Novel cancer‐specific epidermal growth factor receptor antibody obtained from the serum of esophageal cancer patients with long‐term survival

Abstract Although esophageal cancer has a poor prognosis after recurrence, some patients have shown long‐term survival despite recurrence. We hypothesized that induction of either antitumor Abs or antitumor‐specific CTLs could play a role in long‐term survival (5 years or longer) in patients with recurrence and/or distant metastases. Therefore, we aimed to obtain Abs that specifically bind to cancer cells by using serum samples from patients with a good prognosis. A phage library was prepared using PBMC mRNA of the patients, and cell panning was carried out using an esophageal cancer cell line. Results showed the presence of an epidermal growth factor receptor (EGFR) Ab, KT112, that specifically bound to the cancer cell line. Notably, KT112 bound to only EGFR‐positive cancer cells but failed to bind to normal esophageal cells. Furthermore, KT112 was characterized by responses to EGFR expressed on cancer cells but not to the recombinant extracellular domain of EGFR. Immunohistochemical analysis showed that KT112 reacted with 17.4% of esophageal squamous cell carcinoma tissue but not with any other cancer or normal tissue, suggesting that the Ab recognizes cancer‐specific forms of EGFR and might have contributed to tumor suppression in patients with esophageal cancer. Furthermore, because of its high cancer specificity, KT112 could be a promising therapeutic option (e.g., in Ab‐drug conjugates) for esophageal cancer.


| INTRODUC TI ON
Esophageal cancer accounts for approximately 590,000 deaths per year worldwide, making it the sixth leading cause of cancer-related mortality. Furthermore, the prevalence of the disease is increasing worldwide, with 570,000 new cases diagnosed annually. 1 Most patients with esophageal cancer are diagnosed in the advanced stages of the disease, and the prognosis for such patients is poor despite the availability of multimodal therapies such as surgery, chemotherapy, and radiation therapy, with a 5-year survival rate of less than 20%. [2][3][4] Due to its malignant potential, only a few patients with esophageal cancer have long-term survival after recurrence. In recent years, cancer immunotherapies, such as effector cell therapies, Ab therapies, and immune checkpoint inhibitor therapies, have been shown to be highly effective, [5][6][7] and some cancer patients may have had a good prognosis due to suppression of disease progression by a tumor immune response. In this study, we explored esophageal cancerspecific target molecules, such as EGFR, vascular endothelial growth factor, HER2, MET, and PD-L1, for the development of molecular targeted drugs. 8 Epidermal growth factor receptor, a tyrosine kinase receptor, is known to be involved in various cancer progression events, including signal cell proliferation, migration, and metastasis. Epidermal growth factor receptor has been shown to be upregulated in 30%-90% of esophageal cancers. Furthermore, 70% of esophageal cancers overexpress EGFR. [9][10][11] Currently, four major types of EGFR mAbs-cetuximab, panitumumab, nimotuzumab, and necitumumab-are clinically used for several types of cancers, including lung, head and neck, colon, and pancreatic cancer. In addition to these four approved drugs, various EGFR Abs have been tested in clinical trials, including monotherapy, combination therapy with small molecules, and ADCs, such as mAb A13, AMG595, depatuxizumab (ABT-806), duligotuzumab (MEHD7945A, RG7597), futuximab (Sym004), GC1118, imgatuzumab (GA201), matuzumab (EMD72000), panitumumab (ABX-EGF), zalutumumab, humMR1, and tomuzotuximab. 12 Despite their effectiveness in preclinical studies, the clinical utility of EGFR-targeted therapies has been limited because of toxic side effects and poor clinical responses due to the broad expression profiles of the antigens that can cross various normal tissues. The most common toxicities were skin rashes, diarrhea, constipation, stomatitis, fatigue, and electrolyte abnormalities. Therefore, EGFR Abs with a higher tumor selectivity are desired.
We hypothesized that induction of either antitumor Abs or antitumor-specific CTLs could play a role in the long-term survival (5 years or longer) in esophageal cancer patients with recurrence. Using a phage library prepared from patient PBMC mRNA, we attempted to obtain Abs that specifically bind to cancer cells by cell panning using a cancer cell line. Subsequently, using antigen identification, antigen reactivity, and immunohistochemical analyses, we elucidated that the isolated Ab was a novel EGFR Ab.

| Ethics approval
All study participants gave consent for future analyses of their blood samples for research purposes. The protocol for this prospective study was approved by the ethics committee of Kyowa Kirin Co., Ltd.
(IRB no. 2013_028) and Toho University (Ethical Committee no. . Patients provided written informed consent before enrollment.

| Patients
Serum samples were obtained from 17 patients with advanced esophageal cancer who had survived for 5 years or longer after initial treatment. Of these, four patients had developed recurrent tumors at the time of sampling. Of these four patients, two died 6 months and 5 years after recurrence, while the remaining two survived for 5 years or longer after recurrence. Serum Abs were analyzed using samples from one of the two patients with long-term survival. This patient had stage III (T3N1M0) disease with a high p53 Ab titer. 13 Three years after surgical treatment, the patient had developed cervical lymph node recurrence, but he survived for 5 years or longer after recurrence.

| Evaluation of cell reactivity using serumderived polyclonal Abs
Serum samples were collected using a blood collection tube containing a serum separator (# 367988; BD), allowed to stand for 30 min, and then centrifuged at 1000 g for 10 min. Antibodies were purified from serum samples using Protein G (#17-0618-02; GE Healthcare).
The cells were suspended in D-PBS supplemented with 5% FBS, 1 mM EDTA, and 0.1% NaN 3 (staining buffer) and dispensed into a 96-well plate. After centrifugation, the supernatant was removed and patient Ab and human Ig (#867279694, for blocking purpose; Japan Blood Products Organization) diluted to 100 µg/ml were added to the pellet and incubated for 30 min on ice. After washing, Alexa Fluor 647-labeled goat anti-human Ab (#A21445; Molecular Probes) was added and incubated for 30 min. After washing, the cells were suspended in staining buffer containing 7-AAD staining solution, and the fluorescence intensity of each cell was measured using a flow cytometer.

| Phage display
The scFv library was generated using modified phage display protocols. [14][15][16] Briefly, RNA from blood collected in PAXgene tubes was extracted using the PAXgene Blood RNA Kit (PreAnalytiX) according to the manufacturer's instructions. cDNA was synthesized using the SMARTer RACE cDNA Amplification Kit (#634859; Clontech). The V H and V L chain genes were amplified from the cDNA using PCR. The V H and V L PCR products were inserted into the phagemid vector pCAN-TAB 5E (Amersham Pharmacia) and transformed into Escherichia coli TG1 cells by electroporation. Thereafter, the cells were cultured in 2YTGA (2YT media, glucose, and ampicillin) and M13KO7 helper phages were added and incubated for 1 h. Subsequently, kanamycin was added and incubated at 37℃ for 16 h and the phages were purified using a PEG/NaCl solution and resuspended in 1 ml staining buffer. Phages were added to MRC5 cells or HEK293F cells (negative control cells) and incubated at 4℃ for 1 h. Unbound phages were transferred to OE21 cells and incubated at 4℃ for 1 h. Phages that bound to OE21 cells were eluted, neutralized, and infected into TG1 cells. A total of five selection cycles were carried out.

| Construction and expression of Abs
The scFv regions of KT112 (phage clone) were cloned in-frame into an expression vector with a signal peptide and the human IgG4 Fc region (S228P/S235E/R409K). 17 Immunoglobulin G4 (S228P/S235E/ R409K) is a low-effector-activity Ab format that shows increased sta-

| Flow cytometry
The cells were suspended in staining buffer and dispensed into a 96-well plate. After centrifugation, the supernatant was removed and each Ab was added to the pellet and incubated for 30 min on ice. After washing, R-phycoerythrin-labeled goat anti-human Ab was added and incubated for 30 min. After washing, the cells were suspended in staining buffer containing 7-AAD staining solution, and the fluorescence intensity of each cell was measured using a flow cytometer. The results were analyzed, and the mean fluorescence intensity was calculated using the geometric mean.

| Antigen identification
Membrane proteins were purified using the Mem-PER Plus Membrane Protein Extraction Kit (Thermo Fisher Scientific) from OE21 cells as samples and MRC5 cells as negative controls. Immunoprecipitation was then carried out using KT112 and plasma membrane proteins prepared using the Dynabeads ProteinA Immunoprecipitation Kit (Thermo Fisher Scientific). Proteins recovered by immunoprecipitation were subjected to SDS-PAGE. Bands specifically purified using the KT112 Ab were excised, and in-gel differentiation was performed using Trypsin/Lys-C Mix, Mass Spec Grade (Promega).
Peptides extracted from the gels were used for protein identification by liquid chromatography with tandem mass spectrometry analysis.
The Swiss-Prot database was used for analysis.

| Antibody-dependent cellular cytotoxicity assay and CDC assay
Antibody-dependent cellular cytotoxicity was determined by the lactate dehydrogenase release assay as described previously, 19 using cryopreserved human PBMCs (HemaCare) as effector cells and OE21 cells as target cells at an effector / target ratio of 25:1.

| Internalization assay
Cells were seeded in 96-well plates and cultured overnight at 37℃.
Antibodies were added at dilutions per the required concentration in the medium. Hum-ZAP (Advanced Targeting Systems) was added at 100 ng/well and incubated at 37℃. OE21 cells were cultured for 2 days and T. Tn and HEEpiC cells for 3 days. Viable cells after culture were detected using the Cell Proliferation Kit II (XTT assay; Roche Diagnostics). The XTT-labeled reagent and electron coupling reagent were mixed and added to a 96-well plate and incubated at 37℃ for 4 h. The absorbance was measured at a sample wavelength of 490 nm and a reference wavelength of 630 nm.

| Determination of esophageal SCC cell linebinding Abs in patients with a good prognosis
For identification of esophageal SCC cell line-binding Abs, IgG Abs were purified from the serum of a patient who had survived for 5 years or longer after recurrence ( Figure 1). The purified IgG Abs showed binding to four esophageal SCC cell lines as well as normal esophageal cells (Figure 2A). Notably, the purified IgG Abs also showed binding to a fibroblast cell line (MRC5 cells). Because IgG Abs obtained from serum are polyclonal in nature, an absorption assay was carried out to determine whether the Abs that bound to the esophageal SCC cell lines were the same as the Abs that bound to the MRC5 cells. Flow cytometry following absorption of MRC5 cells incubated with the purified Abs showed no decrease in reactivity of the esophageal SCC cell line (OE21), suggesting that the OE21 cell-binding Abs were different from the MRC5 cell-binding Abs ( Figure 2B). Thus, the presence of esophageal SCC cell-binding Abs was confirmed, which formed the basis for isolation of these Abs using a phage library as described below.

| Isolation of esophageal SCC cell line-binding Abs from a patient phage library
In order to isolate cancer-specific mAbs, we prepared a phage dis- Of these four Abs, three were noncancer cell-selective and also reacted strongly with normal esophageal cells, whereas the remaining Ab (KT112) was cancer cell-selective. The KT112 Ab reacted with all four esophageal SCC cell lines, albeit the reaction with one of the strains was very weak, and only slightly with normal esophageal cells ( Figure 3). Of note, KT112 did not bind to MRC5 cells or HUVECs ( Figure 3).

| Recognition of EGFR by KT112
The membrane fractions of each cell were immunoprecipitated with each Ab and compared using SDS-PAGE ( Figure 4A

| KT112 does not bind to recombinant EGFR extracellular domain
The reactivity of KT112 with the recombinant EGFR extracellular domain was analyzed. Panitumumab, AM1, and cetuximab, which are available as EGFR Abs, were used for comparison. AM1, a cancerspecific EGFR Ab, 21 mainly recognizes EGFR vIII, but also binds to EGFR WT at higher expression levels. The responses to the recombinant EGFR extracellular domain and recombinant EGFR vIII extracellular domain were evaluated using an ELISA ( Figure 5). Unlike cetuximab, panitumumab, and AM1, KT112 did not bind to the recombinant proteins of EGFR and EGFR vIII. The lack of response to recombinant EGFR suggests the possibility of a selective response to EGFR expressed on cancer cells.

| Comparison with known EGFR Abs
As KT112 was thought to have a different mechanism for EGFR reactivity compared with the known EGFR Abs, ADCC and CDC assays between KT112 and cetuximab were carried out. Figure 6 shows ADCC ( Figure 6A) and CDC against OE21 cells ( Figure 6B).
KT112 had potent ADCC activity to OE21, the extent of which was lower than that of cetuximab. KT112 also showed no CDC activity.
Subsequently, a competitive inhibition assay between KT112 and the known EGFR Abs was carried out. Figure 7 shows the reactivity of KT112 in the presence of the competitive Abs ( Figure 7A) and the reactivity of each Ab in the presence of KT112 ( Figure 7B). KT112 was competitively inhibited by cetuximab and panitumumab but not were not competitively inhibited by KT112 ( Figure 7B). This discrepancy might be attributable to the lower affinity of KT112 compared with cetuximab or panitumumab. Alternatively, binding of cetuximab or panitumumab might have altered the conformation of EGFR and thus prevented binding of KT112.

| KT112 has a similar internalization capacity as cetuximab
The internalization capacity of KT112 was confirmed using a saporin-labeled anti-human IgG Ab as a secondary Ab. Following internalization of the primary Ab, the saporin added to the secondary Ab was also taken up into the cell, resulting in inhibition of protein synthesis with inactivation of the ribosome, leading to cell death.
Results of the detection of cell death in normal esophageal cells (HEEpiC) and in the esophageal SCC cell line (OE21 cells) are shown in Figure 8A,B, respectively. Cetuximab, which is a known EGFR Ab, was also evaluated. Interestingly, KT112 was shown to be internalized to a similar level as cetuximab. Notably, cetuximab also bound to HEEpiC, normal esophageal cells that internalize and induce cell death, whereas KT112 did not bind to normal esophageal cells and did not induce cell death.

| KT112 specifically bound to esophageal cancer tissues
The KT112 Ab did not bind to formalin-fixed paraffin-embedded tissue samples but bound to frozen sections. Therefore, frozen array samples were prepared from the esophageal SCC and normal esophageal mucosa for immunohistochemical analyses. Frozen array samples were also used for the various types of cancer and normal tissues, since KT112 was obtained as an Ab that binds to esophageal SCC and lung SCC tissue. A summary of the immunohistochemical

| DISCUSS ION
A novel EGFR Ab, KT112, was obtained from serum samples of esophageal SCC patients with a good prognosis despite recurrence.
This Ab specifically bound to esophageal cells but not to normal cells. of Abs, such as ADCC and CDC. In the cell growth inhibitory assay using cetuximab or KT112, cell growth inhibition was observed only with cetuximab, but not with KT112 (Document S1, Figure S1).
Moreover, it was found that IgG1 was the predominant IgG subclass F I G U R E 7 Competitive inhibition of each epidermal growth factor receptor (EGFR) Ab. The reactivity of each EGFR Ab to OE21 cells was analyzed by flow cytometry. The horizontal axis indicates the Ab name, and the vertical axis indicates mean fluorescence intensity. (A) Reactivity of the KT112 Ab in the presence of a competitive Ab is shown. Concentrations of the competitive Abs are 0.3, 1, 3, 10, and 30 μg/ml. 2,4-Dinitrophenol (DNP) Ab was analyzed only at 30 μg/ml. AM1 Ab was analyzed at 3, 10, and 30 μg/ml. The concentration of the biotinylated KT112 Ab for detection is 0.8 μg/ml. (B) Reactivity of each Ab in the presence of the KT112 Ab. Concentrations of the competitive Abs are 0.3, 1, 3, 10, and 30 μg/ml. Concentrations of the Abs for detection are 0.8 μg/ml for the biotinylated KT112 Ab, 0.1 μg/ ml each for biotinylated cetuximab and panitumumab, and 2 μg/ml for the biotinylated AM1 Ab among cancer cell line-binding Abs isolated from the patient's serum (Document S1, Figure S2). As the IgG1 subclass is known to have an effector function, it is possible that KT112 contributed to antitumor activity in the patient through exertion of effector functions, especially ADCC, as shown in Figure 6.
Indeed, ADCC might be an important therapeutic mechanism of EGFR Abs, as suggested by the significant correlation between CD16 polymorphism, an individual factor that determines ADCC intensity, 26 and the clinical outcome in some clinical trials of cetuximab. 27 Therefore, enhancing ADCC by engineering Ab molecules might be a promising approach to improve anti-EGFR therapies. An ADCC-enhancing EGFR Ab, imgatuzumab (GA201), showed promising results in head and neck SCC based on the therapeutic kinetics of the peripheral immune cells and cytokines; however, it failed to show effectiveness in subsequent phase II trials. 28 Alternatively, using KT112 as a vehicle for ADCs could be a good approach for the treatment of esophageal cancer, considering its high internalizing activity. Several clinical trials have assessed ADCs using EGFR Abs. 21,29,30 As EGFR is expressed systemically, it is considered important to enhance tumor selectivity in order to achieve strong efficacy with low toxicity. Our results indicated that KT112 was highly selective for esophageal SCC. Considering that KT112 was obtained