Anchorage independence altered vasculogenic phenotype of melanoma cells through downregulation in aminopeptidase N /syndecan-1/integrin β4 axis

Loss of vascular structure in the tumor derived from suspended melanoma

We established the lung metastases of adherent or suspended melanoma cells, and examined the xenograft tumor samples to see whether detachment stress played roles in different phenotypes. These were achieved by injection of the adherent or suspended melanoma cells into tail veins of nude mice [13, 14]. Tumor disseminated right away in the circulation that would simulate the post-intravasation stage and allowed cell-trapping at lung tissues for the development of lung metastasis [13–15]. Two groups of nude mice were inoculated with either adherent or suspended melanoma cells and allowed tumor development for 2 months. Upon inspection of the lungs of these mice inoculated with either adherent or suspended melanoma cells, multiple white nodules were grossly developed as metastases derived from amelanotic cells (Figure 1A). No significant difference in the number or the size of nodules in the lung tissues of these two groups of mice.

Pathological dissection with HE stain showed totally different architectures in the nodules generated by adherent or suspended melanoma inoculations. As shown in Figure 1B, networks of connective tissues were both observed in the nodules derived adherent or suspended melanoma cells. In the nodules derived from adherent melanoma cells, dominant vascular tubes infiltrated with red blood cells (RBCs) were seen. These tubular structures of hyaline cavities were embedded and surrounded by loose connective tissues with diluted fibrous structures. These nodule tissues were also immunostained to visualize the endothelial cell marker of the vessel, CD31. It showed that most of RBC-containing cavities were CD31-positive. It indicated the generation of vascular structure in the neoplastic nodules derived from adherent melanoma cells. For the nodules derived from suspended melanoma cells, no dominant vascular cavities were seen. However, some RBCs were infiltrated into the loose connective networks. As to CD31 immunostaining, no clear vascular structures were observed in the nodules derived from suspended melanoma cells. This suggested the loss of vasculogenic phenotypes for melanoma cells after suspension treatment.

Proteomic analysis showed downregulation of ANPEP in suspended melanoma cells

In order to find out the potential targets correlated with the loss of vascular structures in nodules derived suspended melanoma cells, we compared the protein contents in the adherent, suspended, and re-attached melanoma cells by proteomic analysis. The tumor metastases from suspended melanoma cells could be recognized as re-attachment of suspended melanoma cells.

In order to increase the peptide identification, we divided protein lysate as cytosolic and nuclear protein fractions. After protein digestion by Lys-C/trypsin, and followed dimethyl labeling, each part were further separated into 5 fractions by strong cation exchanger. They were analyzed using LTQ Orbitrap mass spectrometer connected with the HPLC system. The peptide targets from each fraction were integrated and their protein identities were determined by the mass of fragmented peptides compared using SwissProt database.

Several protein identities were upregulated or downregulated in suspended melanoma cells or re-attached melanoma cells comparing to those in adherent cells (Table 1). Totally 61 proteins were significantly changed upon cell suspension with 27 protein upregulated and 34 protein downregulated. Totally 97 proteins were significantly changed upon cell reattachment after suspension with 44 protein upregulated and 53 protein downregulated.

In further, we used the webtool DAVID (The Database for Annotation, Visualization and Integrated Discovery [16, 17]) to analyze the changed proteins to associate with the loss of vascular structures. Functional annotations showed enriched clusters at different terms of gene ontology (biological process, cellular component, and molecular function).

As shown in Tables 2 and 3, the changes in protein expression were categorized and mostly enriched at nucleus for the processing of nucleic acid synthesis. The molecular function associated with blood vessel formation or angiogenesis was also highlighted, which was correlated with our observation (Figure 1). As showing in Table 1 and Figure 2, there were 5 downregulated proteins (ANPEP; aminopeptidase N, ANXA2; annexin A2, CAV1; caveolin-1, CSPG4; chondroitin sulfate proteoglycan 4, and TGM2; transglutaminase 2) associated with blood vessel/vasculature development in suspended melanoma cells; while 4 downregulated proteins (ANPEP, CAV1, CHD7; Chromodomain-helicase-DNA-binding protein 7, and TGM2) associated with blood vessel/vasculature development in re-attached melanoma cells. Accordingly, it suggested that these 3 downregulated proteins (ANPEP, CAV1, and TGM2) were potentially remained downregulated and might associate with the loss of vascular structures in tumor metastases. Comparing our previous microarray data (GSE42876 and GSE61671) with the result of current proteomic analysis, the positive and strong correlation that both ANPEP and CAV1 gene products were downregulated in suspended melanoma cells and reattached melanoma cells. No significant change was seen in the result of cDNA microarray analysis for TGM2, which suggested the change in protein expression might not result from altered gene expression but posttranslational regulation. ANPEP is the membrane-bound metalloproteases for general protein hydrolysis and also involved in cancerogenesis and tumor angiogenesis [18]. It is also recognized as CD13, the known cancer-stem cell marker for liver cancers [19].

In addition, we also noticed that integrin β4 was also downregulated in both suspended and re-attached melanoma cells. Although it was not indicated and highlighted in geneset of vessel formation by DAVID analysis, it was already known that integrin involved in different cell activities, including cell proliferation, adhesion, and migration [20], as well participated in several physiological or pathological conditions such as vessel formation or tumor angiogenesis through specific extracellular matrix protein recognition [21]. For instance, integrin α6β4 was recognized as a specific laminin-binding protein [22, 23]. Our previous study suggested melanoma cell suspension contributed to reduced laminin-binding ability through downregulation of SDC1 protein [1]. Our other research work had demonstrated the correlation of SDC1 with ANPEP expression that affected stemness characteristics of hepatoma cells [24]. We hypothesized that the regulation of ANPEP, SDC1, and integrin β4 expression might associate with loss of vasculogenic phenotype upon suspension of melanoma cells.

Laminin-binding integrin beta4 was downregulated by melanoma cell suspension, SDC1 downregulation, and ANPEP downregulation

We examined the expression of ANPEP in adherent, suspended, and reattached melanoma cells by qPCR and flow cytometry. As seen in Figure 3A, ANPEP mRNA expression was significantly downregulated upon cell suspension, while partially restored by re-attachment after suspension. Expression at the cell surface was also examined using flow cytometry. As seen in Figure 3A, surface levels of ANPEP in suspended and re-attached melanoma cells were reduced, which were consistent with qPCR results.

Previously, we found that anchorage independence enabled the decreased SDC1 expression and altered the expressions of several integrin isoforms [1]. Consistent with our previous observation by microarray analysis, qPCR results also suggested that cell suspension upregulated integrin αV, β1, and β3; while integrin α6, β2, and β4 were downregulated (Figure 3B). This indicated the downregulation of integrin α6β4 would correlate with the reduced laminin-binding ability [1]. The protein expressions of integrin isoforms were also examined by western blot. As shown in Figure 3C, integrin β2 and β4 protein expression were reduced by cell suspension. However, integrin α6 protein level was not affected by cell suspension.

Since SDC1 level also affected the laminin-binding ability and it was downregulated in suspended melanoma, we checked whether SDC1 expression level would affect laminin-binding integrin expression. As seen in Figure 4A, the transfection of SDC1-specific shRNA suppressed SDC1 expression, but upregulated SDC2 expression, which was consistent with our previous observation [2]. Integrin β3 expression was upregulated, while integrin β2 expression was marginally reduced by SDC1-specific shRNA transfection. Only integrin β4 expression was significantly downregulated by SDC1-specific shRNA transfection. We suggested that integrin β4 expression would be specifically regulated by SDC1. The protein expressions of integrin isoforms were examined by western blot. As shown in Figure 4B, only integrin β4 protein expression was reduced by suppression of SDC1 expression. Although integrin β2 protein expression was reduced by cell suspension (Figure 3B and 3C), we suggested that integrin β2 expression would be regulated by other factors under anchorage-independence. In addition, SDC1 downregulation by shSDC1 did not change the level of ANPEP expression (Figure 4C). This implied that ANPEP would unidirectionally regulate SDC1 expression and sequentially affect the integrin expression.

In order to investigate whether ANPEP level in melanoma cells affected the expression of integrin isoforms and vascular phenotypes as we observed in suspended or reattached melanoma cells, we transfected ANPEP-specific shRNAs into melanoma cells. As seen in Figure 5A, shRNAs transfection reduced ANPEP expression levels (53% and 39% for shANPEP_a and shANPEP_b, respectively) in melanoma cells. The expression of ANPEP at the cell surface was also suppressed by shRNA transfection as evidenced by flow cytometry (Figure 5A). In addition, the expression levels of SDC1 and integrin isoforms upon suppression of ANPEP expression were examined by qPCR and western blot. As seen in Figure 5B, SDC1 and integrin β4, but not significant for integrin α6 and β2, were downregulated by suppression of ANPEP expression as examined by qPCR. As the protein expression, suppression of ANPEP protein expression reduced SDC1 protein expression (Figure 5C). The relative protein levels were reduced as 86% and 74% for shANPEP_a and shANPEP_b, respectively. Moreover, the protein level of integrin β4 was significantly reduced by ANPEP downregulation (Figure 5C), however, integrin α6 and β2 protein levels were increased and remained unchanged, respectively. Nevertheless, these indicated the expression axis ANPEP/SDC1/integrin β4 existed in melanoma cells, which was downregulated by anchorage independence. Previously, we had shown that the downregulation of SDC1 protein was mediated by the PKCδ activation in suspended melanoma cells [1]. So that we investigated whether ANPEP levels affected with PKCδ activation. As seen in Figure 5C, the levels of PKCδ phosphorylation were increased, while PKCδ protein expressions relatively remained unchanged upon the transfection of ANPEP-specific shRNAs. This suggested the regulatory axis of ANPEP/SDC1/integrin β4 that would be affected by the status of PKCδ activation/inactivation upon detachment stress.

ANPEP downregulation in suspended melanoma conferred the characteristics of invasiveness loss but increased tumor growth

In our previous study, anchorage-independence of melanoma cells by treatment of cell suspension contributed to the loss of cell invasiveness [1]. Since ANPEP expression was downregulated upon cell suspension, we examined the effects on ANPEP-specific shRNAs at cell migration ability. As seen in Figure 6A, cell migration ability was inhibited by transfection of ANPEP-specific shRNAs in the transwell migration assay. This was consistent with reported literatures that ANPEP played role in melanoma cell invasion [25, 26]. Next, we examined the effect of suppression in ANPEP expression at xenograft tumor growth. We subcutaneously inoculated melanoma cells transfected with control vector (shLuc) or ANPEP-specific shRNA (shANPEP-a). As shown in Figure 6B, the subcutaneous tumors derived from shANPEP-transfected melanoma cells were relatively larger than those derived from shLuc-transfected melanoma cells. This suggested downregulation of ANPEP in suspended melanoma cells would promote tumor malignancy. This was consistent with our previous study that anchorage-independence of melanoma cells by cell suspension promoted tumor growth [3].

Other literatures suggested ANPEP was a therapeutic target or prognostic marker of other tumors, that indicated its role in tumor malignancy [18, 19, 25]. However, we observed that the reduced level of ANPEP promoted melanoma tumor growth. In order to reconfirm this result, we overexpress ANPEP protein and examined its effects in protein expressions of ANPEP/SDC1/integrin β4 axis and xenograft tumor growth. As seen in Figure 6C, overexpression of ANPEP reduced the extent of phosphor-PKCδ, but increased the expression of SDC1 and integrin β4. This strengthened our conclusion that ANPEP would regulate the downstream expression of SDC1 and integrin β4 protein. Detachment treatment of melanoma cells would promote extensive xenograft tumor growth (Figure 6B). However, ANPEP overexpression reduced suspended cell-inoculated tumor growth (Figure 6D). This suggested ANPEP might play a role as tumor suppressor in melanoma cells under detachment stress.

ANPEP downregulation contributed to the downregulation of SDC1 and integrin β4 in situ

It was known that integrin proteins involved in different cell activities, including cell proliferation, adhesion, and migration [20], as well participated in several physiological or pathological conditions such as vessel formation or tumor angiogenesis [21]. We further examined whether altered expressions of ANPEP/SDC1/integrin β4 were associated with different vascular phenotypes in vivo. As seen in Figure 7A, HE stain and immunostaining of CD31 showed vascular structures in the tumor derived from adherent melanoma cells. It also showed positive ANPEP, SDC1, and integrin β4 expression around these vascular cavities. Interestingly, Expression of integrin β4 showed different sub-phenotypes. The filled arrow indicated the location of vascular cavities. Some integrin β4-positive cells (dashed eclipses) were surrounded these cavities, while some integrin β4 positive cells were lined into vascular cavities as the same in CD31-positive cavities (the locations were indicated by empty arrows). We suggested these might be the trace of trans-differentiation for integrin β4-positive cells to become vascular structures. For the tumor derived from suspended melanoma cells, no vascular structure was observed by HE stain. No cavities with positive CD31, ANPEP, SDC1, or integrin β4 immunostaining was seen.

In the tumor sample inoculated with shLuc-transfected cells, several vascular cavities were seen with positive CD31, ANPEP, SDC1, and integrin β4 immunostainings (Figure 7B). The large areas of integrin β4-positive cells were also observed (dashed ellipse) and surrounded at the fibrous structure. In the tumor sample inoculated with shANPEP-a transfected cells, we still could find some vascular cavities with RBCs (empty arrow). However, no CD31-positive or ANPEP-positive cells to form these vascular cavities, while weak SDC1 and integrin β4 immunostaining were seen. These cavities seem to be formed by loose connective tissues with diluted fibrous structures, which enabled the RBC diffusion. Nevertheless, we suggested the ANPEP expression correlated with SDC1/ integrin β4 expression, which affects vasculogenic phenotypes to show CD31-positive vascular cavities.

Cell culture and cell transfection

Human melanoma A375 cells were purchased from Bioresource Collection and Research Center (BCRC; Hsinchu, Taiwan) with authentication. Adherent culture were maintained using culture dish (Corning Incorporated Life Sciences, Tewksbury, Massachusetts, USA) and DMEM medium supplemented with 10% (v/v) fetal bovine serum (Biological Industries Ltd., Cromwell, Connecticut, USA). Suspended melanoma cells were established and cultured by impaired attachment of adherent melanoma cells according to the reported procedures [37, 38]. Re-attached melanoma cells were established by seeding suspended cells onto culture dish and keeping at incubator for 1 day.

The plasmids expressing shRNAs against ANPEP (clone ID: TRCN0000050239 and TRCN0000050238) or SDC1 (clone ID: TRCN0000072581) gene were purchased from the National RNAi Core Facility located at Institute of Molecular Biology/Genomic Research Center, Academia Sinica, Taiwan. The gene construct to overexpress ANPEP and the control vector were all purchased from Sino Biological (Beijing, China). Cells were transfected using TurboFect transfection reagent (Thermo Fisher Scientific Inc., Pittsburgh, Pennsylvania, USA) according to the manufacturer’s instruction. The transfected cells were selected and enriched under growth medium containing 5 μg/mL puromycin (for shRNA-transfect cells) or 0.5 mg/mL hygromycin B (for ANPEP-transfected cells).

Animal experiment and IHC analysis

The animal experiments were supervised and regulated by Experimental Animal Care and Use Committee in FuJen Catholic University (permission number A9830). The nude mice (Nu/Nu strain, male, n=5) in 3 weeks old were inoculated intravenously (at least 3 cm from the root of tail) or subcutaneously (at both left and right abdomens) according to the experimental designs. Basically, 1x10⁶ adherent or suspended melanoma cells (in 200 uL of serum-free DMEM) was injected, and the mice were supplemented with free access to food and water in a specific pathogen free environment for 2 months. After animal scarification, the tumor tissues were harvested and fixed in 10% (v/v) buffered formalin. The tissues were then dehydrated in graded ethanol, and embedded in paraffin. The tissues in 3 μm thick sections were generated and stained with hematoxylin and eosin (HE stain) or immunohistochemical staining. The tissues were blocked and then incubated with desired antibodies. The monoclonal antibody (clone 3D8) against ANPEP protein was from Santa Cruz Biotechnology, Inc., Dallas, TX, USA. The primary antibodies against CD31 were from Bioss antibodies Inc. (Woburn, Massachusetts, USA). The primary antibodies against integrin β4 were from eBioscience Inc., San Diego, California, USA. The monoclonal antibody (clone B-A38) against SDC1 was from Novus Biologicals LLC., Littleton, CO, USA. Tissue-bound primary antibody was then visualized with a horseradish peroxidase-conjugated secondary antibody-3,3’-diaminobenzidine system (Rabbit PolyDetector HRP/DAB system; Spring Bioscience, Pleasanton, California, USA). Hematoxylin was used as a counterstain. These slides were scanned by digital microscope PreciPoint M8 (Precipoint, Inc., Freising, Germany). The images were captured by software viewpoint (Precipoint, Inc., Freising, Germany).

Quantitative proteomic analysis of protein contents by dimethyl labeling approach

The protein contents in adherent, suspended, and re-attached melanoma cells were quantified using dimethyl peptide labeling approach. The process of protein digestion, dimethyl peptide labeling, fractionation, LC-MS/MS analysis, and data analysis were done according to the published paper [39]. The cell lysis and protein extraction were separated into cytoplasmic and nuclear fractions using NE-PER™ Nuclear and Cytoplasmic Extraction kit (Thermo Fisher Scientific Inc., Pittsburgh, Pennsylvania, USA). Protein lysates were lyophilized and resuspended in 100 mM triethylammonium bicarbonate and 8M urea. The proteins were then reduced with 10 mM dithiothreitol at room temperature for 30 min and S-alkylated with 50 mM iodoacetamide at room temperature in the dark for 30 min. The resulting proteins were sequentially digested with endopeptidase Lys-C (1:100 w/w) (WAKO, Tokyo, Japan) at room temperature for 2 h, and further digested with sequencing-grade modified trypsin (1:100 w/w) (Promega corporation, Madison, Wisconsin, USA) at room temperature overnight.

Digested peptides from suspended or re-attached melanoma cells were labeled with isotopic formaldehyde (¹³CD₂O, heavy-labeled); while those from adherent melanoma cells were labeled with formaldehyde (CH₂O, light-labeled). After dimethyl labeling, equal amounts of the heavy- and light-labeled peptides were mixed and desalted. To increase peptide identification, the dimethyl-labeled peptides were further separated into 5 fractions using a strong cation exchange StageTip (Empore Cation Exchange-SR membrane; 3M Company, St. Paul, Minnesota, USA) by elution with different salt concentrations (30, 60, 90, 200, and 500 mM ammonium acetate in 15% acetonitrile and 0.1% trifluoroacetic acid). The eluted peptides were desalted, vacuum-dried, and resuspended in 2% acetonitrile and 0.5% acetic acid for mass analysis.

The fractionated peptides were analyzed using a nanoflow LC-MS/MS on a Dionex Ultimate 3000 RSLC nanosystem online-coupled to an LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific Inc., Pittsburgh, Pennsylvania, USA). The peptides were loaded onto a 100 μm × 15 cm tip column with 3-μm ReproSil-Pur 120 C18-AQ reverse-phase beads (Dr. Maisch HPLC GmbH, Ammerbuch-Entringen, Germany). The peptides were eluted at a flow rate of 500 nL/min by gradient elution with two mobile phases: 0.5% acetic acid in water (solution A) and a mixture of 0.5% acetic acid and 80% acetonitrile (solution B). The gradient was 5–10% B (5 min), 10–40% B (60 min), 40–100% B (5 min), and 100% B (10 min). The LTQ Orbitrap XL mass spectrometer was operated in a data-dependent mode with a spray voltage of 2400 V. The full MS scan spectra (m/z 300–1600) were acquired with a solution of 60,000 at m/z 400 in the Orbitrap analyzer with auto gain control value of 5×10⁵. The top 10 most intense precursor ions with charge state ≥ +2 were sequentially isolated with an isolation width of 2 Da and auto gain control value of 1×10⁴. The selected precursor ions were then fragmented using collision-induced dissociation in the linear ion trap with 35% normalized collision energy. The activation q was set to 0.25, and the activation time was set to 30 ms. Dynamic exclusion was set at 90 s with an exclusion list size of 500. A lock mass of m/z 445.120024 was applied for all LC-MS/MS analysis.

Raw files from LC-MS/MS were analyzed using MaxQuant software (version 1.6.1.0). The peak lists were searched with SWISS-PROT sequence database. Cysteine carbamidomethylation was set as a fixed modification. Variable modifications were the N-acetylation of proteins and oxidation of methionine. The search setting included minimum peptide length of 7 amino acids, enzyme specificity with a maximum of 2 missed cleavages, peptide tolerance of 4.5 ppm, MS/MS tolerance of ±0.5 Da, and false discovery rates (FDRs) at 1%. The results from adherent melanoma cells were as control, and those from suspended or re-attached melanoma cells were compared. The potential upregulated or downregulated protein targets were initially filtered with Perseus software (version 1.6.2.3) [40]. The significant cut-off thresholds were based on the Benjamini–Hochberg FDR calculation with threshold value p < 0.05 [41]. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD011883.

The protein identities with significantly upregulated or downregulated in melanoma cells under suspension (or reattachment upon suspension) were functionally annotated using DAVID v6.7 (The Database for Annotation, Visualization and Integrated Discovery [16, 17]). The functional annotation clustering of these protein identities was done with high classification stringency and the parameters of kappa similarity term overlap 3, similarity threshold 0.85, classification with initial and final group membership 3, multiple linkage thresholds 0.50, enrichment thresholds 1.0 and p-value < 0.05.

Polymerase chain reactions and statistical analysis

The levels of mRNA in cultured cells were analyzed by qPCR. The cDNAs were synthesized by MMLV HP reverse transcriptase (Epicentre Inc., Madison, Wisconsin, USA) using freshly prepared RNA as PCR template. Quantitative real-time PCR was performed using VeriQuest Fast SYBR green qPCR reagent (Affymetrix Inc., Santa Clara, California, USA) in a StepOne Plus real-time PCR system (Thermo Fisher Scientific Inc., Pittsburgh, Pennsylvania, USA). The 2^-ΔΔCT method was used to determine the relative gene expression using GAPDH as control. The p-value of < 0.05 or < 0.01 was statistically significant and was indicated in each figure. The forward and reverse primers used were: ANPEP, gccgtgtgca caatcatcgcact and caccagggagcccttgaggtg; SDC1, gctctggggatgactctgac and gtattctcccccgaggtttc; SDC2, ccagccgaagaggatacaaa and gcgttctccaaggtcatagc; Integrin αV, agatctggaccaggatggtt and atctgtggctcctttcattg; Integrin α6, caagatggctacccagatat and ctgaatctgagagggaacca; Integrin β1, gttacacggctgctggtgtt and ctactgctgacttagggatc; Integrin β2, tgcgtcctctctcaggagtg and ggtccatgatgtcgtcagcc; Integrin β3, accactgatgccaagactca and gcatcaacaatgagctggag; Integrin β4, aacgatgaacggtgccacct and ctccacgatgttggacgagt; GAPDH, gagtcaacggatttggtcgt and gatctcgctcctggaagatg.

Western blot and antibodies

To harvest cell lysate for western blot analysis, cells were washed and disrupted by lysis buffer (10 mM Tris-HCl, 5 mM EDTA, pH 8.0, 1% TritonX-100, and protease inhibitors) and kept on ice for 30 min. The lysate was then centrifuged at maximum speed using a desktop centrifuge at 4^oC for 10 min. Protein concentrations were quantified by protein assay kit (Bio-Rad Laboratories Inc., Hercules, California, USA). For the characterization of SDC1 protein, a pretreatment to remove glycosaminoglycan chain by digestion enzymes (all purchased from Sigma-Aldrich China, Inc., Shanghai, PR China) was done. 50 μg cell lysate was pretreated with 0.83 mIU heparinase I, 0.83 mIU heparinase II, 0.83 mIU heparinase III, and 0.83 mIU chondroitinase in 200 uL reaction buffer (20 mM Tris-HCl, pH 7.5, 4 mM CaCl₂, and 0.1 % (w/v) BSA) at 37 ^oC for 16 hr. The proteins were precipitated by trichloroacetic acid and solubilized by solubilizing solution (10 mM Tris-HCl, pH8.0, 5 mM EDTA, and 2 M urea) then subjected to SDS-PAGE.

Western blot was performed according to standard protocol. Briefly, the protein mixture was subjected to SDS-PAGE and transferred onto a PVDF membrane followed by blocking with 5% (w/v) skim-milk. The membrane was then incubated in primary antibodies (1:1000 in 5% skim-milk in TBST) for 2 hr at room temperature, and HRP-conjugated secondary antibody (1:20000) for 1 hr at room temperature followed by enhanced chemiluminescent (Millipore Co., Massachusetts, USA) detection. The primary antibodies against ANPEP, integrin β4, SDC1 were the same as used in IHC. The primary antibodies against integrin α6 and integrin β2 were from Bioss antibodies Inc. (Woburn, Massachusetts, USA). The primary antibodies against PKCα, phospho-PKCα, PKCδ, and phospho-PKCδ were purchased from Cell signaling Inc., Danvers, Massachusetts, USA. The primary antibody against β-actin was purchased from GeneTex Inc., Hsinchu, Taiwan.

Flow cytometry analysis

ANPEP expression analyzed by flow cytometry was performed accordingly [1]. The 1 x 10⁵ cells were incubated with 1 μg anti-ANPEP primary antibody (3D8; Santa Cruz Biotechnology, Inc., Dallas, Texas, USA) overnight at 4^oC, and 0.5 μg FITC-labeled secondary antibody at room temperature for 1 hr. The labeled cells were then analyzed by Partec flow cytometer ML (Partec North America, Inc., Swedesboro, New Jersey, USA).

Transwell migration assay

For transwell migration assay, culture insert (8 um pore size; BD bioscience, Inc., San Jose, California, USA) were precoated with matrigel (BD bioscience, Inc., San Jose, California, USA). 1x10⁵ cells (in DMEM medium with 1% (v/v) fetal bovine serum) were applied to the insert. The bottom wells contained 600 uL of culture medium with 10% (v/v) fetal bovine serum and were left un-agitated in the cell incubator. After 24 hrs, the cells attached to the transwell membrane were fixed with 0.5% (w/v) crystal violet and washed extensively. The stained cells were visualized and counted under microscopy.