HSF1 promotes inflammation induced tumor development through ECM remodeling

Abstract number
Presentation Form
Corresponding Email
[email protected]
Multiparameter and Multimodal Imaging for High Dimensional Data Acquisition
Oshrat Levi (6), Hagar Lavon (6), Rina Wassermann-Dozorets (6), Meirav Pevsner-Fischer (6), Shimrit Mayer (6), Esther Wershof (5), Yaniv Stein (6), Lauren Brown (1), Wenhan Zhang (1), Gil Friedman (6), Reinat Nevo (6), Ofra Golani (6, 8), Lior Katz (2, 7), Rona Yaeger (3), Ido Laish (2, 4), John Porco (1), Erik Sahai (5), Dror Shouval (4, 9), David Kelsen (3), Ruth Scherz-Shouval (6)
1. Department of Chemistry and Center for Molecular Discovery (BU-CMD), Boston University
2. Gastroenterology Institute, Sheba Medical Center, Tel Hashomer, Israel
3. Gastrointestinal Oncology Service, Memorial Sloan Kettering Cancer and Weil Cornell Medical College, USA
4. Sackler Faculty of Medicine, Tel-Aviv University
5. The Francis Crick Institute, London, UK
6. The Weizmann Institute of science, Israel
7. Department of Gastroenterology and Hepatology, Hadassah Medical Center, Israel
8. Department of Life Sciences Core Facilities
9. Pediatric Gastroenterology Unit ,Edmond and Lily Safra Children’s Hospital, Sheba Medical Center, Israel

TME, ECM, CAF, 2photon microscopy, Spatial analysis, MxIF, Ilastik, Fiji

Abstract text

In the colon, long-term exposure to chronic inflammation drives colitis-associated colon cancer (CAC) in patients with inflammatory bowel disease (IBD). Chronic inflammation underlies tumor initiation, promotion, invasion, and metastasis. While the causal and clinical link between chronic inflammation and CAC is well established, we lack a molecular understanding of what is the way in which chronic inflammation leads to develop colon cancer. Within the tumor, cancer cells are surrounded by a variety of non-malignant cells, such as macrophages, endothelial cells, neutrophils, cancer-associated fibroblasts (CAFs), and together with the extracellular matrix (ECM) they compose the tumor microenvironment (TME), also termed the stroma. Even the most aggressive cancers depend and interact with their environment mostly through secreted factors. Unlike cancer cells, stromal cells are genomically stable and do not harbor oncogenic mutations that could drive their co-evolution and functional reprogramming. Rather, stromal reprogramming is thought to be achieved by transcriptional rewiring. Previous work by us and others has shown that the master regulator heat shock factor 1 (HSF1) plays a crucial role in this process, by mediating a transcriptional program in fibroblasts that enables their reprogramming into cancer-associated fibroblasts (CAFs) to promote malignancy. We hypothesized that HSF1 plays a crucial role in inflammation-driven cancer by initiation of a transcriptional program that leads to changes in the extracellular matrix (ECM). 

We found that, in cell culture, cancer-induced ECM assembly by fibroblasts requires HSF1. Using an inflammation-driven cancer model in mice, we measured the changes in proteomic and ECM organization over time. We found that HSF1 drives a transcriptional program that leads to ECM remodeling in early stages and results in development of colon cancer. Loss of HSF1 prevents inflammation-induced ECM remodeling. Further to that, in CAC patients, we found high activation of stromal HSF1 and a proteomic signature similar to that driven by HSF1. Thus, HSF1-dependent ECM remodeling mediates the transition from chronic inflammation to colon cancer.

In this research, light microscopy was the main tool to observe our findings. We started with pathology scoring on whole slide images. In aim to validate the targets we found by using proteomic method (Mass spectrometry), we continued with multiplex immunofluorescence imaging- imaged with confocal microscopy. In parallel, we used 2-photon microscopy to examine how the ECM changes in cell culture and also in the colon of mice, throughout the progression of the disease. For that, we used Ilastik pixel classifier combined, with a code we wrote in Fiji (available here: https://doi.org/10.5281/zenodo.4172577). In addition, we performed decellularization of the colons and recellularization with cancer cells to obtain whether the differences in the ECM attract the cancer cells in a different way. For that aim, we used 2-photon microscopy with HeNe 633 laser, and analyzed the images with Imaris.


This work was recently published in Nature communication: https://doi.org/10.1038/s41467-020-20054-x

figure3Selected main figure from this work:

a Representative H&E, Sirius red (SR), and Masson’s trichrome (MT) staining of colons from WT (upper panels) and Hsf1 null (lower panels) mice following 52 days treatment with AOM-DSS (right panels) or control (sham injection and water; left panels). Scale ba-100 µm; M-mucosa; SM-submucosa; TM-tunica muscularis; C-cancer; S-stroma. bl Representative cross-sections of the colons with second harmonic generation (SHG) images taken from the mucosal side b and analysis cl of fibrillar collagen of mouse colons following 52 days of AOM-DSS treatment or control. Scale bar-100 µm. cf Quantification of the average crypt number c and size e, and Pearson correlation (two-sided) between tumor burden and crypt number d and size. gh Analysis of the average distance to the nearest crypt, within 40 µm, calculated based on SHG images of fibrillar collagen of mouse colons following 52 days of AOM-DSS treatment or control. Representative distance heatmaps of the SHG images presented in b are shown in gi Pearson correlation between the average distance to the nearest crypt and tumor burden. jk Analysis of the number of neighboring crypts within 20 µm, calculated based on SHG images of fibrillar collagen of mouse colons following 52 days of AOM-DSS treatment or control. Representative distance heatmaps of the SHG images presented in b are shown in k. #NN-number of nearest neighbors. l Pearson correlation between the number of neighboring crypts and tumor burden. c, e, hk results are shown as mean ± SEM, h p value was analyzed using two-sided Welch’s t-test, e, h, and k were analyzed by an unpaired Student’s t-test.