Neural Regulation of Colorectal Cancer: Implications for Immunomodulation

Abstract

Colorectal cancer (CRC) is still the third most common cancer and the second deadliest cancer worldwide with more than 930,000 deaths each year. The burden of CRC has been estimated to rapidly increase to 3.2 million new cases and 1.6 million deaths per year by 2040. Therefore, advancements in early detection and more accurate patient stratification are critical to enable personalized treatment approaches and improve therapeutic outcomes. CRC develops from epithelial cells in the mucosal layer of the colon or rectum. The progression from normal epithelium to polyp, adenoma and eventually carcinoma is characterized by the accumulation of genetic and epigenetic changes. Next to these factors intrinsic to cancer cells, the tumor microenvironment (TME), composed of a variety of cell types, such as endothelial cells, fibroblasts, and immune cells, and components of the extracellular matrix, engages in a complex crosstalk to drive CRC initiation, progression and therapy resistance. Neural cells are one of the latest identified members of the TME, and the impact of neurons and glia is only beginning to be appreciated in the new research field of cancer neuroscience. The colon is highly innervated both extrinsically by parasympathetic and sympathetic nerves and intrinsically by the enteric nervous system (ENS). Together, the extrinsic and intrinsic innervation regulate gastrointestinal homeostasis and function by controlling gut motility, mucosal secretion and absorption, and local blood flow through interactions with other cell types in the gut.

Although the interest in CRC neuroscience is growing, most studies remain limited in scope, not providing mechanistic understanding of the impact of neural signaling on the tumor, or not investigating the potential contribution of other TME components. As a result, the cellular and molecular mechanisms through which specific neural cells influence CRC development and interact with other components of the TME remain poorly understood. Gaining deeper insight into these interactions is essential for understanding disease progression and identifying novel therapeutic targets. The aim of this PhD thesis was to explore the phenotype, prognostic value and functional role of neurons and glial cells in colorectal carcinogenesis and the colorectal TME.

In Chapter 2, we provide an overview of the literature regarding the emerging role of neurons and glia in CRC, discussing both extrinsic and intrinsic innervation of the colon and their influence on carcinogenesis and cancer therapeutics. Key questions that were identified for future studies include: What are the molecular mechanisms involved in the communication between neural cells and CRC cells? Does neural–cancer crosstalk occur directly and/or through interplay with other TME components? Which therapeutic options can be considered to intervene in this crosstalk? To explore the direct communication between neural cells and CRC cells, we studied the effect of neural cues on patient-derived CRC cell lines, as presented in Chapter 3. Using different neural signals and various CRC lines, we investigated which neural cells might affect colorectal carcinogenesis and which subtypes of CRC patients are susceptible to such cues. Most neural signals increased the colony formation ability of CRC cells. The enhanced clonogenicity after stimulation with the muscarinic activator Bethanechol was influenced by BRAF mutation status. On the contrary, epinephrine treatment resulted in a pronounced decrease in clonogenicity and CRC cell viability, which was not dependent on genomic status, whereas vasoactive intestinal peptide only decreased cell viability in BRAF wildtype cells. All neural signals tended to induce migration in CRC cells, but the effects were dependent on the genomic characteristics, as migration seemed increased for MSS cells and for BRAF mutated cells. These results emphasize the importance of targeting specific neural cells and their signals and underscore the need for patient stratification in all cancer neuroscience studies.

To study the effect of neuronal innervation on CRC initiation and progression, we used mouse models with increased and decreased neuronal density. In Chapter 4, we state our experiments aimed to use the previously described hyperinnervation NSE-Noggin mouse model, but in our hands NSE-Noggin mice did not present with a hyperganglionic phenotype. Based on fur appearance, intact NSE-Noggin-IRES-EGFP construct, increased RNA expression of Noggin, and Noggin-EGFP protein expression, the correct genotype and transgene overexpression of NSE-Noggin mice was confirmed. Nevertheless, the total number of enteric neurons in the colonic myenteric plexus of transgenic mice did not differ significantly from wild types. Possible reasons for the absence of the hyperinnervated phenotype as compared to previous studies include differences in study design, influence of microbiota, and/or other environmental variables.

In Chapter 5, we present our results using the hypo-innervation mouse model Hand2fl/+;Wnt1-Cre2 to study the effect of a reduced neuron density on colorectal carcinogenesis. RNA sequencing of isolated tumors depicted a clear difference in tumor transcriptome between hypo-innervated and control mice and gene set enrichment analysis indicated that the cancer hallmark ‘avoiding immune destruction’, among others, was enriched. The colon of hypo-innervated mice was characterized by a reduced B cell population, specifically germinal center (GC) B cells, and the immunoglobulin profile of B cells was found to be less mature. B cells localized mostly in the colonic submucosa near neuronal processes containing varicose release sites, suggesting the possibility of direct communication. Furthermore, B cells subjected to epinephrine in vitro showed an increase in proliferation and maturation. Epinephrine-treated B cells also displayed increased IL-10 and altered immunoglobulin secretion as well as an altered transcriptomic profile. They were transcriptionally more similar to plasma cell populations from human CRC tissues as compared to those of the normal colon, and the epinephrine-induced B cell mRNA expression profile was associated with a worse survival for CRC patients. These are, to our knowledge, the first data demonstrating a direct interaction between neurons and B cells in colorectal carcinogenesis.

To translate our findings on the role of neural cells in CRC to the human setting, we used two large population-based CRC cohorts to characterize the innervation of human colorectal tumors and examine its prognostic potential. In Chapter 6, protein gene product 9.5 (PGP9.5) and neurofilament (NF) immunoreactive nerve fibers were found within the tumor stroma and mostly characterized by the neuronal subtype markers vasoactive intestinal peptide and neuronal nitric oxide synthase, suggesting that inhibitory neurons are the most prominent neuronal subtype in the CRC stroma. NF immunoreactivity was associated with a worse CRC-specific survival in the study cohort independent of other prognostic factors, but these results were not observed in the in-cohort validation group.

For Chapter 7, we also utilized two population-based CRC cohorts to characterize the phenotype of enteric glia throughout CRC development and to study the prognostic value of the glial marker glial fibrillary acidic protein (GFAP) in CRC. GFAP positive enteric glia were identified within carcinoma tissue stroma, but were absent in normal mucosa or adenoma tissue, whereas S100B staining was detected in all stages. High-density GFAP staining was associated with improved survival in the study cohort, though this finding could not be validated. These findings suggest that CRC upregulates GFAP expression in enteric glia, indicative of glial reactivity, but prognostic value of GFAP could not be confirmed.

To explore the molecular characteristics of enteric glia in CRC, we applied fluorescenceactivated cell sorting to isolate enteric glia from human adjacent normal and cancerous colon tissues followed by single cell RNA sequencing in Chapter 8. The presence of seven enteric glial subtypes in both normal colon tissue and colon cancer tissue was uncovered. Transcriptional changes were observed between enteric glial cells isolated from normal versus CRC tissue, e.g. increased HLA-DRA expression, indicating a role for enteric glia in antigen-presentation and immune response modulation within the TME. Moreover, a significant shift in cluster distribution towards inflammation- and injury response- glia was noted in glia present in tumor samples. This study suggests that enteric glia take part in immune responses in CRC.

The general discussion in Chapter 9 provides an overview of the results obtained in this thesis and puts the findings in the context of recent literature. It reflects on the role of neural innervation in CRC development and progression, and contemplates on the multidirectional communication between neurons, glia, immune cells and colorectal tumors. By combining the insights from the different studies that are part of this thesis, it became clear that the field needs holistic approaches that integrate neural subtypes, patientspecific factors, and the full complexity of TME, as both the type of innervation and cancer subtype can influence treatment response.