The A549 human lung cancer cell line (cat. no. CCL-185) and the MDA231 breast cancer cell line (cat. no. HTB-26) were procured from American Type Culture Collection, and cultured in DMEM (Thermo Fisher Scientific, Inc.) with 10% FBS (MilliporeSigma) in accordance with the instructions provided by the suppliers. The cells were maintained in a 5% CO₂ humidified atmosphere at 37°C.

The proliferation rate of cultured cells was analyzed using MTT (MilliporeSigma) as previously described (9). The assay was performed in triplicate where applicable.

Total RNA (1 µg) from cells, extracted using ISOGEN (Nippon Gene Co., Ltd.), was converted into cDNA using a GeneAmp RNA-PCR kit (Thermo Fisher Scientific, Inc.). Reverse transcription was performed at 37°C for 2 h. qPCR was performed using TB Green Premix Ex Taq II (Takara Bio, Inc.), including TB Green as an intercalator that emits fluorescence when bound to double-stranded DNA. The thermocycling conditions were as follows: 95°C for 1 min and 50 cycles of 95°C for 5 sec and 60°C for 30 sec. Quantification based on the cycle threshold (Ct) values (2^−ΔΔCq method) (10) was performed using an Applied Biosystems 7900 HT Fast Real-time PCR System (Thermo Fisher Scientific, Inc.) as previously described (11). The primers used to amplify the target genes are listed in Table SI. GAPDH was used to normalize the expression levels in quantitative analyses. For the siRNAs listed in Table I, the analyses were conducted once, but in triplicate in case the migration rate was <50%. For a few siRNAs that impaired the viability of the cells, it was not possible to perform the analyses (see below; Table I).

Total proteins were extracted from cells using RIPA buffer (FUJIFILM Wako Pure Chemical Corporation) containing protease inhibitor mix Complete^™ (Roche Diagnostics). The protein concentrations were determined using a BCA Protein Assay Kit (Pierce; Thermo Fisher Scientific, Inc.). The proteins were boiled at 100°C, loaded (20 mg per lane), separated by SDS-PAGE (5–20%) and transferred to PVDF membranes. After 1 h of blocking with 3% (w/v) BSA (Merck KGaA) at room temperature, the membranes were incubated overnight with primary antibodies at 4°C. After being rinsed twice with TBS buffer (pH 8.0) containing 0.1% Tween-20, the membranes were incubated with HRP-labeled secondary antibodies against mouse IgG (100-fold dilution; cat. no. 7076; Cell Signaling Technology, Inc.) at room temperature for 1 h. An enhanced chemiluminescence solution (GE Healthcare) was used for color development. β-actin was used as the internal standard. The experiment was performed in triplicate. Antibodies against protein kinase DNA-activated catalytic subunit (PRKDC)/DNA-dependent protein kinase catalytic subunit (DNA-PK) (1,000-fold dilution; cat. no. 12311; Cell Signaling Technology, Inc.), proteasome subunit β type-4 (PSMB4) (200-fold dilution; cat. no. sc-390878; Santa Cruz Biotechnology, Inc.) and β-actin (200-fold dilution; cat. no. sc-47778; Santa Cruz Biotechnology, Inc.) were used. ImageJ software ver. 1.54 g (National Institutes of Health) was used for densitometry.

The migration assay was performed using the Boyden chamber method using polycarbonate membranes with a pore size of 8 µm (Chemotaxicell; Kurabo Bio-Medical Department; Kurabo Industries, Ltd.) as previously described (11). The membranes were coated with fibronectin (50 µg/ml) on the outer side at room temperature for 1 h and dried for 2 h at room temperature. The cells (2×10⁴ cells/well) were then seeded into the upper chambers containing 500 µl migrating medium (DMEM containing 0.5% FBS), and the upper chambers were placed into the lower chambers containing 700 µl DMEM with 10% FBS. After incubation at 37°C for 24 h, the media in the upper chambers were aspirated, and the non-migrated cells on the inner sides of the membranes were removed using a cotton swab. The cells that had migrated to the outer side of the membranes were fixed with 100% ethanol at room temperature for 5 min, stained with 0.1% crystal violet solution at room temperature for 15 min and observed under a light microscope. The number of migrated cells in five fields per chamber was averaged. The migration rate was verified in triplicate for each gene where applicable (Table I).

The selection of genes was based on our previous expression microarray analysis for CUP (GSE42392) (8), which was compared with microarray datasets for several cancer types obtained from the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/) (12). The datasets used were GSE781 (13), GSE1456 (14), GSE2109 (15), GSE2742 (16), GSE3149 (17), GSE3167 (18), GSE4127 (19), GSE4176 (20), GSE5787 (21), GSE6791 (22) and GSE8218 (23).

Predesigned siRNAs (MISSION^® siRNA) targeting 50 genes and a nonspecific target (MISSION^® siRNA Universal Negative Control; Table I) were purchased from MilliporeSigma. Cells were transfected with each siRNA (10 nM) using RNAiMAX (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions as previously described (24), incubated at 37°C for 48 h, and then trypsinized and seeded immediately for the migration assay.

shRNAs targeting PRKDC or PSMB4 were constructed using oligonucleotides encoding siRNA directed against the respective genes (5′-CCTGAAGTCTTTACAACATATCTC-3′ and 5′-CCGCAACATCTCTCGCATTATCTC-3′ for PRKDC and PSMB4 shRNA, respectively). The oligonucleotides were cloned into RNAi-Ready pSIREN RetroQZsGreen (Takara Bio, Inc.), which is a self-inactivating retroviral expression vector. This dual expression vector encodes green fluorescent protein (GFP) under the control of the cytomegalovirus promotor and shRNA under the control of the U6 promoter. The viral vectors constructed were designated as pSIREN-shPRKDC, pSIREN-shPSMB4 and a control vector (pSIREN-RetroQZsGreen vector without oligonucleotide inserted). Each of the pSIREN-RetroQZsGreen constructs (45 µg) was co-transfected with a pVSV-G vector (Takara Bio, Inc.) constituting the viral envelope (molar ratio of 3:1 for pSIREN-RetroQZsGreen constructs:pVSV-G vector) into GP2-293 packaging cells (2×10⁷ cells; Takara Bio, Inc.) using the FuGENE6 transfection reagent (Roche Diagnostics). After 48 h at 37°C, the culture medium was collected, and the viral particles were concentrated by centrifugation at 15,000 × g for 3 h at 4°C. The viral pellet was then resuspended in fresh DMEM. The viral vector titer was calculated using a Retrovirus Titer Set (Takara Bio, Inc.) according to the manufacturer's instructions. The minimal MOI was then determined to be 4 for all the constructs by counting the GFP-positive A549 cells infected by serial dilutions of the virus-containing media. At 3 days after the infection with the virus in A549 cells (1×10⁵ cells), the cells were trypsinized, propagated for 1 week and resuspended in DMEM (5×10⁶ cells/ml). Subsequently, 1 ml of the resuspended solution was processed for cell sorting using a FACSCalibur Flow Cytometer (BD Biosciences). This device consists of a flow cytometer and a fluorescence activated cell sorter (FACS) in which GFP-positive cells excited by the 488 nm laser line emit green fluorescence at 509 nm, which is optimally detected by the FL-1 detector, and the GFP-positive cells were captured by a catcher tube for sorting (25). The software used for analysis was CellQuest software Ver. 3.3 (BD Biosciences). The subsequent experiment (LN metastasis assay) was performed immediately after the sorted cells reached the number required for inoculation into mice (1×10⁷ cells for each construct) during the culture for ~1 week.

A total of 60 nude mice (BALB/c nu/nu; 5-week-old; female; mean weight, 17.8 g; CLEA Japan, Inc.) were used for the in vivo experiments. Experimemts were approved by the Institutional Animal Care Committee of the Kindai University Faculty of Medicine (approval no. KDMS-25-019; Osaka-Sayama, Japan). The mice were housed at the Animal Laboratory, Kindai University Faculty of Medicine (Osaka-Sayama, Japan), and maintained in a specific pathogen-free vivarium at 18–23°C with 50% humidity under a 12/12 h dark/light cycle. The animals had free access to drinking water and were fed a pelleted basal diet ad libitum.

The in vivo study consisted of two parts that estimated two effects on the potential of the cells to metastasize from footpad to popliteal LN: Experiment 1, the effect of downregulation of PSMB4 and PRKDC; and experiment 2, the effect of inhibitors of PSMB4 and PRKDC. For each of them, a single experiment was performed and 30 mice were used. For experiment 1, the cells (A549 cells bearing a control vector, pSIREN-shPRKDC, or pSIREN-shPSMB4) were resuspended in Hanks' Balanced Salt Solution (MilliporeSigma) and injected subcutaneously (1×10⁶ cells) into the hind footpads of mice (10 mice per group) on day 0 in accordance with the method described for a previous study (26). On the final day (day 20), mice were euthanized using CO₂ flowing from a compressed cylinder into a chamber (under ambient conditions) containing the mice so that 100% CO₂ gradually filled the chamber (30%/min) (27). Next, the primary tumors and popliteal LNs were visualized by fluorescence imaging and images were captured using a macro-imaging station consisting of a SBIG cooled CCD camera model ST-7XME (Santa Barbara Imaging) mounted onto a dark box. The integrated density of the fluorescence spots was quantified using ImageJ software ver. 1.54 g (National Institutes of Health). For experiment 2, the DNA-PK inhibitor NU7441 (10 mg/kg; FUJIFILM Wako Pure Chemical Corporation) (28) and the proteasome inhibitor bortezomib (1 mg/kg; Takeda Pharmaceutical Company, Ltd.) (29), and vehicle (DMSO) were administered intravenously alone (9–10 mice/group) 1 week after inoculation of A549 cells bearing a control vector, and this administration was repeated twice a week for 3 weeks (2nd to 4th weeks). The LNs were resected (on day 35), and their maximum diameters were measured. Fluorescence images were also captured for the primary tumors and popliteal LNs. The animal details and protocol were otherwise as aforementioned.

Data are presented as the mean ± SD, unless otherwise noted. An unpaired t-test was used for comparisons between two groups. Comparisons among multiple groups were conducted using one-way ANOVA, followed by the Student-Newman-Keuls test. These analyses were performed as a single experiment. Statistical analyses were performed using SigmaPlot v13.0 (Grafiti LLC). P<0.05 was considered to indicate a statistically significant difference.

In our previous microarray study, 44 genes that were upregulated in CUPs compared with non-CUPs as a whole were identified (8). Of the 44 genes, 40 were selected as candidates for the next screening using the migration assay (Table I). Three pseudogene-like genes (LOC392501, LOC442171 and LOC646417) with unknown functions and one gene without an available siRNA (LAPTM5) were excluded. As additional candidates for the screening, 10 genes that were screened out by the comparison of gene expression profiles between CUP and individual cancer types of known primary site were selected (8). These 10 genes were most highly expressed in CUP of all cancer types examined, and each of the genes showed ≥1.5-fold enriched expression in CUP compared with the second most enriched cancer type for each gene (Table SII). A total of 50 genes were selected (Table I) for the migration assay.

For all 50 aforementioned genes, specific siRNAs were obtained from the same commercial source (MISSION^® siRNA; Table I). For screening, A549 cells were transfected with siRNAs, and genes were evaluated for suppressed migration after siRNA knockdown. Images for cell migration assays are shown in Figs. 1 and S1. The migration assay showed that A549 cells transfected with some siRNAs exhibited poor migration, meaning the number of A549 cells migrating to the outer side of the porous membrane in the Boyden chamber was decreased compared with that of the cells transfected with control siRNA (Fig. 1A). For each of the 50 siRNAs transfected into A549 cells, the present study aimed to determine the cell viability, the fold-change in the expression of the targeted gene and the cell migration rate (Table I). The results revealed that transfection with siRNAs targeting SERF2, PRKDC, SH3GLB1, CAPNS1, PSMB4, RPS7 and RPL11 reduced the migration of A549 cells by >50% (Fig. 1A; Table I).

Our previous study identified six genes encoding ribosomal proteins (RPLP2, RPL18A, RPL36, RPS10, RPS7 and RPL11) as CUP-associated (−upregulated) genes (8). siRNAs against four of these ribosomal protein genes (RPLP2, RPL18A, RPL36 and RPS10) almost completely impaired the viability of A549 cells, and it was not possible to perform the expression assay and migration assay using siRNAs of these genes (Table I). The other two siRNAs (targeting RPS7 and RPL11) abolished the viability of MDA231 cells (Table I), indicating that reduced expression of any single ribosomal protein molecule could be critical for cell survival.

The present study subsequently focused on the siRNAs against the five genes (SERF2, PRKDC, SH3GLB1, CAPNS1 and PSMB4) based on their ability to reduce migration to less than half of that of the control without affecting the cell viability (>50% of the control; Table I). The present study further analyzed their influence on the migration of another cell line (MDA231). Only siRNAs specific for PRKDC and PSMB4 markedly reduced the migration of the cells (Fig. 1B; Table I). These siRNAs did not affect the viability of MDA231 cells, resulting in a cell viability rate of 75.2% for PRKDC-specific siRNA and >90% for PSMB4-specific siRNA. Furthermore, a marked reduction in the expression of the respective genes was observed. The siRNAs targeting PRKDC and PSMB4 reduced the expression levels of the corresponding genes (0.107- and 0.033-fold changes, respectively), which was similar to the trends observed for A549 cells (Table I). Western blotting confirmed that the expression of PRKDC and PSMB4 was reduced by the siRNA-induced mRNA knockdown (Fig. 2).

To determine whether downregulation of the two genes, PRKDC and PSMB4, might affect the potential of the cells to metastasize to the LNs in vivo, cell lines that stably expressed both GFP and shRNA targeting the respective genes were engineered. A control cell line transfected with a control vector that expressed GFP alone was also established. The stable cell lines for these constructs were isolated by FACS using GFP as a reporter (Fig. S2A-C) and they showed stable shRNA-based knockdown of PRKDC and PSMB4 during the passage culture for ≥3 months. Western blotting confirmed that the PRKDC and PSMB4 proteins expressed in A549 cells were reduced for the respective constructs compared with the control vector (Fig. S3A). The growth rate remained unaltered in the A549 cells transfected with shRNA targeting PSMB4 (>90% of the rate in the control cells), but the cells transfected with shRNA targeting PRKDC exhibited an ~25% reduction in the growth rate relative to that in the control cells at 72 h (Fig. S3B). To evaluate the metastasis-promoting potential of these two genes, cells (1×10⁶) containing the respective shRNA constructs were subcutaneously injected into the footpads of BALB/c nude mice. On day 20, the fluorescence intensities of GFP expressed in the footpad and popliteal LNs were measured (Fig. 3A). The fluorescence intensities were lower after injection of cells expressing PRKDC-shRNA than after injection of cells transfected with a control vector (~14.8%), whereas fluorescence intensities following injection of the cells expressing PSMB4-shRNA were comparable to those following injection of cells containing the control vector (Fig. 3B and C). A proportionate reduction in fluorescence intensities was also observed in the popliteal LNs in both the PRKDC-shRNA and PSMB4-shRNA groups, with popliteal LN/footpad fluorescence intensity ratios of 39.4 and 45.5%, respectively, compared with the control group (Fig. 3C). It was attempted to perform a similar assay using MDA231-shRNA constructs; however, this cell line failed to become engrafted into the footpads of the mice (data not shown).

The present study subsequently examined whether known inhibitors of PRKDC (DNA-PK) or PSMB4 (proteasome) could alter the migration of cells. The inhibitors were injected intravenously into the tails of mice that had been inoculated with A549 cells harboring a control vector expressing GFP. NU7441, a DNA-PK inhibitor, did not significantly change the fluorescence intensity in either the footpads or popliteal LNs (Fig. 4A and B). By contrast, bortezomib, a proteasome inhibitor, markedly retarded the migration of the cells to the popliteal LNs, although the cell growth rate in the footpads was comparable to that in the control animals, resulting in a lower popliteal/footpad fluorescence intensity ratio compared with that in the control group (P=0.093) and in the animals injected with NU7441 (P=0.017) (Fig. 4B).

A marked difference in the size of the metastatic LNs was observed between the groups that were and were not treated with bortezomib (Fig. 4C). Compared with the control group (vehicle), the bortezomib-treated group showed significantly smaller (undeveloped) tumors at the metastatic sites (LNs; P<10⁻⁴). The mean LN maximum diameters in the group treated with vehicle and in the group treated with NU7447 (DNA-PK inhibitor) were 6.3- and 4.5-fold greater than the diameter of intact LNs, whereas they were only 1.4-fold greater in the group treated with bortezomib.