Pancreatic ductal adenocarcinoma induces neural injury that promotes a transcriptomic and functional repair signature by peripheral neuroglia

Abstract

Perineural invasion (PNI) is the phenomenon whereby cancer cells invade the space surrounding nerves. PNI occurs frequently in epithelial malignancies, but is especially characteristic of pancreatic ductal adenocarcinoma (PDAC). The presence of PNI portends an increased incidence of local recurrence, metastasis and poorer overall survival. While interactions between tumor cells and nerves have been investigated, the etiology and initiating cues for PNI development is not well understood. Here, we used digital spatial profiling to reveal changes in the transcriptome and to allow for a functional analysis of neural-supportive cell types present within the tumor-nerve microenvironment of PDAC during PNI. We found that hypertrophic tumor-associated nerves within PDAC express transcriptomic signals of nerve damage including programmed cell death, Schwann cell proliferation signaling pathways, as well as macrophage clearance of apoptotic cell debris by phagocytosis. Moreover, we identified that neural hypertrophic regions have increased local neuroglial cell proliferation which was tracked using EdU tumor labeling in KPC mice, as well as frequent TUNEL positivity, suggestive of a high turnover rate. Functional calcium imaging studies using human PDAC organotypic slices confirmed nerve bundles had neuronal activity, as well as contained NGFR+ cells with high sustained calcium levels, which are indicative of apoptosis. This study reveals a common gene expression pattern that characterizes solid tumor-induced damage to local nerves. These data provide new insights into the pathobiology of the tumor-nerve microenvironment during PDAC as well as other gastrointestinal cancers.

Figure 1.. Digital spatial profiling of the tumor-nerve microenvironment in KPC mice reveals transcriptomic upregulation of apoptotic processes and JUN signaling gene signatures.

(A) Immunofluorescence of a whole mount tissue section containing PDAC tumor tissue, adjacent acinar tissue and spleen. Nuclei (blue), F4/80 (green), PanCK (teal), tyrosine hydroxylase (TH; red). Scale bar (1 cm, bottom right). High magnification of boxed region highlighted in A showing TH+ axonal fibers, as well as TH+ bundles (red, right panel). (B) Representative scheme of Nanostring’s GeoMx digital spatial analysis pipeline. (C) Principal component analysis of comparing TH+ axonal fibers and larger caliber TH+ bundles. 9 total ROI’s from N = 5 independent bundles and N = 4 axonal fibers. (D) Hierarchical sample clustering based on differential gene expression matrix of transcription profile on individually isolated TH+ axons fibers and TH+ bundles. (E) Volcano plot and MA plot highlighting significant differential gene expression upregulated (red) and downregulated (blue) in TH+ bundles, as well as the average expression. (F) Gene ontology clustering of the most upregulated pathways based on differential gene expression analysis. (G) Highlighted gene expression changes in gene classes involved in GO processes including JUN kinase, apoptosis, SMAD/BMP signaling. Scale bar log2 fold change in expression.

Figure 2.. Nerve associated macrophage enriched regions express gene signatures involved in phagocytosis.

Image of TH+ nerve bundles selected for nanostring analysis immunostained for macrophages (F4/80+; green), epithelial cells (PanCK; white), nuclei (blue), tyrosine hydroxylase (TH; red) (A) without F4/80- and (B) with F4/80+ labeled cells. (C) Principal component analysis of comparing TH+ bundles and TH+ bundles containing F4/80+ cells. 10 total ROI’s from N = 5 independent bundles and N = 5 bundle + F4/80 ROI’s. (D) Heatmap of differentially expressed genes compared from 10 total ROI’s in F4/80 positive and F4/80 negative TH+ bundles in cell type gene clusters (macrophages, fibroblasts, epithelial). Scale bar represents log2 fold change. (E) Volcano plot and MA plot highlighting significant differential gene expression upregulated (red) and downregulated (blue) in TH+, F4/80+ bundles, as well as the average expression. (F) Gene ontology clustering and Kegg Pathways analysis of the most upregulated pathways based on differential gene expression analysis. (G) Highlighted gene expression changes in gene classes involved in GO processes including phagocytosis, chemokine ligands and receptors, and TAM genes.

Figure 3.. Proliferation of local neuroglia in the tumor nerve microenvironment.

(A) Immunofluorescence of KPC mice injected with EdU, sacrificed prior to any tumor development. Pancreas was stained for DAPI (blue), TH (red) and EdU (green). Images show representative regions with low presence of TH+ fibers and (B) TH+ fiber high regions of pancreas tissue. (D-E) Immunofluorescence of nerve bundles from PDAC tumors from KPC mice stained for DAPI (blue), TH (red), EdU (green) and PanCK (white). (E) High magnifications images for colocalization analysis of EdU+ cells within TH+ Bundles. (F) Quantification of the density of TH fibers in acinar regions and intra-tumoral regions from KPC mice. N = 9 regions from 3 KPC (G-H) Immunofluorescence of TH nerves, near PanCK epithelial cells (< 50 microns) or far (> 50 microns). Scale bar (50 microns). Unpaired students t-test, *** p-value < .001. (I) Quantification of the percentage of EdU+ percentage of EdU incorporation into cell nuclei within TH+ regions in tumor near PanCK+ or far from PanCK+ cell types. N = 9 total regions from 3 KPC mice stratified into 6 regions near PaNCK+ and 3 not near PanCK+ cell types. Unpaired students t-test was not significant.

Figure 4.. Neural injury and non-myelinating Schwann cell signatures identified in murine transcriptomic analysis are expressed in human PDAC and gastrointestinal tumors.

(A) Gene expression analysis of undefined Schwann cell markers, non-myelinating Schwann cells, and acinar genes from TH+ regions in tumor fibers, bundles and acinar regions. (B) Gene ontology analysis using WikiPathways of the most upregulated pathways based on differential gene expression analysis of TH+ fibers in tumor tissues and TH+ fibers in acinar tissue (C-D) Graphical analysis of tumor injury model and non-myelinating Schwann cell transcriptional activation during neural injury. (E-L) Histological analysis of images containing JUND protein signatures from the Human Protein Atlas (HPA) for multiple tumor types and their corresponding healthy organ. (E) normal pancreas (F) PDAC low magnification (G) PDAC high magnification (H) liver (I-J) liver cancer (K) stomach (L-M) stomach cancer (Dotted box denotes low magnification while dotted line denotes nerve contour. Scale bar (200 um) in figure E applies to F, H, I, L, Scale bar in figure G applies to J, M.

Figure 5.. Functional imaging of the tumor nerve microenvironment using organotypic tumor slices from human PDAC donor resections.

(A-C) Graphical representation of cutting and loading organotypic tumor slices with Fluo-4 Ca²⁺ imaging dye to produce live cell recordings. (D) Loaded PDAC tumor slices using in-situ cytolabeling to identify regions containing NGFR+ tumor bundles, CD11b+ immune cells, and Fluo-4 loaded cell types. Scale bar 200 microns (upper right). (E-H) Still frame high magnification images recordings of nerve bundles with (E) merged fluorescent channels and (F) split channel views of containing NGFR+ bundles, (G) CD11b+ immune cells, (H) Fluo-4 labeled cells. Scale bar 30 microns (bottom right). (I) Pseudocolor scale of nerve bundles (I) before and (J) after stimulation of nerve axons using KCl. Pseudocolor scale bar for Ca²⁺ levels shown in I (bottom left). (K) Raw traces of live cell recordings of an endoneural CD11b+ immune cell during KCl stimulation. (L-M) Still frame images of live recordings of spontaneous activity of cell activity within NGFR+ labeled nerve bundles. Scale bar in I applies to L and M. (N) Raw traces of live cell activity records from responding and non-responding cells. (O) Raw traces of cell types containing high sustained levels of calcium in NGFR+ cells, responding cells, and baseline Ca²⁺ levels. Graph shown as non-normalized calcium levels.