Lung cancer is the leading cause of cancer-related deaths in the Western world and
is classified into two groups: small cell lung cancer (SCLC) and non-SCLC (NSCLC) [1].
SCLC, a neuroendocrine tumor, is distinguished from NSCLC by its rapid tumor growth,
high degree of invasiveness and early development of widespread metastases [2]. SCLC
is distinctly different from extrapulmonary small cell carcinoma with respect to disease
progression, prognosis, and etiology [3]. Without proper treatment, the life expectancy of
SCLC patients is less than four months. Although the five-year relative survival rate has
improved by 7% over the last few decades, it remains extremely poor [4].
A variety of molecular markers have been implicated in the pathogenesis and prognosis
of SCLC [5,6]. Paracrine or autocrine signal transduction pathways are widely used
to explain dysregulated SCLC growth [6]. In addition, tumor protein p53, retinoblastoma
protein, NOTCH, MYC, and phosphatidylinositol 3-kinase (PI3K) are aberrantly mutated
in SCLC; however, well-established etiological factors, such as EGFR mutations that occur
in NSCLC, have not been identified [7–10]. SCLC has a very aggressive course and is characterized
by genomic instability, increased vascularity, and a high metastatic potential [11].
Consequently, most SCLC patients already present with metastatic disease outside of the
chest, at the time of diagnosis, which results in premature death [12]. In addition, most
SCLC patients are current or former heavy smokers, which is associated with a high tumor
mutational burden, with C:G > A:T transversions being the most common type of base
substitutions [13,14]
The c-Kit proto-oncogene encodes a transmembrane tyrosine kinase growth factor receptor
that belongs to the platelet-derived growth factor receptor (PDGFR) family [15,16]. Its ligand stem cell factor (SCF) is a hematopoietic growth factor that promotes the proliferation of
multiple hematopoietic stem cells [17,18]. In addition, c-Kit activity is dysregulated in
various cancers [19,20]. Previous studies reported that the expression of c-Kit containing
oncogenic mutations is either dysregulated and/or up-regulated in various cancers, which
results in SCF-independent c-Kit activation and an aggressive form of cancer [20]. Interestingly,
a variety of evidence indicates that SCLC cell lines and tumors express both the
c-Kit receptor and SCF mRNA, suggesting that these gene products constitute an autocrine
loop that mediates tumor cell survival and growth [21,22]. Although SCLC is significantly
correlated with smoking, it does not contain oncogenic c-Kit mutations. Immunohistochemical
staining showed that overexpression of c-Kit occurs in 70% of SCLC patients [23,24].
Imatinib, which was developed to target BCR-ABL, platelet-derived growth factor receptor
(PDGFR), and c-Kit, is currently used to treat chronic myeloid leukemia, acute lymphoid
leukemia, gastrointestinal stromal tumor (GIST), and hypereosinophilic syndrome [25–28].
A variety of in vitro and in vivo studies demonstrated that imatinib exhibits therapeutic
efficacy against SCLC [29,30]. However, in phase 2 clinical trials, imatinib failed to exhibit
significant therapeutic efficacy as shown by a lack of objective responses [31–33]. Thus, an
alternative approach to target c-Kit in SCLC is needed. In this study, we generated and
characterized 4C9, a human antibody targeting c-Kit. We developed an antibody-drug
conjugate (ADC) using DM1, a microtubule inhibitor, coupled with N-succinimidyl-4-(Nmaleimidomethyl)
cyclohexane-1-carboxylate (SMCC) to generate 4C9-DM1, and then
evaluated its therapeutic efficacy in vitro and in vivo.

2. Results

2.1 4C9 Antibody Specifically Binds to c-Kit

First, we examined whether the 4C9 antibody specifically binds to c-Kit on cell surface.
FACS analysis revealed that 4C9 binds to various SCLC cell lines, including NCI-H526,
NCI-H1048, and NCI-H889, in a dose-dependent manner (Figure 1A). Interestingly, 4C9
antibody binding was saturated at 50 ng/mL in c-Kit-positive SCLC cell lines. In addition,
the expression of c-Kit was higher in NCI-H889 cells compared with that in NCI-H526 and
NCI-H1048 cells, which is consistent with a previous report [34]. However, 4C9 did not
show cross-reactivity with NCI-H446 and NCI-H2170 cells, which are c-Kit-negative SCLC
cell lines. We further examined the specific binding of 4C9 to c-Kit using siRNA knockdown
experiment. Western blot analysis showed that c-Kit siRNA efficiently decreased
protein expression in NCI-H1048 cells (Figure 1B). FACS analysis demonstrated that c-Kit
expression on the cell surface was also down-regulated by c-Kit siRNA. Taken together,
4C9 binds specifically to the extracellular domain of c-Kit on the cell surface.

Figure 1. Determination of specific binding of the 4C9 antibody to the surface of SCLC cells. (A)
SCLC cell lines were incubated with 4C9 antibody in a dose-dependent manner and analyzed by
FACS. (B) NCI-H1048 cells were transfected with control c-Kit si-RNA and incubated for 72 h.
Binding of 4C9 (1 μg/mL) to NCI-H1048 cells was determined by FACS. The si-RNA-mediated
down-regulation of c-Kit protein expression was evaluated byWestern blot analysis. Alpha-tubulin
was used as a loading control.

Next, we investigated whether the 4C9 antibody could inhibit the binding of SCF,
a ligand for c-Kit. Competitive enzyme-linked immunosorbent assay (ELISA) results
showed that the binding of 4C9 antibody to c-Kit was not affected by SCF, even at high
concentrations (Figure 2A), suggesting that 4C9 antibody binds to c-Kit independent of
SCF. In addition, we assessed whether 4C9 could inhibit SCF-mediated phosphorylation
of c-Kit. Using GIST-T1 cells, pretreatment with 4C9 antibody resulted in decreased c-
Kit phosphorylation in a dose-dependent manner; however, the phosphorylation levels
of the ERK and Akt, downstream molecules of c-Kit, were not changed by treatment
with 4C9 antibody (Figure 2B). Interestingly, total c-Kit levels decreased by 4C9 antibody
treatment (Figure 2B). Hence, a decrease in phosphorylated c-Kit levels may result from
decreased expression of c-Kit. A stability assessment of c-Kit indicated that the 4C9 antibody
dramatically decreased total c-Kit levels in a time-dependent manner in both GIST cell
lines and in some SCLC cell lines (Supplementary Figure S1), which may be associated
with ubiquitination-dependent degradation. Nevertheless, this needs further elucidation.
In contrast to GIST-T1, 4C9 did not reduce SCF-mediated c-Kit phosphorylation or c-Kit
stability in the NCI-H526 and NCI-H1048 cell lines (Figure 2C). Furthermore, the 4C9
antibody did not inhibit phosphorylation of ERK and Akt induced by SCF, which suggests
that the 4C9 antibody does not function as an antagonist of SCF/c-Kit signaling.

 

Figure 2. Characterization of the 4C9 antibody. (A) Human c-Kit (20 ng/well) was coated onto
96-well plates and the binding of the 4C9 antibody was investigated in the presence of human SCF at
the indicated concentrations. The results represent the mean  SD of three independent experiments.
(B,C) GIST-T1, NCI-H526, or NCI-H1048 cells were treated with 4C9 at the indicated concentrations
in the presence or absence of SCF (100 ng/mL). The phosphorylation of c-Kit, Akt, and ERK was
assessed by Western blot analysis. In NCI-H1048 cells, phosphorylation of Y568/570 and Y823 by
SCF treatment was not detected. Alpha-tubulin was used as a loading control. The results represent
the mean  SD of three independent experiments.

2.2. Generation and characterization of ADC (4C9-DM1)

Even though c-Kit is overexpressed in SCLC cell lines, naked antibodies cannot be
applied to treat SCLC because of the limited contribution of SCF/c-Kit signaling in the
pathogenesis of SCLC and failure of imatinib [32,33]. Therefore, the development of an
ADC would be a reasonable option to effectively treat SCLC. One important factor to
consider when creating an ADC is the internalization efficiency after the complex formation
of the antibody with the target molecule. Therefore, we first investigated whether the 4C9
antibody is internalized in SCLC cell lines. FACS analysis exhibited that the internalization
efficiency of the 4C9 antibody was 91% in NCI-H526, 76.6% in NCI-H1048, and 68.6%
in NCI-H889 cells (Figure 3A), which suggests that the 4C9 antibody could be used as
an efficient carrier for the specific delivery of toxin to treat SCLC. Next, we generated
an ADC using SMCC-DM1, which consists of a noncleavable linker and a microtubule
inhibitor. Because DM1 absorbs ultraviolet light at 252 nm [35], the absorbance of the naked
antibody (4C9) and ADC (4C9-DM1) at 252 nm was analyzed. The absorbance of 4C9-DM1
was higher than that of 4C9 at 252 nm (Figure 3B). The drug-antibody ratio (DAR) was
determined as described previously [36] and the DAR was approximately 2.16. SDS-PAGE
analysis exhibited that the conjugation of SMCC-DM1 to the 4C9 antibody resulted in a
slight size shift (Supplementary Figure S2). Since N-hydroxysuccinimide ester of the SMCC
linker reacts with primary amines of lysine residues in the antibody, the target binding
affinity of ADC may be affected. An ELISA demonstrated that the binding affinities of
4C9 and 4C9-DM1 to c-Kit were similar (Figure 3C). A quantitative analysis of the binding
affinity using surface plasmon resonance (SPR) indicated that the binding affinity of 4C9 to
human c-Kit was 5.5 × 10^-9 M (Ka = 2.27 × 10⁴ M^-1s^-1 and Kd = 1.27 × 10^-4 s^-1), and 4C9-DM1 also showed similar binding affinity (5.46 × 10^-9 M; Ka = 2.14 × 10⁴ M^-1s^-1 and Kd = 1.16 × 10^-4 s^-1) (Figure 3D). Taken together, these results indicate that conjugation
using SMCC-DM1 to the 4C9 antibody did not affect the binding affinity of 4C9 antibody
for c-Kit.

Figure 3. Characterization of 4C9-DM1. (A) The internalization of the 4C9 antibody into various
SCLC cell lines was determined using FACS analysis. SCLC cells were incubated with cycloheximide
(75 μg/mL) and blocked with Fc blocker for 10 min to inhibit Fc receptor-mediated internalization.
SCLC cells were incubated in the presence or absence of the 4C9 antibody at 4 'C or 37 'C for 1–4 h
and subjected to flow cytometry. The fluorescent signal of the 4C9/c-Kit complex on the cell surface
decreased after incubation at 37 'C. (B) Optical absorbance of 4C9-DM1 compared with that of the
naked 4C9 antibody at 252 nm. The binding affinity of 4C9 and 4C9-DM1 to c-Kit was compared
using ELISA (C) and SPR (D).

2.3. 4C9-DM1 Exhibits Antitumor Activity In Vitro and In Vivo
Next, we analyzed the in vitro cytotoxicity using various SCLC cell lines (NCI-H526,
NCI-H889, NCI-H1048, NCI-H446, and NCI-H2170) and a breast cancer cell line (MDAMB-
453). The c-Kit negative cell lines (NCI-H446, NCI-H2170, and MDA-MB-453) were
used as negative controls. 4C9-DM1 exhibited in vitro cytotoxicity against NCI-H526, NCIH889,
and NCI-H1048 with half-maximal inhibitory concentration (IC50) values ranging
from 158 pM to 4 nM. The in vitro cytotoxic activity of 4C9-DM1 against c-Kit-positive
cancer cell lines was 4- to >300-fold higher than that against c-Kit-negative cancer cell
lines (Figure 4A,B and Table 1). Interestingly, the cytotoxic activity of DM1 against the
c-Kit-positive cancer cells was 7- to >77 fold higher when applied as ADC rather than as
payload alone. By contrast, its cytotoxic activity was 2–5 times lower against c-Kit-negative
cancer cells (Table 1), suggesting the inclusion of a payload as an ADC may reduce off-target
toxicity. DM1 induces cell cycle arrest at the G2/M phase by inhibiting the assembly of
microtubules, resulting in apoptosis of actively dividing cells. Therefore, we analyzed the
effect of 4C9-DM1 on the cell cycle using NCI-H526, a c-Kit positive SCLC cell line, and
NCI-H446, a c-Kit negative SCLC cell line. As shown in Figure 4C, 4C9-DM1 significantly
increased the cell population in the G2/M phase in a time-dependent manner but 4C9
and IgG-DM1 did not. In addition, 4C9, 4C9-DM1, and IgG-DM1 did not alter the cell
population in the G2/M phase in c-Kit negative NCI-H446 cells, suggesting that cell cycle
arrest in NCI-H526 cells is specifically mediated by DM1 delivered by complex formation
with 4C9-DM1 and c-Kit.

Figure 4. 4C9-DM1 exhibits cytotoxicity against SCLC cells in vitro. (A,B) c-Kit positive or negative
SCLC cell lines were seeded into 96-well plates and incubated with 4C9 or 4C9-DM1 in a dosedependent
manner for 3–5 days. Live cells were stained with Hoechst 33342 (10 μM) at 37 'C for 30
min and quantitated using a Celigo Imaging Cytometer. Treatment with 4C9-DM1 decreased cell
viability in a dose-dependent manner. The results represent the mean  standard error of the mean
of at least three independent experiments. (C) 4C9-DM1 induced cell cycle arrest at the G2/M phase.
SCLC cell lines were seeded into 96-well plates and incubated with 4C9 (1 g/mL), IgG-DM1 (1
μg/mL), or 4C9-DM1 (1 μg/mL) for 24 and 48 h. Then, the cells were stained with propidium iodide
and analyzed using a Celigo Imaging Cytometer (** and *** vs. control; ## and ### vs. IgG-DM1). The
results represent the mean  standard error of the mean of at least three independent experiments.
The results are presented as the mean  standard error of the mean. The means were compared using
an unpaired Student’s two-sided t-test. ** p < 0.01, *** p < 0.001, ## p < 0.01, ### p < 0.001

 

Based on the in vitro cytotoxicity assay, the in vivo efficacy of 4C9-DM1 was examined
using mouse models xenotransplanted with NCI-H526. Although IgG-DM1 did not
exert an effect, 4C9-DM1 significantly suppressed NCI-H526 tumor growth in a dosedependent
manner (Figure 5A and Supplementary Figure S3A). Tumor growth inhibition
(TGI) rates of 4C9-DM1 at doses of 1, 3, and 5 mg/kg were 40%, 45%, and 59%, respectively,
compared with that of the vehicle control. The 4C9 antibody alone partially inhibited
tumor growth, but this effect was not statistically significant. Body weight losses due
to the administered materials were not observed (Figure 5A and Supplementary Figure
S3B), suggesting that there was no concern related to toxicity. Chemotherapy, including
etoposide, cisplatin, carboplatin, and lurbinectedin are used to treat SCLC patients as the
standard of care [37,38]. Although combination therapy using chemotherapeutic drugs is
effective, most SCLCs rapidly recur within 1 year [39]. Therefore, we determined whether the combination chemotherapy could enhance the therapeutic efficacy
of SCLC. As shown in Figure 5B and Supplementary Figure S3C, 4C9-DM1, lurbinectedin,
and carboplatin/etoposide exhibited similar antitumor activities; the TGI rates of these
groups were 50% at day 18, compared with that of the vehicle control. Interestingly, the
combination of 4C9-DM1 with lurbinectedin or carboplatin/etoposide synergistically suppressed
tumor growth. The TGI rates of both groups were 85%, compared with that of the
vehicle control at day 18 with subsequent regrowth. The combinatorial treatment with
4C9-DM1 plus lurbinectedin induced body weight loss of approximately 10%; however,
body weight increased again after cessation of the treatment (Figure 5B and Supplementary
Figure S3D). This suggests that this combination may be used to treat SCLC with
manageable

Figure 5. 4C9−DM1 suppresses SCLC tumor growth in a xenograft mouse model. (A,B) Antitumor
activity of 4C9−DM1 was evaluated in an in vivo xenograft mouse model. NCI−H526 cancer cells
were implanted into immune-deficient mice as described in the Methods section. Mice with established
tumors were randomized into different treatment groups when the tumor volume reached ~
200 mm3 (n = 6). The animals were intravenously administered vehicle, 4C9, IgG−DM1, or
4C9−DM1. Carboplatin (60 mg/kg on days 1 and 11) and etoposide (3 mg/kg on days 1–5 and days
11–15) were intraperitoneally administered or combined with 4C9−DM1. Additionally, lurbinectedin
(0.08 mg/kg on days 1, 8, and 15) was intravenously administered or combined with
Figure 5. 4C9?DM1 suppresses SCLC tumor growth in a xenograft mouse model. (A,B) Antitumor
activity of 4C9-DM1 was evaluated in an in vivo xenograft mouse model. NCI-H526 cancer cells were
implanted into immune-deficient mice as described in the Methods section. Mice with established
tumors were randomized into different treatment groups when the tumor volume reached ~200
mm3 (n = 6). The animals were intravenously administered vehicle, 4C9, IgG-DM1, or 4C9-DM1.
Carboplatin (60 mg/kg on days 1 and 11) and etoposide (3 mg/kg on days 1–5 and days 11–15) were
intraperitoneally administered or combined with 4C9-DM1. Additionally, lurbinectedin (0.08 mg/kg
on days 1, 8, and 15) was intravenously administered or combined with 4C9-DM1 as indicated. Green
arrows indicate the administration of vehicle, IgG-DM1, 4C9, or 4C9-DM1, and blue and red arrows
indicate the administration of lurbinectedin and carboplatin, respectively (*, **, and *** vs. their
respective corresponding vehicle control; § and §§§ vs. their respective corresponding 4C9-DM1
control; † vs. carboplatin/etoposide; ‡ vs. lurbinectedin). The results are presented as the mean 
standard error of the mean. The means were compared using an unpaired Student’s two-sided t-test.
* p < 0.05, ** p < 0.01, *** p < 0.001, † p < 0.05, ‡ p < 0.01, § p < 0.05, §§§ p < 0.001.