Human PDAC tissue microarrays, as well as human breast and gastric cancer tissue microarrays (including 15 surgically resected breast/gastric tumor tissues), were obtained from Shanghai BioChip Co., Ltd. PDAC tissue microarray chips contained 157 tumor tissue samples (Table I) and 121 peritumoral tissues, which were free of signs of malignant transformation and distant from the tumor. The information on patient survival ranged from 1.2 to 7.0 years. TNM staging (7th edition) and American Joint Committee on Cancer staging data (7th edition) (22) were available for the majority of PDAC samples (the data were not available for 23 samples). Expression levels of tumor-associated markers, including Ki-67, p53, CD8 and programmed death ligand 1 (PDL1), were obtained from the database of the pathology department of the National Engineering Center of Shanghai BioChip Co., Ltd. All patients with PDAC were diagnosed by positive histology of invasive ductal carcinoma. Overall survival (OS) was calculated as the time interval between the date of surgery and the date of death or last follow-up visit. Not all deceased subjects were subjected to pathological analysis, and therefore the rate of PDAC-associated death is unknown.

The level 3 information (aggregated, normalized and/or segmented data) of The Cancer Genome Atlas (TCGA, http://portal.gdc.cancer.gov/) containing pancreatic adenocarcinoma datasets of 178 individual tumors were downloaded for expression analysis (survival data were available for 177 patients). The Root Mean Squared Error normalized mRNA count (‘count’), which represents PALB2 gene expression, was evaluated. For Gene Set Enrichment Analysis (GSEA), the latest official tool was downloaded from software.broadinstitute.org/gsea (version 3.0). The cBio Cancer Genomics Portal (cbioportal.org) was used to analyze PALB2 mRNA expression across different types of human cancer, based on TCGA public database.

Thin slices (4-µm-thick sections) from 10% formalin-fixed (24 h at room temperature), paraffin-embedded tissue specimens were used. The sections were deparaffinized in xylene and rehydrated in a descending alcohol series (100, 95, 90, 75 and 70%), using routine procedures. Antigen retrieval was achieved by digesting the sections with proteinase K (0.2 mg/ml at 37°C for 10 min) before IHC, followed by repeated washing steps and blocking with 3% H₂O₂ for 30 min at room temperature. Sections were incubated with a rabbit polyclonal anti-PALB2 antibody (1:2,000; cat. no. ab202970; Abcam) for 12 h at 4°C and then with an HRP-conjugated anti-rabbit secondary antibody (1:5,000; cat. no. P044801; Dako; Agilent Technologies, Inc.) for 30 min at room temperature. Tissue microarray slides were scanned using a Leica Aperio digital slide scanner (Leica Microsystems GmbH). Tumor cells displaying staining in the nucleus were categorized as positively stained. The percentage of PALB2⁺ tumor cells was calculated as the ratio of stained to unstained tumor cells. The percentages of positive cells were classified into five scores: 0, 0%; 1, 1–5%; 2, 6–30%; 3, 31–60%; and 4, 61–100% positively stained cells. The scoring of each tissue section was conducted independently by two pathologists. In cases of inconsistent results (~15% of samples), the pathologists re-evaluated the sample in question together to achieve a consensus. Tissue sections with scores 0–1 were considered as low expression, whereas scores of 2–4 were considered as high expression. According to the information from TCGA, gastric cancer tissues are known to express low PALB2 protein levels, while breast cancer tissues are known for high PALB2 expression; therefore, the reliability of staining with the commercial PALB2 antibody was tested in a gastric cancer tissue as a negative control, and a breast cancer tissue as a positive control.

SW1990, PANC1 and CFPAC1 cell lines were purchased from The Cell Bank of Type Culture Collection of the Chinese Academy of Sciences. SW1990, PANC1 and CFPAC1 cells were maintained at 37°C with 5% CO₂ and cultured in DMEM supplemented with 10% FBS (both Gibco; Thermo Fisher Scientific, Inc.). A siRNA targeting PALB2 (si-PALB2) and a scrambled negative control siRNA (si-NC) were purchased from Santa Cruz Biotechnology, Inc. (cat. nos. sc-93396 and sc-37007). Transfection was performed using standard protocols of Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.). Lipofectamine 3000 reagent and siRNAs were diluted separately with OPTI-MEM (Gibco; Thermo Fisher Scientific, Inc.) in a centrifuge tube, then Lipofectamine 3000 and siRNAs were mixed and incubated for 15 min at room temperature. Subsequently, DNA-lipid complex was added to the cells and incubated for 48 h at 37°C. The transfected cells were used for subsequent experiments.

Cell proliferation was assessed by performing the Cell Counting Kit-8 assay (Dojindo Molecular Technologies, Inc.). Briefly, cells were seeded into 96-well plates. After 0, 24 or 48 h, the number of viable cells was quantified after incubation with the CCK-solution for 2 h by measuring the optical density at a wavelength of 450 nm using a microplate reader. For the wound healing assay, cells were seeded into 6-well plates and transfected with si-PALB2 or si-NC. At 24 h post-transfection, a scratch wound was made in the confluent cell monolayer using a 200-µl pipette tip. Cells were cultured in serum-free medium. The wound was observed at 0 and 48 h with a light microscope (×100 magnification) after scratching. Cell migration area was measured using ImageJ (v1.52q; National Institutes of Health).

Total RNA from cells was isolated using TRIzol^® (Invitrogen; Thermo Fisher Scientific, Inc.). Total RNA (1 µg) was reverse transcribed into cDNA using a ReverTra Ace qPCR RT kit (Toyobo Life Science) according to the manufacturer's protocol. Subsequently, qPCR was performed using SYBR Green Gene Expression Assays (Applied Biosystems; Thermo Fisher Scientific, Inc.) and an AB7500 Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.). Thermocycling conditions included denaturation at 95°C for 15 sec, followed by annealing at 60°C for 10 sec and elongation at 72°C for 20 sec, for 40 cycles. The following primers were used for qPCR: Snail family transcriptional repressor 1 (Snail) forward, 5′-ACCACTATGCCGCGCTCTT-3′ and reverse, 5′-GGTCGTAGGGCTGCTGGAA-3′; snail family transcriptional repressor 2 (Slug) forward, 5′-ATGAGGAATCTGGCTGCTGT-3′ and reverse, 5′-CAGGAGAAAATGCCTTTGGA-3′; zinc finger E-box binding homeobox 1 (Zeb1) forward, 5′-GCACCTGAAGAGGACCAGAG-3′ and reverse, 5′-TGCATCTGGTGTTCCATTTT-3′; Vimentin forward, 5′-GACGCCATCAACACCGAGTT-3′ and reverse, 5′-CTTTGTCGTTGGTTAGCTGGT-3′; and GAPDH forward, 5′-ACCACAGTCCATGCCATCAC-3′ and reverse, 5′-TCCACCACCCTGTTGCTGTA-3′. mRNA expression levels were quantified using the 2^−∆∆Cq method (23) and normalized to the internal reference gene GAPDH.

Total protein was extracted from pancreatic cancer cell lines using RIPA lysis buffer (cat. no. 89900; Thermo Fisher Scientific, Inc.). Protein concentration was measured using BCA kit and western blotting was performed according to standard protocols. Briefly, 30 µg protein/lane was separated via 8% SDS-PAGE and transferred to a PVDF membrane. The membrane was blocked with 5% BSA (cat. no. 9998; Cell Signaling Technology, Inc.) for 2 h at room temperature. Subsequently, the membrane was incubated with the following primary antibodies: Anti-PALB2 (1:1,000; cat. no. ab202970; Abcam) and anti-β-actin (1:2,000; cat. no. sc-130656; Santa Cruz Biotechnology, Inc.) for 12 h at 4°C. A mouse anti-rabbit IgG HRP-conjugated secondary antibody (1:2,000; cat. no. sc-2357; Santa Cruz Biotechnology, Inc.) was used and the membrane was incubated for 30 min at room temperature. TBS-Tween (0.5% Tween) was used for membrane washing, and an ECL kit (Bio-Rad Laboratories, Inc.) was used for visualization. Protein expression levels were semi-quantified using ImageJ software v1.52q (National Institutes of Health) with β-actin as the loading control.

Statistical analyses were performed using GraphPad Prism (v8.0; GraphPad Software, Inc.) or SPSS (v24.0; IBM Corp.) softwares. Continuous variables were compared using the unpaired Student's t-test or Mann-Whitney U test. The χ² test was used to analyze the distribution of categorical variables between PALB2⁻ and PALB2⁺ groups. OS was plotted as a Kaplan-Meier survival curve with 95% CIs, and differences between subgroups were compared using log-rank tests. Cox regression analysis was performed to identify independent prognostic factors. P<0.05 was considered to indicate a statistically significant difference.

To establish the IHC method and test whether the selected antibody selected was capable of yielding congruent data in agreement with current knowledge, stomach and breast carcinoma tissues were measured for PALB2 expression as negative and positive controls, respectively. Weak staining signals were observed for the negative control stomach cancer tissue (Fig. S1; upper panels), while strong staining signals were observed for the positive control breast carcinoma tissues (Fig. S1; lower panels). The results supported the suitability of the antibody and the IHC protocol, and the staining pattern was consistent with the results obtained using the public TCGA database.

Subsequently, PALB2 expression in PDAC samples was analyzed using characterized tissue samples arranged in an array format. The staining patterns were evaluated by two independent pathologists and a consensus was reached on the resulting data. The number of PALB2⁻ samples was slightly higher compared with that of PALB2⁺ samples within the array of 157 tumor samples analyzed (Table I). In PALB2⁺ samples, compared with in the para-tumor normal tissues, staining was stronger in PDAC tissues (Fig. 1A). PALB2⁺ cell nuclei in PDAC samples and peritumoral tissues were identified and counted in relation to unlabeled cells. Overall, the positive rates for PALB2 were 24.8% in para-tumor healthy tissues and 40.8% in PDAC tissues. The distribution of the samples with respect to the IHC score of 0–4 for the PDAC and peritumor tissue samples are presented in Table SI. Representative images of PALB2 IHC labeling for each score are shown in Fig. 1B.

To assess whether PALB2 expression was associated with classical cancer markers, the expression pattern of PALB2 expression in tumor and peritumoral tissues was assessed. The results indicated that PALB2 expression was significantly lower in peritumoral tissues compared with in PDAC tissues (Fig. 2A). A direct comparison of Ki-67, p53, CD8 and PDL1 staining patterns in tissues with positive or negative PALB2 expression indicated no significant differences, suggesting that the expression levels of Ki-67, p53, CD8 and PDL1 were independent from PALB2 expression (Fig. 2B-E),

The association between patient clinical characteristics, including tumor PALB2 expression, and OS was analyzed in 157 patients. Multivariate and univariate Cox proportional hazard regression analyses identified that N stage, age and PALB2 expression were significantly associated with OS in patients with PDAC (Table II). Kaplan-Meier analysis revealed that patients with positive PALB2 expression had a poorer OS rate compared with patients with negative PALB2 expression (P=0.0384; Fig. 3A). Furthermore, patients were divided into two groups: Low PALB2 expression (positive rate ≤5%; n=119) and high PALB2 expression (positive rate >5%; n=38). Kaplan-Meier curves demonstrated a significant negative association between high PALB2 expression and OS (P=0.0195; Fig. 3B). Therefore, high PALB2 expression may indicate a poor prognosis in patients with surgically resectable PDAC. By contrast, co-expression of p53 and PALB2 was not significantly associated with survival (P=0.1260; Fig. 3C), whereas co-expression of PALB2 and Ki-67 indicated a poor prognosis in patients with PDAC (P=0.0415; Fig. 3D). In addition, co-expression of PALB2 and the cancer immune markers CD8 and PDL1 were not associated with survival (P=0.1497; Fig. 3E and P=0.5971; Fig. 3F).

To investigate the spectrum of cancers associated with PALB2 expression on a larger scale, TCGA public database was analyzed. The mRNA expression levels of PALB2 in each type of cancer compared with their respective para-tumor tissues were presented in the log scale format (Fig. 4A). Most types of cancer in the database displayed upregulated PALB2 mRNA expression, including breast, cervical and pancreatic cancer, which is consistent with the detection of PALB2 protein expression in the present study. The 177 patients with PDAC from TCGA database were divided into two groups: Relatively high PALB2 expression group (n=88) and relatively low PALB2 expression group (n=89). The Kaplan-Meier curves demonstrated a significant negative association between high PALB2 expression and long-term survival rate (P<0.001; Fig. 4B). Multivariate Cox regression analysis were conducted on the PALB2 mRNA expression data extracted from TCGA database (Table II). The multivariate analysis identified tumor location, N stage and PALB2 mRNA expression as tumor parameters with a significant association with OS (Table II). The summary of clinicopathological information of patients with PDAC from TCGA database are shown in Table SII.

To investigate the potential biological role of PALB2 in PDAC, PALB2 expression was knocked down using a siRNA in PDAC cells. The western blotting results indicated successful PALB2-knockdown by si-PALB2 compared with si-NC in SW1990 PDAC cells (Fig. 5A). Cell proliferation assays indicated similar proliferation rates in SW1990 cells regardless of PALB2-knockdown (Fig. 5B). Migration was assessed by performing wound healing experiments. Cell migration area was measured using ImageJ software. The results indicated that PALB2-knockdown inhibited PDAC cell migration depending on cell line. Migration was significantly decreased by PALB2-knockdown in SW1990 cells and CFPAC1 cells, but it was independent of PALB2 expression in PANC1 cells. (Figs. 5C and S2). Additionally, PALB2-knockdown significantly decreased the cell proliferation rate of CFPAC1 cells, but had no effect on PANC1 cells (Fig. S2).

To identify biological signaling pathways associated with PALB2 expression, a GSEA was conducted on TCGA gene expression datasets from patients with different levels of PALB2 mRNA expression. Among the 189 available datasets, the top and bottom 20 datasets were selected. The most significantly enriched signaling pathways were selected based on the normalized enrichment scores. The results indicated that samples with relatively high PALB2 mRNA expression were enriched for genes associated with the ‘epithelial mesenchymal transition’ (EMT) signaling pathway, along with genes involved in the TNF-α, TGF-β, p53, NOTCH and mTORC1 signaling pathways (Fig. 5D). To further validate the GSEA results, the expression levels of characteristic EMT-associated genes were quantified via RT-qPCR in PALB2-knockdown PDAC cells in vitro. SW1990 cells displayed PALB2-dependent migration effects. PALB2-knockdown significantly decreased Slug, Zeb1 and vimentin gene expression, but induced a small but non-significant increase on Snail expression. (Fig. 5E).