MRC-5, HEK293, HEK293F, K-562, Huh-7, and HepG2 cells were grown in Dulbecco's modified Eagle's medium (Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% (vol/vol) heat inactivated fetal bovine serum (Thermo Fisher Scientific, Inc.). NCI-N87, A549, and MDA-MB-231 cells were grown in RPMI-1640 supplemented with 10% heat inactivated fetal bovine serum. HEK293F cells were grown in Expi293^™ Expression medium (Gibco; Thermo Fisher Scientific, Inc.). All cell lines, with the exception of Huh-7 and HEK293F, were obtained from the ATCC (Manassas, VA, USA). Huh-7 cells were obtained from The Japanese Collection of Research Bioresources Cell Bank (Osaka, Japan) and HEK293F cells were purchased from Thermo Fisher Scientific, Inc.

BL21 (DE3; #RH217; Real Biotech Corporation, Banqiao, Taiwan) on ice for 10 min were transformed with the pET22b (+) vector (#69744; Merck KGaA, Darmstadt, Germany) containing AMF cDNA, according to the manufacturer's protocol. The transformed E. coli cells in luria broth (Lennox; Sigma-Aldrich; Merck KGaA) were incubated for 4 h at 37°C with agitation (250 rpm), and induced with 0.1 mM isopropyl β-D-1-thiogalactopyranoside. HEK293F cells were transfected with pcDNA3.1-AMF vector (Thermo Fisher Scientific, Inc.) containing AMF cDNA, using polyethylenimine (PEI; Polysciences, Inc., Warrington, PA, USA). The transiently transfected cells were grown in Expi293^™ Expression medium at 37°C for 4 days. The transformed E. coli lysate and the transfected HEK293F conditioned culture medium were applied to HisTrap FF (GE Healthcare Bio-Sciences, Pittsburgh, PA, USA) equilibrated with 50 mM Tris buffer (pH 7.4) containing 0.5 M sodium chloride. The column-bound fractions were eluted with 50 mM Tris buffer (pH 7.4) containing 0.5 M sodium chloride and 250 mM imidazole, and then dialyzed with PBS (pH 7.4) three times at 201 × g for 10 min at 4°C using VIVA spin 20 (Sartorius AG, Göttingen, Germany) with 30 kDa molecular weight threshold.

The cell lysate of E. coli cells transformed with the pET22b-AMF vector and rhAMF purified from the transformed E. coli cells were separated by electrophoresis on a 10% SDS Tris-glycine polyacrylamide gel. The protein bands were stained with Coomassie brilliant blue R-250 (Bio-Rad Laboratories, Inc., Hercules, CA, USA).

The E. coli cell lysates and the rhAMF-containing column chromatography fractions were separated by 10% SDS-PAGE and blotted onto a polyvinylidene difluoride membrane using iBlot apparatus (Invitrogen; Thermo Fisher Scientific, Inc.). The blot was blocked with 5% skimmed milk in PBS (pH 7.4) for 2 h at room temperature. The blocked membrane was incubated for 1 h at room temperature with a mouse monoclonal antibody recognizing the His-tag conjugated to rhAMF (#70796-3; dilution, 1:1,000; Novagen; Merck KGaA), then with a horseradish peroxidase (HRP)-conjugated anti-mouse immunoglobulin G (IgG) antibody (KPL, Inc., Gaithersburg, MD, USA) for 1 h at room temperature. The blot was incubated in the presence of tetramethylbenzidine substrate (KPL, Inc.), the reaction was stopped by adding a stop solution (#50-85-01; KPL, Inc.) and the signal was visualized using ChemiDoc XRS+ with Image Lab^™ Software (Bio-Rad Laboratories, Inc.).

The enzyme activity of a commercial rhAMF (MyBioSource, San Diego, CA, USA) and the purified rhAMFs were measured using a glucose-6-phosphate isomerase activity assay kit (Abcam, Cambridge, UK), according to the manufacturer's protocol.

AMF secreted in the culture medium and AMF in the plasma of Balb/c nude mice were measured using a glucose-6-phosphate isomerase human ELISA kit (Abcam), according to the manufacturer's protocol.

Cells were suspended at a density of 1×10⁶ cells/ml in PBS (pH 7.4) and incubated in the presence of a fluorescein isothiocyanate-conjugated anti-AMF receptor (AMFR) antibody (#ab76841; dilution, 1:1,000, Abcam) for 1 h at 4°C in the dark. Subsequent to incubation, the cells were washed 3 times with PBS and incubated in the presence of goat-anti-rabbit IgG H&L (Alexa Fluor 488^®; #ab150077, dilution 1:1,000, Abcam) for 1 h at 4 in the dark. The cells were washed 2 times with stain buffer (#554656; BD Biosciences, Franklin Lakes, NJ, USA). Subsequently, the cell surface fluorescence was analyzed by flow cytometry using a FACScaliber (BD Biosciences).

Cancer cell motility was measured by a modification of the Boyden chamber assay, using a CytoSelect 96-well cell migration assay kit (Cell Biolabs Inc., San Diego, SD, USA). The MDA-MB-231, A549, NCI-N87 and HepG2 cells (1×10⁶ cells/ml) were incubated in the upper chamber, and 10 µg/ml E. coli-derived rhAMF was added to the lower chamber, which was separated by an 8 µm pore membrane. After 24 h incubation, cells on the upper surface of the membrane were removed by scraping, and the cells on the lower surface were detached using cell detachment solution. The detached cells were lysed and quantified using CyQuant GR fluorescent dye (Cell Biolabs, Inc., San Diego, CA, USA), according to the manufacturer's protocol.

Human monoclonal antibodies against AMF were selected from a naive human scFv (single chain fragment variable) phage display library (a total of ~10¹¹ members, A & R Therapeutics Co., Daejeon, Korea). A total of 4 rounds of biopanning were performed and four clones of anti-AMF antibodies of 9A-4H, 7F4, 8F and 4H were isolated. During each round, AMF-specific phages were selected out of the pool by washing away non-specific binders using PBS-T (PBS containing 0.05% Tween 20). Following 4 rounds, a mixture of highly specific phage clones was separated from a mono clone phage and selected on the basis of affinity and specificity against AMF using indirect ELISA. Briefly, putative AMF-specific phages were added into the well of 96-well plates coated with rh-AMF and incubated for 2 h at room temperature. The wells were washed three times with PBS-T and then anti-M13 HRP antibody (#ab50370; Abcam) was added and incubated for 1 h at room temperature. Tetramethylbenzidine substrate was added and the reaction was stopped at 10 min by adding the ABTS^® Peroxidase stop solution kit (#50-85-01; KPL, Inc.). The plate was measured at a wavelength of 450 nm. The phage vector from each clone was subjected to sequencing to determine the nucleotide sequence of VL and VH, and then the nucleotide sequences were analyzed and grouped. The VH and VL gene elements of a selected phage were amplified using polymerase chain reaction and then subcloned into mammalian expression vectors (pYG300H and pYG300L, respectively; A & R Therapeutics Co.), each of which contained a constant region of heavy chain and light chain of human IgG1 coding sequence for fully human IgG1 expression. Sequencing was performed using ABI PRISM 3730XL Analyzer (Applied Biosystems; Thermo Fisher Scientific, Inc.) to confirm in-frame insertion of each variable region into the expression vectors.

The antibody-antigen interaction was examined by surface plasmon resonance spectrometry (SR7500DC; Reichert, Inc., Depew, NY, USA). Prior to antigen immobilization, a Polyethylene Glycol/Carboxyl sensor chip (#13206061; Reichert, Inc.) was first activated with a mixture of 0.1 M 1-ethyl-3- (3-dimethylamminopropryl) carbodiimide hydrochloride and 0.05 M N-hydroxysuccinimide at a flow rate of 20 µl/min for 10 min. Following the activation of the chip surface, 50 µg/ml antigen was flowed over the chip surface. The antigen was immobilized to the surface of the chip via free amine coupling to the immobilized succinimide, followed by quenching the remaining activated succinimide ester with 1 M ethanolamine (pH 8.5). The chip was equilibrated with PBS-T. For the analysis of antibody association, solutions (0, 12.5, 25, 50, 100, 200, 400 and 800 nM) of anti-AMF antibodies, 9A-4H, 4H, 7F4 and 8F, in PBS-T was flowed over the sensor chip at a rate of 30 µl/min for 5 min. For the molecular dissociation, washing buffer was flowed over the sensor chip at a rate of 30 µl/min for 5 min. The chip was regenerated with 20 mM HCl. The data was analyzed using the Scrubber 2.0 software (BioLogic Software; Bio-Rad Laboratories, Inc.).

The six-week-old BALB/c female mice (weighing ~20 g; Charles River Laboratories International, Inc., Yokohama, Japan) were housed under specific pathogen free conditions. Food and water were sterilized prior to use; the light cycles were 12 h light/dark and the temperature was maintained between 20 and 22°C. The animal protocol was reviewed by the Korea University Institutional Animal Care & Use Committee (Seoul, Korea). A549 or NCI-N87 cells suspended in PBS (pH 7.4) were injected subcutaneously in the right flank of the mice at a concentration of 2×10⁷ cells/ml. The intravenous administration of sunitinib (4 or 40 mg/kg, orally administered, once a day for 4 weeks; Sigma-Aldrich; Merck MGaA), cetuximab (4 or 40 mg/kg, intraperitoneal, twice a week for 4 weeks; Namyang Pharm Co., Ltd., Seoul, Korea), trastuzumab (15 mg/kg, intravenously twice a week for 2 weeks; Namyang Pharm Co., Ltd., Seoul, Korea) or human monoclonal anti-AMF antibodies (9A-4H, 7F4, 8F and 4H; 15 mg/kg, intravenously twice a week for 2 weeks) against E. coli-derived rhAMF, began when the average volume (V₀) of the tumors reached ~100 mm³. Tumor volumes (V_t) were measured 3 times a week for 3 weeks using a vernier caliper, and then calculated by the formula 0.5 × height × length × width. At the end of the experiment the mice were sacrificed by CO₂ asphyxiation.

GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, USA) was used for statistical analysis. One-way ANOVA and Dunnett's post hoc t-test were used for multiple comparisons. Spearman correlation analysis was used to compare the correlation between plasma AMF concentration and tumor weight in A549-xenografted mice. P<0.05 was considered to indicate a statistically significant difference.

To select the best material for the experiments, rhAMFs from 3 different sources were analyzed and compared. The enzyme activity of rhAMFs purified from the E. coli lysate transformed with the pET22b-AMF vector, rhAMFs present in the culture medium of HEK293F cells transiently transfected with the human AMF cDNA in the pcDNA3.1-AMF vector, and commercial rhAMF were 4.64, 3.50 and 3.64 mU/mg (Table I), respectively. The enzyme activity of rhAMFs prepared from E. coli and HEK293F cells were similar to that of the commercial rhAMF, which was included in the assay for comparison. The identity of the final product purified by the column chromatography method was validated by SDS-PAGE and western blot analyses (Fig. 1A and B). The lanes of the gel loaded with the commercial rhAMF, the soluble fraction from the pET22b-AMF-transformed E. coli lysate, and the 250 mM imidazole elution fraction revealed clear bands stained with Coomassie brilliant blue R-250 at ~55 kDa, which is the expected molecular weight of rhAMF. The bands were identified to be rhAMF by western blot analysis using a monoclonal antibody against the His-tag. The commercial rhAMF was not detected by the western blot analysis since it does not contain a His-tag (Fig. 1B). Since the enzyme activity of E. coli-derived rhAMF was not lower than that of rhAMFs from other sources, and the preparation of E. coli-derived rhAMF is easier than that from HEK293F cells, the following experiment was conducted using E. coli-derived rhAMF.

E. coli-derived rhAMF was examined for its chemoattractive activity upon cancer cells using a Boyden chamber. As shown in Fig. 2, the migration of MDA-MB-231 and A549 cells was significantly increased by treatment with rhAMF. The cell motility of MDA-MB-231 and A549 cells as a result of treatment with rhAMF increased by ~1.6- and 1.9-fold, respectively, compared with that of the control cells. However, the migration of NCI-N87 and HepG2 cells was not significantly changed by treatment with rhAMF.

In order for a signal to be transmitted into cells by extracellular AMF, the presence of AMFR is required. The relative amount of AMFR was measured in the normal human fibroblast MRC-5 and embryonic kidney HEK293 cells, and in the human gastric carcinoma NCI-N87, hepatocellular carcinoma HepG2, lung carcinoma A549, chronic myelogenous leukemia K-562, hepatocellular carcinoma Huh-7, and breast cancer MDA-MB-231 cells using flow cytometry (Fig. 3). The percentage values of cells that express AMFR in HEK293 and MRC-5 cells were 3.57 and 4.93%, respectively. In the case of cancer cells, while the percentage values of cells that express AMFR in K-562, Huh-7, and HepG2 cells were similar to those in the normal cell lines, those in NCI-N87, A549, and MDA-MB-231 cells were 8.71, 8.86, and 11.32%, respectively, which were 1.8 to 3.2 times higher compared with those in the normal cells. In contrast to the expression of AMFR, there were huge differences in the amount of AMF secreted by these cells (Fig. 4). The AMF-specific productivities of NCI-N87 and HepG2 cells were statistically greater (P<0.001) compared with the other groups of cells. While the AMF-specific productivities of NCI-N87 and HepG2 cells were 0.097±0.006 and 0.119±0.011 pg/cell/day, respectively, those of A549, K-562, Huh-7 and MDA-MB-232 cells were no higher than 0.037±0.011 pg/cell/day. The AMF-specific productivities of the normal cell lines, MRC-5 and HEK293, were 0.012±0.001 and 0.000±0.000 pg/cell/day, respectively. These results were in contrast to those demonstrated in Fig. 2, in that NCI-N87 and HepG2 cells secreted endogenous AMF much more actively than the other cell lines, including MDA-MB-231 and A549 cells, whose motilities were more greatly affected by treatment with rhAMF.

To investigate whether there was a correlation between tumor growth and plasma concentration of AMF, A549-xenografted Balb/c nude mice were treated with sunitinib (21), a small-molecule multi-targeted receptor tyrosine kinase inhibitor, and cetuximab (22,23), a chimeric anti-EGFR monoclonal antibody (Fig. 5). Blood and tumors were collected at the end of the experiments, the plasma concentration of AMF and the tumor weights were measured, and the correlation between plasma AMF concentration and tumor weight was analyzed by the Spearman's rank correlation method (Fig. 5). The results reveal that the plasma AMF concentration became lower as the tumor weight decreased upon treatment with sunitinib and cetuximab. As demonstrated with a Spearman's rank coefficient of 0.785 obtained from the analysis of the correlation between the plasma AMF concentration and tumor weight, there was a relatively good correlation between the two factors.

In total, 4 human monoclonal antibodies (9A-4H, 4H, 7F4 and 8F) specific for AMF were selected from a naive human scFv phage display library using biopanning. The binding kinetics of the anti-AMF human monoclonal antibodies with E. coli-derived and HEK293F-derived rhAMF were measured using surface plasmon resonance (Table II). The equilibrium dissociation constant (K_D) values of these 4 monoclonal antibodies with E. coli-derived rhAMF were similar to those with HEK293F-derived rhAMF, however, the K_D value of 8F with E. coli-derived rhAMF may not be obtained due to a lack of detectable dissociation from the immobilized ligand. Among these 4 antibodies, 9A-4H and 7F4 had somewhat lower affinity for the two rhAMFs than the 8F and 4H antibodies. Since the binding kinetics of these antibodies with rhAMFs derived from two different sources were similar, the following experiments, to investigate the effect of AMF on the migration of cancer cells, were performed with E. coli-derived rhAMF, which may be easily prepared compared with the method using the human cell line.

Among the 4 monoclonal antibodies (9A-4H, 4H, 7F4, and 8F), only 9A-4H and 7F4 significantly inhibited the growth of tumors in NCI-N87-xenografted mice (Fig. 6A). Monoclonal antibody 9A-4H caused the greatest inhibition among the four monoclonal antibodies, with an inhibition rate slightly higher than that caused by trastuzumab, a HER2 inhibitor, which was included as a positive control. The concentration of AMF in plasma collected from the mice at the end of the experiments demonstrated a similar pattern to the inhibition of tumor growth (Fig. 6B). The mean plasma AMF concentration of the Balb/c nude mice treated with 9A-4H and trastuzumab were significantly decreased compared with that of the control Balb/c nude mice (P<0.01). Although it was not statistically significant, the mean concentration of AMF in the plasma of the mice treated with 7F4 was decreased compared with the control mice.