Our study was approved by the Ethics Committee of Huzhou Central Hospital (Approval number 20141103-01, Date of approval: November 27, 2014). Written consent was obtained from patients enrolled in this study. A total of 176 patients (males: 108, females: 68) with histologically confirmed PDAC at Shanghai Changhai Hospital were recruited for this study. The mean patient age was 60.6 years (range 32–75). Patient diagnoses were independently reviewed by two pathologists and classified by WHO criteria. Follow-up data were completed on May 1, 2019.

The human pancreatic cancer cell lines PANC-1, BxPC3, SW1990 and ASPC-1 were purchased from the Cell Bank of the Chinese Academy of Sciences. BxPC3 and ASPC-1 were cultured in RPMI1640 medium. SW1990 was cultured in Leibovitz’s L-15 Medium. PANC-1was maintained in Dulbecco’s Modified Eagle Medium (DMEM). All the medium were supplemented with 10% heat-inactivated fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, United States), penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C in a humidified atmosphere of 5% CO₂. All the cells were free of mycoplasma contamination.

Lentivirus vectors for ELK1 and LGMN knockdown or overexpression were constructed by Hanyin Biotech (Hanyin Biotech, Shanghai, China). The lentivirus was packaged using psPAX2 and pMD2G (Hanyin Biotech, Shanghai, China). To obtain stable cells with ELK1 or LGMN knockdown or overexpression, lentivirus supernatant was added to cells, followed by selection with 1 μg/ml puromycin for 2 weeks. siRNA sequences for transcription factors are listed in Supplementary Table S1.

Anti-human LGMN antibody (AF2199, MAB2199; R&D), anti-human ELK1 antibody (ab27708; Abcam), anti-actin antibody (ab8227; Abcam), anti-goat IgG (ab6741; Abcam) and anti-rabbit IgG (#7074; Cell Signaling Technology) were used in this study.

Lysates (50 μg per lane) collected from pancreatic cancer cells or exosomes were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. After incubation with 5% nonfat milk for 30 min at 25°C, the membranes were further incubated with primary antibodies (1:500 dilution) overnight at 4°C. Horseradish peroxidase-conjugated secondary antibodies (1:3,000 dilution) were added and incubated for 60 min at 25°C. Immune-reactive proteins were captured by enhanced chemiluminescence (ECL).

Total RNA was extracted from pancreatic adenocarcinoma cancer cells using TRIzol reagent according to the manufacturer’s instructions (Invitrogen). cDNA was transcribed from 1 μg of total RNA. qRT-PCR was performed with SYBR Premix Ex Taq (TaKaRa, Dalian, China). PCR primers are listed in Supplementary Table S2 (Shenggong Biotech, Shanghai, China).

The cycling conditions were as follows: initial denaturation at 95°C for 5 min, followed by 36 cycles of denaturation at 95°C for 10 s and annealing at 60°C for 30 s. The relative mRNA expression levels were calculated using the comparative Ct (ΔΔCt) method. Actin was used as an internal control.

We examined cell proliferation using CCK8 assays (Dojindo, Japan).

Cells were seeded onto the upper chamber of Matrigel-coated transwell inserts with a pore size of 8 μm in serum-free medium. FBS (10%) was added to the lower chamber as a chemoattractant. After 24 h, the upper surface of the insert was gently scratched with a cotton swab. Cells invading the lower chamber were fixed with 4% paraformaldehyde and stained with crystal violet. The numbers of invading cells were counted under a microscope. Five random microscopic fields were analyzed for each insert.

Cell apoptosis was analyzed by annexin V staining. Briefly, cells were seeded for 24 h and then transfected with the appropriate lentivirus for another 48 h. The cells were then harvested, washed twice with PBS, stained with annexin V and DAPI in binding buffer, and detected by FCM after a 15-min incubation at room temperature in the dark. Early apoptotic cells (annexin V+/DAPI−) and late apoptotic cells (annexin V+/DAPI+) were quantified.

Mouse experiments of our study was approved by the Animal Ethics Committee of Huzhou Institute of Food and Drug Inspection (Approval number HSYJ2017001, Date of approval: December 14, 2017). Male BALB/c athymic nude mice, 4–6 weeks old and weighing 20–22 g, were purchased from the SLAC (Shanghai, China). The following ASPC-1 cells (5 × 10⁶) were subcutaneously injected into mice: (Bray et al., 2018) ASPC-1-NC (negative control), (Wood, 2014) ASPC-1 ELK1-KD (knockdown), and (Liu et al., 2015) ASPC-1 ELK1-KD/LGMN-OE (overexpressing). Tumor volumes were monitored every week. After 6 weeks, all the mice were euthanized, and the tumors were collected.

Histological and immunohistochemical analyses were performed as previously described (Liu et al., 2003). Rabbit anti-ELK1 antibody (ab32106, lot#: GR259320-21; Abcam) and rabbit anti-LGMN antibody (ab232870, lot#: GR97368-48; Abcam), diluted 1:50 in blocking buffer, were used as primary antibodies. Normal rabbit immunoglobulin G (Abcam) was included as a negative control. Three pathologists from Shanghai Changhai Hospital individually scored samples in a blinded manner before drawing conclusions.

ELK1 and LGMN were selected as markers respectively. The expression of these immunosuppressive proteins in tissues were evaluated via immunohistochemical analysis. Two pathologists to score the intensity and the percentage of positive cells in the tumor tissue independently. Staining intensity: 0 (No staining); 1 (Light yellow); 2 (Light brown); 3 (Brown). The percentage of positive cells: 0 (<5%); 1 (5–25%); 3 (>25–50%); 4 (≥50–75%); 4 (>75%), these two grades were multiplied and specimens were assigned to four groups according to the achieved score: 0–3, negative; 4–6, weak positive (+); 7–9, moderate positive (++); 10–12, strong positive (+++). Negative control staining was carried out with cold PBS in place of primary antibody. Known positive tissues were used as positive controls. Five fields were randomly taken from each sheet and photographed (magnification ×200–×400)

Survival was calculated starting from the date of surgery to the date of death or the last follow-up. Survival curves for LGMN were plotted using the Kaplan–Meier method and compared using the log-rank test. The median time and hazard ratio are shown with the 95% confidence interval (CI). Data are presented as the mean ± SD or as the number and percentage. The differences between groups and categorical variables were compared by the chi-square test. For normally distributed data, continuous variables were compared via an independent samples t test. Statistical analysis was performed with SPSS 15.0 (Chicago, IL, United States). Significance was defined as a p value <0.05.

Although we previously found that high expression of LGMN is involved in the progression of PDAC, the upstream regulation of LGMN remains unknown (18). In order to study the upstream regulation of LGMN in PDAC, we used JASPR and PROMP to search for potential transcription factors in LGMN promoter. SP1, ELK1, GATA3, NFAT1, E2F1 and c-JUN scored high in both software programs. Therefore, we knocked down each of these factors by siRNAs (Table 1) and determined their effect on LGMN expression in PANC-1 cells (Figures 1A–F). As shown in the figure, the mRNA level of LGMN decreased significantly after knockdown of ELK1 and NFAT1 (*p = 0.0057, **p = 0.0021, Figure 1G), LGMN protein also decreased after the inhibition of ELK1 (*p = 0.0003, Figure 1H and Supplementary Figure S5). Therefore, we constructed the LGMN promoter, and luciferase detection showed that knockdown of ELK1 significantly reduced the transcriptional activity of LGMN mRNA (*p = 0.029, Figure 1I).

We then analyzed the expression of ELK1 and LGMN in four different pancreatic cancer cell lines. RT-PCR results showed that ELK1 was relatively high in PANC-1 and ASPC-1 cells compared with BXPC3 and SW1990 cells (Figure 2A), LGMN was relatively high in PANC-1, ASPC-1 and SW1990 cells compared with BXPC3 cells (Figure 2B). A positive correlation between ELK1 and LGMN was observed in the mRNA level of pancreatic cancer cells (R ² = 0.412, p = 0.2430, Figure 3C).As shown in results, ELK1 protein expression was relatively high in PANC-1 and ASPC-1 cells compared with BXPC3 and SW1990 cells (Figure 2D), LGMN protein expression was higher than that in control (Figure 2D). We chose to overexpress ELK1 in BXPC3 and SW1990 cells and found the protein level of ELK1 increased (Supplementary Figure S2A). We also knocked down ELK1 in PANC-1 and ASPC-1 cells and found the protein level of ELK1 and LGMN decreased, while the expression of LGMN was saved in cells supplemented with over-expressed LGMN (Supplementary Figure S2B). Taken together, these results indicated that transcription factor ELK1 promoted the expression of LGMN in pancreatic cancer cells.

To elucidate the role of ELK1 in PDAC, we examined the proliferation, invasion and survival of pancreatic cancer cells. CCK8 experiment showed that knockdown of ELK1 gene significantly inhibited the proliferation of PANC-1 and ASPC-1 cells (*p < 0.001, **p < 0.001, Figures 3A,B). In addition, Matrigel Transwell invasion experiment indicated that inhibition of ELK1 weakened the invasive capacity of PANC-1 and ASPC-1 cells (*p < 0.001, **p = 0.0013, Figures 3C,D). Fluorescence-activated cell sorting (FACS) showed that the percentage of apoptotic cells increased after knocking down of ELK1 in PANC-1 and ASPC-1 cells (*p = 0.0015, **p < 0.001, Figures 4A,B), and yet LGMN rescued ELK1-KD cells and restored the proliferation, invasion and survival of PANC-1 and ASPC-1 cells (Figures 3, 4). Furthermore, overexpression of ELK1 promoted proliferation, invasion and survival of SW1990 and BXPC3 cells (Figure 5). The CCK8 assay showed that overexpression of ELK1 gene significantly promoted the proliferation of SW1990 and BXPC3 cells (*p < 0.001, **p < 0.001, Figures 5A,B). Matrigel Transwell invasion experiment indicated that overexpression of ELK1 enhanced the invasive capacity of SW1990 and BXPC3 cells (*p = 0.0101, **p = 0.0048, Figures 5C,D). Fluorescence-activated cell sorting (FACS) showed that the percentage of apoptotic cells decreased after SW1990 and BXPC3 cells overexpressed ELK1 (*p = 0.0005, **p < 0.001, Figures 5E,F). In conclusion, our study revealed that ELK1 promoted the proliferation, invasion and survival of pancreatic cancer cells by LGMN.

To further confirm the effects of ELK1 on PDAC development in vivo, we constructed Xenograft models by subcutaneous injection ASPC-1 NC cells, ASPC-1 ELK-KD cells and ASPC-1 ELK1-KD/LGMN-OE cells. Tumor volume analysis showed that tumors grew slower in ELK1-KD group than that in ELK1-KD/LGMN-OE group (*p < 0.001, **p < 0.001, Figures 6A–C). Consistent with the above results, tumors were lighter in ELK1-KD group than that in ELK1 -KD/LGMN-OE group (p < 0.001, p < 0.001, Figure 6D). Collectively, these in vivo experiment confirmed our in vitro data and identified ELK1 promoted pancreatic cancer cell proliferation, invasion and survival through LGMN.

To evaluate the potential clinical significance of ELK1 and LGMN in the prognosis and diagnosis of PDAC, we used immunostaining to detect their expression in 176 PDAC patients. According to pathological findings, PDAC patients were divided into three groups, stage T1, T2, T3. We found that the expressions of LGMN and ELK1 were reduced in ELK1-KD group, while LGMN was saved in ELK1-KD/LGMN-OE group (Figures 7A,B). Representative expression patterns in pancreatic adenocarcinoma samples are shown in Figure 8A.

Besides, according to the expression of ELK1 and LGMN in pancreatic adenocarcinoma samples, all patients were distributed into two groups: the low expression group and the high expression group (Tables 1, 2). The immunohistochemical staining results revealed that ELK1 and LGMN levels were significantly associated with clinical stage, degree of differentiation and lymph node infiltration (Tables 1, 2).

To assess the relationship of ELK1 expression with patient prognosis, the log-rank test and Kaplan–Meier analysis were used to evaluate the effect of ELK1 expression on patient survival. Patients with high levels of ELK1 and LGMN expression in tumor tissues experienced significantly shorter OS than patients with low ELK1 expression (n = 173, p < 0.0001, Figures 8B–D). The mean survival time of patients with low ELK1 expression was 27.93 months (n = 96, 95% CI: 25.61–30.24), whereas the survival time of patients with high ELK1 expression was 18.65 months (n = 77, 95% CI: 16.92–20.38). The log-rank test (univariate analysis) demonstrated that patients with low LGMN expression had a longer overall survival (OS) time than patients with high LGMN expression (χ² = 36.644, p < 0.0001). Moreover, the mean survival time of patients with low LGMN expression was 26.23 months (n = 119, 95% CI: 24.14–28.41), whereas the mean survival time of patients with high LGMN expression was 18.33 months (n = 54, 95% CI: 16.73–19.94). The log-rank test (univariate analysis) demonstrated that the patients with low LGMN expression had a longer OS time than patients with high LGMN expression (χ² = 28.434, p < 0.0001). Multivariate Cox regression analysis was also performed to explore whether ELK1 is an independent prognostic factor for survival. As presented in Figures 8B,C, ELK1 expression and LGMN expression were identified as independent prognostic factors (ELK1: HR = 2.45, 95% CI: 1.79–3.37, p < 0.0001; LGMN: HR = 2.40, 95% CI: 1.69–3.40, p < 0.0001).