Prrx1 isoform switching regulates pancreatic cancer invasion and metastatic colonization

Abstract

The two major isoforms of the paired-related homeodomain transcription factor 1 (Prrx1), Prrx1a and Prrx1b, are involved in pancreatic development, pancreatitis, and carcinogenesis, although the biological role that these isoforms serve in the systemic dissemination of pancreatic ductal adenocarcinoma (PDAC) has not been investigated. An epithelial-mesenchymal transition (EMT) is believed to be important for primary tumor progression and dissemination, whereas a mesenchymal-epithelial transition (MET) appears crucial for metastatic colonization. Here, we describe novel roles for both isoforms in the metastatic cascade using complementary in vitro and in vivo models. Prrx1b promotes invasion, tumor dedifferentiation, and EMT. In contrast, Prrx1a stimulates metastatic outgrowth in the liver, tumor differentiation, and MET. We further demonstrate that the switch from Prrx1b to Prrx1a governs EMT plasticity in both mouse models of PDAC and human PDAC. Last, we identify hepatocyte growth factor ( HGF) as a novel transcriptional target of Prrx1b. Targeted therapy of HGF in combination with gemcitabine in a preclinical model of PDAC reduces primary tumor volume and eliminates metastatic disease. Overall, we provide new insights into the isoform-specific roles of Prrx1a and Prrx1b in primary PDAC formation, dissemination, and metastatic colonization, allowing for novel therapeutic strategies targeting EMT plasticity.

Keywords: EMT; MET; Prrx1a; Prrx1b; metastasis; pancreatic cancer.

Figure 1.

Prrx1a fosters MET, and Prrx1b promotes EMT in pancreatic cancer cells. (A) Morphological changes in KPC2 control, Prrx1a-overexpressing, and Prrx1b-overexpressing cancer cells and percentage of spheroid cyst formation in 3D culture. (*) P < 0.05. (B) Relative Cdh1 (E-cadherin) gene expression in KPC2 control, Prrx1a-overexpressing, and Prrx1b-overexpressing cells cultured in 3D. (*) P < 0.05. (C) E-CADHERIN (ECAD) protein expression by Western blot in KPC2 cells treated with Prrx1 isoform-specific siRNAs. Experiments were performed in triplicate, and band intensities were normalized by densitometry to β-ACTIN. (D,E) Invasion assay of control, Prrx1a-overexpressing, and Prrx1b-overexpressing KPC1 and KPC2 cells (D) and KPC1 cells treated with sicontrol, siPrrx1a, and siPrrx1b (E). (*) P < 0.05; (**) P < 0.005; (***) P < 0.0005. (F,G) Single-cell pancreatosphere formation assay with KPC1-overexpressing and KPC2-overexpressing cells (F) and knockdown KPC1 cells (G). (*) P < 0.05; (**) P < 0.005. (H) Soft agar colony formation assay with KPC1-overexpressing and KPC2-overexpressing cells. (*) P < 0.001. Values are shown as mean ± SEM. Bar, 100 μm.

Figure 2.

Differential expression of PRRX1A and PRRX1B in primary and metastatic pancreatic cancer. (A) Reactive ducts in primary mPDAC (referred to as KP^flCY tumors). (First panel) Hematoxylin and eosin (H&E) staining. The boxed area is magnified in the right panel. (Second panel) IHC staining for PRRX1A and PRRX1B in serial sections is shown. PRRX1A is expressed in the cytoplasm (white arrowheads) and nucleus (black arrowheads). PRRX1B expression is restricted to the nucleus (black arrowheads). (Third and fourth panels) Immunofluorescence (IF) staining for PRRX1A/B (red), E-CADHERIN (ECAD) (white), YFP (green), and DAPI (blue). Boxed regions highlight regions where PRRX1A is coexpressed with E-cadherin. PRRX1B-expressing cells in reactive ducts are devoid of E-cadherin (white arrowheads). (B) IHC staining for PRRX1A and PRRX1B in differentiated tumor areas. (C) IHC staining for PRRX1A and PRRX1B in poorly differentiated regions. (D, first panel) H&E staining of a representative small liver metastasis in a KP^flCY mouse. (Second and third panels) IHC staining for PRRX1A and PRRX1B and IF staining for PRRX1A, PRRX1B, E-CADHERIN, YFP, and DAPI in the serial sections of small liver metastases. Arrowheads indicate PRRX1A⁺ or PRRX1B⁺ cells (black or white). (E, first panel) H&E staining of a representative large liver metastasis. (Second panel) IHC for PRRX1A and PRRX1B. Positive cells are indicated by black arrowheads. (Third panel) IF staining for PRRX1A, PRRX1B, E-CADHERIN, YFP, and DAPI. PRRX1A and PRRX1B expression in large liver metastases. (F) Percentage of PRRX1A- or PRRX1B-positive cells in small or large liver metastatic foci. PRRX1A- or PRRX1B-positive cells were counted manually. Small and large metastatic foci were selected in a blinded fashion and photographed. Small, n = 6; large, n = 10. Total number of cancer cells per focus or field: PRRX1A (small, 21.5 ± 3.4 cells per foci; large, 99.9 ± 13.3 cells per field) and PRRX1B (small, 20.5 ± 3.3 cells per foci; large, 100.0 ± 7.6 cells per field; mean ± SEM). (*) P < 0.0001, Welch's t-test. Bar, 50 μm.

Figure 3.

Differential roles of PRRX1A and PRRX1B in tumor differentiation of primary pancreatic cancer. (A) Experimental design for orthotopic transplantation of KP^flCY control and Prrx1a-overexpressing and Prrx1b-overexpressing cells. (B, top panels) Representative primary tumor histology in experimental groups by H&E staining. Boxed regions are magnified in the bottom panels. (Second and third panels) IHC staining for PRRX1A and PRRX1B. (Bottom panels) Quadruple IF costaining of PRRX1A/B (red), E-CADHERIN (ECAD) (white), YFP (green), and DAPI (blue) in Prrx1a and Prrx1b tumors. White arrowheads indicate cancer cells devoid of E-cadherin at the invasive front of the Prrx1b tumor. (C) Primary tumor volume in three tumor groups. n = 8. Values of mean ± SEM are indicated below each box plot histogram. (D) IF staining for Ki-67 (red), YFP (green), and DAPI (blue) in three tumor groups. (Top panels) Representative pictures of the indicated tumors. (Bottom panels) Percentage of double Ki-67/YFP-positive cells. n = 6. (E) Number of circulating tumor cells (CTCs) in three tumor groups. The number of CTCs (mean ± SEM) is indicated below the box plot histograms. Bar, 50 μm.

Figure 4.

Activation of Prrx1a and suppression of Prrx1b foster metastatic colonization. (A) Experimental design for orthotopic transplantation using KPC2 control, Prrx1a-overexpressing, and Prrx1b-overexpressing cells. (B) Representative histology of liver metastases of KPC2 control, Prrx1a, and Prrx1b tumors. Bar, 100 μm. (C) Frequency of small/large liver metastases in three tumor groups. Small metastases contain <100 cancer cells, and large metastases contain >100 cancer cells. (D) Experimental design for intraportal vein injection (liver metastasis model) of KPC2 cells. (E, top and middle panels) Representative histology and IHC staining for E-CADHERIN (ECAD) in metastatic liver disease in three experimental groups. Bar, 100 μm. (Bottom panels) IHC staining for Ki-67 in three groups. Bar, 20 μm. (F) Frequency of small/large liver metastases in three groups. (G) Percentage of mice harboring large metastases. (H) Percentage of Ki-67-positive cells in each liver metastasis in three groups. n = 6 for each group. (I) Experimental design for orthotopic transplantation using KPC2-Prrx1a or KPC2-Prrx1b cells with or without doxycycline. (J) Representative histology of liver metastases. Bar, 100 μm. (K) Frequency of liver metastases in four experimental groups (on/off Prrx1a or Prrx1b). Doxycycline treatment was initiated 12 d after transplantation. (L) Frequency of large metastases in four experimental groups.

Figure 5.

Suppression of Prrx1a/b attenuates pancreatic cancer formation, hematogenous dissemination, and metastatic colonization. (A) Experimental design for orthotopic transplantation using KPC2 control and Prrx1a/b knockdown cells. (B) Primary tumor volumes of KPC2 control and shPrrx1a/b groups. (C) Box plot histograms indicate the number of CTCs between control and Prrx1a/b knockdown cells. (D) Representative pictures of liver metastases and the frequency of metastases in two groups. Mean ± SEM is indicated below the box plot histograms. Bar, 100 μm.

Figure 6.

Differential correlation of PRRX1A and PRRX1B isoforms with tumor differentiation status in human PDAC. (A) Representative IHC staining for PRRX1A and PRRX1B in well-differentiated, moderately differentiated, or poorly differentiated human PDAC (left panels) and correlation of PRRX1A (low/high) and PRRX1B (low/high) with the differentiation status (right panels). (B) Correlation of PRRX1A low/high and PRRX1B low/high groups in human primary PDAC. n = 100. (C) Primary pancreatic tumor volume of PRRX1A low/high and PRRX1B low/high groups (n = 100). (D, left panels) Representative IHC staining for PRRX1A and PRRX1B in pancreatic and metastatic liver tumors of the same PDAC patient. (Right panels) Correlation of PRRX1A low/high and PRRX1B low/high groups in primary PDAC and liver metastases. Bar, 50 μm.

Figure 7.

Prrx1b–Hgf signaling regulates pancreatic cancer cell invasion. (A) Prrx1a and Prrx1b target genes identified by ChIP-seq. Genes of enriched pathways (KEGG) are indicated. (B) Relative mHgf gene expression in control and Prrx1a-overexpressing and Prrx1b-overexpressing KP^flCY cells (left) and control and Prrx1a and Prrx1b knockdown KPC2 cells (right). (*) P < 0.05; mean ± standard deviation (SD). (C) Invasion assay in control and Prrx1a and Prrx1b knockdown cells using mrHGF. (D) Double IHC staining for PRRX1B (brown) and mHGF (blue) in KP^flCY pancreatic tumors (orthotopic transplantation model). Bars, 50 μm. (E) Quantification of secreted mHGF protein by ELISA. n = 3; mean ± SEM. (F) Secreted mHGF protein in the supernatant of control and Prrx1a/b knockdown KPC2 cells. n = 5; mean ± SEM.

Figure 8.

Cancer cell-derived HGF is required for PDAC and initiation of metastasis in vivo. (A) Experimental design of the randomized trial using Panc1 cells in an orthotopic transplantation model (early trial). Human IgG antibody (control), ficlatuzumab (Fic), gemcitabine (Gem), and the combination ficlatuzumab with gemcitabine (Fic+Gem) were administrated through intraperitoneal injection. (B) Representative histology of pancreatic tumors of IgG control and ficlatuzumab groups. (Top panel) H&E. (Middle panel) IHC for hHGF. (Bottom panel) IHC for E-CADHERIN (ECAD). Black arrowheads show positive staining for the indicated proteins. (C) Pancreatic tumor volumes are depicted for treatment groups in box plot histograms. Each volume represents the mean ± SEM. (D) Experimental design of the randomized trial using Panc1 cells in an orthotopic transplantation model (late trial). (E) Representative H&E staining of liver metastases from each treatment group. (F) Frequency of liver metastases in four treatment groups. (G) Number of liver metastases per mouse in four treatment groups. Each number presents the mean ± SEM. Bar, 100 μm.

Figure 9.

Model of the functional roles of Prrx1a and Prrx1b in pancreatic cancer progression and metastasis.