incomplete understanding of the genomic aberrations and biology associated with SCLC.

As SCLC is a neuroendocrine (NE) tumor, it exhibits molecular features associated with neuroendocrine cells. Some NE-associated molecules, including synaptophysin (SYP), chromogranin A (CHGA), and neural cell adhesion molecule (NCAM or CD56), are used as diagnostic and prognostic markers in SCLC patients (2–4). The transcriptional machinery that drives SCLC differentiation and proliferation has been extensively investigated. Achaete-scute homolog 1 (ASCL1 or ASH1), a basic helix-loop-helix (bHLH) transcription factor, is a master regulator of neuroendocrine differentiation in lung development (5), and of cell proliferation and survival in SCLC (6, 7). Neurogenic differentiation 1 (NeuroD1), a bHLH transcription factor, regulates SCLC survival and migration by functioning as a regulatory hub of signaling pathways associated with these processes (8). In addition, insulinoma-associated protein 1 (INSM1), a zinc-finger protein with high expression in SCLC cells (9), contributes to NE differentiation by activating ASCL1 expression (10).

Lysine-specific demethylase 1A (LSD1) is a flavin-dependent monoamine oxidase that demethylates monomethylated or dimethylated lysine 4 or lysine 9 of histone H3, thereby epigenetically regulating the activation or repression of gene transcription in different contexts (11–13). LSD1 knockout mice exhibited an arrest in embryonic development prior to embryonic day 7.5 (13, 14). Moreover, studies using a pituitary-specific LSD1-knockout allele revealed that LSD1 is required for late cell lineage determination and differentiation during pituitary gland organogenesis (13). LSD1 is also required for normal hematopoiesis in adults (15). A conditional LSD1 knockout allele in adult hematopoietic stem cells induced the loss of terminally differentiated cells and pancytopenia (16). LSD1 forms a large, transcriptional repressor complex with GFI1B, CoREST, HDAC1, and HDAC2 that controls hematopoietic differentiation (17). The interaction between LSD1 and GFI1B is mediated by the SNAG domain of GFI1B, and this domain is also required for the transcriptional repression of GFI1B. Elevated LSD1 expression levels are associated with a poor prognosis in many types of solid cancers, including prostate cancer, breast cancer, esophageal cancer, and Ewing sarcoma (18–24). As LSD1 is also strongly overexpressed in a number of hematologic malignancies, it has gained attention as a potential therapeutic target, especially in acute myeloid leukemia (AML; ref. 25). Small-molecule or siRNA-mediated inhibition of LSD1 induces differentiation and inhibits proliferation of AML cells (25–29). We previously reported that T-3775440, a cyclopropylamine-derived irreversible LSD1 inhibitor, inhibited the proliferation of a subset of AML cell lines by disrupting the interaction between LSD1 and GFI1B, consequently inducing the transdifferentiation of AML cells (30).

Herein, we report that T-3775440 exhibits anticancer activity in SCLC cells by disrupting the interaction between LSD1 and the SNAG domain proteins INSM1 and GFI1B, thereby inhibiting LSD1-mediated NE transcription. Our findings provide new insights into SCLC biology that might facilitate the development of novel molecularly targeted therapies for patients with SCLC.

Cells were purchased from ATCC in 2014 and maintained in RPMI1640 media (Wako) supplemented with 10% (v/v) heat-inactivated FBS (HyClone Laboratories Inc.). All experiments were performed within 16 passages from the original frozen stocks. NCI-H1417, NCI-H510A, and NCI-H526 cells were authenticated using short tandem repeat DNA profiling in 2016. Mycoplasma test was performed by Central Institute for Experimental Animals and all cell lines were confirmed to be negative for mycoplasma. T-3775440, N-(4-{(1S,2R)-2-[(Cyclopropylmethyl)amino]cyclopropyl}phenyl)-1-methyl-1H-pyrazole-4-carboxamide hydrochloride was synthesized at Takeda Pharmaceutical Company Ltd. GSK-LSD1 was purchased from Santa Cruz Biotechnology.

Cells were seeded in 96-well plates and treated with T-3775440. Cell viability was determined using the CellTiter-Glo assay (Promega) as described previously (31). EC₅₀ values for cell growth inhibition were calculated using GraphPad Prism software (GraphPad Software Inc).

The screen was conducted at Eurofins (http://www.eurofins.com). Briefly, after a 10-day treatment with the compound, cell proliferation was determined according to the signal intensity of the incorporated nuclear dye.

Total RNA was extracted from cells using the RNeasy Mini Kit (Qiagen), and cDNA was synthesized using the SuperScript VILO cDNA Synthesis Kit (Thermo Fisher) according to the manufacturer's protocol. Gene expression assays (Applied Biosystems) were used as follows: INSM1 (Hs00357871_s1), GFI1B (Hs01062469_m1), ASCL1 (Hs04187546_g1), NEUROD1 (Hs01922995_s1), SYP (Hs00300531_m1), CHGA (Hs00900375_m1), GRP (Hs01107047_m1), KRT19 (Hs00761767_s1), UMODL1 (Hs00543151_m1), ZEB1 (Hs00232783_m1), PTPN14 (Hs00193643_m1), VIM (Hs00185584_m1), and GAPDH (Hs02758991_g1). The C_t value of each target gene was normalized to the C_t value of GAPDH as described previously (31).

SCLC cells were lysed in lysis buffer [1% NP-40, 0.25% deoxycholic acid, 50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, and 1 mmol/L EDTA) supplemented with cOmplete EDTA-free Protease Inhibitor Cocktail and PhosSTOP phosphatase inhibitor (Roche Diagnostics). LSD1 was immunoprecipitated from the lysates with anti-LSD1 antibody and Protein G Sepharose 4 Fast Flow (GE Healthcare). Immunoblot assays were conducted as described previously (31). The following antibodies were used in the immunoprecipitation or immunoblot analysis: anti-LSD1 (pAb-067-050) (Diagenode), anti-INSM1 (sc-271408), anti-GFI1B (sc-28356), and normal rabbit IgG (sc-2027; Santa Cruz Biotechnology).

Surface plasmon resonance (SPR) biosensing experiments were performed as described previously (30). Briefly, the LSD1/CoREST complex was immobilized on the sensor chip using a standard amine coupling procedure. INSM1 peptide (PRGFLVKRSKKSTPVSYRVR, purchased from SCRUM Inc.) or GFI1B peptide (PRSFLVKSKKAHTYHQPPRVQ, purchased from SCRUM Inc.) were injected at concentrations of 50 nmol/L and 1 μmol/L, respectively, for 120 seconds at a flow rate of 50 μL/minute. To inactivate the LSD1/CoREST complex prior to peptide injection, T-3775440 or GSK-LSD1 was injected at concentarations of 10 μmol/L for 180 seconds at a flow rate of 50 μL/minute. BIAevaluation ver. 4.1.1 (GE Healthcare) was used for subtraction data processing.

Nonsilencing ON-TARGETplus SMARTpool (L-HUMN-XX-0005) and LSD1 ON-TARGETplus SMARTpool (siLSD1_#1, L-009223-00-0005) were purchased from GE Dharmacon. The siLSD1_#2 (108660), siINSM1_#1 (s7474), siINSM1_#2 (s7475), siGFI1B_#1 (s15850), and siGFI1B_#2 (s15851) siRNAs were purchased from Ambion. AllStars Hs Cell Death Control siRNA (siCellDeath) was purchased from Qiagen. The cells were transfected with siRNA nanoparticles prepared at Takeda Integrated Technology Research Laboratories (30) or siRNA using RNAiMAX Transfection Reagent (Invitrogen) according to the manufacturer's protocol.

NCI-H510A, NCI-H526, and NCI-H1417 cells were treated with 500 nmol/L of T-3775440 for 3 days (n = 3). Total RNA was extracted using an RNeasy Plus Kit (Qiagen). The cells were labeled and scanned at Macrogen (http://www.macrogen.com) using a SurePrint G3 Human Gene Expression 8 × 60K v2 Microarray Kit (Agilent). Ingenuity Pathway Analysis (IPA, Qiagen) was used to search for possible biological pathways. All microarray data are deposited in the NCBI GEO repository (accession number: GSE100169).

SCLC cells were subcutaneously implanted in female BALB/cAJcl-nu/nu mice (CLEA Japan Inc.). The tumor volume was measured as described previously (30). The treatment/control (T/C) values (%) were statistically analyzed using Williams' test. All of the animal studies were conducted in accordance with the guidelines of the Takeda Institutional Animal Care and Use Committee in a facility accredited by the American Association for Accreditation of Laboratory Animal Care.

The tumors were excised and preserved in RNAlater (Ambion). Total RNA was extracted using an RNeasy Mini Kit (Qiagen), and cDNA was synthesized using a SuperScript VILO cDNA Synthesis Kit (Thermo Fisher) according to the manufacturer's protocol. Quantitative PCR was conducted as described previously (31).

T-3775440 is an irreversible inhibitor of LSD1 (30). We examined the effect of long-term treatment (10 days) with T-3775440 on 240 cancer cell lines derived from a wide range of tumor types. T-3775440 inhibited the growth of five leukemia cell lines, one ovarian cancer cell line (ES-2), and one SCLC cell line (DMS53) (Fig. 1A). However, the antiproliferative effects on ES-2 cells were not reproducible (data not shown). To confirm the inhibitory effect observed in the one SCLC cell line, we evaluated several other SCLC cell lines. As a result, T-3775440 inhibited the proliferation of three additional SCLC cell lines (NCI-H510A, NCI-H526, and NCI-H1417) in a concentration-dependent manner (Fig. 1B–D). The inhibitory effect of T-3775440 on cell growth was only observed after 4 or more days of treatment. The EC₅₀ values of T-3775440 in NCI-H510A, NCI-H526, and NCI-H1417 cells were highly similar, ranging from 18 to 61 nmol/L (Supplementary Fig. S1). The maximum growth inhibition rate associated with T-3775440 (10 μmol/L) was greater in NCI-H1417 and NCI-H510A cells (82% and 86% inhibition, respectively) compared with NCI-H526 cells (57% inhibition).

In a previous report, we demonstrated that T-3775440 inhibited the proliferation of a subset of AML cell lines by disrupting the interaction between LSD1 and GFI1B and inducing transcriptional reprograming and transdifferentiation (30). Therefore, we investigated whether T-3775440 also affects the interaction between LSD1 and its binding partners in SCLC cells. We focused on the potential interaction between LSD1 and INSM1, as INSM1 is strongly expressed in SCLC cells (Fig. 2A; Supplementary Figs. S2 and S3) and similar to GFI1B, INSM1 possesses a SNAG domain (Fig. 2B). Furthermore, INSM1 is reported to interact with LSD1 in pituitary endocrine cells (32). To evaluate the interaction between LSDI and INSM1 in NCI-H510A cells, we conducted immunoprecipitation experiments with an antibody against LSD1. INSM1 coimmunoprecipitated with anti-LSD1 in DMSO-treated control cells, but not in T-3775440–treated cells, suggesting that LSD1 interacts with INSM1 in SCLC cells and that T-3775440 disrupts this interaction (Fig. 2C). In addition, LSD1 interacted with CoREST, HDAC1, and HDAC2 in NCI-H510A cells. However, these interactions were unaffected by T-3775440. To test whether T-3775440 directly disrupts the LSD1–INSM1 interaction, we developed a SPR biosensing assay using recombinant LSD1 proteins and a peptide consisting of the SNAG domain of INSM1. Binding of the INSM1 or GFI1B peptide to the LSD1 protein significantly elevated the SPR signal, whereas T-3775440 pretreatment diminished this elevation (Fig. 2D). The publicly available LSD1 inhibitor GSK-LSD1 showed a similar effect (Supplementary Fig. S4). These results indicate that these irreversible LSD1 inhibitors, cyclopropylamine derivatives, directly disrupt the interaction between LSD1 and INSM1.

Next, we investigated whether INSM1 is involved in proliferation of SCLC cells. NCI-H510A, NCI-526, and NCI-H1417 cells were transfected with siRNA targeting INSM1 (siINSM1) or nonsilencing siRNA (siControl). INSM1 knockdown was confirmed using quantitative PCR (qPCR) and immunoblot analysis (Fig. 2E; Supplementary Fig. S5A). Cell proliferation was significantly inhibited in NCI-H510A and NCI-H1417 cells transfected with siINSM1 compared with siControl (Fig. 2F and G), but this effect was not observed in NCI-H526 cells (Fig. 2H). Compared with T-3775440 treatment, the siINSM1-mediated antiproliferative effect was marginal because of poor transfection efficiency, as shown in siCellDeath-transfected cells and difference in the kinetics of INSM inhibition between siRNAs and drug treatment.

As INSM1 knockdown did not significantly affect NCI-H526 cell proliferation, we hypothesized that the antiproliferative effects of T-3775440 on NCI-H526 cells were mediated by INSM1-independent mechanisms. NCI-H526 cells treated with T-3775440 underwent a distinct morphologic change (Fig. 3A). DMSO-treated NCI-H526 cells formed sheet-like cell aggregates, whereas T-3775440–treated cells formed spheroid structures. Unlike NCI-H510A and NCI-H1417 cells, NCI-H526 cells unexpectedly expressed high levels of GFI1B, and these levels were as high as in some AML cell lines (Fig. 3B). Similar to what we previously demonstrated in AML cells (30), immunoprecipitation assays revealed that LSD1 interacted with GFI1B, CoREST, and HDAC1/2 in NCI-H526 cells and that T-3775440 disrupted the interaction between LSD1 and GFI1B (Fig. 3C). To determine whether GFI1B is involved in T-3775440–associated changes in the morphology and growth of NCI-H526 cells, we inhibited the expression of LSD1, INSM1, and GFI1B using siRNA. siRNA-mediated knockdown of LSD1 or GFI1B, but not INSM1, mimicked the effects of T-3775440 on cell morphology (Fig. 3D–K; Supplementary Fig. S5B) and inhibited proliferation of NCI-H526 cells (Fig. 3L). These results suggest that the LSD1–GFI1B complex plays a critical role in the morphology and proliferation of NCI-H526 cells and that the effects of T-3775440 on NCI-H526 cells in this context are primarily mediated by the disruption of the LSD1–GFI1B complex.

To identify the genes affected by LSD1 inhibition in T-3775440–sensitive SCLC cell lines, we evaluated NCI-H1417, NCI-H510A, and NCI-H526 cells treated with T-3775440 for 3 days using DNA microarray analysis (Supplementary Table S1). Ninety-two genes (183 probes) and 25 genes (39 probes) were significantly upregulated and downregulated (P < 0.05), respectively, with a 1.5-fold change or greater in T-3775440–treated cells compared with control cells (Fig. 4A and B; Supplementary Table S2). These genes were considered to contain putative pharmacodynamic markers of T-3775440 in SCLC cells. To identify the molecular mechanism underlying the antiproliferative effect of T-3775440, we further evaluated the differentially expressed genes. Gene ontology (GO) analysis revealed that T-3775440–affected genes were associated with neuronal development and differentiation (Fig. 4C and D). Furthermore, Ingenuity Pathway Analysis (IPA) revealed that the transcription factors ASCL1 and NeuroD1, which are commonly expressed in SCLC cells (refs. 5, 7, 8, 33; Supplementary Fig. S6A and S6B), are upstream regulators of T-3775440–affected genes (Fig. 4E). T-3775440 inhibited the expression of ASCL1 and NeuroD1 at both the mRNA and protein levels in NCI-H1417 and NCI-H510 cells, but not in NCI-H526 cells (Fig. 4F–H; Supplementary Fig. S7A and S7B). T-3775440 also inhibited the expression of the NE markers SYP and CHGA, and gastrin-releasing peptide (GRP), a well-established serum marker of SCLC, in NCI-H1417 and NCI-H510A cells, but not in NCI-H526 cells (Fig. 4I–K; Supplementary Fig. S7C–S7E). In addition, the effects of the siRNA-mediated knockdown of ASCL1 or NeuroD1 on the proliferation of NCI-H1417 and NCI-H510A cells were investigated. All of the three siASCL1s significantly inhibited proliferation in NCI-H1417 and NCI-H510A cells (Fig. 4L–N). However, one of the three siNeuroD1s (siNeuroD1_#2) did not significantly inhibit the proliferation of NCI-H1417 cells, and growth-inhibitory effects of siNeuroD1s were less pronounced than those of siASCL1s in H1417 and H510A cells. These results suggest that the antiproliferative effects of T-3775440 are mainly mediated by a decrease in ASLC1 expression. In contrast, neither ASCL1 nor NeuroD1 knockdown affected the proliferation of NCI-H526 cells (Fig. 4O).

Next, we investigated the potential role of INSM1 in T-3775440–induced changes in gene transcription. In both T-3775440–treated and INSM1 knocked down NCI-H1417 cells, expression levels of the NE-associated genes ASCL1, NeuroD1, SYP, CHGA, and GRP decreased (Fig. 4F, 4G, 4I–K, and Fig. 5A–F), and expression levels of keratin 19 (KRT19), uromodulin like 1 (UMODL1), and vimentin (VIM) increased (Supplementary Fig. S8A–C; Supplementary Table S2). These results suggest that T-3775440–mediated changes in gene expression were at least partially attributable to the perturbation of INSM1 function via the disruption of its interaction with LSD1.

We evaluated the antitumor activity of T-3775440 in vivo using SCLC tumor xenograft mouse models that received two cycles (5 days on/2 days off) of once-daily oral doses of T-3775440. T-3775440 significantly inhibited tumor growth at 10 or 30 mg/kg, with treatment/control values (T/C) values of 54% and 45%, respectively, in a NCI-H510A model on day 15 (Fig. 6A). It also inhibited tumor growth at 15 and 30 mg/kg with T/C values of 34% and 20%, respectively, in a NCI-H526 model (Fig. 6B). Body weight loss was observed in mice treated with 30 mg/kg T-3775440 (Supplementary Fig. S9A and S9B) and the treatment had to be discontinued in one of five mice bearing NCI-H510A on day 10, when the humane endpoint was reached. None of the mice that received 10 or 15 mg/kg T-3775440 required treatment discontinuation. Consistent with the in vitro results, in vivo tumor growth inhibition was not detectable in the first few days after treatment, but was apparent by the second week of treatment. In addition, expression levels of the putative pharmacodynamic (PD) markers KRT19, UMODL1, ZEB1, and PTPN14 increased 8–24 hours after a single dose of T-3775440 (30 mg/kg) compared with the vehicle control (Fig. 6C–F). Plasma concentrations of T-3775440 reached peak levels 15 minutes to 2 hours after T-3775440 administration (T_max = 1.08 hours) and returned to near baseline levels after 24 hours (Supplementary Fig. S10; Supplementary Table S3).

In this study, we demonstrated that the irreversible LSD1 inhibitor T-3775440 exerts anticancer effects in SCLC in vitro and in vivo. Previous reports have demonstrated that LSD1 is overexpressed in many types of human cancers, including lung, breast, prostate, and blood cancers (18, 19, 25, 34, 35). However, preclinical data supporting a therapeutic role for LSD1 inhibitors have been limited to AML. We recently reported that T-3775440 exerts anticancer effects in a subset of AML cell lines (30). During the preparation of this manuscript, Mohammad and colleagues reported that the LSD1 inhibitor GSK2879552 exerted antitumor effects in a subset of SCLC cell lines (36). The three T-3775440–sensitive SCLC cell lines identified in this study (NCI-H510, NCI-H1417, and NCI-H526) were also sensitive to GSK2879552. Consistent with the reported effects of GSK2879552, the antiproliferative effects of T-3775440 were primarily cytostatic. Mohammad and colleagues utilized similar cell growth assays, and both LSD1 inhibitors contain a cyclopropylamine moiety. Together, these results suggest that a common mechanism underlies the antitumor effects of T-3775440 and GSK2879552 in SCLC cells. However, the precise mechanism underlying the antitumor effects of GSK2879552 remains unclear.

We found that T-3775440 disrupted the interaction between LSD1 and the SNAG domain transcription factor INSM1 (Fig. 2C and D). INSM1 knockdown inhibited proliferation of NCI-H510A and NCI-H1417 cells (Fig. 2E and F), and mimicked the effects of T-3775440 on gene expression patterns (Figs. 4 and 5). These results suggest that the antiproliferative effects of T-3775440 in SCLC cells are at least partially dependent on the inhibition of INSM1-mediated gene transcription. It has been reported that INSM1 interacts with LSD1 via its SNAG domain and that the resulting complex controls the differentiation of endocrine cells in the anterior pituitary gland (32). Our findings suggest that the LSD1–INSM1 complex plays a similar role in neuroendocrine differentiation in SCLC cells. Indeed, T-3775440 treatment disturbed neuroendocrine lineage-associated transcription in SCLC cells, as evidenced by a decrease in the expression of neuroendocrine markers such as CHGA and GRP (Fig. 4J and K). These findings were further supported by Gene Ontology analysis (Fig. 4C and D).

Furthermore, our results suggest that the downstream effects of T-3775440–mediated INSM1 inhibition are mainly dependent on ASCL1, a master regulator of neuroendocrine differentiation (Fig. 4E), because ASCL1 expression was downregulated in T-3775440–treated cells as well as in INSM1 knockdown cells (Figs. 4F and H and 5A and B). Notably, INSM1 has been reported to bind to the promoter region of ASCL1 and positively regulate ASCL1 mRNA expression (10). Therefore, ASCL1 downregulation may be a direct downstream effect of T-3775440–mediated INSM1 inhibition. Moreover, ASCL1 knockdown inhibited SCLC growth (Fig. 4L and M), a finding consistent with previous reports showing that ASCL1 is essential for the growth and survival of neuroendocrine lung cancers (6, 7) Although NeuroD1 is reported to promote SCLC cell survival and metastasis (8), NeuroD1 knockdown had a less pronounced effect on SCLC cell proliferation than ASCL1 knockdown in our experiment. These results suggest that T-3775440 targets the LSD1/INSM1 complex–ASCL1 axis and suppresses SCLC cell growth. Lenhart and colleagues recently reported that the BET inhibitor JQ1 inhibits SCLC proliferation via downregulation of ASCL1 (37). This supports our mechanistic insight that T-3775440–mediated INSM1 inhibition leads to growth inhibition via ASCL1 downregulation.

It should be noted that the anticancer effects of T-3775440 would be mediated by a similar mechanism in SCLC and AML cells. T-3775440 disrupted the interaction between LSD1 and INSM1 and altered INSM1-dependent neuroendocrine-related transcription in SCLC cells. Similarly, T-3775440 disrupted the interaction between LSD1 and GFI1B, and altered GFI1B-dependent megakaryocyte/erythroid–related transcription in megakaryoblastic leukemia/erythroleukemia cells (30). Unexpectedly, we found that GFI1B is highly expressed in NCI-H526 cells. Similar to what was observed in AML cells, T-3775440 disrupted the interaction between LSD1 and GFI1B in NCI-H526 cells and altered the corresponding downstream transcriptional program, thereby inhibiting cell growth. These results suggest that INSM1 and GFI1B play similar roles via their interaction with LSD1 in the maintenance of the neuroendocrine lineage in SCLC cells, although the morphologic change caused by T-3775440 were only observed in GFI1B-expressing NCI-H526 cells. Moreover, among 23 pulmonary tissue samples resected from SCLC patients, many of which overexpressed INSM1, one sample overexpressed GFI1B rather than INSM1, suggesting that GFI1B plays a clinically relevant role in patients with SCLC (Supplementary Fig. S11A and S11B; ref. 38).

Our findings indicate that the anticancer effects of T-3775440 are primarily mediated by the inhibition of the interaction between LSD1 and INSM1 or GFI1B, and by subsequent changes in downstream transcription associated with neuroendocrine phenotype. Stricker and colleagues reported that the redifferentiation of glioblastoma-derived iPS cells to non-neural lineage cells suppressed the malignant behavior (39). These results suggest that concordance between genetic mutation and differentiation status is important for the maintenance of the malignant phenotype. Thus, disturbance in the neuroendocrine lineage induced by T-3775440 may override the activity of genetically altered pathways in driving unconstrained proliferation. However, we cannot eliminate the possibility that the anticancer effects of T-3775440 are also mediated by catalytic inhibition of LSD1.

Ogasawara and colleagues reported that the cyclopropylamine-derived LSD1 inhibitors form covalent bond with flavin adenine dinucleotide (FAD) in the active site cavity of LSD1, where several residues of the histone tail substrate can be accommodated (40). Lin and colleagues have additionally reported that the sequence of the SNAG domain is similar to that of the histone H3 tail and that the interaction of LSD1 and Snail1, another SNAG domain protein, can be blocked by a histone H3 peptide (41). These data suggest that cyclopropylamine-derived LSD1 inhibitors specifically disrupt the interaction between LSD1 and SNAG domain proteins through covalent bond formation with FAD of LSD1. Indeed, interaction of CoREST or HDAC1/2 proteins with LSD1 is not affected by T-3775440 (Fig. 2C), suggesting its selectivity for SNAG domain proteins. Other SNAG domain proteins interact with LSD1 and are selectively recruited to specific loci to ensure the fine-tuning of transcription and downstream effects. For example, the SNAG domain protein Snai1 recruits LSD1 to epithelial gene promoters, mediating transcriptional repression of epithelial markers during epithelial–mesenchymal transition (41, 42). Thus, it would be worth investigating whether T-3775440 affects the interaction with other SNAG domain proteins and their functions in different cell types.

We demonstrated the antitumor effects of T-3775440 in vivo in tumor xenograft models generated from INSM1-overexpressing NCI-H510 cells and GFI1B-overexpressing NCI-H526 cells (Fig. 6A and B). T-3775440 upregulated the same subset of genes in both cell types in vitro and in vivo, suggesting that T-3775440 functions via the same mechanism in both settings. In addition, prolonged pharmacodynamic responses were observed even 24 hours after a single dose of T-3775440, despite the rapid clearance of the compound (T_max of approximately 1 hour). This effect might be attributable to the irreversible mode of action of T-3775440 and/or a delayed response to its downstream transcriptional effects. At efficacious doses, T-3775440 was generally tolerated in mice; however, thrombocytopenia was observed in T-3775440–treated mice (data not shown). This observation is consistent with that of previous reports in AML models and is considered as a mechanism-driven, reversible effect of T-3775440 (15, 30). Additional in vivo studies are needed to determine the optimal T-3775440 dose and dosing schedule in SCLC.

In conclusion, we demonstrated that the LSD1 inhibitor T-3775440 exerts anticancer effects in SCLC. Mechanistically, T-3775440 disrupted the interaction between LSD1 and the SNAG domain proteins INSM1 and GFI1B, thereby distorting the neuroendocrine-related gene expression program of SCLC. Our findings provide a novel insight into the functional interaction between LSD1 and SNAG domain proteins in SCLC and suggest that targeting these interactions using LSD1 inhibitors is a potential strategy to treat patients with SCLC. Moreover, it is generally difficult to directly target transcription factors using small-molecule inhibitors. However, emerging evidences suggest that the inhibition of factors involved in super-enhancer regulation may represent a strategy to target transcription factor–driven cancers, such as BRD4 inhibition (43, 44). Our results provide an example of the regulation of intractable oncogenic programs driven by sequence-specific transcription factors by targeting obligate epigenetic coregulators.

No potential conflicts of interest were disclosed.

Conception and design: S. Takagi, K. Nakamura

Development of methodology: S. Matsumoto, Y. Kamada

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Takagi, Y. Ishikawa, A. Mizutani, S. Iwasaki, Y. Kamada

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Takagi, A. Mizutani, S. Iwasaki

Writing, review, and/or revision of the manuscript: S. Takagi, S. Iwasaki, K. Nakamura

Study supervision: S. Takagi, T. Nomura, K. Nakamura

We would like to thank all the LSD1 project members for helpful discussions; Syu Morita for technical assistance; and Koji Yamamoto for processing of Agilent raw data. We would also like to express cordial gratitude to Daisuke Tomita and Shinichi Imamura for providing us with T-3775440 and Hiroshi Miyake and Christopher Claiborne for their guidance and support during the course of this work.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.