In vitro model of inflammatory, hypoxia, and cancer stem cell signaling in pancreatic cancer using heterocellular 3-dimensional spheroids

Introduction As one of the most aggressive cancers worldwide, pancreatic cancer is associated with an extremely poor prognosis. The pancreatic tumor microenvironment consists of cancer cells and other tumor associated cells. Cross-talk between these different cell types through various signaling molecules results in the development of a more aggressive and malignant phenotype. Additionally, due to the highly dysregulated vasculature of tumors, the inner tumor core becomes hypoxic and eventually necrotic. Therefore, there is a need for the development of a physiologically relevant in vitro model that recapitulates these dynamic cell-cell interactions and the 3-dimensional (3D) structure of pancreatic tumors. Methods Four different 3D co-culture spheroid models using different combinations of Panc-1 tumor cells, J774.A1 macrophages, and NIH-3T3 fibroblast cell lines were reproducibly developed using the hanging drop technique in order to mimic the tumor microenvironment and to evaluate the differences in expression of various inflammatory, hypoxia, and cancer stem cell markers, including IL-8, TNF-α, TGF-β, HIF-1α HIF-2α, SCF, and LDH-A. Additionally, immunofluorescence studies were employed to investigate whether these spheroids tested positive for a cancer stem cell population. Results Pronounced differences in morphology as well as expression of signalling markers were observed using qPCR, indicative of strong influences of co-culturing different cell lines. These models also tested positive for cancer stem cell (CSCs) markers based on immunofluorescence and qPCR analysis. Conclusion Our results demonstrate the potential of 3D co-culture spheroid models to capture the inflammatory and hypoxic markers of pancreatic tumor microenvironment. We further demonstrate the presence of cancer cells with stem cell markers, similar to actual pancreatic cancer tumor. These spheroids present excellent in vitro system to study tumor-immune-stromal cell interactions as well as test deliverability of potential therapeutics in the tumor microenvironment with accurate physical and physiological barriers.


Conflict of Interests:
The authors declare that they have no conflicts of interest .   18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42 1. Introduction 92 Pancreatic cancer is one of the deadliest cancers associated with the 4 th highest number of 93 deaths amongst all cancers worldwide, with the incidence rate nearly equaling the mortality rate. 94 In 2018, it is estimated that 55,440 adults will be diagnosed with pancreatic cancer and 44,330 95 patients will die of the disease (1). With the growing epidemic of type I diabetes worldwide, the 96 number of patients with pancreatic cancer is expected to increase significantly in the coming years. 97 Despite tremendous efforts to improve patient survival, including surgical resection, 98 chemotherapy, and radiation therapy, these options have not been successful in prolonging patient 99 life beyond a few months in pancreatic cancer (1-5). The poor prognosis in pancreatic cancer 100 patient is largely associated with late presentation of the symptoms when the disease has already 101 progressed to advanced stage where significant metastasis has occurred combined with lack of 102 effective therapies that can improve patient outcomes. Additionally, there is a need to develop 103 suitable in vitro and in vivo disease models that can improve the drug development process. 104 The pancreatic tumor microenvironment is represented by an outer proliferative and inner 105 necrotic zone, along with stromal components including the extra-cellular matrix, blood vessels, 106 signaling molecules, and other cells including tumor-associated macrophages (TAMs), tumor-107 associated fibroblasts (TAFs), and endothelial cells form (6, 7). Animal models for pancreatic 108 cancer recapitulate the tumor microenvironment to some extent but are difficult to develop, time 109 consuming and very expensive (8). Two dimensional (2D) cell monolayers are simple to culture 110 and provide convenient testing platforms for screening anti-cancer drugs but they are not truly 111 representative of the tumor microenvironment, morphologically or functionally (9). More recent 112 efforts have shifted focus on three dimensional (3D) co-culture spheroids that serve as a robust in 113 vitro model that exhibit several features of pancreatic cancer microenvironment (10).      To measure the size of these spheroids, they were washed with 1X PBS upon harvesting, 181 fixed with 4% Formalin and mounted on microscope depression slides (Fisher Scientific, PA) 182 using the Shandon Immu-Mount (Thermo Scientific, PA), and covered with a cover-slip. The Carl 184 software was then used to measure and record the diameter. 5μm optical slices of spheroids were 185 imaged through Z stacking on the confocal microscope and the total number of slices per spheroid 186 was determined. This information was then used to calculate and record spheroid depth.   TAMs, the trend seen in the spheroid models was not surprising (41). Highest levels of TNF-α at low levels of 4%, 2.5% and 5% in the Panc-1 homocellular spheroid, Panc-1:NIH/3T3 spheroid 280 and the 3-in-1 heterocellular spheroid respectively. Similarly, TGF-b has anti-tumorigenic effects 281 at the early stages of tumor growth but at latter stages promotes tumor cell proliferation, 282 desmoplasia and metastasis (42). Since TGF-b is produced by both tumor and stromal cells, the 283 trend seen across spheroid models in our PCR experiment agreed with expected results (43).

284
Highest levels of TGF-β (Figure 3.c.) of 56% were observed in the Panc-1:NIH/3T3 spheroid 285 followed by 42% in the Panc-1 homocellular spheroid. It is interesting to note here that the Panc- Due to increased hypoxia and glycolytic metabolism due to the Warburg effect, there is an 302 associated low pH tumor microenvironment and an increase in HIF-2a expression (44). This 303 increased expression is associated with dense stroma and so we hypothesized that the fibroblast 304 containing spheroids may exhibit hypoxia like molecular feature. Indeed, qPCR analysis of HIF-305 2α level (Figure 4.b.) reveals highest levels of 147% in the Panc-1:NIH/3T3 spheroid followed by 306 15% in the 3-in-1 heterocellular spheroid (45). Both results were found to be statistically 307 significant. Third highest HIF-2α levels were observed in the Panc-1:J774.A1 spheroid at 8.95%.

308
Levels in normoxic and hypoxic monolayers were 6.27% and 5.15% and at 4.34% in the Panc-1 309 homocellular spheroids. These results assert that the multicellular spheroids are able to recapitulate 310 the hypoxic microenvironment of a tumor, which is a hallmark property of cancer.

313
Although we would have expected to see highest levels of LDH-A in hypoxic Panc-1 cells, hypoxic 314 cells very quickly revert their phenotype to normoxic when exposed even briefly to oxygen. Since 315 these cells were withdrawn from the hypoxia chamber before processing PCR samples, the brief 316 exposure to oxygen may have obfuscated results seen in the monolayer samples (46). It is however 317 interesting to note that the highest levels among spheroids were seen in the densest 3-in-1 318 heterocellular spheroid. The Panc-1 homocellular spheroid, Panc-1:NIH/3T3 spheroid and Panc-319 1:J774.A1 spheroid expression was 71%, 22% and 19% respectively. All results were found to be 320 statistically significant. Binding of SCF to its ligand, c-kit, induces the up-regulation of HIF-1a 321 through activation of PI3K/Akt and Ras/MEK/ERK pathways (47). Highest levels of SCF, shown 322 in Figure 4.d., of 20% were seen in the Panc-1:J774.A1 spheroid, followed by 10% in the 3-in-1 323 heterocellular spheroid. These results were found to be statistically significant. Next highest levels were seen in the Panc-1:NIH/3T3 spheroid at 2.2%. Differences between SCF expression levels in 325 the normoxic, hypoxic and homocellular spheroid were minimal and with no statistical 326 significance. It is extremely interesting to note that correlation between HIF-1a and SCF 327 expression across our models is suggestive of SCF induced HIF-1a upregulation. Immunostaining

337
The hanging drop method was used to successfully and reproducibly generate all four self-338 aggregating spheroid models, which were then characterized for their morphology and size.  Positive staining of CD24 and SCF surface markers through immunofluorescence is indicative of 359 the presence of CSC within these spheroid models. SCF levels being significantly higher in the 360 co-culture models, as evidenced in the PCR study, offer further support to the hypothesis that 361 cancer stem cell formation is largely dependent on microenvironmental cells and cross-talk.

362
This model has the potential to replicate the barrier to drug delivery seen in tumors.