Localization of KRAS downstream target ARL4C to invasive pseudopods accelerates pancreatic cancer cell invasion

  1. Akikazu Harada
  2. Shinji Matsumoto
  3. Yoshiaki Yasumizu
  4. Kensaku Shojima
  5. Toshiyuki Akama
  6. Hidetoshi Eguchi
  7. Akira Kikuchi  Is a corresponding author
  1. Department of Molecular Biology and Biochemistry, Graduate School of Medicine, Osaka University, Japan
  2. Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives (OTRI), Osaka University, 2-2 Yamadaoka, Japan
  3. Laboratory of Experimental Immunology, WPI Frontier Immunology Research Center, Osaka University, Japan
  4. Gene Expression Laboratory (GEL-B), Salk Institute for Biological Studies, United States
  5. Gastroenterological Surgery, Graduate School of Medicine, Osaka University, Japan

Abstract

Pancreatic cancer has a high mortality rate due to metastasis. Whereas KRAS is mutated in most pancreatic cancer patients, controlling KRAS or its downstream effectors has not been succeeded clinically. ARL4C is a small G protein whose expression is induced by the Wnt and EGF–RAS pathways. In the present study, we found that ARL4C is frequently overexpressed in pancreatic cancer patients and showed that its localization to invasive pseudopods is required for cancer cell invasion. IQGAP1 was identified as a novel interacting protein for ARL4C. ARL4C recruited IQGAP1 and its downstream effector, MMP14, to invasive pseudopods. Specific localization of ARL4C, IQGAP1, and MMP14 was the active site of invasion, which induced degradation of the extracellular matrix. Moreover, subcutaneously injected antisense oligonucleotide against ARL4C into tumor-bearing mice suppressed metastasis of pancreatic cancer. These results suggest that ARL4C–IQGAP1–MMP14 signaling is activated at invasive pseudopods of pancreatic cancer cells.

eLife digest

Most cases of pancreatic cancer are detected in the later stages when they are difficult to treat and, as a result, survival is low. Over 90% of pancreatic cancers contain genetic changes that increase the activity of a protein called KRAS. This hyperactive KRAS drives cancer growth and progression. Attempts to treat pancreatic cancer using drugs that reduce the activity of KRAS have so far failed.

The KRAS protein can accelerate growth in healthy cells as well as in cancer and it does this by activating various other proteins. Drugs that target some of these other proteins could be more effective at treating pancreatic cancer than the drugs that target KRAS. One of these potential targets is called ARL4C. ARL4C is active during fetal development, but it is often not present in adult tissues. Harada et al. investigated whether the protein is important in pancreatic cancer, and what other roles it has in the body, to better understand if it is a good target for cancer treatment.

First, Harada et al. used cells grown in the lab to show that ARL4C contributes to the aggressive spread of human pancreatic cancers. Using mice, Harada et al. also showed that blocking the activity of ARL4C in pancreatic cancers helped to slow their progression.

Harada et al.’s results suggest that ARL4C could be a good target for new drugs treating pancreatic cancers. Given that this protein does not seem to have important roles in the cells of adults, targeting it is unlikely to have major side effects. Further investigation of ARL4C in more human-like animal models will help to confirm these results.

Introduction

Pancreatic cancer is extremely aggressive and exhibits poor prognosis, with a 5 year survival of only 5 % (Klein, 2013). Most pancreatic cancer-related deaths are due to metastatic disease, and more than 80 % of patients have either locally advanced or metastatic disease (Hidalgo, 2010; Klein, 2013). Genome sequencing analysis has revealed the mutational landscape of pancreatic cancer and KRAS mutations are considered an initiating event in pancreatic ductal cells (Collins et al., 2012; Waddell et al., 2015). Irrespective of our improved understanding of tumor biology, the treatment outcome has not changed for many years. Therefore, new innovative treatment options need to be tested based on better understanding of the characteristics of pancreatic cancer.

ARL4C is a member of the ADP-ribosylation factor (ARF)-like protein (ARL) family, which belongs to the ARF protein subgroup of the small GTP-binding protein superfamily (Engel et al., 2004; Matsumoto et al., 2017; Wei et al., 2009). Cytohesin2/ARF nucleotide-binding site opener (ARNO), a GDP/GTP exchange factor of ARF family proteins, has been identified as a direct effector protein (Hofmann et al., 2007). ARL4C is expressed through activation of Wnt–β-catenin and EGF–RAS signaling and plays important roles in both epithelial morphogenesis and tumorigenesis (Matsumoto et al., 2017; Matsumoto et al., 2014). Because aberrant activation of the Wnt–β-catenin and/or EGF–RAS pathways are frequently observed in various types of cancers, ARL4C is indeed expressed in a number of cancers (Fujii et al., 2015; Fujii et al., 2016). In colon and lung cancer cells, ARL4C promotes cell proliferation through ARF6, RAC, RHO, and YAP/TAZ. On the other hand, in liver cancer cells, ARL4C promotes cell proliferation through phosphatidylinositol three kinase δ (PI3Kδ) (Harada et al., 2019). Thus, ARL4C would activate different downstream pathways in a cancer cell context-dependent manner. These prompted us to study the involvement of ARL4C, as a KRAS downstream molecule, in aggressiveness of pancreatic cancer, and IQ-domain GTPase-activation protein 1 (IQGAP1) was identified as a binding protein of ARL4C.

IQGAPs are an evolutionally conserved family of proteins that bind to a diverse array of signaling and structural proteins (Hedman et al., 2015). Mammalian IQGAP1 is a well-characterized member of the IQGAP family and a fundamental regulator of cytoskeletal function (Briggs and Sacks, 2003). IQGAP1 is highly expressed in the tumor lesions and suggested to be involved in cancer cell metastasis (Johnson et al., 2009; Sakurai-Yageta et al., 2008). Here, we show that ARL4C bound to IQGAP1 and recruited IQGAP1 and membrane type1-matrix metalloproteinase (MT1-MMP, also called MMP14) (Sakurai-Yageta et al., 2008) to invasive pseudopods in a phosphatidylinositol (3,4,5)-trisphosphate (PIP3)-dependent manner and accelerated invasion. In addition, ARL4C antisense oligonucleotide (ASO) suppressed the lymph node metastases of pancreatic cancer cells orthotopically implanted into the pancreas of immunodeficient mice. These results suggest that the ARL4C–IQGAP1–MMP14 signaling axis promotes pancreatic cancer aggressiveness and that ARL4C is a novel molecular target for the treatment of pancreatic cancer.

Results

ARL4C is expressed in human pancreatic cancer

Whether ARL4C is expressed in pancreatic cancer patients was examined using immunohistochemistry. Fifty-seven pancreatic ductal adenocarcinoma (PDAC) patients, who did not receive preoperative chemotherapy, were used in this study (Supplementary file 1 table 1; Source data 1). ARL4C staining in the tumor lesions was calculated as a continuous variable, and the patients were classified into two groups (high and low), depending on ARL4C expression levels (Figure 1A). ARL4C expression was considered high when the total area of the tumor stained with anti-ARL4C antibody exceeded 5 %. High expression of ARL4C was observed in 47 cases (82%), but minimally detected in non-tumor regions of pancreatic ducts (Figure 1A). Anti-ARL4C antibody used in this study was validated in western blotting and immunohistochemical assay (IHC) (Figure 1—figure supplement 1A and B). A significant difference was observed between low and high ARL4C expression based on perineural invasion (Supplementary file 1 table 1). Because the perineural invasion is considered as one of the causes of the recurrence and metastasis after pancreatic resection (Liang et al., 2016), ARL4C expression may be correlated with the ability of cancer cell invasion. Consistently, ARL4C expression was correlated with decreased overall survival (Figure 1B). Analysis of TCGA and GTEx datasets revealed that ARL4C is highly expressed in tumor tissue than in non-diseased tissue (Figure 1C). In addition, when ARL4C high and low expression groups were separated based on the top 75 % of mRNA values of ARL4C in TCGA dataset, high expression of ARL4C indicated a poor prognosis (Figure 1D). Univariate and multivariate analysis revealed that higher ARL4C expression is an independent prognostic factor (Table 1). Taken together, these results indicate that high expression of ARL4C is correlated with the aggressiveness and poor prognosis of pancreatic cancer.

Figure 1 with 1 supplement see all
ARL4C is expressed in human pancreatic cancer.

(A) PDAC tissues (n = 57) were stained with anti-ARL4C antibody and hematoxylin. The percentages of ARL4C expression cases in the non-tumor regions and tumor lesions are shown. (B) The relationship between overall survival and ARL4C expression in patients with PDAC. (C) ARL4C mRNA levels in pancreatic adenocarcinoma and normal pancreatic tissues were analyzed using TCGA and GTEx datasets. The results shown are scatter plots with the mean ± s.e.m. p Values were calculated using a two-tailed Student’s t-test. (D) TCGA RNA sequencing and clinical outcome data for pancreatic cancer were analyzed. (E) Lysates of the indicated pancreatic cancer cells were probed with the indicated antibodies. (F) S2-CP8 and PANC-1 cells were treated with 10 μM PD184161 or 10 μM U0126, and ARL4C mRNA levels were measured by quantitative real-time PCR. Relative ARL4C mRNA levels were normalized to those of GAPDH and expressed as fold changes compared with the levels in control cells. Lysates were probed with the indicated antibodies. (G–I) S2-CP8 cells and PANC-1 cells were transfected with the indicated siRNAs, and ARL4C mRNA levels were measured by quantitative real-time PCR. Relative ARL4C mRNA levels were normalized to those of Β2Μ and expressed as fold changes compared with the levels in control cells. Lysates were probed with the indicated antibodies. EGR1 was used as an established transcription target gene of RAS signaling. (B,D) Data were analyzed using Kaplan–Meier survival curves, and a log-rank test was used for statistical analysis. (F–I) Data are shown as the mean ± s.d. of three biological replicates. p Values were calculated using a two-tailed Student’s t-test (G) or one-way ANOVA followed by Bonferroni post hoc test (F,H,I). Scale bars in (A) 50 μm. **, p < 0.01. See Figure 1—source data 1.

Table 1
Univariate analysis and multivariate analysis of overall survival by Cox’s Proportional Hazard model.
Univariate analysis
ParametersHazard ratio95% CIP value
ARL4C(low/high)3.511.0611.700.040
Sex(Male/Female)1.100.542.240.80
Age( < 65/≧65)1.050.472.350.91
Tumor Location(Head/Body or Tail)0.410.180.940.036
pStage(IA-IIA/IIB-III)2.511.175.410.019
pT(1-2/3)5.291.2322.700.025
pN(0/1)2.511.175.410.019
ly(0/1–3)2.741.176.460.021
v(0/1–3)2.051.004.200.049
ne(0/1–3)28,2585.25E-361.52E + 440.83
Multivariate analysis
ParametersHazard ratio95% CIP value
pT(1-2/3)3.720.7817.70.099
pN(0/1)1.800.794.100.16
ARL4C(low/high)3.561.0312.30.044
  1. Hazard ratios with 95 % confidence intervals (CIs) were calculated using a Cox regression model and P values were calculated using a log-rank test. CI, confidence interval; pT, primary tumor; pN, regional lymph node; ly, lymphatic invasion; v, venous invasion; ne, perineural invasion.

Pancreatic intraepithelial neoplasia (PanIN) lesions were observed in 26 specimens. ARL4C was expressed in 20 of 26 cases (77%) of PanIN, suggesting that ARL4C is expressed in early stages of PDAC (Figure 1—figure supplement 1C). The results are consistent with our recent observations that ARL4C is frequently expressed in atypical adenomatous hyperplasia, which is the possible precursor lesions and develops to lung adenocarcinoma (Kimura et al., 2020).

In cultured pancreatic cancer cell lines, ARL4C was highly expressed in PANC-1 and S2-CP8 cells and it was barely detected in BxPC-3 cells (Figure 1E). Consistent with the previous results with IEC6 rat intestinal epithelial cells and colorectal and lung cancer cells (Fujii et al., 2015; Matsumoto et al., 2014), the MEK inhibitors PD184161 and U0126 and siRNAs for β-catenin and KRAS decreased ARL4C expression in S2-CP8 and PANC-1 cells (Figure 1F–H). In addition, simultaneous knockdown of KRAS and β-catenin further suppressed ARL4C expression (Figure 1I). Taken together, these results suggest that ARL4C is expressed in pancreatic cancer cells through activated RAS–MAP kinase and Wnt–β-catenin pathways.

ARL4C expression is involved in the invasion of pancreatic cancer cells

ARL4C ASO-1316 has been shown to inhibit growth of xenograft tumors induced by colon and lung cancer cells (Harada et al., 2019; Kimura et al., 2020). However, ARL4C ASO-1316 had little effect on sphere formation of pancreatic cancer cell (Figure 2—figure supplement 1A and B and B) and did not induce cell death, which is assessed by propidium iodide (PI) staining (Figure 2—figure supplement 1C). Since the clinicopathological analysis of human pancreatic cancer specimens indicates that ARL4C expression may be correlated with invasive ability, migratory and invasive abilities of S2-CP8 and PANC-1 cells were studied in Boyden chamber assays. ARL4C ASO-1316 inhibited the migratory and invasive abilities with dominant effects on invasion (Figure 2A and B; Figure 2—figure supplement 1D). Inhibition of migratory and invasive abilities by ARL4C ASO, targeting the non-coding region of ARL4C mRNA, was not observed in the cells expressing ARL4C-GFP ectopically (Figure 2C and D; Figure 2—figure supplement 1E). Thus, ARL4C could be involved in migration and invasion of pancreatic cancer cells.

Figure 2 with 5 supplements see all
ARL4C expression is especially involved in the invasion of pancreatic cancer cells.

(A-D) S2-CP8 cells (A,B) or S2-CP8 cells expressing GFP or ARL4C-GFP (C,D) were transfected with control or ARL4C ASO-1316 and subjected to migration (A,C) and invasion (B,D) assays. Migratory and invasive abilities are expressed as the percentage of the same cells transfected with control ASO. (E) A schematic illustration of 3D invasion into collagen I gel using a 3D cell culture chip is shown. There is a chemical concentration gradient across the gel channel and cells can invade into the gel. The right panel shows a fluorescent confocal image (top) and a 3D reconstructed image (bottom). (F) S2-CP8 cells were transfected with control or ARL4C ASO-1316 and subjected to a 3D collagen I gel (2 mg/mL) invasion assay. The distances from the edge of the gel interface of all cells invading into the collagen gel were measured. (G) The same assay as in (F) was performed in the presence of different concentrations of collagen I. (H) S2-CP8 cells stably expressing ARL4C-tdTomato were observed with time-lapse imaging. Arrowheads indicate the tips of invasive pseudopods and yellow circles indicate the cytoplasm (20 μm away from the tip of pseudopods). The region in the yellow dashed squares is shown enlarged in the bottom image. Fluorescence intensities of the cytoplasm and invasive pseudopods were measured and plotted as a function of time. (I) S2-CP8 cells were subjected to a 3D collagen I gel invasion assay and stained with phalloidin and Hoechst 33342. The angle of pseudopods to the direction of cell invasion toward FBS was calculated (n = 105). The results were plotted to a polar histogram. (J) S2-CP8 cells expressing ARL4C-tdTomato were subjected to a 3D collagen I gel invasion assay with DQcollagen I, and stained with phalloidin and Hoechst 33342. The regions in the yellow dashed squares (a, pseudopod; b, cell body) are enlarged. (K) S2-CP8 cells transfected with control ASO or ARL4C ASO-1316 were subjected to a 3D collagen I gel invasion assay with DQcollagen I. The percentages of cells with DQcollagen I-positive pseudopods compared with the total number of cells were calculated. (A–D,F,G,K) Data are shown as the mean ± s.d. of three biological replicates. p Values were calculated using a two-tailed Student’s t-test. Scale bars in (F,G) 100 μm; (H) 20 µm; (J) 10 µm; (K) 5 µm. n.s. not significant. *, p < 0.05; **, p < 0.01. See Figure 2—source data 1.

ARL4C has been shown to be localized to membrane protrusions of non-tumor cells, such as IEC6 and Madin-Darby canine kidney (MDCK) cells (Matsumoto et al., 2014). ARL4C-tdTomato was localized to protrusive structures extending from S2-CP8 cells under Matrigel-coated 2D culture conditions (Figure 2—figure supplement 1F). At the structures, focal adhesion proteins such as paxillin, phosphorylated paxillin, FAK, and phosphorylated FAK were localized with ARL4C-tdTomato, also with F-actin (Figure 2—figure supplement 1F). Therefore, we defined the membrane protrusions as actin-based structures that contain the adhesion sites, of which length is longer than 10 μm and diameter is shorter than 10 μm. ARL4C is unique in that it is locked to the GTP-bound active form, and ARL4CQ72L-GFP, in which the amino acid at the same position in a constitutively active RAS mutant was mutated, showed a similar distribution to ARL4C-GFP. However, ARL4CT27N-GFP, which is an inactive form (Hofmann et al., 2007), was not present in the protrusions (Figure 2—figure supplement 1G). These results suggest that ARL4C is present in the tips of membrane protrusions where it is expressed as wild type.

Invadopodia are well-known membrane protrusions that localize at the ventral surfaces of cells and are active in extracellular matrix (ECM) degradation during cancer invasion (Murphy and Courtneidge, 2011). To analyze invadopodia, pancreatic cancer cells were grown on gelatin-coated glass coverslips (Figure 2—figure supplement 2A). Dark areas represent gelatinolytic activity of invadopodia and are equal to invadopodia structures. BxPC-3 cells exhibited invadopodia clearly, whereas S2-CP8 and PANC-1 cells did not (Figure 2—figure supplement 2A). Thus, some pancreatic cancer cells do not form typical invadopodia in gelatin surface but can invade into ECM through probably other structures. Meanwhile, components of invadopodia, such as cortactin and ARPC2, were localized at the tips of protrusions defined above, with ARL4C (Figure 2—figure supplement 2B), suggesting that the protrusions might contribute to invasive phenotypes of pancreatic cancer cells and ARL4C functions there. Therefore, we referred to the protrusive structures as ‘invasive pseudopods’, because they seem to be analogous to invadopodia (Jacquemet et al., 2013; Murphy and Courtneidge, 2011; Yu and Machesky, 2012).

ARL4C knockout did decrease numbers of invasive pseudopods but slightly, while knockdown of ARPC2, which regulates formation of pseudopods as one of the components of Arp2/3 complex, clearly reduced the number of pseudopods (Figure 2—figure supplement 2C-G). Next it was tested whether ARL4C is involved in the presence of invadopodia markers in the tips of pseudopods. ARL4C knockout did not affect ARPC2 staining statistically and reduced the staining of cortactin only modestly (Figure 2—figure supplement 2H and I). It is quite likely that ARL4C contributes to invasive properties through other than the formation of pseudopods. Therefore, ARL4C may be necessary for functions of invasive pseudopods rather than their formation. This prompted us to look further into the invading process.

For visualization of cancer cells invading through the ECM (Poincloux et al., 2009), a 3D microfluidic cell culture with type I collagen (Farahat et al., 2012; Shin et al., 2012) (3D gel invasion assay) was performed (Figure 2E). At 0 time the same numbers of cells treated with control and ARL4C ASO were placed in the starting position (Figure 2—figure supplement 3A), and after 72 hr directional invading ability was compared. Whereas control S2-CP8 cells invaded into type I collagen, ARL4C ASO decreased invasive ability (Figure 2F). ARL4C KO cells also decreased invasive ability (Figure 2—figure supplement 3B). When the collagen concentration was reduced, S2-CP8 cells invaded irrespective of ARL4C knockdown (Figure 2G), suggesting that their invasive ability is not required for cells to move into the ECM when collagen fiber-formed 3D net structures are sparse. Furthermore, in the 3D gel invasion assay ARL4C-tdTomato accumulated in the tips of invasive pseudopods (Figure 2H; Figure 2—video 1). Fluorescence intensities of ARL4C-tdTomato in the edges of the pseudopods and cytoplasm (20 μm away from the tip of pseudopod), respectively, were measured over time, and then the intensities were plotted as a function of time. The results indicate that ARL4C dynamically appeared and disappeared in the pseudopods, but it did not accumulate in the cytoplasm (Figure 2H). Using time-lapse imaging the angle of pseudopods to the direction of cell movement towards FBS was observed. Most of them were located in the angle of –45 to +45 degrees in the polar histogram, suggesting that invasive pseudopods play a role in purposeful directional invasion (Figure 2I; Figure 2—video 2).

To visualize the relationship between the localization of ARL4C and matrix degradation, the steady-state activity of cell-derived protease was measured as the dequenched signal emitted from collagen I fibers with dye-quenched (DQ) FITC (DQcollagen I) (Wolf et al., 2007) in the 3D gel invasion assay. Protease-induced fluorescence dequenching was detected in the collagen fibers crossing the tips of the pseudopods but not in the cell body (Figure 2J). Protease activity was decreased when ARL4C was depleted (Figure 2K), suggesting that ARL4C is involved in degradation of the ECM through its localization to the tips of invasive pseudopods and plays an important role in the invasion of pancreatic cancer cells.

IQGAP1 is an ARL4C-interacting protein

ARL4C recruits cytohesin2 to the plasma membrane through their direct interaction in HeLa cells (Hofmann et al., 2007). In S2-CP8 cells, ARL4C did not bind to cytohesin2 (Figure 3—figure supplement 1A), and knockdown of cytohesin2 had no effect on the migratory or invasive ability (Figure 3—figure supplement 1B). Furthermore, cytohesin2 was distributed throughout the cytosol in S2-CP8 cells, whereas it was localized to the cell periphery of HeLaS3 cells (Figure 3—figure supplement 1C). Whereas ARL4C ASO inhibited RAC1 activity in A549 cells (Fujii et al., 2015), the ASO did not affect RAC1 activity in S2-CP8 cells (Figure 3—figure supplement 1D). Although ARL4C induces the nuclear import of YAP/TAZ in HCT116 cells (Harada et al., 2019), ARL4C knockdown did not inhibit it in pancreatic cancer cells (Figure 3—figure supplement 1E). These results suggest that cytohesin2 neither functions downstream of ARL4C nor is involved in migration or invasion of pancreatic cancer cells and prompted us to explore an uncharacterized effector protein of ARL4C.

ARL4C-FLAG-HA–binding proteins were precipitated and the precipitates were analyzed by mass spectrometry (Figure 3A). Among the possible interacting proteins, IQGAP1 was further studied (Figure 3A; Supplementary file 1 table 2; Source data 2) because its expression is associated with the aggressiveness of various types of cancer (Johnson et al., 2009). Ectopically expressed and endogenous ARL4C were associated with endogenous IQGAP1 in S2-CP8 cells (Figure 3B and C). ARL4C-FLAG-HA and ARL4CQ72L-FLAG-HA formed a complex with GFP-IQGAP1 to the similar levels, but ARL4CT27N-FLAG-HA showed diminished binding to GFP-IQGAP1 in X293T cells (Figure 3D).

Figure 3 with 2 supplements see all
IQGAP1 is a novel ARL4C-interacting protein.

(A) The ARL4C-interacting proteins in X293T cells were analyzed by mass spectrometry. The results are listed in Supplementary file 1 table 2 and Source data 2. Arrowheads indicate the identified proteins, including IQGAP1 (red). (B,C) Lysates of S2-CP8 cells expressing ARL4C-GFP (B) or S2-CP8 WT cells (C) were immunoprecipitated with anti-GFP antibody (B) or anti-ARL4C antibody (C), and the immunoprecipitates were probed with the indicated antibodies. (D) Lysates of X293T cells expressing the indicated proteins were immunoprecipitated with anti-GFP antibody, and the immunoprecipitates were probed with the indicated antibodies. (E) S2-CP8 cells were stained with the indicated antibodies. Images of ARL4C and IQGAP1 were merged. (F) S2-CP8 cells expressing ARL4C-tdTomato were subjected to a 3D collagen I gel invasion assay and were stained with the indicated antibodies. Images of ARL4C and IQGAP1 were merged. (G,H) S2-CP8 cells expressing GFP or GFP-IQGAP1 were transfected with the indicated siRNAs and subjected to migration (G) and invasion (H) assays. Migratory and invasive abilities are expressed as the percentage of the same cells transfected with control siRNA. (I) S2-CP8 cells depleted of the indicated proteins were subjected to an invasion assay. Invasive activities are expressed as the percentage of control cells. (J) PDAC tissues were stained with anti-IQGAP1 antibody and hematoxylin. (K) The relationship between overall survival and IQGAP1 expression in PDAC patients was analyzed. (L) Scatter plot showing the correlation between the mRNA expression levels of ARL4C (X-axis) and IQGAP1 (Y-axis) in pancreatic cancer patients obtained from TCGA datasets using the R2: Genomics Analysis and Visualization Platform. r indicates the Pearson’s correlation coefficient. (G-I) Data are shown as the mean ± s.d. of three biological replicates. p Values were calculated using a two-tailed Student’s t-test (G,H) or one-way ANOVA followed by Bonferroni post hoc test (I). (K) The data were analyzed by Kaplan–Meier survival curves, and a log-rank test was used for statistical analysis. (E,F) The regions in the yellow dashed squares are shown enlarged in the left bottom images. The right bottom images are shown with a false color representation of fluorescence intensity. More than 50 cells were imaged and the representative image is shown. False color representations were color-coded on the spectrum. Scale bars in (E) 10 μm; (F) 20 μm; (J) 50 µm. KD, knockdown. RFI, relative fluorescence intensity. n.s., not significant. *, p < 0.05; **, p < 0.01. See Figure 3—source data 1.

Using another anti-ARL4C antibody for the immunocytochemical study (Figure 3—figure supplement 2A and B), ARL4C and IQGAP1 were shown to accumulate to invasive pseudopods at endogenous level in S2-CP8 and PANC-1 cells under Matrigel-coated 2D culture conditions (Figure 3E; Figure 3—figure supplement 2C). Colocalization of ARL4C and IQGAP1 at invasive pseudopods was observed in 94 % of cells with ARL4C accumulation to the pseudopods. In 3D culture conditions, IQGAP1 was found at the tips of invasive pseudopods, similar to ARL4C-tdTomato (Figure 3F). IQGAP1 siRNA inhibited the migratory and invasive abilities in S2-CP8 and PANC-1 cells, and the cells expressing GFP-IQGAP1 were resistant to IQGAP1 siRNA (Figure 3G and H; Figure 3—figure supplement 2D and E). Simultaneous knockdown of ARL4C and IQGAP1 decreased the invasive ability, but the inhibitory degree was similar to that induced by knockdown of either ARL4C or IQGAP1 (Figure 3I). Thus, IQGAP1 and ARL4C regulate invasion in identical signaling pathways.

IQGAP1 was highly expressed in 31 of 57 PDAC patients (54%) (Figure 3J). The anti-IQGAP1 antibody was validated by Western blotting and immunocytochemical and immunohistochemical analyses (Figure 3—figure supplement 2E-G). Although higher expression of IQGAP1 was not associated with clinical parameters (Supplementary file 1 table 3), IQGAP1 expression correlated with decreased overall survival (Figure 3K). Similar results were obtained from the analysis of TCGA and GTEx datasets (Figure 3—figure supplement 2H and I). TCGA dataset revealed that expression of ARL4C mRNA in pancreatic cancer patients is positively correlated with that of IQGAP1 mRNA (Figure 3L). Of 47 PDAC patients with high ARL4C expression, IQGAP1 was highly expressed in 27 patients (Supplementary file 1 table 4). Higher expression of ARL4C in the patients positive for IQGAP1 was associated with perineural invasion (Supplementary file 1 table 4). The overall survival of the patients who were double positive for ARL4C and IQGAP1 tended to be worse although it is not statistically significant (Figure 3—figure supplement 2J). Therefore, the relationship between ARL4C and IQGAP1 expression on patient survival using public datasets was analyzed. Overall survival was significantly decreased in the order of low ARL4C/low IQGAP1, high ARL4C/low IQGAP1, and high ARL4C/high IQGAP1, although the result of low ARL4C/high IQGAP1 could not conclude because of the small case numbers (n = 2) (Figure 3—figure supplement 2K). Thus, simultaneous expression of ARL4C and IQGAP1 would be correlated with aggressiveness of pancreatic cancer.

The polybasic region of ARL4C is required for its binding to IQGAP1

ARL4C is modified by myristate at the N terminus and has a polybasic region (PBR), comprising nine Lys or Arg residues, at the C terminus (Donaldson and Jackson, 2011). ARL4CG2A, whose N-terminal myristoylation site (Gly2) is mutated to Ala, and ARL4CΔPBR were expressed in S2-CP8 cells. In contrast to ARL4C-GFP, ARL4CG2A-GFP and ARL4CΔPBR-GFP were not accumulated at invasive pseudopods where cortactin was present, but distributed throughout the cytosol (Figure 4A and B; Figure 4—figure supplement 1A-C), and both mutants severely decreased the binding activity to GFP-IQGAP1 (Figure 4C). The C-terminal region of KRAS includes the PBR and the CAAX motif, which is farnesylated, and fusion of the KRAS C-terminal region triggers the localization of the proteins to the cell surface membrane (Hancock et al., 1990). The KRAS C-terminal region was fused to the ARL4C mutants, which were referred to as ARL4C-GFP-Cterm. Both ARL4CG2A-GFP-Cterm and ARL4CΔPBR-GFP-Cterm were localized to invasive pseudopods where cortactin was present (Figure 4B; Figure 4—figure supplement 1A). However, although ARL4CG2A-FLAG-HA-Cterm formed a complex with GFP-IQGAP1, ARL4CΔPBR-FLAG-HA-Cterm did not (Figure 4D), suggesting that membrane localization of ARL4C is not sufficient for its binding to IQGAP1. Taken together, the PBR is necessary for ARL4C to associate with IQGAP1, as well as for recruiting ARL4C to invasive pseudopods.

Figure 4 with 1 supplement see all
The PBR of ARL4C is required for ARL4C and IQGAP1 binding.

(A) A schematic representation of four ARL4C-GFP mutants is shown. (B) S2-CP8 cells were transfected with the indicated mutants of ARL4C-GFP. The percentages of cells with ARL4C-GFP mutant accumulated at invasive pseudopods compared with the total number of cells were calculated. (C,D) Lysates of X293T cells expressing the indicated proteins were immunoprecipitated with anti-GFP antibody and the immunoprecipitates were probed with anti-HA and anti-GFP antibodies. (E) S2-CP8 WT or ARL4C KO cells transfected with control or the indicated mutants of ARL4C-GFP were stained with anti-IQGAP1 antibody and phalloidin. The percentages of cells with IQGAP1 accumulated at invasive pseudopods compared with the total number of cells were calculated. (F) S2-CP8 WT or IQGAP1 KO cells were transfected with ARL4C-GFP. The percentages of cells with ARL4C-GFP accumulated at invasive pseudopods compared with the total number of cells were calculated. (G) S2-CP8 cells stably expressing GFP or the indicated mutants of ARL4C-GFP were transfected with control or ARL4C ASO and subjected to invasion assays. Invasive ability is expressed as the percentage of the same cells transfected with control ASO. (B,E–G) Data are shown as the mean ± s.d. of three biological replicates. p Values were calculated using a two-tailed Student’s t-test (F,G) or one-way ANOVA followed by Bonferroni post hoc test (B,E). (B,E,F) The regions in the yellow dashed squares are shown enlarged in the left bottom images. The right bottom images are shown in a false color representation of fluorescence intensity. False color representations were color-coded on the spectrum. Scale bars in (B,E,F) 10 μm. KO, knockout. RFI, relative fluorescence intensity. n.s., not significant. *, p < 0.05; **, p < 0.01. See Figure 4—source data 1.

The localization of IQGAP1 to invasive pseudopods was lost in ARL4C KO cells, but not vice versa (Figure 4E and F; Figure 2—figure supplement 2C and D; Figure 4—figure supplement 1D-G). The similar results were obtained in ARL4C knockdown cells (Figure 4—figure supplement 1H). In ARL4C KO cells, ARL4C-GFP and ARL4CG2A-GFP-Cterm rescued the recruitment of IQGAP1 to the plasma membrane, unlike ARL4CG2A-GFP, ARL4CΔPBR-GFP, and ARL4CΔPBR-GFP-Cterm (Figure 4E). Therefore, for IQGAP1 to be recruited to invasive pseudopods, the localization of ARL4C to the plasma membrane and the binding to IQGAP1 through the PBR might be necessary. In addition, inhibition of invasive ability by ARL4C ASO-1316 was cancelled by expression of ARL4CG2A-GFP-Cterm but not by that of ARL4CG2A-GFP, ARL4CΔPBR-GFP, or ARL4CΔPBR-GFP-Cterm (Figure 4G; Figure 4—figure supplement 1I). Thus, the binding of ARL4C and IQGAP1 in invasive pseudopods could be essential for the invasive ability.

ARL4C recruits IQGAP1 to invasive pseudopods in a PI(3,4,5)P3-dependent manner

PI(4,5)P2 (PIP2) and PI(3,4,5)P3 (PIP3) are required for ARL4C membrane targeting (Heo et al., 2006). The pleckstrin homology (PH) domain functions as a protein- and phospholipid-binding structural protein module (Maffucci and Falasca, 2001). The PH domains of PLCδ and GRP1 prefer to bind to PIP2 and PIP3, respectively (Lemmon, 2008). GFP-PLCδPH was detected throughout the cell surface membrane, whereas GFP-GRP1PH was accumulated in invasive pseudopods (Figure 5A).

Figure 5 with 1 supplement see all
ARL4C recruits IQGAP1 to invasive pseudopods in a PIP3-dependent manner.

(A) S2-CP8 cells were transfected with GFP-PLCδPH or GFP-GRP1PH. The percentages of cells with GFP-PLCδPH or GFP-GRP1PH accumulated at invasive pseudopods compared with the total number of cells were calculated. (B) S2-CP8 cells expressing FRB-CFP, mRFP-FKBP-5-ptase domain, and ARL4C-FLAG-HA were treated with or without rapamycin or LY294002 and stained with anti-HA and anti-IQGAP1 antibodies. The percentages of cells with IQGAP1 or ARL4C-FLAG-HA accumulated at invasive pseudopods compared with the total number of cells were calculated. (C) S2-CP8 cells expressing FRB-CFP and mRFP-FKBP-5-ptase domain were treated with or without rapamycin or LY294002 and subjected to an invasion assay. Invasive abilities are expressed as the percentage of control cells. (D) S2-CP8 cells expressing ARL4C-mCherry and GFP-GRP1PH were stained with anti-IQGAP1 antibody. Images of GFP-GRP1PH, ARL4C-mCherry, and IQGAP1 were merged. (E) A schematic representation of ARL4C-GFP mutants is shown. (F) S2-CP8 cells were transfected with the indicated mutants of ARL4C-GFP. The percentages of cells with ARL4C-GFP mutant accumulated at invasive pseudopods compared with the total number of cells were calculated. (G) ARL4C KO cells expressing control or the indicated mutants of ARL4C-GFP were stained with anti-IQGAP1 antibody. Quantification was performed as in (B). (H) S2-CP8 cells stably expressing GFP or the indicated mutants of ARL4C-GFP were transfected with control or ARL4C ASO and subjected to an invasion assay. Invasive abilities are expressed as the percentage of the same cells transfected with control ASO. (A,F) Enlarged images of the regions in the yellow dashed squares and a false color representation of fluorescence intensity are shown on the right. (a) and (c) show the pseudopods, and (b) and (d) show the cell body. (B,D,G) The regions in the yellow dashed squares are shown enlarged in the left bottom images. The right bottom images are shown in a false color representation of fluorescence intensity. (A–C,F–H) Data are shown as the mean ± s.d. of three biological replicates. p Values were calculated using a two-tailed Student’s t-test (A,H) or one-way ANOVA followed by Bonferroni post hoc test (B,C,F,G). (A,B,D,F,G) False color representations were color-coded on the spectrum. Scale bars in (A,B,D,F,G) 10 μm. KO, knockout. RFI, relative fluorescence intensity. n.s., not significant. **, p < 0.01. See Figure 5—source data 1.

The levels of PIP2 and PIP3 in the plasma membrane were decreased by a rapamycin-inducible PIP2-specific phosphatase (Inp54p) (Suh et al., 2006) and a PI3 kinase inhibitor LY294002 (Petrie et al., 2012), respectively. S2-CP8 cells were treated with rapamycin and LY294002 for 30 min to examine the localization of ARL4C and IQGAP1, and for 24 hr to test invasive ability. PIP3 depletion decreased the membrane targeting of ARL4C and IQGAP1 and reduced the invasive ability, but PIP2 depletion did not (Figure 5B and C). IQGAP1 and ARL4C-mCherry colocalized with GRP1PH in invasive pseudopods (Figure 5D), suggesting that both proteins accumulate in the cell peripheral regions containing PIP3 and promote invasion.

To reveal the importance of PIP3 for the localization area of ARL4C and IQGAP1, PLCδPH or GRP1PH was fused to the C terminus of ARL4CG2A-GFP (Figure 5E). While both ARL4CG2A-GFP-GRP1PH and ARL4CG2A-GFP-PLCδPH formed a complex with GFP-IQGAP1, the former construct was localized to invasive pseudopods, but the latter construct was present throughout the cell surface membrane (Figure 5F; Figure 5—figure supplement 1A). Consistently, in ARL4C KO cells extending invasive pseudopods, the localization of IQGAP1 to invasive pseudopods was rescued by ARL4CG2A-GFP-GRP1PH but not by ARL4CG2A-GFP-PLCδPH (Figure 5G). Furthermore, ARL4C ASO-1316 inhibited the invasive ability of S2-CP8 cells expressing ARL4CG2A-GFP-PLCδPH but not those expressing ARL4CG2A-GFP-GRP1PH (Figure 5H; Figure 5—figure supplement 1B). Taken together, these results suggest that PIP3-dependent membrane targeting of ARL4C recruits IQGAP1 to invasive pseudopods and promotes invasion.

ARL4C is involved in the focal delivery of MMP14 to invasive pseudopods through IQGAP1

IQGAP1 is involved in the trafficking of MMP14-containing vesicles to the leading structures of cancer cells (Sakurai-Yageta et al., 2008). TCGA dataset showed that expression of MMP14 mRNA in pancreatic cancer patients is positively correlated with that of both ARL4C and IQGAP1 mRNA (Figure 6—figure supplement 1A). In addition, MMP14 expression was associated with poor prognosis (Figure 6—figure supplement 1B).

Cell surface MMP14-GFP accumulated in invasive pseudopods containing IQGAP1 and ARL4C-FLAG-HA (Figure 6A). MMP14-GFP extremely disappeared from invasive pseudopods of ARL4C knockdown and KO cells and the phenotype was rescued by expression of ARL4C-FLAG-HA (Figure 6B; Figure 6—figure supplement 1C and D). IQGAP1 KO caused the loss of MMP14-GFP from the pseudopods, and FLAG-HA-IQGAP1 expression rescued this phenotype (Figure 6C). The failure of MMP14 membrane targeting in ARL4C KO cells was rescued by expression of ARL4CG2A-FLAG-HA-Cterm but not by that of ARL4CG2A-FLAG-HA, ARL4CΔPBR-FLAG-HA, or ARL4CΔPBR-FLAG-HA-Cterm (Figure 6B). In addition, PIP3 depletion, but not PIP2 depletion, suppressed the membrane localization of MMP14 (Figure 6D). Therefore, in co-operation with ARL4C and IQGAP1, MMP14 is likely to be trafficked to invasive pseudopods with PIP3 accumulation.

Figure 6 with 1 supplement see all
ARL4C is involved in focal delivery of MMP14 to invasive pseudopods through IQGAP1.

(A) S2-CP8 cells expressing MMP14-GFP and ARL4C-FLAG-HA were stained with anti-MMP14 without permeabilization, followed by permeabilization and staining with anti-HA and anti-IQGAP1 antibodies. (B) S2-CP8 WT or ARL4C KO cells expressing MMP14-GFP and the indicated mutants of ARL4C-FLAG-HA were stained with anti-MMP14 without permeabilization. The percentages of cells with MMP14 accumulated at invasive pseudopods compared with the total number of cells were calculated. (C) The same assay as in (B) was performed except with S2-CP8 WT or IQGAP1 KO cells expressing MMP14-GFP and FLAG-HA-IQGAP1. (D) S2-CP8 cells expressing MMP14-GFP, FRB-CFP, and mRFP-FKBP-5-ptase domain were treated with 100 nM rapamycin or 50 µM LY294002 for 30 min. Staining and quantification were performed as in (B). (E) S2-CP8 cells depleted of the indicated proteins were subjected to an invasion assay. Invasive activities are expressed as the percentage of control cells. (F–H) S2-CP8 cells (F,G) or S2-CP8 cells expressing MMP14-mCherry or MMP14ΔC-mCherry (H) depleted of the indicated proteins were subjected to a 3D collagen I gel invasion assay with DQcollagen I. The distances from the edge of the gel interface of all cells that invaded into the gel were measured (F,H). The percentages of cells with DQcollagen I-positive pseudopods compared with the total number of cells were calculated (G). (I) PDAC tissues were stained with the indicated antibodies and hematoxylin. The regions in the black dashed squares are shown enlarged in the solid squares. Nine patient samples were imaged and the representative images are shown. (A–D) The regions in the yellow dashed squares are shown enlarged in left bottom and a false color representation of fluorescence intensity is shown in right bottom. False color representations were color-coded on the spectrum. (B–H) Data are shown as the mean ± s.d. of three biological replicates. p Values were calculated using a two-tailed Student’s t-test (H) or one-way ANOVA followed by Bonferroni post hoc test (B–G). Scale bars in (A–D) 10 μm; (F,H) 100 µm; (G) 5 µm; (I) 100 µm. KO, knockout; KD, knockdown. RFI, relative fluorescence intensity. n.s., not significant. *, p < 0.05; **, p < 0.01. See Figure 6—source data 1.

Consistent with these results, the inhibited invasive ability after double knockdown of ARL4C and MMP14 or IQGAP1 and MMP14 was similar to that seen after single knockdown of ARL4C, IQGAP1, or MMP14 (Figure 6E; Figure 6—figure supplement 1E and F). Knockdown of ARL4C, IQGAP1, or MMP14 decreased invasive ability in 3D microfluidic cell culture (Figure 6F) and the protease activity was also reduced (Figure 6G). Previous work has shown that MMP14ΔC(Δ563–582) lacking the cytoplasmic region fails to be endocytosed (Jiang et al., 2001). Here, MMP14ΔC was retained in invasive pseudopods of ARL4C-KO cells (Figure 6—figure supplement 1G), and the ARL4C knockdown-mediated decreases in cell invasion and collagen degradation were rescued by MMP14ΔC (Figure 6H; Figure 6—figure supplement 1H and I). Thus, ARL4C-dependent recruitment of MMP14 to invasive pseudopods is required for cell invasion.

Pancreatic cancer tissues were stained with anti-ARL4C, anti-IQGAP1, and anti-MMP14 antibodies in the serial section. Notably, ARL4C and MMP14 were expressed more highly in invasive cancer cells rather than in PanIN lesions, although IQGAP1 was thoroughly expressed in tumor lesions including PanIN area (Figure 6I). Using triple immunofluorescence imaging assay, it was confirmed that three proteins are simultaneously expressed in PDAC cells invading the surrounding interstitial tissues (Figure 6—figure supplement 1J). Taken together, these results support the idea that the ARL4C–IQGAP1–MMP14 signaling axis participates in pancreatic cancer cell invasion.

ARL4C ASO inhibits pancreatic tumor metastasis in vivo

To show that ARL4C is indeed involved in cancer cell invasion in vivo, the effects of subcutaneous injection of ARL4C ASO-1316 on an orthotopic transplantation model were tested. S2-CP8 cells expressing luciferase were injected into the pancreas of nude mice, and control ASO or ARL4C ASO-1316 was subcutaneously injected from day 3 (Figure 7A). After 2 and 3 weeks, ARL4C ASO-1316 suppressed the luminescence signal compared with control ASO (Figure 7B), and ARL4C expression was decreased immunohistochemically (Figure 7C). Whereas ARL4C ASO-1316 did not reduce the size of the primary tumor in the pancreas, the ASO decreased the numbers of lymph node metastases and tended to improve the survival (Figure 7D and E; Figure 7—figure supplement 1A).

Figure 7 with 1 supplement see all
ARL4C ASO inhibits pancreatic tumor metastasis in vivo.

(A) S2-CP8/Luciferase cells were implanted into the pancreas of nude mice, and control ASO (n = 6) or ARL4C ASO-1316 (n = 7) was subcutaneously administered. (B) Bioluminescence images of the intraperitoneal tumors are presented (left) and quantification of the tumor burden is shown (right). The data are presented as the mean ± s.e.m. of the fold change in luminescent intensity relative to that of week 1 treated with control ASO. (C) Sections from the pancreatic tumors from control ASO- or ARL4C ASO-1316-treated tumor-bearing mice were stained with anti-ARL4C antibody and hematoxylin. (D) Representative images of the tumors in the pancreas (left bottom) and metastatic mesenteric lymph nodes (top and right bottom) are shown. (E) Primary tumor weight (left) and metastatic mesenteric lymph node number are presented (right). Data are shown as the mean ± s.e.m. (F,G) Four hr after subcutaneous injection of 6-FAM-ARL4C ASO-1316 into tumor-bearing mice, the fluorescence intensities of various organs were measured (F), and the sections prepared from the pancreas were stained with the indicated antibodies (G). Area indicated by yellow dashed square is enlarged on the right panel (F). (H) Total RNA was extracted from tumors of control ASO- or ARL4C ASO-1316-treated tumor-bearing mice. Tumor block was cut into three pieces from each tumor of 2 mice per group. ARL4C mRNA levels were measured by quantitative real-time PCR. Relative ARL4C mRNA levels were normalized to those of B2M and expressed as fold changes compared with the levels in one of the control samples. (I) Sections from the pancreatic tumors were stained with anti-IQGAP1 antibody and hematoxylin. The two panels on the right show enlarged images of the yellow dashed squares. Positive staining of IQGAP1 is color-coded as yellow (weakly positive) or red (strongly positive). The percentage of the strongly positive IQGAP1 area was calculated. Data are shown as the mean ± s.e.m. Twenty fields were analyzed from 3 mice per group. (J) Sections from the pancreatic tumor were stained with the indicated antibodies. The numbers of tumor cells in the lymphatic vessels (indicated with yellow arrowheads) were counted. Data are shown as the mean ± s.e.m. Thirty lymphatic vessels were analyzed from 3 mice per group. (K,L) RNA sequencing was performed for S2-CP8-derived primary tumors, and the results of principal component analysis (K) and hierarchical clustering (L) are shown. (M) Differentially expressed genes were subjected to Ingenuity Pathway Analysis (IPA). The top five disease or function annotations of the positive and negative Z-score groups are shown. Bars indicate the -log10 (p value). Inhibited pathways are represented by blue-colored bars while activated pathways are shown by red-colored bars. (B,E,H-J) p Values were calculated using a two-tailed Student’s t-test. Scale bars in (D) 5 mm; (C,G,I) 50 µm; (J) 20 µm. n.s., not significant. *, p < 0.05; **, p < 0.01. See Figure 7—source data 1 and Figure 7—source data 2.

When 6-FAM–labeled ARL4C ASO-1316 was subcutaneously injected into tumor-bearing mice, the fluorescence was extremely detected in the pancreas and slightly observed in the kidney which is due to renal excretion (Figure 7F). 6-FAM–labeled ARL4C ASO-1316 was highly accumulated in tumor lesions but not in the neighboring normal tissues (Figure 7G), indicating that ASO was incorporated into tumor lesions after systemic injection. In primary pancreatic tumors, ARL4C ASO-1316 reduced ARL4C expression at protein and mRNA levels (Figure 7H; Figure 7—figure supplement 1B) and decreased the localization of IQGAP1 to the cell surface area (Figure 7I; Figure 7—figure supplement 1C). Tumor cells were observed in lymphatic vessels of peritumoral areas of control ASO-treated mice but not in those of ARL4C ASO-treated mice (Figure 7J). In addition, tumor cells surrounding peritumoral lymphatic vessels were also decreased, which is consistent with our hypothesis that ARL4C is required for cell invasive activity.

To compare molecular characteristics between pancreatic tumors from mice injected with control ASO and ARL4C ASO-1316, RNA sequence analysis was performed for primary tumors (Figure 7—source data 2). Principal component analysis (PCA) indicated a clear difference in the gene expression profiles of tumors from control ASO- and ARL4C ASO-1316–treated mice (Figure 7K). Furthermore, hierarchical clustering revealed a drastic change in expression of genes due to ARL4C ASO-1316 injection (Figure 7L). Two hundred and three differentially expressed genes (DEGs) were detected, and by subjecting them to Ingenuity Pathway Analysis (IPA), the top five significantly enriched terms of the biological process of molecular function in the inhibition and activation of the pathways were obtained (Figure 7M). In particular, DEGs linked to the inhibition of the pathways in ARL4C ASO-1316–treated mice were predicted to be involved in terms such as cell migration and invasion (Figure 7M). Taken together, these results suggest that ARL4C ASO inhibits the invasion of tumor cells into lymphatic vessels in vivo, and the gene profiles of tumors treated with ARL4C ASO in vivo support the putative functions of ARL4C in pancreatic cancer invasion.

Discussion

Pancreatic cancer represents one of the leading causes of cancer death, despite advances in cancer therapy (Keleg et al., 2003). Major problem of pancreatic cancer is uncontrollable invasion and metastasis. In this study, we found that the ARL4C–IQGAP1–MMP14 signaling axis is involved in pancreatic cancer invasion. Because ARL4C expression is induced by Wnt and EGF signaling, it is reasonable that ARL4C would be expressed in a β-catenin- and RAS-dependent manner in pancreatic cancer cells. ARL4C is a unique small G protein because it is constitutively active, regardless of wild-type (Burd et al., 2004; Matsumoto et al., 2017). The long interswitch region of ARL4C may prevent the retractile conformation change in the GDP-bound state (Burd et al., 2004; Pasqualato et al., 2002). ARL4C could be a constitutively active form without active mutations, and its activity may be controlled by transcriptional regulation.

ARL4C binds to cytohesin2 (Hofmann et al., 2007), leading to activation of ARF6–RAC–RHO–YAP/TAZ signaling in colon and lung cancer cells (Fujii et al., 2015; Kimura et al., 2020). Because ARL4C did not bind to cytohesin2 but to IQGAP1 in pancreatic cancer cells, it is likely that ARL4C regulates different downstream signaling pathways in a cancer cell context-dependent manner. Invadopodia are the unique structures observed at the ventral sites of certain types of cancer cells, such as BxPC-3, breast cancer MDA-MB-231 cells, and head and neck squamous carcinoma SCC61 cells (Dalaka et al., 2020; Murphy and Courtneidge, 2011), but the typical invadopodia are not observed in S2-CP8 and PANC-1 cells. Invasive pseudopods that we defined in these pancreatic cancer cells highly expressing ARL4C consisted of similar molecules, including cortactin, ARPC2, IQGAP1, and MMP14, which are involved in invadopodia functions (Caswell and Zech, 2018; Jacquemet et al., 2013; Murphy and Courtneidge, 2011). ARL4C depletion severely suppresses the localization of IQGAP1 and MMP14 to pseudopods and inhibits invasive ability, but does affect the localization of cortactin as well as the structure itself only moderately. Therefore, major function of ARL4C in invasive pseudopods would be to recruit MMP14 by binding to IQGAP1 rather than pseudopod formation.

Both myristoylation and the PBR of ARL4C support plasma membrane targeting (Heo et al., 2006). In our results, both motifs were necessary for the localization of ARL4C to the plasma membrane, whereas the PBR, rather than myristoylation, was indispensable for the activity of the ARL4C–IQGAP1–MMP14 signaling axis. Phosphoinositides have been implicated in many aspects of cell physiology (Di Paolo and De Camilli, 2006). PIP3 is localized to the leading edge of migrating cells and invadopodia of cancer cells (Saykali and El-Sibai, 2014) and recruits cytosolic proteins containing lipid-binding domains, such as the PH domain, to the plasma membrane (Toker and Cantley, 1997). ARL4C in pancreatic cancer cells preferred PIP3 to PIP2. Because PI3 kinase is one of the direct effector proteins of RAS (Castellano and Downward, 2011; Rodriguez-Viciana et al., 1994), RAS-dependent PI3 kinase activation and ARL4C expression could co-operatively function to promote pancreatic cancer invasion.

In conclusion, this study clarified that invasion of pancreatic cancer cells is promoted by ARL4C, of which expression is induced by KRAS and Wnt signaling, and that association of ARL4C with IQGAP1 and MMP14 at the tips of invasive pseudopods are essential for the invasive ability. The novel functions of ARL4C were confirmed by the mouse model. The inhibition of ARL4C expression by ARL4C ASO could directly inhibit invasive ability of pancreatic cancer cells and may indirectly affect the genes involved in invasion perhaps through the interaction between tumors and surrounding tissues. Because histological damage to the non-tumor regions was not observed after the administration of ARL4C ASO-1316 (Harada et al., 2019), ARL4C might represent an appropriate target for pancreatic cancer therapy.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Mus musculus, male)BALB/cAJcl-nu/nuCLEATen-week-old
Cell line (Homo sapiens)Lenti-X 293TTakara Bio Inc
Cell line (Homo sapiens)HeLaS3K.Matsumoto (Nagoya University, Aichi, Japan)RRID:CVCL_0058
Cell line (Homo sapiens)S2-CP8Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku UniversityRRID:CVCL_F971
Cell line (Homo sapiens)PANC-1RIKEN Bioresource Center Cell BankRRID:CVCL_0480
Cell line (Homo sapiens)BxPC-3American Type Culture CollectionRRID:CVCL_0186
Cell line (Homo sapiens)HPAF-IIAmerican Type Culture CollectionRRID:CVCL_0313
Transfected construct (Homo sapiens)ARL4C-EGFPThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CG2A-EGFPThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CT27N-EGFPThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CQ72L-EGFPThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CΔPBR-EGFPThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CG2A-EGFP-CtermThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CΔPBR-EGFP-CtermThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CG2A-EGFP-GRP1PHThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4CG2A-EGFP-PLCδPHThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4C-mCherryThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4C-tdTomatoThis paperVarious mutants of ARL4C
Transfected construct (Homo sapiens)ARL4C-FLAG-HAThis paperVarious mutants of ARL4C
Biological sample (Homo sapiens)Resected specimens of 57 patients with PDACOsaka University
AntibodyARL4C (rabbit polyclonal)Atlas Antibodies#HPA028927(WB 1:1000, IHC 1:50)
AntibodyClathrin (mouse monoclonal)BD Biosciences#610,500(WB 1:1000)
AntibodyEGR1 (rabbit monoclonal)Cell Signaling Technology#4,153 S(WB 1:1000)
Antibodyβ-catenin (mouse monoclonal)BD Biosciences#610,154(WB 1:1000)
AntibodyRas (G12D) (rabbit monoclonal)Cell Signaling Technology#14,429 S(WB 1:1000)
AntibodyHsp90 (mouse monoclonal)BD Biosciences#610,419(WB 1:1000)
AntibodyHA (mouse monoclonal)BioLegend#901,502(WB 1:1000)
AntibodyHA (rat monoclonal)Roche#1867423001(ICC 1:100)
AntibodyGFP (rabbit polyclonal)Life Technologies/Thermo Fisher Scientific#A6455(WB 1:4000)
AntibodyGFP (mouse monoclonal)Santa Cruz Santa Cruz Biotechnology#sc-9996(WB 1:1000)
AntibodyFLAG (mouse monoclonal)WAKO#014–22,383(WB 1:1000)
AntibodyIQGAP1 (mouse monoclonal)Santa Cruz Santa Cruz Biotechnology#sc-376021(WB 1:1000, IHC 1:800, ICC 1:100)
AntibodyMMP14 (rabbit monoclonal)Abcam#ab51074(WB 1:1000, IHC 1:200, ICC 1:100)
AntibodyCytohesin2 (rabbit polyclonal)Proteintech Group, Inc#67185–1-Ig(ICC 1:100)
AntibodyRac1 (mouse monoclonal)BD Biosciences#610,651(WB 1:1000)
AntibodyCdc42 (rabbit polyclonal)Cell Signaling Technology#2,466 S(WB 1:1000)
AntibodyCK19 (rabbit monoclonal)Abcam#ab52625(IHC 1:100)
AntibodyMitochondria (mouse monoclonal)Merck Millipore#MAB1273(IHC 1:100)
AntibodyLYVE-1 (rabbit polyclonal)Abcam#ab14917(IHC 1:100)
AntibodyYAP/TAZ (rabbit monoclonal)Cell Signaling Technology#8,418 S(ICC 1:100)
AntibodyPaxillin (mouse monoclonal)BD Biosciences#610,052(ICC 1:100)
AntibodyFAK(mouse monoclonal)BD Biosciences#610,087(ICC 1:100)
AntibodyP-Paxillin (Y118)(rabbit polyclonal)Cell Signaling Technology#2,541 S(ICC 1:100)
AntibodyP-FAK (Y397)(rabbit monoclonal)Life Technologies/Thermo Fisher Scientific#44,625 G(ICC 1:100)
AntibodyCortactin(mouse monoclonal)Merck Millipore#05–180(ICC 1:100)
AntibodyARL4C (rabbit polyclonal)This paperSAJ5550275(ICC 1:100)
Recombinant DNA reagentpEGFPC2-IQGAP1 (plasmid)K.Kaibuchi (Nagoya University, Japan)
Recombinant DNA reagentpEGFP-mCyth2 (plasmid)J.Yamauchi (Tokyo University of Pharmacy and Life Science, Japan),
Recombinant DNA reagentpAcGFP-mPlcd1PH (plasmid)M.Matsuda (Kyoto University, Kyoto, Japan)
Recombinant DNA reagentCSII-CMV-MCS-IRES2-Bsd (plasmid)H.Miyoshi (RIKEN Bioresource Center, Ibaraki, Japan)
Recombinant DNA reagentpEGFPN3-hARL4C (plasmid)A.Kikuchi (Osaka University, Osaka, Japan)
Recombinant DNA reagentpEGFPN3-hGRP1 (plasmid)This paperFull length cDNAs of GRP1 ORF were reversely transcribed from mRNA extracted from MCF-7 cells.
Recombinant DNA reagentpEGFPN3-hMMP14 (plasmid)This paperFull length cDNAs of MMP14 ORF were reversely transcribed from mRNA extracted from U2OS cells.
Recombinant DNA reagentmRFP-FKBP-5-ptase-domAddgene67,516
Recombinant DNA reagentPM-FRB-CFPAddgene67,517
Sequence-based reagentsiRNA: randomized controlThis paper5'-CAGTCGCGTTTGCGACTGG-3'
Sequence-based reagentsiRNA: human IQGAP1#1This paper5'-GCTGCACATAGTTGCCTTT-3'
Sequence-based reagentsiRNA: human IQGAP1#2This paper5'-CCCTAATGTAGAATGTCAT-3'
Sequence-based reagentsiRNA: human CYTH2#1This paper5'-GGATGGAGCTGGAGAACAT-3'
Sequence-based reagentsiRNA: human CYTH2#2This paper5'-GCAGTTTCTATGGAGCTTT-3'
Sequence-based reagentsiRNA: human ARPC2#1This paper5'-GCCTATATTCACACACGTA-3'
Sequence-based reagentsiRNA: human ARPC2#2This paper5'-CCTATATTCACACACGTAT-3'
Sequence-based reagentsiRNA: human MMP14#1This paper5'-GCAGCCTCTCACTACTCTT-3'
Sequence-based reagentsiRNA: human MMP14#2This paper5'-CCGACATCATGATCTTCTT-3'
Sequence-based reagentsiRNA: human KRAS#1This paper5'-GCATCATGTCCTATAGTTT-3'
Sequence-based reagentsiRNA: human KRAS#2This paper5'-GTTGGAGCTGATGGCGTAG-3'
Sequence-based reagentsiRNA: human CTNNB1#1This paper5'-CCCACTAATGTCCAGCGTT-3'
Sequence-based reagentsiRNA: human CTNNB1#2This paper5'-GCATAACCTTTCCCATCAT-3'
Sequence-based reagentAntisense oligonucleotide: randomized controlThis paperT(Y)^a^g^A(Y)^g^a^G(Y)^t^a^5(Y)^c^c^A(Y)^t^c (Lower case = DNA; N(Y) = AmNA; 5(Y) = AmNA_mC; ^ = Phosphorothioated)
Sequence-based reagentAntisense oligonucleotide: ARL4C-1316This paperG(Y)^5(Y)^A(Y)^t^a^c^c^t^c^a^g^g^T(Y)^A(Y)^a (Lower case = DNA; N(Y) = AmNA; 5(Y) = AmNA_mC; ^ = Phosphorothioated)
Sequence-based reagenthuman GAPDH_FThis paperPCR primers5'-TCCTGCACCACCAACTGCTT-3'
Sequence-based reagenthuman GAPDH_RThis paperPCR primers5'-TGGCAGTGATGGCATGGAC-3'
Sequence-based reagenthuman B2M_FThis paperPCR primers5'-TGCTGTCTCCATGTTTGATGTATC-3
Sequence-based reagenthuman B2M_RThis paperPCR primers5'-TCTCTGCTCCCCACCTCTAAG-3'
Sequence-based reagenthuman ARL4C_FThis paperPCR primers5'-AGGGGCTGTGAAGCTGAGTA-3’
Sequence-based reagenthuman ARL4C_RThis paperPCR primers5'-TTCCAGGCTGAAAAGCAGTT –3'
Sequence-based reagenthuman ARPC2_FThis paperPCR primers5'-AGATTTCGATGGGGTCCTCT-3'
Sequence-based reagenthuman ARPC2_RThis paperPCR primers5'-CCGGAAGATTTTCAAGGTCA-3'
peptide, recombinant proteinFLAG peptideSigma-AldrichF3290
Commercial assay or kitLipofectamine2000 transfection reagentLife Technologies/Thermo Fisher Scientific11668019
Commercial assay or kitLipofectamine LTX reagentLife Technologies/Thermo Fisher Scientific15338100
Commercial assay or kitRNAiMAXLife Technologies/Thermo Fisher Scientific13778075
Commercial assay or kitViaFectPromega Corp.E4981
Commercial assay or kitTrypLE Express EnzymeThermo Fisher Scientific12604013
Commercial assay or kitPrimeSTAR Max DNA PolymeraseTakara Bio IncR045A
Commercial assay or kitIn-Fusion HD Cloning KitClontech639,649
Commercial assay or kitDakoReal EnVision Detection SystemDakoK500711-2
Commercial assay or kitPeroxidase-Blocking SolutionDakoS202386-2
Commercial assay or kitG-BlockGenoStaffGB-01
Commercial assay or kitBlocking One Histonacalai tesque06349–64
Commercial assay or kitrat tail type I collagenCorning Inc354,236
Commercial assay or kitDQ-collagen type IInvitrogenD12060
Commercial assay or kit3D microfluidic cell culture chipAIM BiotechDAX-1
Commercial assay or kitQCM Gelatin Invadopodia Assay (Red)Merck MilliporeECM671
Commercial assay or kitpoly-D-lysineSigma-AldrichP6407
Commercial assay or kitMatrigel Growth Factor ReducedCorning Inc354,230
Commercial assay or kit6.5 mm Transwell with 8.0 µm Pore Polycarbonate Membrane InsertCorning Inc3,422
Commercial assay or kitBioCoat Matrigel Invasion Chambers with 8.0 µm PET MembraneCorning Inc354,480
Commercial assay or kitAnnexin V-FITC Apoptosis Detection Kitnacalai tesque15342–54
Commercial assay or kitprotein A Sepharose beadsGE Healthcare17078001
Commercial assay or kitDynabeads Protein GThermo Fisher ScientificDB10003
Commercial assay or kitPierce Silver Stain for Mass SpectrometryThermo Fisher Scientific24,600
Commercial assay or kitO.C.T. CompoundSakura Finetek
Commercial assay or kitNucleoSpin RNAMACHEREY-NAGEL GmbH & Co. KG740,955
Commercial assay or kitReverTra Ace qPCR RT Master MixTOYOBOFSQ-201
Commercial assay or kitPierce BCA Protein Assay KitThermo Fisher Scientific23,227
Chemical compound, drugPD184161Sigma-AldrichPZ0112
Chemical compound, drugU0126Promega Corp.V1121
Chemical compound, drugRapamycinCell Signaling Technology9,904
Chemical compound, drugLY294002Cell Signaling Technology9,901
Chemical compound, drugVivoGlo luciferinPromega Corp.P1043
Software, algorithmHALOIndica LabsRRID:SCR_018350
Software, algorithmNanoZoomer-SQHamamatsu Photonics K.K.
Software, algorithmUCSC Xena browserhttp://xena.ucsc.eduRRID:SCR_018938
Software, algorithmKaplan–Meier plotterhttp://www.kmplot.comRID:SCR_018753
Software, algorithmGraphPad Prism 8GraphPad Software.RRID:SCR_002798
Software, algorithmExcel ToukeiESUMI Co., Ltd.
Software, algorithmImarisBitplaneRRID:SCR_007370
Software, algorithmImage JNational Institutes of HealthRRID:SCR_003070
Software, algorithmLiving Image 4.3.1 SoftwareCaliper Life SciencesRRID:SCR_014247
Software, algorithmikra v1.2.2https://zenodo.org/record/3606888 (Yu et al., 2019)
Software, algorithmiDEP.90http://bioinformatics.sdstate.edu/idep90/ (Ge et al., 2020)
Software, algorithmIngenuity Pathway AnalysisIPA; QiagenRRID:SCR_008653
OtherLSM880 laser scanning microscopeCarl Zeiss
OtherBZ-9000Keyence
OtherIVIS imaging systemXenogen Corp.
OtherHoechst33342InvitrogenH1399
OtherAlexa Fluor 488 PhalloidinInvitrogenA12379
OtherAlexa Fluor 546 PhalloidinInvitrogenA22283
OtherAlexa Fluor 647 PhalloidinInvitrogenA22287

Materials and chemicals

Request a detailed protocol

HeLaS3 cells were kindly provided by Dr. K. Matsumoto (Nagoya University, Aichi, Japan) in May 2002. S2-CP8 pancreatic cancer cells were purchased from Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University, in April 2014. Lenti-X 293T (X293T) cells were purchased from Takara Bio Inc (Shiga, Japan) in October 2011. PANC-1 cells were purchased from RIKEN Bioresource Center Cell Bank (RIKEN BRC, Tsukuba, Japan) in October 2014. BxPC-3 cells were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA) in May 2018. HPAF-II cells were purchased from ATCC in July 2017. S2-CP8, X293T, HeLaS3, and HPAF-II cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS). PANC-1 and BxPC-3 cells were grown in RPMI-1640 supplemented with 10 % FBS. All cell lines were authenticated using short tandem repeat profiling by BEX CO., LTD (Tokyo, Japan) and tested negative for Mycoplasma using e-Myco Mycoplasma PCR Detection Kit (iNtRON Biotechnology, Inc, Gyeonggi-do, Korea).

S2-CP8 cells stably expressing GFP, ARL4C-EGFP, ARL4CG2A-EGFP, ARL4CT27N-EGFP, ARL4CQ72L-EGFP, ARL4CΔPBR-EGFP, ARL4CG2A-EGFP-Cterm, ARL4CΔPBR-EGFP-Cterm, ARL4CG2A-EGFP-GRP1PH, ARL4CG2A-EGFP-PLCδPH, ARL4C-mCherry, ARL4C-tdTomato, EGFP-IQGAP1, and luciferase were generated using lentivirus as described previously (Kimura et al., 2016). BxPC-3 cells stably expressing EGFP or ARL4C-EGFP were generated using lentivirus. Lentiviral vector CSII-CMV-MCS-IRES2-Bsd harboring a cDNA was transfected with the packaging vectors pCAG-HIV-gp and pCMV-VSV-G-RSV-Rev into X293T cells using Lipofectamine2000 transfection reagent (Life Technologies/Thermo Fisher Scientific, Carlsbad, CA, USA). To generate S2-CP8 stable cells above, 1 × 105 parental cells/well in a 12-well plate were treated with lentiviruses and 5 μg/mL polybrene, centrifuged at 1200 x g for 30 min, and incubated for 24 h. The cells were selected and maintained in the medium containing 10 μg/mL Blasticidin S.

ARL4C or IQGAP1 knockout cells were generated as previously described (Fujii et al., 2016). The target sequences for human ARL4C, 5’-CTTCTCGGTGTTGAAGCCGA-3’, and human IQGAP1, 5’-CACCGTGGGGTCTACCTTGCCAAAC-3’ were designed with the help of the CRISPR Genome Engineering Resources (http://www.genome-engineering.org/crispr/). The plasmids expressing hCas9 and single-guide RNA (sgRNA) were prepared by ligating oligonucleotides into the BbsI site of pX330 (addgene #42230). The plasmid pX330 with sgRNA sequences targeting ARL4C, IQGAP1 and Blasticidin resistance was introduced into S2-CP8 cells using Lipofectamine LTX reagent (Life Technologies/Thermo Fisher Scientific) according to manufacturer’s instructions and the transfected cells were selected in medium containing 5 μg/mL Blasticidin S for 2 days. Single colonies were picked, mechanically disaggregated, and replated into individual wells of 24-well plates.

ARL4C ASO-1316 and 6-carboxyfluorescein (FAM)-labeled ARL4C ASO-1316 were synthesized by GeneDesign (Osaka, Japan) as described (Harada et al., 2019). The sequences of the ASOs are listed in Supplementary file 1 table 5. S2-CP8 cell were transfected with ASOs at 10 nmol/L using RNAiMAX (Life Technologies/Thermo Fisher Scientific) in antibiotics-free medium. The transfected cells were then used for experiments conducted at 48 hr after transfection.

Anti-ARL4C polyclonal antibody (SAJ5550275) for immunoprecipitation and immunocytochemistry was generated in rabbits by immunization with recombinant human ARL4C. Antibodies used in this study are shown in Supplementary file 1 table 6.

The following drugs were used: PD184161 (Sigma-Aldrich Co, St. Louis, MO, USA); U0126 (Promega Corp., Madison, WI, USA); Rapamycin (Cell Signaling Technology, Beverly, MA, USA); LY294002 (Cell Signaling Technology); and VivoGlo luciferin (Promega Corp.).

Plasmid construction pEGFPC2-IQGAP1, pEGFP-mCyth2, pAcGFP-mPlcd1PH, and CSII-CMV-MCS-IRES2-Bsd were kindly provided by K. Kaibuchi (Nagoya University, Japan), J. Yamauchi (Tokyo University of Pharmacy and Life Science, Japan), M. Matsuda (Kyoto University, Kyoto, Japan), and H. Miyoshi (RIKEN Bioresource Center, Ibaraki, Japan), respectively.

To generate plasmid DNA with mutated codons or deletions, site-directed mutagenesis method was performed using PrimeSTAR Max DNA Polymerase (Takara Bio Inc, Shiga, Japan). To generate plasmid DNA with insertions, PCR amplified fragments and linearized vector by restriction enzyme digestion were assembled using In-Fusion HD Cloning Kit (Takara Bio Inc).

pEGFPN3-ARL4C was constructed as previously described (Matsumoto et al., 2014). Full length cDNAs of GRP1 and MMP14 ORF were reversely transcribed from mRNA extracted from MCF-7 cells and U2OS cells, respectively. Linear double strand oligonucleotides of the C-terminal 22 amino acids of KRAS, which includes the PBR and CAAX motifs, were synthesized, and the oligonucleotides were inserted into C terminal of ARL4C-EGFP or ARL4C-FLAG-HA using In-Fusion HD Cloning Kit (Takara Bio Inc).

Standard recombinant DNA techniques mentioned above were used to construct the following plasmids: pEGFPN3-ARL4C, pEGFPN3-ARL4CG2A, pEGFPN3-ARL4CT27N, pEGFPN3-ARL4CQ72L, pEGFPN3-ARL4CΔPBR, pEGFPN3-ARL4CG2A-EGFP-PLCδPH, pEGFPN3-ARL4CG2A-EGFP-GRP1PH, pEGFPN3-ARL4CG2A-EGFP-Cterm, pEGFPN3-ARL4CΔPBR-EGFP-Cterm, pEGFPC1-CHD, pEGFPC1-IQ, pEGFPC1-WW, pEGFPC1-IR, pEGFPC1-GRD, pEGFPC1-RGCT, pcDNA3-ARL4C-FLAG-HA, pcDNA3-ARL4CG2A-FLAG-HA, pcDNA3-ARL4CΔPBR-FLAG-HA, pcDNA3-ARL4CG2A-FLAG-HA-Cterm, pcDNA3-ARL4CΔPBR-FLAG-HA-Cterm, pcDNA3-FLAG-HA-IQGAP1, pmCherryN1-ARL4C, pmCherryN1-MMP14, pmCherryN1-MMP14ΔC(Δ563–582), pCAG-ARL4C-tdTomato. To construct lentiviral vectors harboring EGFP, ARL4C-EGFP, ARL4CG2A-EGFP, ARL4CT27N-EGFP, ARL4CQ72L-EGFP, ARL4CΔPBR-EGFP, ARL4CG2A-EGFP-Cterm, ARL4CΔPBR-EGFP-Cterm, ARL4CG2A-EGFP-PLCδPH, ARL4CG2A-EGFP-GRP1PH, EGFP-IQGAP1, ARLC-mCherry, MMP14-mCherry, MMP14ΔC-mCherry, ARL4C-tdTomato were cloned into CSII-CMV-MCS-IRES2-Bsd provided by Dr. H. Miyoshi (RIKEN Bioresource Center, Ibaraki, Japan).

Patients and cancer tissues

Request a detailed protocol

The present study involved 57 presurgical untreated patients with PDAC and ages ranging from 47 to 87 years (median, 70 years) who underwent surgical resection at Osaka University between April 2001 and April 2015. Tumors were staged according to the Union for International Cancer Control (UICC) TNM staging system. Resected specimens were fixed in 10 % (vol/vol) formalin, processed for paraffin embedding, and were sectioned at 5 μm thickness and stained with hematoxylin and eosin (H&E) or immunoperoxidase for independent evaluations. The protocol for this study was approved by the ethical review board of the Graduate School of Medicine, Osaka University, Japan (No. 13455), under the Declaration of Helsinki, and written informed consent was obtained from all patients. The study was performed in accordance with Committee guidelines and regulations.

Immunohistochemical studies

Request a detailed protocol

Immunohistochemical studies were performed as previously described (Fujii et al., 2015) with modification. Briefly, all tissue sections were stained using a DakoReal EnVision Detection System (Dako, Carpentaria, CA, USA) in accordance with the manufacturer’s recommendations. Formalin-fixed, paraffin-embedded tissue specimens for examination were sectioned at 5 μm thickness. Heat-induced epitope retrieval was performed using Decloaking Chamber NxGen (Biocare Medical, Walnut Creek, CA, USA). Tissue peroxidase activity was blocked with Peroxidase-Blocking Solution (Dako) for 30 min, and the sections were then incubated with G-Block (GenoStaff, Tokyo, Japan) or Blocking One Histo (nacalai tesque, Kyoto, Japan) for 30 min or 10 min, respectively, to block nonspecific antibody binding sites. Tissue specimens were treated with anti-ARL4C (1:100), anti-IQGAP1 (1:800), or anti-MMP14 (1:100) antibody for 3 hr at room temperature. Then, the specimens were detected by incubating with goat anti-rabbit or anti-rabbit/mouse IgG-HRP for 1 h and subsequently with DAB (Dako). The tissue sections were then counterstained with 0.1 % (wt/vol) hematoxylin. ARL4C expression was considered high when the total area of the tumor stained with anti-ARL4C antibody exceeded 5 %. IQGAP1 expression was considered high when the total area of the tumor stained with anti-IQGAP1 antibody exceeded 40 %.

IQGAP1 staining positivity in PDAC patients was measured using HALO (Indica Labs, Corrales, NM, USA). The threshold for positive or negative staining was based on the optical density of the staining: regions above the positivity threshold were scored according to the optical density threshold set in the module; weakly positive is shown in yellow and strongly positive in red. The samples were viewed and analyzed using NanoZoomer-SQ (Hamamatsu Photonics K.K., Shizuoka, Japan).

Clinical data analyses using open sources

Request a detailed protocol

The data on ARL4C and IQGAP1 mRNA expression in pancreatic adenocarcinoma were obtained from the UCSC Xena browser (http://xena.ucsc.edu). Tumors and normal samples in the UCSC Xena browser were derived from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) projects. Differential analysis was performed using a two-tailed Student’s t-test. The correlations of overall survival rates with ARL4C, IQGAP1, and MMP14 expression in pancreatic cancer in TCGA datasets were analyzed using a Kaplan–Meier plotter (http://www.kmplot.com) and visualized using GraphPad Prism 8 (GraphPad Software. San Diego, CA, USA). High and low expression groups were classified by auto select best cutoff. p Values and r values were calculated using GraphPad Prism.

3D gel invasion assay using a 3D microfluidic cell culture chip

Request a detailed protocol

Collagen gels were made by diluting and neutralizing rat tail type I collagen (Corning Inc, Corning, NY, USA) in PBS and 12.1 mM NaOH, and were adjusted to 2 mg/mL. DQ-collagen type I (Life Technologies/Thermo Fisher Scientific, Carlsbad, CA, USA) was mixed with collagen gels at a final concentration of 25 μg/mL. The gel channel of 3D microfluidic cell culture chip (AIM Biotech, Biopolis Rd, Singapore) was filled with collagen solution and incubated at 37 °C for at least 1 hr to polymerize collagen. After hydration of medium channels, a cell suspension (1 × 104 cells) in serum-free cell culture medium with 0.2 % BSA was injected into one of the ports at the medium channel. The opposite medium channel was filled with cell culture medium containing 10 % FBS to create a chemoattractant gradient across the collagen gel. The cells were then incubated for 3 days and fixed for 15 min at room temperature in PBS containing 4 % (w/v) paraformaldehyde. Then, the cells were permeabilized and blocked in PBS containing 0.5 % (w/v) Triton X-100 and 40 mg/mL BSA for 30 min and stained with the indicated antibodies. The samples were viewed and analyzed under an LSM880 laser scanning microscope (Carl Zeiss, Jana, Germany). Reconstruction of confocal z-stack images into 3D animations and analysis of 4D images were performed using Imaris (Bitplane, Belfast, UK).

Invadopodia assay

Request a detailed protocol

QCM Gelatin Invadopodia Assay (Red) (Merck Millipore, Burlington, MA, USA) was used in accordance with the manufacturer’s protocol. Briefly, poly-L-lysine–coated coverslips were treated with glutaraldehyde. The coverslips were then incubated with Cy3-labeled gelatin, followed by culture medium quenching of free aldehydes. Cells (6 × 104 cells) were seeded onto the gelatin-coated coverslips and incubated for 4 hr. After incubation, the cells were fixed for 20 min at room temperature in phosphate-buffered saline (PBS) containing 4 % (w/v) paraformaldehyde and permeabilized in PBS containing 0.2 % (w/v) Triton X-100 for 10 min. After being blocked in PBS containing 0.2 % (w/v) BSA for 30 min, the cells were immunohistochemically stained. The samples were viewed and analyzed under an LSM880 laser scanning microscope (Carl Zeiss, Jana, Germany).

2D culture on poly-D-lysine– or matrigel-coated dishes

Request a detailed protocol

Cells grown on glass coverslips coated with poly-D-lysine (Sigma-Aldrich) or Matrigel Growth Factor Reduced (Corning) were fixed for 10 min at room temperature in PBS containing 4 % (w/v) paraformaldehyde and permeabilized in PBS containing 0.1 % (w/v) saponin (Sigma-Aldrich) or 0.2 % (w/v) Triton X-100 for 10 min. The cells were then blocked in PBS containing 0.2 % (w/v) BSA for 30 min. They were then incubated with primary antibodies for 3 hr at room temperature and with secondary antibodies in accordance with the manufacturer’s protocol (Life Technologies/Thermo Fisher Scientific). For cell surface MMP14 staining, samples were incubated with anti-MMP14 antibody for 3 hr at room temperature without permeabilization. The samples were viewed and analyzed under an LSM880 laser scanning microscope (Carl Zeiss).

Migration and invasion assays

Request a detailed protocol

Migration and invasion assays were performed using a modified Boyden chamber (6.5 mm Transwell with 8.0 µm Pore Polycarbonate Membrane Insert; Corning) and a Matrigel-coated modified Boyden chamber (BioCoat Matrigel Invasion Chambers with 8.0 µm PET Membrane; Corning), respectively as described previously (Kurayoshi et al., 2006; Matsumoto et al., 2014). In the standard conditions, S2-CP8 cells (2.5 × 104 cells) were seeded in the upper side of Boyden Chamber. In GFP-expressing S2-CP8 cells, after 4 h (migration assay) or 24 hr (invasion assay, except for Figure 6E) incubation with control ASO, 122 cells (average) and 126 cells (average), respectively, were observed in the lower side chamber in the one field of view under fluorescence microscope (BZ-9000, Keyence, Osaka, Japan) using a 10 x air objective. In Figure 6E, cells were observed after 20 hr incubation with ASO. Migration and invasion rates of cells expressing ARL4C, IQGAP1, and MMP14 mutants were calculated as the percentages of the same cells transfected with control ASO or siRNA.

3D type I collagen gel culture

Request a detailed protocol

Collagen gels were made by diluting and neutralizing rat tail type I collagen (Corning) in PBS and 12.1 mM NaOH, and were adjusted to 2 mg/mL. Then, 140 µL of cell-embedded collagen gels (1 × 106 cells/mL) were overlaid onto glass coverslips in a 24-well plate and allowed to polymerize for at least 1 hr at 37 °C and 5 % CO2. After polymerization, growth medium was added on top of the collagen gel. The cells were then incubated for 3 days and fixed for 15 min at room temperature in PBS containing 4 % (w/v) paraformaldehyde. Then, the cells were permeabilized and blocked in PBS containing 0.5 % (w/v) Triton X-100 and 40 mg/mL BSA for 30 min and incubated with primary antibodies for 3 h at room temperature and secondary antibodies in accordance with the manufacturer’s protocol (Life Technologies/Thermo Fisher Scientific). The samples were viewed and analyzed under an LSM880 laser scanning microscope using a 20 x air objective (Carl Zeiss). In the standard conditions (for Figure 3L) with BxPC-3/ARL4C-GFP cells treated with control ASO, the number of cells with pseudopods and the total number of cells were 15 (average) and 76 (average), respectively, in the one field of view under an LSM880 laser scanning microscope (Carl-Zeiss) using a 20 x air objective. The percentages of cells with pseudopods compared with the total number of cells in the presence of control siRNA or IQGAP1 siRNA were calculated.

Inducible recruitment of phospholipid phosphatases mRFP-FKBP-5-ptase-dom and PM-FRB-CFP plasmids were obtained from Addgene (deposited by the laboratory of T. Balla). S2-CP8 cells were then transiently transfected with both mRFP-FKBP-5-ptase-dom and PM-FRB-CFP (0.5 μg/well of a six-well plate for each vector) with ViaFect (Promega Corp.). After 24 hr culture, the cells were treated with 100 nM rapamycin or 50 μM LY294002 for 30 min before fixation.

Cytotoxic assay

Request a detailed protocol

Cells transfected with control ASO or ARL4C ASO-1316 were cultured on Matrigel coated dish for 3.5 days, and dissociated using TrypLE Express (Thermo Fisher Scientific). Suspension of cells was stained with Hoechst 33342 or propidium iodide (PI) using Annexin V-FITC Apoptosis Detection Kit (nacalai tesque). The samples were viewed and analyzed under an LSM880 laser scanning microscope (Carl Zeiss), and the number of PI-positive cells was divided by the total number of nuclei stained with Hoechst 33342.

Isolation of ARL4C-interacting protein

Request a detailed protocol

Confluent X293T cells transiently transfected with ARL4C-FLAG-HA in two 10 cm culture dishes were harvested and lysed in 800 μL of lysis buffer (25 mM Tris-HCl [pH7.5], 50 mM NaCl, 0.5 % TritonX-100) with protease inhibitors (nacalai tesque). After 10 min of centrifugation, lysates were incubated with 40 μL of 50 % slurry of anti-FLAG Affinity Gel (Sigma-Aldrich) for 30 min, and then add another 40 μL and incubated for 30 min. Beads were washed three times with 1 mL of lysis buffer. Recovered beads were incubated once with FLAG peptide (0.5 mg/mL) to elute proteins in 80 μL of PBS for 30 min at 4 °C. Then, the supernatant was precleaned with 40 μL of 50 % slurry of protein A Sepharose beads (GE Healthcare, Chicago, IL, USA) for 30 min at 4 °C. The precleaned lysates were incubated with 2 μg of anti-HA antibody (Santa Cruz, Dallas, TX, USA) and 50 μL of 50 % slurry of protein A Sepharose beads for 1 hr at 4 °C. Beads were washed three times with 1 mL of lysis buffer, and bound complexes were dissolved in 50 μL of Laemmli’s sample buffer. The ARL4C-FLAG-HA-interacting proteins were detected by Pierce Silver Stain for Mass Spectrometry (Life Technologies/Thermo Fisher Scientific). Six bands (arrowheads in Figure 3A) were cut from the gel and analyzed by mass spectrometry.

Immunoprecipitation

Request a detailed protocol

Immunoprecipitation was performed as described previously with modification (Matsumoto et al., 2014). For Figure 3C S2-CP8 cells (60 mm diameter dish) were lysed in 300 µL of lysis buffer (25 mM Tris–HCl pH 7.5, 50 mM NaCl, 0.5 % Triton-X100) with protease inhibitors (nacalai tesque) for 10 min on ice. After centrifugation, the supernatant was collected and pre-cleaned using 30 µL of Dynabeads Protein G (Thermo Fisher Scientific). After pre-cleaning, lysates were rotated with complex of Dynabeads (50 μL) and antibody (3.6 μg) for 10 min at room temperature. The beads were then washed with lysis buffer three times, and finally suspended in Laemmli’s sample buffer.

The RAC1 activity assay

Request a detailed protocol

The RAC1 activity assay was performed as described (Matsumoto et al., 2014). Briefly, cells were lysed in 400 μL of RAC1 assay buffer (20 mM Tris–HCl [pH 7.5], 150 mM NaCl, 1 mM dithiothreitol, 10 mM MgCl2, 1 % Triton-X100) with protease inhibitors (nacalai tesque) containing 20 μg of glutathione-S-transferase (GST)-CRIB. After the lysates were centrifuged at 20,000 g for 10 min, the supernatants were incubated with glutathione-Sepharose (20 μl each) for 2 h at 4 °C. The beads were then washed with RAC1 assay buffer three times, and finally suspended in Laemmli’s sample buffer. The precipitates were probed with the anti-RAC1 antibody.

Imaging of ASO accumulation in tumor-bearing mice

Request a detailed protocol

Orthotopic transplantation was performed as described previously (Kim et al., 2009). Ten days after the transplantation, 150 μg/animal (approximately 7.5 mg/kg) of 6-FAM-ARL4C ASO-1316 was subcutaneously administered. Four h after the injection, the fluorescence intensities of various organs were measured ex vivo using the IVIS imaging system (Xenogen Corp.). After ex vivo imaging, unfixed mouse pancreas tissues were frozen in an O.C.T. Compound (Sakura Finetek, Tokyo, Japan)/sucrose mixture [1:1 (v/v) OCT and 1 x PBS containing 30 % sucrose]. Freshly frozen tissues were sectioned at 10 μm and fixed for 30 min at room temperature in PBS containing 4 % (w/v) paraformaldehyde. The cells were then permeabilized and blocked in PBS containing 0.5 % (w/v) Triton X-100 and 40 mg/mL BSA for 30 min and stained with the indicated antibodies. The samples were viewed and analyzed under an LSM880 laser scanning microscope (Carl Zeiss).

Orthotopic xenograft tumor assay

Request a detailed protocol

An orthotopic transplantation assay was performed as described previously (Kim et al., 2009) with modification. Ten-week-old male BALB/cAJcl-nu/nu mice (nude mice; CLEA, Tokyo, Japan) were anesthetized and received an orthotopic injection of S2-CP8 cells into the mid-body of the pancreas using a 27 G needle (5 × 105 cells suspended in 100 μL of HBSS with 50 % Matrigel). ASOs (50 μg/mouse, approximately 2.5 mg/kg) were administered subcutaneously twice a week from day 3. To evaluate the knockdown efficiency of ARL4C ASO-1316, tumor tissues were harvested from tumor-bearing mice 8 days after transplantation. Total RNAs were isolated using NucleoSpin RNA (MACHEREY-NAGEL GmbH & Co. KG, Dueren, Germany), and complementary DNAs were synthesized using ReverTra Ace qPCR RT Master Mix (TOYOBO, Osaka, Japan). For extraction of tissue proteins, tumor samples were lysed in 150 μL of lysis buffer (20 mM Tris–HCl [pH 8.0], 137 mM NaCl, 10 % glycerol, 1% NP40) and homogenized using Biomasher II (KANTO CHEMICAL CO.,Inc, Tokyo, Japan). Debris was removed by centrifugation and finally suspended in Lammli’s buffer. Protein concentration was determined with Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). The band intensities of western blotting were calculated using Image J (National Institutes of Health, USA). To assess the effect of ARL4C ASO-1316 on tumor progression, tumor burden was measured once a week using the IVIS imaging system (Xenogen Corp., Alameda, CA, USA). For the in vivo imaging, 100 μL of VivoGlo luciferin (30 mg/mL) was intraperitoneally administered and the bioluminescence imaging was performed 8 min later. The region of interest (ROI) was selected and the radiance values were measured with Living Image 4.3.1 Software (Caliper Life Sciences, Hopkinton, MA, USA). The mice were euthanized 28 days after transplantation. Tumor weights and numbers of mesenteric lymph nodes (diameter of lymph nodes > 1 mm) were measured. All protocols used for the animal experiments in this study were approved by the Animal Research Committee of Osaka University, Japan (No. 26-032-048).

RNA sequencing

Request a detailed protocol

Sequenced reads were preprocessed by Trim Galore! v0.6.3 and quantified by Salmon v0.14.0 with the flags gcBias and validateMappings. GENCODE vM21 annotation was used as the transcript reference. The quantified transcript-level scaled TPM was summarized into a gene-level scaled TPM by using the R package tximport v1.6.0. All procedures were implemented using the RNAseq pipeline ikra v1.2.2 [https://zenodo.org/record/3606888 (Yu et al., 2019)] with the default parameters. Downstream analysis was conducted with an integrative RNAseq analysis platform, iDEP.90. After normalization with VST, principal component analysis was conducted. Hierarchical clustering was performed on the top 1,000 genes in terms of their standard deviation. Finally, DEGs were selected with a log2 fold change >1 and false discovery rate <0.1.

Ingenuity pathway analysis (IPA)

Request a detailed protocol

DEGs identified from RNA sequence data were subjected to Ingenuity Pathway Analysis (IPA; Qiagen, Hilden, Germany). This analysis examines DEGs that are known to affect each biological function and compares their direction of change to what is expected from the literature. To infer the activation states of implicated biological functions, two statistical quantities, Z-score and p value, were used. A positive or negative Z-score value indicates that biological functions are predicted to be activated or inhibited in the ARL4C ASO-1316–treated group relative to the control ASO-treated group. A negative Z-score means that the indicated biological functions are inhibited by ARL4C ASO-1316. The p value, calculated with the Fisher’s exact test, reflects the enrichment of the DEGs on each pathway. For stringent analysis, only biological functions with a |Z-score| > 2 were considered significant.

Statistics and reproducibility

Request a detailed protocol

Biological replicates are replicates on independent biological samples versus technical replicates that use the same starting samples. All experiments in this study were repeated using biological replicates. A minimum of three biological replicates were analyzed for all samples, and the results are presented as the mean ± s.d. or s.e.m. The cumulative probabilities of overall survival were determined using the Kaplan–Meier method; a log-rank test was used to assess statistical significance. The Student’s t-test or Mann–Whitney test was used to determine if there was a significant difference between the means of two groups. One-way analysis of variance (ANOVA) with Bonferroni tests was used to compare three or more group means. Statistical analysis was performed using Excel Toukei (ESUMI Co., Ltd., Tokyo, Japan) and GraphPad Prism 8 (GraphPad Software, La Jolla, CA, USA); p < 0.05 was considered statistically significant. In box and whiskers plots, the top and bottom horizontal lines represent the 75th and the 25th percentiles, respectively, and the middle horizontal line represents the median. The size of the box represents the interquartile range and the top and bottom whiskers represent the maximum and the minimum values, respectively.

Others

Request a detailed protocol

The siRNAs and primers used in these experiments are listed in Supplementary file 1 tables 7 and 8, respectively. 2.5D Matrigel growth assay and quantitative PCR were performed as described previously (Matsumoto et al., 2019; Sato et al., 2010).

Data availability

-Sequencing data have been deposited in DDBJ under accession codes DRA011537. -All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 1-7, Figure 2-figure supplement 1, Figure 2-figure supplement 2, Figure 2-figure supplement 3, Figure 3-figure supplement 1, Figure 3-figure supplement 2, Figure 4-figure supplement 1, Figure 6-figure supplement 1, Figure 7-figure supplement 1, Supplementary File 1 Table 1, and Supplementary File 1 Table 2.

The following data sets were generated
    1. Akikazu H
    (2021) DRASearch
    ID DRA011537. Effects of ARL4C ASO on an orthotopic transplantation model.
The following previously published data sets were used

References

  1. Software
    1. Ge X
    2. Son E
    3. Yao R
    (2020) iDEP, version iDEP.90
    South Dakota State University Bioinformatics Research Group.
    1. Hidalgo M
    (2010) Pancreatic cancer
    The New England Journal of Medicine 362:1605–1617.
    https://doi.org/10.1056/NEJMra0901557

Decision letter

  1. Lynne-Marie Postovit
    Reviewing Editor; University of Alberta, Canada
  2. Erica A Golemis
    Senior Editor; Fox Chase Cancer Center, United States
  3. Ivan Robert Nabi
    Reviewer; Life Sciences Institute, University of British Columbia Faculty of Medicine
  4. Hon Leong
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This work reveals a hitherto unknown mechanism by which KRAS and ARL4C regulate the invasion of pancreatic cancer cells. These results may reveal new targets for the prevention of pancreatic cancer metastasis.

Decision letter after peer review:

Thank you for submitting your article "Recruitment of KRAS downstream target ARL4C to membrane protrusions accelerates pancreatic cancer cell invasion." for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Erica Golemis as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Ivan Robert Nabi (Reviewer #2); Hon Leong (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

The current manuscript describes a functional relationship between ARLC4, IQGAP1, and MMP14 in pancreatic cancer cells, such that ARLC4 is up-regulated in response to MAPK and Wnt signaling and colocalizes with IQGAP1 and MMP14 at membrane protrusions. These changes in expression and localization were associated with an increase in cellular invasion. Finally, the authors demonstrated that systemically delivered antisense oligonucleotides targeting ARL4C may prevent pancreatic cancer metastasis. Overall, this study reveals a hitherto undescribed mechanism for the regulation of pancreatic cancer cell invasion. Many of the in vitro cell biology experiments linking ARL4C to invasion are thorough; however, a more extensive and focused analysis of membrane protrusion localization is still required. Moreover, ties to clinical correlates as well as the in vivo metastasis assays could be improved upon.

Many strengths were noted:

Reviewers commended the comprehensive analysis of ARL4C in migration and 3D gel invasion, matrix degradation, and expression in tumors including mutational analyses that defined ARL4C domains that mediate these interactions. The potential to target ARL4C with ASO was deemed to be interesting and novel. The IHC data strongly supports expression of ARL4C in perineural lesions, which is a strong risk factor for poor OS and HR in pancreatic cancer.

Essential revisions:

A number of issues were raised that need to be addressed with experimentation, prior to publication. Specifically, studies that more concretely demonstrate that ARL4C recruits IQGAP1 and MMP14 to membrane protrusions are required. In addition, several controls would be needed throughout, to enhance to clinical and in vivo data sets. Specific requirements are as follows:

1) While ARL4C, IQGAP1 and MMP14 are located to protrusions, where they are clearly involved in matrix degradation, the loss of protein distribution to the protrusions (with ARL4C loss) could very well be due to the reduced abundance of protrusions associated with reduced migration and invasion of the cells. While a role for ARL4C in cell invasion is clearly demonstrated, the conclusion that ARL4C's "unique distribution to membrane protrusions is required for cancer cell invasion" is not supported by the data as presented.

This could be improved on by a variety of experiments. For example:

– All images of protrusions should include F-actin labeling to define protrusions. If the authors are arguing that ARL4C serves to recruit IQGAP1 and MMP14 to protrusions, then they need to show actin protrusions lack IQGAP1 and MMP14 when ARLC4 is knocked out.

– Studies that better demonstrate the role of protrusions in productive and/or directional migration are required. For example, the morphology of PDAC cells in the microfluidic assay suggests that the "cell protrusions" may in fact be focal adhesions/anchor points. Time-lapse imaging to show that these protrusions form via new offshoots perpendicular to the side of a cell versus a simple "zig-zag" motile pattern would be more convincing. Furthermore, time-lapse imaging of cells as they migrate towards the FBS media through the collagen matrix would enable one to determine if cell membrane protrusions are productive and have purposeful direction towards FBS. Hence, studies are needed that better characterize the direction and type of protrusions that are made.

– While ARLC4 knockouts are used for the localization studies, knock downs are used to study invasion. Both methods should be used for both assays.

– Please include a more thorough description of Mass Spec results so that the strength of the ARLC4-IQGAP1 interaction can be better appreciated.

– The 3D gel invasions demonstrate more cells in the control ASO migrating towards the FBS media, but shows there are also fewer cells in the ARLC4C ASO cells. Furthermore, there are more apoptotic bodies in the ARL4C ASO treatment cells. This suggests that the amount of siRNA/transfection agent used is toxic and interfering with migration and viability. An irrelevant binding control ASO should be used as well as a positive control ASO.

2) While ARLC4 shows promise as a target for PDAC, the preclinical studies should be improved, to strengthen correlations made and to ensure all controls are present in the in vivo assays. This could be improved on by a variety of experiments and/or edits. For example:

– Please include a table showing all patient information, as well as what the staining was for each patient.

– The images shown in Figure 1 suggest that ARLC4 protein levels in cancer cells is rather binary; however, the expression data suggests that this is not the case. This should be addressed, with a more thorough scoring scheme as well as analysis of the extent to which protein levels correlate with transcript levels. In particular, this is important when setting the cutoffs for survival curves, as an example.

– As presented in the manuscript, the clinical correlations made between IQGAP1 and ARL4C are not particularly well supported. The study would benefit from a more thorough correlational analysis as was shown for IQGAP1 and MMP14 and MMP14 and ARLC4 (supplemental 1).

– For the in vivo experiments, it is important that the extent of ARL4C knock down is established at the level of protein and transcript. In addition, the gene expression alterations should be shown in a table format. The effect of ASO1316 on ARL4C levels should also be shown in Figure 2G.

– In Figure 7H, the ASO control lesion reveals no tumor cells within the LYVE-1-delineated lumen. However, there is also a lack of tumor cells around this particular ROI. A more controlled analysis would evaluate all lymphatics that have the same amount of tumor cells surrounding them, and to look for lack of cells within the lymphatic lumen, if this is to be consistent with ASO's proposed inhibitory activity on cell invasion (and not cell proliferation or cytotoxicity).

Reviewer #1:

The current manuscript describes a functional relationship between ARL4C, IQGAP1 and MMP14 in pancreatic cancer cells; such ARLC4 is up-regulated in response to MAPK and Wnt signaling and colocalizes with IQGAP1 and MMP14 at membrane protrusions. This was associated with an increase in cellular invasion. Finally, the authors demonstrated that systemically delivered antisense oligonucleotides targeting ARL4C may prevent pancreatic cancer metastasis. Overall, this study reveals a hitherto undescribed mechanism for the regulation of pancreatic cancer cell invasion. The in vitro cell biology experiments are thorough and generally convincing. However, ties to clinical correlates as well as the in vivo metastasis assays could be improved upon.

Please include a table showing all patient information, as well as what the staining was for each patient.

The images shown in Figure 1 suggest that ARL4C protein levels in cancer cells is rather binary; however the expression data suggests that this is not the case. This should be addressed, with a more thorough scoring scheme as well as analysis of the extent to which protein levels correlate with transcript levels. In particular, this is important when setting the cutoffs for survival curves, as an example.

As presented in the manuscript, the clinical correlations made between IQGAP1 and ARL4C are not particularly well supported. The study would benefit from a more thorough correlational analysis as was shown for IQGAP1 and MMP14 and MMP14 and ARL4C (supplemental 1).

Please include a more thorough description of mass spec results so that the strength of the ARL4C-IQGAP1 interaction can be better appreciated.

For the in vivo experiments, it is important the extent of ARL4C knock down is established at the level of protein and transcript. In addition, the gene expression alterations should be shown in a table format.

Reviewer #2:

The authors undertook to assess the role of ARL4C-IQGAP1-MMP14 signaling downstream of Ras in pancreatic cancer progression and metastasis. Strengths are the comprehensive analysis of ARL4C in migration and 3D gel invasion, matrix degradation, and expression in tumors including mutational analysis that defines ARL4C domains that mediate these interactions and functional roles. Extension of these studies to use of anti-sense oligonucleotide to target ARL6 and particularly the in vivo data showing tumor regression are very interesting. A weakness of the study is the extensive analysis of protein localization to membrane protrusions, used to support the conclusion that ARL4C recruits IQGAP1 and MMP14 to protrusions. While these three proteins are located to protrusions, where they are clearly involved in matrix degradation, the loss of protein distribution to the protrusions could very well be due to the reduced abundance of protrusions associated with reduced migration and invasion of the cells. While a role for ARL4C in cell invasion is clearly demonstrated, the conclusion that ARL4C "unique distribution to membrane protrusions is required for cancer cell invasion" is not sufficiently supported by the data.

Figure 2G. The effect of ASO1316 on ARL4C expression is not shown.

Figure 2H. Not clear what the line graphs are showing.

Figure 3. All images of protrusions need to include F-actin labeling to define protrusions. If the authors are arguing that ARL4C serves to recruit IQGAP1 and MMP14 to protrusions, then they need to show actin protrusions lacking these proteins, not just count how many densities they observe. More likely, ARL4C and IQGAP1 are required for protrusion formation, something that is actually shown in Figure 3L for IQGAP1 knockdown, although not for ARL4C.

Figure 3I this is not clear.

Figure 3J. What is the relationship between low/high IQGAP1 and low /high ARL4C groups oin PDAC patient survival?

Figure 4. Show effect of ARL4C knockout on protrusions (E,F) but then resort to ASO to study invasion. What is the invasive capability of the KO cells? Same in Figure 5 G, H.

Figure 6. Not at all clear what the images are showing or what the merge is showing.Reviewer #3:

Kikuchi, Hamada, and colleagues continue their work on ARL4C and its role in promoting metastasis in pancreatic cancer by showing it acts by interacting with IQGAP1 which then interacts with MMP14. When this occurs, the entire complex is directed to membrane protrusions in pancreatic cancer cell lines. This is initiated by PIP3 and not PIP2 activation. Anti-sense oligos specific for ARL4C results in decreased lymphatic metastasis in a mouse model of pancreatic cancer, which is a surprising finding given that this is a transcriptionally based therapy. These pre-clinical results combined with their findings of elevated ARL4C in various PDAC tissue sections offer a novel opportunity to halt PDAC metastasis by antagonizing cell membrane protrusions caused by this ARL4C+IQGAP1+MMP14 complex.

Strengths:

1. If the main finding of this work is that cell protrusions are key for invasion/metastasis and ARL4C is a key protein involved in PDAC, then this is a very interesting and worthy paper. This is because the concept of a cell based organelle/feature that promotes metastasis is more likely to occur than a series of intricate molecular pathways. This concept is agnostic to any pathway involved and therefore the complex (ARL4C-IQGAP1-MMP14) observed here may or may not be broadly applicable to other cancers, but it is at least offers some molecular insights as to how it may happen.

2. PIP2/3 activation is a highly plausible activation pathway because it can work parallel to other classic outside/inside pathways such as beta1integrin activation that also can occur in cancer cells.

3. The IHC data strongly supports expression of ARL4C in perineural lesions, which is a strong risk factor for OS and HR in pancreatic cancer.

Suggestions for additional experiments:

1. Time-lapse imaging of cells as they migrate towards the FBS media through the collagen matrix would be helpful. It's not clear if the cell membrane protrusions are productive and have purposeful direction towards FBS.

2. The use of patient-derived PDX material would have added further momentum towards this idea that interrupting the ARL4C-IQGAP1-MMP14 complex leads to decreased cell membrane protrusion formation.

3. The use of organoids to understand the impact of these transgenes/mutations on cell membrane protrusions would have been closer to what is observed pre-clinically and clinically.

4. PDAC has a tremendous amount of fibrosis and is not a heavily vascularized tumor. Understanding the efficacy of ASO extravasation into tumors and how it precludes lymphatic metastasis would be of broad interest.

Open-ended questions:

1. The microfluidics chamber to assess uni-directional invasion by PDAC cells is innovative but the morphology of these cells suggests that the "cell protrusions" observed resemble focal adhesions/anchor points. Time-lapse imaging to show that these protrusions form via new offshoots perpendicular to the side of a cell versus a simple "zig-zag" motile pattern would have been convincing.

2. Only 20% of cells portray these IQGAP1-ARL4C rich protrusions. What is the metastatic efficiency like if these cells are removed from the total pool of cells within the tumor? Would there still be metastatic colony formation?

3. What are the cellular/biophysical barriers for PDAC cells as they intravasate into perineural space? Does this truly require breakdown of basement membrane or is another type of matrix/cellular barrier present? Such as myelin, fibroblasts, etc.?

4. What is the expression level of ARL4C in pre-PDAC lesions? Such as in PanIN2/3? What is/are the pioneer factors that induce ARL4C expression leading to PDAC? Hypoxia?

I have some concerns/comments regarding some of the findings and what else is missing.

1. Lian and Mulligan (Oncogene 2020) showed that "invasive processes" also contribute to perineural and neural invasion. These were driven by RET kinase activity and subsequent Src kinase activity. RET also needs to be analyzed in the IHC experiments since this is first published description of protrusions/invadopodia involved in perineural invasion in PDAC.

2. The 3D gel invasions demonstrate more cells in the control ASO migrating towards the FBS media but there are also fewer cells in the ARLC4C ASO cells. Furthermore, there are more apoptotic bodies in the ARL4C ASO treatment cells. This suggests that the amount of siRNA/transfection agent used is toxic and interfering with migration and viability. An irrelevant binding control ASO should be used as well as a positive control ASO.

3. "ARL4C and IQGAP1 were shown to accumulate to membrane protrusions at endogenous level in S2-CP8 and PANC-1 cells" the inset of the cell chosen doesn't appear to be a membrane protrusion, it may appear more as a focal adhesion anchorage point of the cell as it moves in that direction or away from that point. The same could be said for Figure 2C (Supplement#2 for Figure 3).

4. The cell protrusions formed by cells in the microfluidics chamber are of a radial projection. Do the authors contend that the cell protrusions form regardless of direction? What is the purpose or effectiveness of this kind of protrusion formation radial as opposed to the side of the cell facing the FBS?

5. In Figure 6I, there is a PanII lesion (large) that has abundant IQGAP expression and a minor amount of ARL4C protein expression. However, there is minimal MMP14 expression, save for some puncta. This suggests that ARC4C recruitment to IQGAP does not necessarily lead to MMP14 co-localization. Hence, is MMP14 a more important factor in the proposed mechanism than ARL4C and IQGAP?

6. What is the function of the cortical compartmentalization of ARL4C and IQGAP? (signal that is on the sides of the cells rather than the focal adhesions/cell protrusions)

7. Is there any impact on siRNA KD of IQGAP/ARL4C on protrusion formation as analyzed in Figure 3L when analyzed on the cells shown in Figure 4B (which only shows accumulation of ARL4C at protrusions and not if there is a change in the total number of protrusions).

8. Many of the experiments rely on overexpression of MMP14-GFP. Are the same results observed (Figure 6) when de novo MMP14 levels are evaluated?

9. Figure 7H is curious to me. The ASO control lesion reveals no tumor cells within the LYVE-1 lumen. However, there is also a lack of tumor cells around this particular ROI. A more controlled analysis would evaluate all lymphatics that have the same amount of tumor cells surrounding it (human mitochondria stain) and to look for lack of cells within the lymphatic lumen if this is to be consistent with the ASO's proposed inhibitory activity on cell invasion and not cell proliferation or cytotoxicity.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Recruitment of KRAS downstream target ARL4C to membrane protrusions accelerates pancreatic cancer cell invasion" for further consideration by eLife. Your revised article has been evaluated by Erica Golemis (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

The study was revised and was improved; however, the conclusion that ARL4C recruits IQGAP1 and MMP14 to protrusions is still not adequately supported. Figures relating to these structures should define not only the presence of the protein of interest in protrusions but also the actin labeling and number of protrusions per cell. If these protrusions are defined by the absence of focal adhesions, then focal adhesion protein expression in these structures should also be shown. Alternatively, the focus in the manuscript on ARL4C regulation of these poorly defined membrane protrusions should be reduced and emphasis placed on regulation of cancer cell migration, invasion and invadopodia formation (on collagen) in addition to the in vivo data.

Reviewer #1:

Previous reviews were responded to well, with the addition of new data/and or more thorough descriptions of methodology and data.

I am satisfied with this revision.

Reviewer #2:

The manuscript is interesting and defines a role for ARL4C in pancreatic cell migration, invasion and tumor progression. I still however have serious issues about the data on membrane protrusions that is presented, and indeed the definition of membrane protrusions that I outline here:

In the abstract: ARL4C recruited IQGAP1 and its downstream effector, MMP14, to membrane protrusions. Specific localization of ARL4C, IQGAP1, and MMP14 was the active site of invasion, which induced degradation of the extracellular matrix.

From the text: Membrane protrusions were defined as actin-based structure of which length is longer than 10 μm and diameter is shorter than 10 μm. Also, cell protrusions were not stained with focal adhesion marker (Figure 2—figure supplement 1F).

What are these protrusions? Leading edge or trailing edge. Absence of focal adhesions and images from 2F and others showing focal adhesion and actin-rich lamellipodia at the opposing end of the cell suggests that they are trailing edge. If so how to reconcile with the regulation of invadopodia which are necessarily leading edge? While well-defined focal adhesions are not present in invadopodia, invadopodia still retain integrin-based matrix adhesions. Pseudopodia and lamellipodia in cancer cells plated on cover slips all contain focal adhesions. If the focal adhesion free membrane protrusions that the authors are studying are indeed protrusive structures analagous to the invadopodia they study in cells plated on collagen, then they must show this using live cell imaging. If they are protrusive structures, how to reconcile the absence of focal adhesions proteins with the extensive literature defining a role of focal adhesions/contacts/ matrix adhesions in pseudopodia protrusion and tumor cell migration and invasion? If these protrusions are retracting trailing edge structures, not analogous to invadopodia, what is their role in invasion and migration? Overall, these protrusions and their role in cell migration and invasion need to be better defined.

Specifically, protrusions shown in untreated WT have clear actin densities – but protrusions in treated cells lacking enrichment of a protein of interest (ARL6, IQGAP1, MMP-14 …) do not. This raises the question as to whether the targeted treatments that inhibit migration and invasion are also preventing formation of protrusions? Also, that quantification is based on number of "cells presenting protein enriched protrusions" is troubling. Do the various treatments alter the number of protrusions per cell? Do they alter the actin density of the protrusions as seems evident from some of the data presented. If so are the treatments altering the nature of the protrusion or ARL4C recruitment to the protrusions?

For example, PI3K is well known to be required for actin-dependent pseudopod protrusion – so the presentation that LY294002 prevents ARL4C and IQGAP1 accumulation at "protrusions" is not surprising. However, it raises serious questions as to what exactly are the protrusions that are being measured.

Without defining exactly what these membrane protrusions and a clear demonstration that the focal adhesion-free protrusions on glass are analagous to protrusive invadopodia, the idea that "Recruitment of KRAS downstream target ARL4C to membrane protrusions accelerates pancreatic cancer cell invasion" is interesting but not supported by the data presented and does not provide a clear mechanistic understanding of the role of ARL4C in cancer invasion. At best there is a correlation of association of ARL4C and as yet to be defined membrane protrusive structures.

Here is a list of protrusion data in the paper. Each of these figures should define not only the presence of the protein of interest in protrusions but also the actin labeling and number of protrusions per cell. If these protrusions are defined by the absence of focal adhesions then focal adhesion protein expression in these structures should also be shown. Alternatively, the focus in the manuscript on ARL4C regulation of these poorly defined membrane protrusions should be reduced and emphasis placed on regulation of cancer cell migration, invasion and invadopodia formation (on collagen) in addition to the in vivo data.

Figure 2K ARL4C ASO ◊ reduced number of collagen positive protrusions. What about actin protrusions?

Figure 2 Supp 1 The T27N ARL4C-GFP mutant is said not to accumulate at protrusions – but does it prevent protrusion formation? need to also quantify #actin protrusions

Supp1H,J – ARL4C ASO and KO reduce protrusions by about 5-10% but invasion by 80-90%

Figure 2 Supp 2C,E – ARL4C-GFP does not induce invadopodia in BxPC3 cells but does induce protrusions.

Figure 3M IQGAP siRNA reduces actin protrusions, even the increased number induced by OE of ARL4C-GFP

Figure 4B just look at ARL4C protrusions – are there actin-rich protrusions?

Figure 4E WT protrusions have clear actin densities – but others do not – based on what are these structures considered to be pseudopodia/ lamellipodia and not trailing edges of cells. There is a clear lack of actin densities in mock G2A DPBR cells and the protrusions selected in the WT and Mock cells are elongated protrusions at the opposite end of the cells from actin-rich lamellipodia and pseudopodia.

Figure 4 Supp 1B – ARL4C ASO reduces IQGP-1 accumulation to protrusions – what about #protrusions; what about actin accumulation in protrusions? In A looks as though ARL4SO is reducing actin accumulation in protrusions. Same in Figure 6 Supp 1C – mmp-14 down so is actin.

Figure 5 B – PI3K is well known to be required for actin-dependent pseudopod protrusion – so the presentation that LY294002 prevents Arl4C and IQGAP1 accumulation at "protrusions" is not surprising. However, it raises serious questions as to what exactly are the protrusions that are being measured. This has to be clarified – are they actin-rich? Are they leading or trailing edge?

Figure 5F, G again are there actin -rich protrusions under the treated conditions? Are the authors measuring recruitment of ARL6 and IQGAP1 to protrusions – or are they measuring that treatments are disrupting protrusions (i.e. pseudopodia and invadopodia)? The latter is more likely and consistent with their migration and invasion data.

Figure 6 B, C,D same as above for MMP-14.

Figure 6 Supp1G, H – WT vs ARL4C KO and WT vs ∆C show similar minimal trends and one is significant and one is not – is this supposed to be a real effect?

https://doi.org/10.7554/eLife.66721.sa1

Author response

Essential revisions:

A number of issues were raised that need to be addressed with experimentation, prior to publication. Specifically, studies that more concretely demonstrate that ARL4C recruits IQGAP1 and MMP14 to membrane protrusions are required. In addition, several controls would be needed throughout, to enhance to clinical and in vivo data sets. Specific requirements are as follows:

1) While ARL4C, IQGAP1 and MMP14 are located to protrusions, where they are clearly involved in matrix degradation, the loss of protein distribution to the protrusions (with ARL4C loss) could very well be due to the reduced abundance of protrusions associated with reduced migration and invasion of the cells. While a role for ARL4C in cell invasion is clearly demonstrated, the conclusion that ARL4C's "unique distribution to membrane protrusions is required for cancer cell invasion" is not supported by the data as presented.

This could be improved on by a variety of experiments. For example:

– All images of protrusions should include F-actin labeling to define protrusions. If the authors are arguing that ARL4C serves to recruit IQGAP1 and MMP14 to protrusions, then they need to show actin protrusions lack IQGAP1 and MMP14 when ARLC4 is knocked out.

We agree to the comment that membrane protrusions should be shown by F-actin labeling. ARL4C and IQGAP1 are involved in the regulation of actin cytoskeleton. Two dimensional (2D) experiments are suitable for intracellular localization of proteins. In this study we defined membrane protrusions as actin-based structure of which length is longer than 10 μm and diameter is shorter than 10 μm. ARL4C KO affected the structure of protrusions slightly and decreased numbers of cells with protrusions by about 10%. However, IQGAP1 and MMP14 disappeared in most of protrusions where ARL4C is knocked out. Moreover, in Figure 4E not only ARL4C KO cells but also all the KO cells expressing various ARL4C mutants show actin protrusions with or without IQGAP1 accumulation. The results are described in the text (page 6, lines 15 and 16, page 11, lines 12 and 13, page 13, lines 7 through 10) and shown in Figure 4E, Figure 4—figure supplement 1A and B, and Figure 6—figure supplement 1C and D.

– Studies that better demonstrate the role of protrusions in productive and/or directional migration are required. For example, the morphology of PDAC cells in the microfluidic assay suggests that the "cell protrusions" may in fact be focal adhesions/anchor points. Time-lapse imaging to show that these protrusions form via new offshoots perpendicular to the side of a cell versus a simple "zig-zag" motile pattern would be more convincing. Furthermore, time-lapse imaging of cells as they migrate towards the FBS media through the collagen matrix would enable one to determine if cell membrane protrusions are productive and have purposeful direction towards FBS. Hence, studies are needed that better characterize the direction and type of protrusions that are made.

Focal adhesions are defined as paxillin-containing, multi-protein structures that form mechanical links between intracellular actin bundles and the extracellular substrate. “Cell protrusions” that we pointed out throughout this study were not stained with anti-paxillin antibody. Paxillin was clearly observed in cell adhesion sites, which were different from cell protrusions. Therefore, they are unlikely to be focal adhesions. The results are described in the text (page 6, lines 16 and 17) and shown in Figure 2—figure supplement 1F.

Using time-lapse imaging we revealed that the protrusions are extended in the direction of cell invasion. We also calculated the angle of protrusions to the direction of cell movement towards FBS. Most of them were located in the angle of -45 to +45 degrees in the polar histogram. These results suggest that membrane protrusions play a role in purposeful direction. The results are described in the text (page 7, lines 15 through 18) and shown in Figure 2I and Figure 2-video 2.

– While ARLC4 knockouts are used for the localization studies, knock downs are used to study invasion. Both methods should be used for both assays.

According to the comments, we performed the localization assays with ARL4C knockdown and the invasion assays with ARL4C knockout. Depletion of ARL4C by either method showed the same phenotypes. The results are described in the text (page 7, lines 5 through 7, page 11, lines 13 and 14, page 13, lines 7 through 10) and shown in Figure 2—figure supplement 1J, Figure 4—figure supplement 1B, and Figure 6—figure supplement 1D.

– Please include a more thorough description of Mass Spec results so that the strength of the ARLC4-IQGAP1 interaction can be better appreciated.

We apologize for our poor description. In Supplementary File 1 Table 2-source data 1, the numbers of peptides recognized by mass spectrometry for each protein were described with available information.

– The 3D gel invasions demonstrate more cells in the control ASO migrating towards the FBS media, but shows there are also fewer cells in the ARLC4C ASO cells. Furthermore, there are more apoptotic bodies in the ARL4C ASO treatment cells. This suggests that the amount of siRNA/transfection agent used is toxic and interfering with migration and viability. An irrelevant binding control ASO should be used as well as a positive control ASO.

As a control ASO, we used randomized nucleotides for the sequence of ARL4C ASO. As already shown in Figure 2—figure supplement 1A. ARL4C ASO did not affect pancreatic cancer cell growth compared with control ASO. In addition, both ASOs did not induce cell death, which is assessed by propidium iodide (PI) staining. Thus, ASOs used in this study are not toxic for pancreatic cancer cells. The results are described in the text (page 6, lines 5 and 6) and shown in Figure 2—figure supplement 1C.

The reviewer criticized that there are few cells in ARL4C ASO-treated cells in the 3D gel invasion assay. At 0 time we placed the same numbers of cells treated with control and ARL4C ASOs in the starting position, and counted the numbers of invading cells after 72 h. Under the conditions, ARL4C ASO inhibited invasion ability of pancreatic cancer cells. The results are described in the text (page 7 lines 3 through 6) and shown in Figure 2—figure supplement 1I.

2) While ARLC4 shows promise as a target for PDAC, the preclinical studies should be improved, to strengthen correlations made and to ensure all controls are present in the in vivo assays. This could be improved on by a variety of experiments and/or edits. For example:

– Please include a table showing all patient information, as well as what the staining was for each patient.

According to the comment, all patients (57 cases) information was described in Supplementary File 1 Table 1-source data 1. This statement is described in the text (page 4, lines 18 through 20).

– The images shown in Figure 1 suggest that ARLC4 protein levels in cancer cells is rather binary; however, the expression data suggests that this is not the case. This should be addressed, with a more thorough scoring scheme as well as analysis of the extent to which protein levels correlate with transcript levels. In particular, this is important when setting the cutoffs for survival curves, as an example.

In Figure 1A and B, ARL4C expression was considered high when the total area of the tumor stained with anti-ARL4C antibody exceeded 5%. Under the definition, 82% of cases were judged as high expression. According to the comments, the ARL4C expression levels were scored as continuous variables based on the percentages of ARL4C-staining areas to total tumor areas. The detailed scores were described in Supplementary File 1 Table 1-source data 1.

It is hard to assess which protein levels correlate with transcription levels in public datasets. Therefore, dataset cases were separated into ARL4C high and low expression groups based on the top 75% of mRNA values of ARL4C, and 131 of 174 pancreatic cancer cases were classified as a high expression group, of which proportion was similar to that of immunohistochemical study for ARL4C. These statements are described in the text (page 4, lines 20 through 23, page 5, lines 7 through 9) and shown in Figure 1D.

– As presented in the manuscript, the clinical correlations made between IQGAP1 and ARL4C are not particularly well supported. The study would benefit from a more thorough correlational analysis as was shown for IQGAP1 and MMP14 and MMP14 and ARLC4 (supplemental 1).

As pointed out, the overall survival of PDAC patients who were double positive for ARL4C and IQGAP1 tended to be worse but not statistically significant (Figure 3—figure supplement 2J). According to the comment, the correlation of the expression between ARL4C and IQGAP1 was examined in the TCGA dataset. The dataset showed that expression of ARL4C mRNA in pancreatic cancer patients is positively correlated with that of IQGAP1 mRNA. The results are described in the text (page 10, lines 3 through 5) and shown in Figure 3K.

– For the in vivo experiments, it is important that the extent of ARL4C knock down is established at the level of protein and transcript. In addition, the gene expression alterations should be shown in a table format. The effect of ASO1316 on ARL4C levels should also be shown in Figure 2G.

The effects of ARL4C ASO on ARL4C levels in pancreatic cancer cells in vitro were already demonstrated in Figure 2—figure supplement 1B of the original manuscript. We have also reported that ARL4C ASO-1316 suppresses the levels of ARL4C mRNA and protein in hepatic cancer HLE cells and inhibits in vivo tumor formation induced by HLE cells (Harada, T. et al. Mol. Cancer Ther. 2018). Thus, the efficacy of ARL4C ASO-1316 was shown in different types of cancer cells.

The mRNA and protein levels of ARL4C in the pancreatic tumors obtained from the mice treated with control or ARL4C ASO were analyzed by immunohistochemistry, qPCR, and Western blotting. ARL4C ASO-1316 indeed reduced ARL4C expression in tumor lesions. The results are described in the text (page 14, lines 17 and 18, page 15, lines 1 and 2) and shown in Figure 7C, 7H, and Figure 7—figure supplement 1B.

The gene expression alterations in RNA-seq analysis based on log2 (fold change) > 1.0 or log2 (fold change) < -1.0 and FDR < 0.1, which was affected by ARL4C ASO treatment, was shown in Figure 7-data source 2.

– In Figure 7H, the ASO control lesion reveals no tumor cells within the LYVE-1-delineated lumen. However, there is also a lack of tumor cells around this particular ROI. A more controlled analysis would evaluate all lymphatics that have the same amount of tumor cells surrounding them, and to look for lack of cells within the lymphatic lumen, if this is to be consistent with ASO's proposed inhibitory activity on cell invasion (and not cell proliferation or cytotoxicity).

First of all, in the orthotopic transplantation model, lymphatic vessels of the host were located around the tumor mass implanted. According to the comment, we carefully checked many lymphatics and confirmed again that tumor cell numbers surrounding lymphatic vessels are decreased by ARL4C ASO. We believe this makes sense, because ARL4C ASO reduces the numbers of tumor cell that leave the primary site due to inhibition of cell invasive ability. Since ARL4C ASO did not affect primary tumor sizes, it is consistent with our hypothesis that ARL4C is involved in invasion but not in proliferation of pancreatic cancer cells. These statements are described in the text (page 15, lines 4 through 6). We also confirmed that ARL4C ASO does not induce cell death as shown in Figure 2—figure supplement 1C.

Reviewer #2:

Figure 2H. Not clear what the line graphs are showing.

We apologize for our insufficient description. Fluorescence intensities of ARL4C-tdTomato in the edges of cell protrusions and cytoplasm, which are indicated by white arrowheads and yellow closed circles (20 μm away from the tip of protrusion), respectively, were measured over time, and then the intensities were plotted as a function of time. The results indicate that ARL4C is dynamically appeared and disappeared in the protrusions, but it did not accumulate in the cytoplasm. The statements are described in the text (page 7, lines 11 through 15).

Figure 3I this is not clear.

We examined the IQGAP1 expression levels in pancreatic cancer immunohistochemically and found that 31 cases of 57 PDAC patients (54%) highly expressed IQGAP1 as shown in Figure 3I and Supplementary file 1 Table 3. The left image of original Figure 3I indicates that both IQGAP1 and ARL4C were expressed in the serial section. The right graph showed that IQGAP1 high expression cases are observed in 27 cases of 47 ARL4C high expression cases, and in 4 cases of 10 ARL4C low expression cases. However, these data do not show significant results, probably due to limited numbers of cases. Therefore, we removed the image showing ARL4C expression and showed IQGAP1 expression in tumor lesions and non-tumor regions in Figure 3I. Instead, we analyzed the relationship between ARL4C and IQGAP1 using public datasets. The statements are described in the text (page 10, lines 3 through 5) and the results are shown in Figure 3K.

Figure 3J. What is the relationship between low/high IQGAP1 and low /high ARL4C groups oin PDAC patient survival?

The overall survival of PDAC cases who highly expressed both ARL4C and IQGAP1 was not statistically significant compared with that of the cases with high ARL4C and low IQGAP1, probably due to the limited case numbers (Figure 3—figure supplement 2J). Although we examined only 57 PDAC cases, these are rare cases without the use of chemotherapy and benefit for immunohistochemically staining, because chemotherapy is usually administered in pancreatic cancer patients prior to operation. I would appreciate it if the reviewer could understand the reasons of the limited case numbers that we used.

Therefore, we analyzed the relationship between ARL4C and IQGAP1 expression on patient survival using public datasets. The results demonstrated that overall survival was significantly decreased in the order of low ARL4C/low IQGAP1, high ARL4C/low IQGAP1, and high ARL4C/high IQGAP1, although the result of low ARL4C/high IQGAP1 could not conclude because of the small sample size. The results are described in the text (page 10, lines 10 through 14) and shown in Figure 3—figure supplement 2K.

Figure 4. Show effect of ARL4C knockout on protrusions (E,F) but then resort to ASO to study invasion. What is the invasive capability of the KO cells? Same in Figure 5 G, H.

To respond to the reviewer’s comment, we showed that knockdown and knockout of ARL4C little affect the extension of membrane protrusions. The results are described in the text (page 6, lines 22 through 24) and shown in Figure 2—figure supplement 1H. Then we demonstrated that depletion of ARL4C by either method inhibits invasion ability. The results are described in the text (page 7, lines 5 through 7) and shown in Figure 2F and Figure 2—figure supplement 1J. In addition, we examined the effect of ARL4C knockout and knockdown on localization of IQGAP1 and MMP14 and found that depletion of ARL4C suppressed their localization to the tip of membrane protrusion. The results are described in the text (page 11, lines 12 through 14; page 13, lines 7 through 10) and shown in Figure 4—figure supplement 1A and B and Figure 6—figure supplement 1C and D.

Thus, we would like to emphasize that invasion ability and localization of IQGAP1 and MMP14 at the membrane protrusions are inhibited by knockdown and KO by ARL4C, but the structure of the protrusions is little changed.

Figure 6. Not at all clear what the images are showing or what the merge is showing.

We believe that the reviewer pointed out Figure 6I. In this experiment, we tried to show that ARL4C, IQGAP1, and MMP14 are simultaneously expressed in invading pancreatic cancer cells by staining the serial section of human specimen using triple immunofluorescence imaging assay. Let me firstly explain our interpretation of Figure 6I and Figure 6—figure supplement 1H of the original manuscript

In Figure 6I of the original manuscript, PanIN lesion was shown at the upper half of the figure. The cell in the yellow dashed boxes was an invasive PDAC cell. ARL4C and MMP14 were expressed more highly in invasive cancer cells rather than in PanIN lesions, although IQGAP1 was thoroughly expressed in tumor lesions including PanIN areas. Merged images on the right bottom indicated that invasive cancer cells express the three proteins simultaneously. Low-power images were shown in Figure 6—figure supplement 1H of the original manuscript, which demonstrates a group of cells invaded the surrounding interstitial tissues, and concurrently expressed three proteins. However, these statements were not well described.

Therefore, we replaced the figures in the revised manuscript and again emphasized that ARL4C, IQGAP1, and MMP14 are expressed together in invading cells. The results are described in the text (page 14, lines 3 through 8) and shown in Figure 6I and Figure 6—figure supplement 1J.

Reviewer #3:

Suggestions for additional experiments:

1. Time-lapse imaging of cells as they migrate towards the FBS media through the collagen matrix would be helpful. It's not clear if the cell membrane protrusions are productive and have purposeful direction towards FBS.

2. The use of patient-derived PDX material would have added further momentum towards this idea that interrupting the ARL4C-IQGAP1-MMP14 complex leads to decreased cell membrane protrusion formation.

3. The use of organoids to understand the impact of these transgenes/mutations on cell membrane protrusions would have been closer to what is observed pre-clinically and clinically.

We agree to the reviewer’s comment that the experiments with PDX materials and organoids would strengthen our model in vivo. However, since to use these materials is beyond the aim of this study, we would like to perform these experiments in the future.

4. PDAC has a tremendous amount of fibrosis and is not a heavily vascularized tumor. Understanding the efficacy of ASO extravasation into tumors and how it precludes lymphatic metastasis would be of broad interest.

Yes, as the reviewer said, fibrosis of pancreatic cancer blocks the transport of medicine from the vessels to cancer cells. Honestly, we have not yet completely understood how ASO reaches to the tumors after subcutaneous administration. Since the orthotopic implantation tumors do not have a large amount of fibrosis around tumors, it would be good to use the KPC (KrasG12D, p53mutant, and Cre) pancreatic cancer mouse model for evaluation of ARL4C ASO as future experiments.

Open-ended questions:

1. The microfluidics chamber to assess uni-directional invasion by PDAC cells is innovative but the morphology of these cells suggests that the "cell protrusions" observed resemble focal adhesions/anchor points. Time-lapse imaging to show that these protrusions form via new offshoots perpendicular to the side of a cell versus a simple "zig-zag" motile pattern would have been convincing.

2. Only 20% of cells portray these IQGAP1-ARL4C rich protrusions. What is the metastatic efficiency like if these cells are removed from the total pool of cells within the tumor? Would there still be metastatic colony formation?

As shown by time-lapse imaging in Figure 2-video 1 and Figure 2H, cell protrusions are dynamically formed. Therefore, there are some cells with protrusions and other cells without protrusions at one point. We demonstrated still images of cultured cells in most figures to show detailed localization of proteins clearly. We believe that most of cancer cells expressing ARL4C extend protrusions, to which ARL4C is localized, in tumors in vivo, and these cells could invade into the stroma and metastasize.

3. What are the cellular/biophysical barriers for PDAC cells as they intravasate into perineural space? Does this truly require breakdown of basement membrane or is another type of matrix/cellular barrier present? Such as myelin, fibroblasts, etc.?

Perineural invasion has a prevalence more than 70% of pancreatic cancer patients, and is associated with poor prognosis. However, the mechanisms underlying perineural invasion are poorly understood. Peripheral nerve structure consists of the endoneurium, perineurium, and epineurium. The endoneurium is the innermost layer and consists of nerve fibers, composed of axons surrounded by Schwann cells; it also contains blood vessels, resident macrophages and fibroblasts. The perineurium, which surrounds the endoneurium, is a layer of cylindrical cells tightly interconnected, forming a protective barrier. The epineurium is the outermost layer surrounding several nerve bundles; it includes an elastin and collagen sheath, blood and lymphatic vessels, resident macrophages, mast cells, and fibroblasts. Therefore, there would be several possible mechanisms that PDAC cells intravasate into perineural space. As the reviewer suggested, pancreatic cancer cells need to break epineurium and perineurium cells and degrades elastin, collagen, and myelin to damage endoperium. More studies are absolutely required for the understanding of whole picture of perineural invasion mechanisms. Since we do not have any data, I would like to refrain from discussing the mechanisms of perineural invasion.

4. What is the expression level of ARL4C in pre-PDAC lesions? Such as in PanIN2/3? What is/are the pioneer factors that induce ARL4C expression leading to PDAC? Hypoxia?

This is a great suggestion. According to the comment, we examined ARL4C expression in 26 cases of PanIN and found that 20 cases are positive for ARL4C, suggesting that ARL4C is expressed in precancerous stage of pancreatic cancer. The results are described in the text (page 5, lines 13 through 17) and shown in Figure 1—figure supplement 1C. These results are consistent with our recent observations that ARL4C is frequently expressed in atypical adenomatous hyperplasia, which is the possible precursor lesions and develops to lung adenocarcinoma Kimura, K. et al., Cancer Sci, 2020. Therefore, we need to examine the possibility of ARL4C expression of the early stage of cancer development more systematically.

So far, we have found that ARL4C is expressed downstream of the Wnt/β-catenin and Ras pathways in colon cancer, lung adenocarcinoma, and pancreatic cancer (Oncogene 2017, Oncotarget 2018, and this study). These pathways are activated by mutations of APC, β-catenin, and Ras. However, we also showed that ARL4C is expressed by de-methylation of 3’-UTR of the ARL4C gene in lung and oral squamous cell carcinoma (Oncotarget 2018). It would be interesting to search for another factor that induces ARL4C expression.

I have some concerns/comments regarding some of the findings and what else is missing.

1. Lian and Mulligan (Oncogene 2020) showed that "invasive processes" also contribute to perineural and neural invasion. These were driven by RET kinase activity and subsequent Src kinase activity. RET also needs to be analyzed in the IHC experiments since this is first published description of protrusions/invadopodia involved in perineural invasion in PDAC.

We are currently investigating the functional relationship between the ARL4C pathway and other signaling pathways. We would like to examine RET expression in PDAC in the next project.

2. The 3D gel invasions demonstrate more cells in the control ASO migrating towards the FBS media but there are also fewer cells in the ARLC4C ASO cells. Furthermore, there are more apoptotic bodies in the ARL4C ASO treatment cells. This suggests that the amount of siRNA/transfection agent used is toxic and interfering with migration and viability. An irrelevant binding control ASO should be used as well as a positive control ASO.

3. "ARL4C and IQGAP1 were shown to accumulate to membrane protrusions at endogenous level in S2-CP8 and PANC-1 cells" the inset of the cell chosen doesn't appear to be a membrane protrusion, it may appear more as a focal adhesion anchorage point of the cell as it moves in that direction or away from that point. The same could be said for Figure 2C (Supplement#2 for Figure 3).

4. The cell protrusions formed by cells in the microfluidics chamber are of a radial projection. Do the authors contend that the cell protrusions form regardless of direction? What is the purpose or effectiveness of this kind of protrusion formation radial as opposed to the side of the cell facing the FBS?

5. In Figure 6I, there is a PanII lesion (large) that has abundant IQGAP expression and a minor amount of ARL4C protein expression. However, there is minimal MMP14 expression, save for some puncta. This suggests that ARC4C recruitment to IQGAP does not necessarily lead to MMP14 co-localization. Hence, is MMP14 a more important factor in the proposed mechanism than ARL4C and IQGAP?

As pointed out by the reviewer, tumor lesions in Figure 6I of the original manuscript could be PanIN lesions. In this experiment, we tried to show that ARL4C, IQGAP1, and MMP14 are simultaneously expressed in invading pancreatic cancer cells by staining the serial section of human specimen using triple immunofluorescence imaging assay. Let me explain our interpretation. As suggested, PanIN lesion was shown at the upper half of the figure as indicated. The cell in the yellow dashed boxes was an invasive PDAC cell. ARL4C and MMP14 were expressed more highly in invasive cancer cells rather than in PanIN lesions, although IQGAP1 was thoroughly expressed in tumor lesions including PanIN areas. Merged images on the right bottom indicated that invasive cancer cells express the three proteins simultaneously. Low-power images were shown in Figure 6—figure supplement 1H of the original manuscript, which demonstrates a group of cells invaded the surrounding interstitial tissues, and concurrently expressed three proteins. However, these statements were not well described.

Therefore, we replaced the figures in the revised manuscript and again emphasized that ARL4C, IQGAP1, and MMP14 are expressed together in invading cells. The results are described in the text (page 14, lines 3 through 8) and shown in Figure 6I and Figure 6—figure supplement 1J.

6. What is the function of the cortical compartmentalization of ARL4C and IQGAP? (signal that is on the sides of the cells rather than the focal adhesions/cell protrusions)

ARL4C is localized to the plasma membrane through myristoylation. In addition, we found that the binding of PIP3 and polybasic region of ARL4C is required for the localization to cell protrusions, and PIP3 is enriched in protrusions rather than PIP2. Through the mechanism, IQGAP1 is recruited to cell protrusions. These are important findings in this project.

7. Is there any impact on siRNA KD of IQGAP/ARL4C on protrusion formation as analyzed in Figure 3L when analyzed on the cells shown in Figure 4B (which only shows accumulation of ARL4C at protrusions and not if there is a change in the total number of protrusions).

In this study we defined membrane protrusions as actin-based structure of which length is longer than 10 μm and diameter is shorter than 10 μm. Knockdown and KO of ARL4C affected the structure of protrusions slightly and decreased numbers of cells with protrusions by only about 10%, suggesting that ARL4C is not essential for extension of protrusions. The results are described in the text (page 6, lines 22 through 24) and shown in Figure 2—figure supplement 1H.

8. Many of the experiments rely on overexpression of MMP14-GFP. Are the same results observed (Figure 6) when de novo MMP14 levels are evaluated?

According to the comment, de novo MMP14 expression levels were examined. Endogenous MMP14 were predominantly present in the cytoplasm but hardly detected in the plasma membrane, because MMP14 is recycled between the plasma membrane and cytoplasm. The results are shown in Author response image 1. Therefore, we used MMP14-GFP to clearly show the localization of MMP14 to the plasma membrane.

Author response image 1
Validation of anti-MMP14 antibody.S2-CP8 cells were stained with anti-MMP14 antibody and phalloidin.

9. Figure 7H is curious to me. The ASO control lesion reveals no tumor cells within the LYVE-1 lumen. However, there is also a lack of tumor cells around this particular ROI. A more controlled analysis would evaluate all lymphatics that have the same amount of tumor cells surrounding it (human mitochondria stain) and to look for lack of cells within the lymphatic lumen if this is to be consistent with the ASO's proposed inhibitory activity on cell invasion and not cell proliferation or cytotoxicity.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #2:

What are these protrusions? Leading edge or trailing edge. Absence of focal adhesions and images from 2F and others showing focal adhesion and actin-rich lamellipodia at the opposing end of the cell suggests that they are trailing edge. If so how to reconcile with the regulation of invadopodia which are necessarily leading edge? While well-defined focal adhesions are not present in invadopodia, invadopodia still retain integrin-based matrix adhesions. Pseudopodia and lamellipodia in cancer cells plated on cover slips all contain focal adhesions. If the focal adhesion free membrane protrusions that the authors are studying are indeed protrusive structures analagous to the invadopodia they study in cells plated on collagen, then they must show this using live cell imaging. If they are protrusive structures, how to reconcile the absence of focal adhesions proteins with the extensive literature defining a role of focal adhesions/contacts/ matrix adhesions in pseudopodia protrusion and tumor cell migration and invasion? If these protrusions are retracting trailing edge structures, not analogous to invadopodia, what is their role in invasion and migration? Overall, these protrusions and their role in cell migration and invasion need to be better defined.

We appreciate Reviewer #2’s critique of our definition of “membrane protrusion”. We used the term “membrane protrusion” as the structure extending from cells. In the first revised manuscript (Figure 2—figure supplement 1G) we stated that paxillin, a core component of focal adhesion complex, is not detected in cell protrusions. According to the Reviewer’s critique, we carefully stained cells with anti-paxillin antibody repeatedly under the different blocking conditions and found that paxillin is detected in the structures (Figure 2—figure supplement 1F). In addition, other focal adhesion proteins such as phosphorylated paxillin, FAK, and phosphorylated FAK, and F-actin were also detected in the tips of the protrusions (Figure 2—figure supplement 1F). Therefore, the membrane protrusions that we defined are actin-based structures that contain the adhesion sites. These are described in the text (page 6, lines 18 through 22). We really thank to Reviewer #2’s appropriate comment.

Since the protrusions had focal adhesion sites and contacted the surrounding extracellular matrix, next we investigated the invasive properties of the protrusions. It is known that invadopodia are key structures for cancer cell invasion and have been extensively studied using some cells, including MDA-MB-231 (breast cancer cell) and SCC61 (head and neck squamous carcinoma cell). Therefore, we firstly explored invadopodia formation in pancreatic cancer cell lines. In my understanding, invadopodia are the unique structure which extend vertically from the cancer cell bottom to the extracellular matrix and the structure is detected on a gelatin-coated glass coverslip, using a well-known “invadopodia assay.” As shown in Figure 2—figure supplement 2A, invadopodia were detected in BxPC-3 but not S2-CP8 or PANC-1 cells. Therefore, we thought that pancreatic cancer cells (S2-CP8 and PANC-1 cells) that we used in this study do not form typical invadopodia from the ventral side of the cell body but can invade into extracellular matrix through other structures. This is the reason why we used the term “membrane protrusion” but not the term “invadopodia.” However, we found that invadopodia markers such as cortactin and ARPC2 localize to the tips of protrusions with ARL4C, suggesting that the protrusions might contribute to invasive phenotypes of pancreatic cancer cells and ARL4C functions there (Figure 2—figure supplement 2B).

It has been reported that cells lacking MMP14 display no defects in 2D proliferation or migration across collagen-coated surface, but their capacity to invade is severely impaired [J. Cell Biol. 167, 769, 2004]. The results resemble our findings obtained by ARL4C knockdown. In 3D gel invasion assay conditions, cancer cells clear their path using MMP14 degradative activity and the leading protrusions of invasive cancer cells have the capacity to degrade collagen fibers. Thus, the membrane protrusions that we defined are structurally and functionally similar to invadopodia in that ARL4C recruits MMP14 through IQGAP1 to the tip of the structures where the same component with invadopodia are localized and that cells with the protrusions can penetrate into the collagen gel in 3D collagen assay. Therefore, we referred to the protrusive structures as “invasive pseudopods.” Our results definitively show that ARL4C recruits MMP14 to the tips of invasive pseudopods to degrade the ECM. These are described in the text (page 7, lines 3 through 15).

In addition, cortactin and ARPC2 are also well-known marker for the leading edge of the cell (Nat. Rev. Mol. Cell Biol. 7, 713, 2006, Oncotarget, 7, 46142-46157, 2016). They localize to the tips of pseudopods and contribute to invasion of cells (Sci. Signal. 4, issue 159, pe6, 2011, J Cell Biol. 199, 527, 2012). “Invasive pseudopods” which we have defined show a clear localization of leading edge markers, indicating that the pseudopods are formed at the front of cells to keep cells move forward. This observation can also be confirmed in Figure 2I and Figure 2-video 2 and is the answer for Reviewer #2.

Specifically, protrusions shown in untreated WT have clear actin densities – but protrusions in treated cells lacking enrichment of a protein of interest (ARL6, IQGAP1, MMP-14 …) do not. This raises the question as to whether the targeted treatments that inhibit migration and invasion are also preventing formation of protrusions? Also, that quantification is based on number of "cells presenting protein enriched protrusions" is troubling. Do the various treatments alter the number of protrusions per cell? Do they alter the actin density of the protrusions as seems evident from some of the data presented. If so are the treatments altering the nature of the protrusion or ARL4C recruitment to the protrusions?

Another important critique of Reviewer #2 is whether ARL4C is involved in the formation of invasive pseudopods. As shown in experiments with depletion of ARL4C, the numbers of cells with invasive pseudopods were decreased slightly, while knockdown of ARPC2 which regulates formation of pseudopods as one of the components of Arp2/3 complex clearly reduced the number of pseudopods (Figure 2—figure supplement 2C-G). We tested whether ARL4C is involved in the presence of invadopodia markers or leading edge marker in the tips of pseudopods. As shown in Figure 2—figure supplement 2H and I, ARL4C knockout did not affect ARPC2 staining and reduced the staining of cortactin and F-actin only modestly. Although F-actin can be utilized to show the leading edge of cells, ARPC2 is also a good hallmark of the leading edge. This may be due to loss of IQGAP1 accumulation by ARL4C depletion, because IQGAP1 is functionally associated with actin assembly. Therefore, loss of ARL4C fairly affects the presence of invasive pseudopods nor the properties that they are the leading edge. Taken together, ARL4C may be necessary for functions of invasive pseudopods rather than their formation. It is quite likely that ARL4C contributes to invasive nature through other than pseudopod formation, and it is reasonable to compare the localization of protein of interest in the same pseudopodial structure between control and ARL4C KO cells. Considering other data in this study, major function of ARL4C in invasive pseudopod would be to recruit MMP14 by binding to IQGAP1. These are described in the text (page 7, lines 16 through 24, page 17, lines 2 through 6)

Reviewer #2 also pointed out the quantification method which is based on “cells presenting protein of interest enriched protrusions.” We have checked whether various treatments given to the cells alter the number of pseudopods per cell as the reviewer mentioned. We could not observe remarkable differences by the treatments (Figure 2—figure supplement 2C and D; Figure 2—figure supplement 2F and G; Figure 4—figure supplement 1B and C; Figure 4—figure supplement 1F and G). Reviewer #2 kindly indicated the possibility that LY294002 could affect formation of pseudopods. However, LY294002 as well as rapamycin was used for inducing PIP3 and PIP2 depletion and only incubated with cells for 30 min followed by fixation. For this reason, there were little effects on the structural formation of pseudopods. The details are described in the last part.

We have also measured the intensity of protein accumulated at the tips of invasive pseudopods, focusing only on cells which form pseudopod(s) (Figure 4—figure supplement 1D). The result was the same as the data using the previous method. However, the presence of pseudopods is dynamic as shown in Figure 2H and Figure 2-video 1 and 2, and the intensity of the tips of pseudopods in a static image does not always reflect the degree of function of protein of interest (POI). For example, low intensity of POI in the tips could indicate not only that it is rarely involved in the function of pseudopods, but also that pseudopod itself is in the retracting phase. Therefore, we decided to use the method in the original manuscript for quantification because it can show intuitively whether POI is localized to the tips and evaluate the whole picture of invasive pseudopods. We think the original method can reflect the invasive ability of the cells more precisely and decided to retain the data. As pointed out by Reviewer #2, this quantification is only reasonable on the assumption that various treatments do not alter the formation of the pseudopods and this was confirmed experimentally as we mentioned above.

We stated that overexpression of ARL4C in BxPC-3 cells induces the membrane protrusions in Figure 2—figure supplement 2E and F and Figure 3M in the first revised manuscript. In addition, since BxPC-3 cells barely express ARL4C, it may be difficult to conclude the action of ARL4C by its overexpression in the context of ARL4C-null cells. I am afraid that the results lead to misunderstanding for our conclusion and would like to study extensively the membrane protrusions induced by ARL4C in BxPC-3 cells in the future. Thus, we would like to remove the data using BxPC-3 cell (Figure 2—figure supplement 2B-F; Figure 3M; Figure 3—figure supplement 1D) of the first revised manuscript. We assume that the changes do not influence our conclusions and hope that the editor and Reviewer #2 could understand our explanation.

For example, PI3K is well known to be required for actin-dependent pseudopod protrusion – so the presentation that LY294002 prevents ARL4C and IQGAP1 accumulation at "protrusions" is not surprising. However, it raises serious questions as to what exactly are the protrusions that are being measured.

Without defining exactly what these membrane protrusions and a clear demonstration that the focal adhesion-free protrusions on glass are analagous to protrusive invadopodia, the idea that "Recruitment of KRAS downstream target ARL4C to membrane protrusions accelerates pancreatic cancer cell invasion" is interesting but not supported by the data presented and does not provide a clear mechanistic understanding of the role of ARL4C in cancer invasion. At best there is a correlation of association of ARL4C and as yet to be defined membrane protrusive structures.

Here is a list of protrusion data in the paper. Each of these figures should define not only the presence of the protein of interest in protrusions but also the actin labeling and number of protrusions per cell. If these protrusions are defined by the absence of focal adhesions then focal adhesion protein expression in these structures should also be shown. Alternatively, the focus in the manuscript on ARL4C regulation of these poorly defined membrane protrusions should be reduced and emphasis placed on regulation of cancer cell migration, invasion and invadopodia formation (on collagen) in addition to the in vivo data.

The Reviewer #2 commented that figures should define not only the presence of the protein of interest in protrusions but also the actin labeling and number of protrusions per cell. As described above, we showed that the membrane protrusions we defined in the original manuscript are invasive pseudopods containing focal adhesion sites and invadopodia markers or leading edge markers. In addition, we repeated the experiments with overexpression of ARL4C mutants and LY294002 experiments as follows:

When wild-type ARL4C-GFP and ARL4CQ27L-GFP were expressed in S2-CP8 cells, they were localized to the tips of pseudopods where cortactin and actin were present. However, ARL4CT27N-GFP was not detected in pseudopods although cortactin was observed in the remaining pseudopods. These results suggest that ARL4C is present in the tips of invasive pseudopods where it is expressed as wild type. The results are shown in Author response image 2.

Author response image 2
Effect of ARL4C mutants on pseudopod formation.

A,B,C, S2-CP8 cells were transfected with the indicated mutants of ARL4C-GFP and stained with the indicated antibodies (A).The percentages of cells with invasive pseudopods compared with the total number of cells were calculated (B). Cells were classified according to the number of pseudopods as indicated (C). A, The regions in the yellow dashed squares are shown enlarged in the left bottom images. The right bottom images are shown in a false color representation of fluorescence intensity. False color representations were color-coded on the spectrum. (B), Data are shown as the mean ± s.d. of 3 biological replicates. P values were calculated using one-way ANOVA followed by Bonferroni post hoc test. Scale bars in A, 10 μm. RFI, relative fluorescence intensity. n.s., not significant.

PIP3 is absolutely important for the determination of front-rear polarization of cells and LY294002 treatment would decrease pseudopod formation as the Reviewer #2 commented. In PIP depletion assay shown in Figure 5B, S2-CP8 cells were treated with only 30 min to examine the localization of ARL4C and IQGAP1. Under the conditions, cortactin, a marker of front of the cell, disappeared from the tips of pseudopods along with PIP3 depletion. Since cortactin associates with PIP3, this is not surprising, and serves as a positive control of this assay. Although cortactin lost its localization, protrusive structures still remained 30 min after treatment and it is enough to evaluate the presence of protein of interest. Therefore, we would like to conclude that ARL4C and IQGAP1 disappear from the tips of pseudopods when PIP3 is depleted. When cells were treated with LY294002 for 24 h to test invasive ability, they lost invasive pseudopods and invasive ability. The results are shown in Author response image 3 and described in the text (page 12, lines 17 through 22). We apologize for our insufficient explanation.

Author response image 3
Effect of LY294002 treatment on pseudopod formation.

A,B, S2-CP8 cells were treated with or without 50 µM LY294002 for 30 min or 24 h before fixation, and stained with the indicated antibodies (A).The percentages of cells with protrusive structures compared with the total number of cells were calculated (B). A, The regions in the yellow dashed squares are shown enlarged in the left bottom images. The right bottom images are shown in a false color representation of fluorescence intensity. False color representations were color-coded on the spectrum. (B), Data are shown as the mean ± s.d. of 3 biological replicates. P values were calculated using a two-tailed Student’s t-test. Scale bars in A, 10 μm. RFI, relative fluorescence intensity. n.s., not significant. **, P < 0.01.

Again, we really thank the reviewers for providing comprehensive and insightful comments that have helped strengthen our conclusions.

https://doi.org/10.7554/eLife.66721.sa2

Article and author information

Author details

  1. Akikazu Harada

    1. Department of Molecular Biology and Biochemistry, Graduate School of Medicine, Osaka University, Suita, Japan
    2. Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives (OTRI), Osaka University, 2-2 Yamadaoka, Suita, Japan
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Visualization, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8123-652X
  2. Shinji Matsumoto

    1. Department of Molecular Biology and Biochemistry, Graduate School of Medicine, Osaka University, Suita, Japan
    2. Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives (OTRI), Osaka University, 2-2 Yamadaoka, Suita, Japan
    Contribution
    Conceptualization, Investigation, Methodology, Project administration, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1804-264X
  3. Yoshiaki Yasumizu

    1. Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives (OTRI), Osaka University, 2-2 Yamadaoka, Suita, Japan
    2. Laboratory of Experimental Immunology, WPI Frontier Immunology Research Center, Osaka University, Suita, Japan
    Contribution
    Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9872-4909
  4. Kensaku Shojima

    1. Department of Molecular Biology and Biochemistry, Graduate School of Medicine, Osaka University, Suita, Japan
    2. Gene Expression Laboratory (GEL-B), Salk Institute for Biological Studies, San Diego, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  5. Toshiyuki Akama

    Department of Molecular Biology and Biochemistry, Graduate School of Medicine, Osaka University, Suita, Japan
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  6. Hidetoshi Eguchi

    Gastroenterological Surgery, Graduate School of Medicine, Osaka University, Suita, Japan
    Contribution
    Resources, Writing – review and editing
    Competing interests
    No competing interests declared
  7. Akira Kikuchi

    1. Department of Molecular Biology and Biochemistry, Graduate School of Medicine, Osaka University, Suita, Japan
    2. Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives (OTRI), Osaka University, 2-2 Yamadaoka, Suita, Japan
    Contribution
    Conceptualization, Funding acquisition, Project administration, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    akikuchi@molbiobc.med.osaka-u.ac.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3378-9522

Funding

Ministry of Education, Culture, Sports, Science and Technology (16H06374)

  • Akira Kikuchi

Ministry of Education, Culture, Sports, Science and Technology (18H04861)

  • Akira Kikuchi

Ministry of Education, Culture, Sports, Science and Technology (18H05101)

  • Akira Kikuchi

Yasuda Memorial Medical Foundation

  • Akira Kikuchi

Ichiro Kanehara Foundation for the Promotion of Medical Sciences and Medical Care

  • Akira Kikuchi

Osaka University

  • Akira Kikuchi

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the NGS core facility of the Genome Information Research Center at the Research Institute for Microbial Diseases of Osaka University for the data analysis support.

This study was supported by Eiji Oiki, Yuri Terao, and Center for Medical Research and Education, Graduate School of Medicine, Osaka University. Also this study was supported by Saki Ishino and Center of Medical Innovation and Translational Research, Osaka University.

Ethics

All protocols used for the animal experiments in this study were approved by the Animal Research Committee of Osaka University, Japan (No. 26-032-048). This information is mentioned in the 'Materials and Methods' section.

Senior Editor

  1. Erica A Golemis, Fox Chase Cancer Center, United States

Reviewing Editor

  1. Lynne-Marie Postovit, University of Alberta, Canada

Reviewers

  1. Ivan Robert Nabi, Life Sciences Institute, University of British Columbia Faculty of Medicine
  2. Hon Leong

Version history

  1. Received: January 20, 2021
  2. Preprint posted: February 24, 2021 (view preprint)
  3. Accepted: September 29, 2021
  4. Accepted Manuscript published: September 30, 2021 (version 1)
  5. Version of Record published: November 17, 2021 (version 2)

Copyright

© 2021, Harada et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,635
    Page views
  • 284
    Downloads
  • 6
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Akikazu Harada
  2. Shinji Matsumoto
  3. Yoshiaki Yasumizu
  4. Kensaku Shojima
  5. Toshiyuki Akama
  6. Hidetoshi Eguchi
  7. Akira Kikuchi
(2021)
Localization of KRAS downstream target ARL4C to invasive pseudopods accelerates pancreatic cancer cell invasion
eLife 10:e66721.
https://doi.org/10.7554/eLife.66721

Further reading

    1. Cancer Biology
    Gehad Youssef, Luke Gammon ... Adrian Biddle
    Research Article

    Cancer stem cells (CSCs) undergo epithelial-mesenchymal transition (EMT) to drive metastatic dissemination in experimental cancer models. However, tumour cells undergoing EMT have not been observed disseminating into the tissue surrounding human tumour specimens, leaving the relevance to human cancer uncertain. We have previously identified both EpCAM and CD24 as CSC markers that, alongside the mesenchymal marker Vimentin, identify EMT CSCs in human oral cancer cell lines. This afforded the opportunity to investigate whether the combination of these three markers can identify disseminating EMT CSCs in actual human tumours. Examining disseminating tumour cells in over 12,000 imaging fields from 74 human oral tumours, we see a significant enrichment of EpCAM, CD24 and Vimentin co-stained cells disseminating beyond the tumour body in metastatic specimens. Through training an artificial neural network, these predict metastasis with high accuracy (cross-validated accuracy of 87-89%). In this study, we have observed single disseminating EMT CSCs in human oral cancer specimens, and these are highly predictive of metastatic disease.

    1. Cancer Biology
    2. Medicine
    Dingyu Rao, Hua Lu ... Defa Huang
    Research Article

    Esophageal cancer (EC) is a fatal digestive disease with a poor prognosis and frequent lymphatic metastases. Nevertheless, reliable biomarkers for EC diagnosis are currently unavailable. Accordingly, we have performed a comparative proteomics analysis on cancer and paracancer tissue-derived exosomes from eight pairs of EC patients using label-free quantification proteomics profiling and have analyzed the differentially expressed proteins through bioinformatics. Furthermore, nano-flow cytometry (NanoFCM) was used to validate the candidate proteins from plasma-derived exosomes in 122 EC patients. Of the 803 differentially expressed proteins discovered in cancer and paracancer tissue-derived exosomes, 686 were up-regulated and 117 were down-regulated. Intercellular adhesion molecule-1 (CD54) was identified as an up-regulated candidate for further investigation, and its high expression in cancer tissues of EC patients was validated using immunohistochemistry, real-time quantitative PCR (RT-qPCR), and western blot analyses. In addition, plasma-derived exosome NanoFCM data from 122 EC patients concurred with our proteomic analysis. The receiver operating characteristic (ROC) analysis demonstrated that the AUC, sensitivity, and specificity values for CD54 were 0.702, 66.13%, and 71.31%, respectively, for EC diagnosis. Small interference (si)RNA was employed to silence the CD54 gene in EC cells. A series of assays, including cell counting kit-8, adhesion, wound healing, and Matrigel invasion, were performed to investigate EC viability, adhesive, migratory, and invasive abilities, respectively. The results showed that CD54 promoted EC proliferation, migration, and invasion. Collectively, tissue-derived exosomal proteomics strongly demonstrates that CD54 is a promising biomarker for EC diagnosis and a key molecule for EC development.