A Targetable Secreted Neural Protein Drives Pancreatic Cancer Metastatic Colonization and HIF1α Nuclear Retention

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is an increasingly diagnosed cancer that kills 90% of afflicted patients, with most patients receiving palliative chemotherapy. We identified neuronal pentraxin 1 (NPTX1) as a cancer-secreted protein that becomes overexpressed in human and murine PDAC cells during metastatic progression and identified adhesion molecule with Ig-like domain 2 (AMIGO2) as its receptor. Molecular, genetic, biochemical, and pharmacologic experiments revealed that secreted NPTX1 acts cell-autonomously on the AMIGO2 receptor to drive PDAC metastatic colonization of the liver-the primary site of PDAC metastasis. NPTX1-AMIGO2 signaling enhanced hypoxic growth and was critically required for hypoxia-inducible factor-1α (HIF1α) nuclear retention and function. NPTX1 is overexpressed in human PDAC tumors and upregulated in liver metastases. Therapeutic targeting of NPTX1 with a high-affinity monoclonal antibody substantially reduced PDAC liver metastatic colonization. We thus identify NPTX1-AMIGO2 as druggable critical upstream regulators of the HIF1α hypoxic response in PDAC. Significance: We identified the NPTX1-AMIGO2 axis as a regulatory mechanism upstream of HIF1α-driven hypoxia response that promotes PDAC liver metastasis. Therapeutic NPTX1 targeting outperformed a common chemotherapy regimen in inhibiting liver metastasis and suppressed primary tumor growth in preclinical models, revealing a novel therapeutic strategy targeting hypoxic response in PDAC.

Figure 1. NPTX1 is highly expressed in in vivo selected highly metastatic human and murine pancreatic cancer cells

(A) Schematic drawing of in vivo selection of pancreatic cancer cells with high liver metastatic capacity. (B) Liver metastasis assay in NSG mice injected intrasplenically with PANC1 parental cells or in vivo selected PANC1 LM3 cells (p=0.001, Student’s t-test, n=5 per group). (C) Liver metastasis assay in B6129SF1/J mice injected intrasplenically with KPC parental cells or in vivo selected KPC LM2 cells (p=0.002, Student’s t-test, n=4 per group). (D) Spontaneous liver metastasis assay in NSG mice injected intrapancreatically with PANC1 parental cells or in vivo selected PANC1 LM3 cells (p=0.037, Student’s t-test, n=5 per group). (E) Volcano plots revealing ex vivo transcriptomic profiles of PANC1 LM3/PANC1 and KPC LM2/KPC (n=3 per group).

Figure 2. NPTX1 drives human and murine PDAC liver metastasis

(A) Liver metastasis assay in NSG mice injected intrasplenically with PANC1 LM3 cells expressing either control guide RNA or NPTX1 targeting guide RNAs (_1 and _7). (p=0.003 in sgNPTX1_1, p<0.0001 in sgNPTX1_7, Student’s t-test, n=5 per group). (B) Kaplan-Meier curve comparing the overall survival of mice bearing liver metastasis from PANC1 LM3 cells expressing either control guide or NPTX1 targeting guide (p=0.027, log-rank test, n=5 per group). (C) Relative guide RNA abundance from CRISPR competition liver metastasis assay (p=0.023, Student’s t-test, n=5 per group). (D) Spontaneous liver metastasis assay (p=0.037, Student’s t-test n=5 per group). (E) Liver metastasis assay in NSG mice injected intrasplenically with PANC1 LM3 cells expressing either control hairpin or dox-inducible NPTX1 targeted hairpin. Doxycycline at 200mg/kg body weight or control chow were started on day 14 after the injection of cancer cells (p<0.0001, Student’s t-test, n=5 per group).

Figure 3. NPTX1 is over-expressed in primary PDAC tumors and liver metastases

(A) Relative NPTX1 mRNA abundance in normal pancreas and PDAC samples (p<0.0001, Student’s t-test). (B) NPTX1 ELISA in the serum of PDAC bearing mice (p=0.003, Student’s t-test). (C) NPTX1 ELISA in the serum of PDAC patients (p=0.015, Student’s t-test). (D) Representative images of immunohistochemistry staining of NPTX1 of human liver metastatic foci or primary PDAC tumors (p=0.025, Student’s t-test*). (E) Representative gross liver images and their H&E staining of liver metastasizing PC95 PDAC PDO expressing either control hairpin or NPTX1 targeting hairpins (_4 and _5) and the number of metastatic foci in the livers harboring PC95 PDO expressing either control or NPTX1 targeting hairpins (_4 and _5) (p=0.0031 and 0.0030 for shCTRL vs sh_4 and sh_5, Student’s t-test). (F) In vivo liver metastasis CRISPR competition assay in PC95 PDAC PDX/O (p=0.027, Student’s t-test). (G) Kaplan-Meier curve to estimate disease free survival of PDAC patients from TCGA-PAAD dataset with the median cut off for NPTX1 expression (n=185 in total, p=0.023, log-rank test). *single tailed t-test as the hypothesis was whether the liver tumors’s mean IHC score was higher than that of the primary tumors.

Figure 4. NPTX1 upregulation promotes cell growth under hypoxia

(A) Cell growth assay using PANC1 LM3 cells expressing either control or NPTX1 targeting guides (_1 and _7) under normoxia (p=0.202 and 0.136 for sgCTRL vs sg_1 and vs sg_7, respectively, Student’s t-test). (B) DEVD luciferin-based apoptosis assay of metastasizing PDAC cells expressing either control hairpin or NPTX1 targeting hairpin (_3) (p=0.521 and 0.364 for shCTRL vs shNPTX1_3 on day 14 and 21 respectively, Student’s t-test, n=5 per group). (C) Ki-67 immunofluorescence staining of metastasizing PDAC cells expressing either control or NPTX1 hairpin. (p<0.0001, Student’s t-test). (D) Comparing the growth of PANC1 and PANC1 LM3 cells under either normoxia or hypoxia (0.5% oxygen) (Under normoxia, (p=0.06, Student’s t-test), under hypoxia (1.6 fold increase, p=0.004, Student’s t-test)). (E) Western blot of the conditioned media harvested either from PANC1 parental cells or PANC1 LM3 cells. (F) Western blot of PANC1 or PANC1 LM3 whole cell lysates harvested either under normoxia or hypoxia (0.5% oxygen). NPTX1 expression was induced upon the exposure to hypoxia in both PANC1 and PANC1 LM3 cells (5.7 fold increase in LM3 cells and 5.8 fold increase in PANC1 cells, p<0.0001 and p=0.0051 respectively, Student’s t-test). (G) Comparing the growth of LM3 cells derived or PANC1 cells derived conditioned media treated PANC1 cells under hypoxia (0.5% oxygen) (p=0.0013, Student’s t-test). (H) Comparing the growth of PANC1 cells treated with recombinant NPTX1 at the increasing concentrations where 1x rNPTX1 is 1ng/ml* (p=0.0005, Student’s t-test). (I) Comparing the growth of PANC1 LM3 cells expressing either control guide or NPTX1 targeting guide with or without rNPTX1 treatment (p=0.002, ctrl vs rNPTX1 in sgCTRL cells and p=0.009, ctrl vs rNPTX1 in sgNPTX1 cells respectively, Student’s t-test). *: 1x concentration was determined by ELISA experiments in PDAC mice and patient samples (Figure 3BC).

Figure 5. AMIGO2 is a cell surface NPTX1 receptor and drives PDAC liver metastasis

(A) Volcano plot comparing the transcriptome of ex vivo PANC1 LM3 liver metastasizing tumors expressing either control hairpin or NPTX1 targeting hairpin. (B) Co-immunoprecipitation under hypoxia (IP) IP: FLAG-AMIGO2 Immunoblot: anti NPTX1 and anti AMIGO2 (C) Proximity ligation assay using PANC1 LM3 cells treated either his-tag control protein or his-tagged rNPTX1. (D) Normalized fluorescence (ΔFnorm) plot from MST assay using rEGFR or rAMIGO2. (E) Hypoxia cell growth assay in PANC1 LM3 cells expressing either control hairpin or AMIGO2 targeting hairpins (_2 and _3) (p=0.002 and p<0.0001 in shCTRL vs sh_2 and shCTRL vs sh_3, Student’s t-test, Bonferroni correction). (F) Hypoxia cell growth assay in PANC1 LM3 cells expressing either control hairpin or AMIGO2 targeting hairpins with/without rNPTX1 treatment (p=0.002, p=0.95, and p=0.41 Ctrl vs rNPTX1 in shCTRL, sh_2, and sh_3 respectively). (G-H) Liver metastasis assay using PANC1 LM3 cells expressing either control hairpin or AMIGO2 targeting hairpins (_2 in (G), _3 in (H)) (p<0.0001 for both shAMIGO2_2 and _3, Student’s t-test, n=5 per group).

Figure 6. AMIGO2 mediates HIF1a nuclear retention via specific HIF1a residues

(A) Western blot of HIF1a protein in PANC1 LM3 cells expressing either control hairpin or AMIGO2 hairpin fractionated into the cytoplasmic (cyt) and the nuclear (nuc) fractions. The first column is from control whole cell lysate sample under normoxia. (B) Immunofluorescence staining of HIF1a using in vitro PANC1 LM3 cells expressing either control hairpin or NPTX1 targeting hairpin under hypoxia (0.5% O₂). (C) Hypoxia dual luciferase reporter assay using PANC1 LM3 cells expressing either control hairpin or AMIGO2 targeting hairpins (_2 and _3) under hypoxia (0.5% oxygen) (p<0.0001 for both shCTRL vs _2 and _3, Mann-Whitney test). (D) Volcano plot comparing expression level of known HIF1a target genes (84 genes in total) in shCTRL and shAMIGO2 PANC1 LM3 ex vivo tumors (blue dots: negative log2 fold change and p<0.05 and red dots: positive log2 fold change and p<0.05). (E) Hypoxia dual luciferase reporter assay using shCTRL and shAMIGO2 PANC1 LM3 cells over-expressing either empty vector, full-length HIF1a, or the phosphomimetics (S641, S643, or both) (p<0.0001, Mann-Whitney test). (F) Immunofluorescence staining of HIF1a using in vitro cultured shCTRL or shAMIGO2 PANC1 LM3 cells expressing either empty vector or the dual phosphomimetics under hypoxia (0.5% O₂). (G) Liver metastasis assay using shAMIGO2 PANC1 LM3 cells expressing either empty vector, full-length HIF1a, or the HIF1a dual phosphomimetics (p=0.026 in EV vs plxS641E/S643E, Student’s t-test, n=4 per group).

Figure 7. Therapeutic targeting of NPTX1-AMIGO2 axis in primary and metastatic PDAC

(A) Hypoxia cell viability assay (MTS assay) using PANC1 LM3 cells treated with isotype control antibodies or NPTX1 monoclonal antibodies at 50ug/ml (p<0.0001, Mann-Whitney test) (0.5% O₂). (B) Liver metastasis assay using PANC1 LM3 cells treated with either 31B01-NPTX1 antibody (murine IgG1 isotype) at 10mg/kg twice a week I.P., gemcitabine 100mg/kg weekly I.P. or both (p=0.028 in PBS/IgG vs PBS/31B01, Student’s t-test, n=5 per group. The antibody treatment was started 1 hour after tumor cell injection.). The representative liver sections with H&E staining are shown (p=0.0004 in PBS/IgG vs PBS/31B01, Student’s t-test). (C) Orthotopic pancreas tumor growth assay using in vivo selected PC69P PDX/O treated either isotype control antibody of NPTX1-31B01 antibody (IgG1 isotype) at 10mg/kg twice a week I.P. (p=0.003, Student’s t-test, n=5 for control group and n=4 for 31B01 group. The antibody treatment was started on day 7 after confirming the establishment of primary pancreatic tumors). (D) Subcutaneous pancreas tumor growth assay using Mia PaCa-2 cells treated with either isotype control antibody of humanized 31B01-NPTX1 antibody (IgG isotype) at 10mg/kg twice a week I.P., humanized 31B01-NPTX1 antibody at 10mg/kg twice a week I.P., or gemcitabine 60 mg/kg and nab-paclitaxel 30mg/kg twice a week I.P. (p=0.9012, Student’s t-test, n=4 for control group, n=4 for humanized 31B01 group, and n=4 for gemcitabine/n-paclitaxel group. The antibody treatment was started after tumor volume reached 50 mm³). (E) Liver metastasis assay using BxPC-3 LM3 cells treated with either isotype control antibody of humanized 31B01-NPTX1 antibody (IgG isotype) at 10mg/kg twice a week I.P., humanized 31B01-NPTX1 antibody at 10mg/kg twice a week I.P. The representative gross liver images are shown (p=0.0095, Student’s t-test, n=5 for control group and n=5 for humanized 31B01 group. The antibody treatment was started 1 hour after tumor cell injection). (F) Orthotopic pancreas tumor growth assay using Mia PaCa-2 cells treated with either isotype control antibody of humanized 31B01-NPTX1 antibody (IgG isotype) at 10mg/kg twice a week I.P., humanized 31B01-NPTX1 antibody at 10mg/kg twice a week I.P. (p=0.0151, Student’s t-test, n=4 for control group and n=4 for humanized 31B01 group. The antibody treatment was started on day 7 after confirming the establishment of primary pancreatic tumors). The representative gross liver images are shown (p=0.0158 in PBS/IgG vs PBS/humanized 31B01, Student’s t-test). (G) Schematic representation of the proposing model