Human Immunocompetent Model of Neuroendocrine Liver Metastases Recapitulates Patient-Specific Tumour Microenvironment

Abstract

Neuroendocrine liver metastases (LM-NEN) develop in a considerable proportion of patients with gastroenteropancreatic neuroendocrine neoplasms. There is a paucity of experimental models that accurately recapitulate this complex metastatic human liver microenvironment precluding scientific and clinical advancements. Here, we describe the development of a novel personalised immunocompetent precision cut tumour slice (PCTS) model for LM-NEN using resected human liver tissue. The histological assessment throughout the culture demonstrated that slices maintain viability for at least 7 days and retain the cellular heterogeneity of the original tumour. Essential clinical features, such as patient-specific histoarchitecture, tumour grade, neuroendocrine differentiation and metabolic capacity, are preserved in the slices. The PCTS also replicate the tumor-specific immunological profile as shown by the innate and adaptive immunity markers analysis. Furthermore, the study of soluble immune checkpoint receptors in the culture supernatants proves that these immunomodulators are actively produced by LM-NEN and suggests that this process is epithelium-dependent. This model can be employed to investigate these pathways and provides a powerful platform for mechanistic, immunological and pre-clinical studies.

Keywords: ex vivo model; immune checkpoint receptor; neuroendocrine liver metastases; soluble immunomodulators; tissue slices; tumour modeling.

Figure 1

Precision cut tissue slices from LM-NEN require carbogen for tissue preservation beyond 3 days. (A) Schematic overview of experimental design. (B) Representative images of H&E staining of LM-NEN PCTS cultured in carbogen or atmospheric oxygen (n_patients=3). Tumour stroma is highlighted by blue arrows. Areas of necrosis (green arrows) and apoptosis (black arrows) are magnified. (C) PCTS weight (in mg) was measured daily, each point represents a different patient and is the average of 3 technical replicates (slices), (n_patients indicated; *P ≤ 0.05, **P ≤ 0.01). (D) Normalised lactate dehydrogenase release (LDH) detected in PCTS culture supernatant. Each point is the mean ± SD of 3 supernatants per patient (n_patients indicated). Created with BioRender.com.

Figure 2

LM-NEN PCTS cultured in carbogen are viable for up to 15 days. (A) Tissue integrity was evaluated by measuring the slice weight over the duration of the culture period (at day 0, 1, 2, 3, 5, 7 & 15). Each point represents the weight of a single slice. 3 replicates per patient per timepoint are displayed as mean ± SD. Ns, P>0.05. (B) Weight-adjusted LDH leakage in PCTS supernatant throughout the culture was plotted as a percentage of the total LDH present per mg of tissue (determined separately for each patient). (C) Caspase-cleaved cytokeratin 18 (cCK18) indicative of apoptotic specific cell death is shown as a percentage of total cell death (full form cytokeratin 18, CK18). Datapoints in (B, C) represent individual patients and are an average of 3 replicates per patient. Bars indicate mean ± SD of 4 patients (D0-D8) or 1 patient (D15). (D) Overview of HPLC quantification of ATP, ADP and AMP and formula utilised to calculate the energy charge. Retention times for ATP, ADP and AMP (consistent across all runs) are indicated in the graph. (E, F) Intracellular energy charge and weight adjusted intracellular ATP levels in LM-NEN PCTS at indicated timepoints for patients with viable tumour epithelium (045, 051 and 077). Each datapoint represents a single slice (*P ≤ 0.05). (G) Representative images showing apoptotic nuclei stained by TUNEL (green, indicated by white arrows) in tissue slices generated from 3 patients; 3 slices (D0-D7) or 1 slice (D15) were analysed per patient at each indicated timepoint. (H) TUNEL positive nuclei were quanitfied ans presented as %Apoptotic cells of total number of cells per image. Data are shown as mean ± SEM of 3 patients and each dot represents a single patient.

Figure 3

LM-NEN PCTS retain patient specific histoarchitecture in culture. Representative H&E images of tissue slices generated from 5 patients; 3 slices (D0-D7) or 1 slice (D15) were analysed per patient at each indicated timepoint. Tumour epithelium is indicated by red arrows and tumour stroma by blue arrows. Histoarchitecture was compared with clinical histopathology staining (left column) for each patient.

Figure 4

LM-NEN PCTS retain key molecular features associated with tumour grade and neuroendocrine differentiation. (A) The areas of tumour epithelium and stroma (highlighted in red) were measured in the H&E images of each patient throughout the culture using ImageJ as indicated on the right panel (n_patients=5) and the epithelium percentage is shown for each patient per timepoint. Patients 062 & 106 lacked tumour epithelium hence % = 0. (B) Chromogranin A (CgA) staining in brown or green and DAPI (purple) on epithelial LM-NEN tissue slices at indicated timepoints. (C) Representative images for Ki67 staining (positive nuclei in green or brown) and DAPI (in purple) at indicated timepoints. Areas of tumour and stroma are indicated with T and S, respectively. Images in (B, C) at D0-D7 are representative of 3 technical replicates per patient, for D15, 1 replicate. (D) Ki67 was quantified in 3 images per patient and presented as %Ki67 of total number of cells per image. Data are shown as mean ± SD of 3 patients and each dot represents a single patient.

Figure 5

LM-NEN PCTS retain innate and adaptive immunity markers and release soluble checkpoint receptors (solCRs) in the supernatant. (A) Schematic representation of the experimental design. Plasma samples were collected from LM-NEN patients (n=28) and healthy controls (n=17). Additionally, PCTS were generated from patients with LM-NEN (n=5) and culture supernatants collected at day 1, 2 and 3. solCRs were quantified in plasma and supernatants using Luminex. In parallel, gene expression analysis of innate and adaptive immunity markers was performed at day 1. Created with BioRender.com. (B) Heatmap showing the differential gene expression of immune cell markers in tumour versus surrounding tumour-free liver tissue in slices derived from patients with LM-NEN (n=4) and primary liver cancer with a prevalence of stroma, stroma (n=1). (C) Levels of solCRs in LM-NEN plasma (black box and whiskers plots) and slice supernatants at day 1, 2 and 3 (blue line graphs, mean ± SD of 3 replicates per patient). The supernatant was refreshed daily hence the levels are always displayed from y=0.

Figure 6

The release of soluble checkpoint receptors (solCRs) by LM-NEN PCTS is dependent on the presence of tumour epithelium. (A) Histological assessment of the content of stroma and epithelium in LM-NEN PCTS at day 3 or 5 in culture (see Figure S3 for images related to HCC with prevalent stroma). (B) solCRs quantified in supernatants at day 3 in culture derived from LM-NEN slices with prevalent epithelium (n_patients=3) and slices with prevalent stroma (n_patients=4, 2 LM-NEN + 2 primary liver cancer). Each datapoint indicates the average of 3 technical replicates (slice supernatants) per patient. LM-NEN in blue and grey, stroma primary liver cancer in red. P values following the comparison of stroma slices vs epithelial LM-NEN slices are indicated in graphs.