The role of the stroma, encompassing pancreatic stellate cells and extracellular matrix, in the biology and chemoresistance of pancreatic cancer

Our report entitled “3D pancreatic carcinoma spheroids induce a matrix-rich, chemoresistant phenotype offering a better model for drug testing” (PMID: 23446043) has its ten years anniversary. Herein we present the story "After the Paper" on the background of two recent publication from our group.

Like Comment
Read the paper

Rainer L. Heuchel, J.-Matthias Löhr

CLINTEC, Karolinska Institutet, and Karolinska Comprehensive Cancer Center, Stockholm, Sweden.


Our report entitled “3D pancreatic carcinoma spheroids induce a matrix-rich, chemoresistant phenotype offering a better model for drug testing” (BMC Cancer. 2013 Feb 27;13:95. doi: 10.1186/1471-2407-13-95. PubMed PMID: 23446043.) has its ten years anniversary. Herein we present overall scenario "After the Paper" on the background of two recent publication from our group that were published in BMC Cancer last year (Norberg KJ, Liu X, Moro CF, Strell C, Nania S, Blümel M, Balboni A, Bozóky B, Östman A, Heuchel RL, Löhr JM (2020): A novel pancreatic tumor and stellate cell 3D co-culture spheroid model. BMC Cancer, 2020 May 27;20(1):475. doi: 10.1186/s12885-020-06867-5.PMID: 32460715) and another one more recently (Liu X, Gündel B, Li X, Liu J, Wright A, Löhr M, Arvidsson G, Heuchel R. 3D heterospecies spheroids of pancreatic stroma and cancer cells demonstrate key phenotypes of pancreatic ductal adenocarcinoma. Transl Oncol. 2021 May 1;14(7):101107. doi: 10.1016/j.tranon.2021.101107. PMID: 33946033).


Pancreatic ductal adenocarcinoma (PDAC) is a devastating disease[1]. Most carriers of the disease get diagnosed when metastases are already established. Those that are qualifying for curative surgery, about 25 percent of patients, will die on the average 2 years later from metastases that were too small for the present imaging modalities at the timepoint of diagnosis. PDAC is currently the 4th most frequent cause of cancer related death and is predicted to become number 2 after lung cancer in 2030 [2]. Therefore, PDAC is a medical emergency [3]. Why is the mortality so much higher than in other cancer types? There are no biomarkers for early diagnosis and the cancer is almost completely resistant to radiation-, chemo- and immunotherapy. Despite significant research and initially many promising preclinical therapy approaches, almost nothing works later in the clinic. The inevitable question is why nothing really translates into the clinic? The answer to that might lie in the fact that as the initial screening we used research tools like 2D cell culture that does not recapitulate a patient’s cancer, which can contain of up to only 10 percent of cancer cells surrounded by an enormous mass of extremely collagen-rich stroma, containing most of all activated pancreatic stellate cells (PSCs)/myofibroblasts creating to together with the cancer cells an immune-cold tumor microenvironment (TME) [4]. Based on a study on the interaction between tumor and stromal cells[5], we had already twenty years ago the idea that a key to understanding therapy resistance might lie in the way cells communicate with one another and their environment, and that there must be big differences when cells are grown on stiff plastic in 2D compared to growth in 3D as e.g. spheroids. Another important thought-provoking impulse was given by the newly developed concept of cell adhesion mediated drug resistance (CAM-DR) for hematopoietic malignancies.

We tested different approaches for 3D growth, like the hanging drop method and growth on agarose, but the most economic and reproducible way was to growth cells on non-cell culture treated plates in medium containing the inert crowding agent methylcellulose, which helps the cells to aggregate without interfering with e.g. gemcitabine treatment[6]. Another advantage of this system was that it was easily expandable from 96-well format to 384-well format for robotized/fully automated high throughput drug screening [7]. Most interesting, however, were the differences in the biological behaviour of the cancer cells grown in 3D versus the conventional 2D. Besides increased chemo-resistance to most drugs tested, we observed a metabolic switch towards higher glycolysis rate/lactate production (Warburg effect), response to hypoxia and a slight increase in extracellular matrix (ECM) production [6], all hallmarks of aggressive cancer. Interestingly, one of the drugs we found effective on PDAC spheroids was a genistein derivative, which has already produced promising results in phase 1 clinical trial [8].

These results were already very encouraging, but we felt that it would be very important to also include stromal cells into the cancer cell spheroids. Our lab had previously generated a human pancreatic stellate cell (PSC) line [9], which we now, tagged with green-fluorescent eGFP, used to create tumor/stroma heterospheroids (Fig.1). The initial aim was now to compare the expression of each individual cell type when grown as monospheroid or as heterospheroid together with a second cell type using the at that time common expression arrays. To this end we had to singularize the cells from the spheroids by enzymatic and physical methods to separate fluorescent stromal and non-tagged cancer cells by fluorescence-activated cell sorting (FACS). However, this whole process was so harsh and long-lasting that we lost about ninety percent of the cells, requiring about fifty 96-well plates of the heterospheroids to collect enough RNA for one expression array. Therefore, we decided to restrict our investigation on a couple of known genes representing PDAC cells and activated PSCs/myofibroblasts to establish and validate the heterospheroid approach. In addition, we performed electron microscopy and immunohistochemical analyses of the different spheroid types using two different human PDAC cell lines (Panc-1, HPAF-II) in combination with human PSCs. We compared the mRNA expression of selected genes between human PSCs grown alone as monospheroids or in combination with Panc-1 or HPAF-II cancer cells as heterospheroids. This allowed to test the expression change of certain genes, but we could only speculate about the individual contribution of the two cell types. This changed when a neighbouring research group made us aware of an experimental set up that would exploit a species difference when analysing species- and cell type specific expression of cells gown together [10]. This was achieved by running semiquantitative rt-PCR using species-specific PCR primers on spheroids made up of human/mouse PDAC cells and mouse/human PSCs, which identified how the expression of a gene changed in each cell type by 3D co-culture. We called this method virtual sorting of gene expression, at least on small scale [11]. Just before we were ready to put our results together into a manuscript, we realized from the electron microscopy pictures that our human PSCs were contaminated with a retrovirus (Fig.2). To exclude any artifacts due this contamination, we decided to re-run all involved experiments with a backup of retrovirus-free hPSCs, showing however the same results. The important take home messages of that work were the human and mouse PSCs behaved very similar when combined with mouse and human PDAC cells. And that the species-specific coding difference between mouse and man could be used to interrogate the gene expression of mouse/human cell mixtures without prior manipulation and separation of the cells, catching the cells so to speak “in flagranti” / “in the act”. This work laid the groundwork for the next step, where we set out to look at the cancer-/stromal cell crosstalk at a big scale employing RNA-sequencing and in silico species separation of the sequence reads followed by gene expression analysis and gene set enrichment analysis. We calculated that only about 4 percent of all reads were lost in this analysis due to sequence identity or other annotation problems. This allowed us to compare the expression profiles of cells in monospheroids and in heterospecies heterospheroids (or simply = co-culture) with the different published expression signatures for PDAC [12] and the newly identified subclasses of PSCs (myofibroblast cancer-associated fibroblast = myCAF, inflammatory cancer-associated fibroblast = iCAF) [13] under high serum condition (10 percent fetal calf serum (FCS)) and under low serum condition (0.1 percent FCS), the latter an approach to represent the nutrient-poor environment, characteristic of PDAC. The presence of PCSs changed the phenotype of Panc-1 cells from the more classical, more differentiated phenotype, to the squamous/basal like, more aggressive phenotype independent of growth condition [14]. PSCs on the other hand were affected by the presence of Panc-1 cells and the growth condition, displaying a more iCaf phenotype under low serum and a more myCAF phenotype under high serum condition and co-culture. We also found that interfering with cholesterol biosynthesis, one of the gene sets with significantly higher expression in Panc-1 cells from heterospheroids under high serum condition, with statin treatment increased apoptosis in Panc1 cells, that was further increased by the presence of PSCs. Recently, pharmacotranscriptomic signatures were generated from PDAC patient-derived organoids, in order to predict a patients’ response to treatment with five common chemotherapeutic drugs [15].  Comparing these signatures with expression profiles of Panc-1 cells under high serum condition predicted higher sensitivity of Panc-1 cells for gemcitabine, paclitaxel and SN38 (active metabolite of irinotecan) when co-cultured with PSCs. Investigating epithelial/cancer cell-specific apoptosis by measuring caspase-cleaved cytokeratin 18, we observed that this was true only for gemcitabine treatment, but not paclitaxel and SN-38 treatment of spheroids, indication, that the presence of stromal cells/PSCs can have a significant influence of cancer cell chemosensitivity. What does all this tell us with regard to the usefulness of the preclinical tool armament in PDAC? PDAC is a very heterogenous cancer meaning that the cancer tissue of one patient harbors different cancer clones with different chemosensitivities. The tumor microenvironment is characterized by a very high cellularity and an excessive content of fibrous collagen, which cannot be produced by PSCs in vitro. Organoids from patients, the latest addition to the preclinical PDAC toolbox [16], are grown in Matrigel, containing large amounts of collagen-IV, but no fibrous collagen-I or -III which does by far not represent the stiff environment PDAC cells are surrounded by in a patient’s tumor and the medium composition counter-selects against PSCs, making it more difficult to include this cell type. PDAC cell lines used to generate spheroids are most probably genetically less heterogenous than organoids but are much easier to grow and combine with other cell types. Their expression/omics profiles are easily obtainable, they can easily be used in high throughput drug screens, where not only single drugs but combinations of several drugs are tested in order to increase the chances of more completely targeting the different cancer cell clones within a patient’s cancer. In addition, using PSCs in these screens might identify vulnerabilities or resistances that without their presence stay undetected.


  1. Kleeff J, Korc M, Apte M, La Vecchia C, Johnson CD, Biankin AV, Neale RE, Tempero M, Tuveson DA, Hruban RH et al: Pancreatic cancer. Nature Reviews Disease Primers 2016, 2(1):16022.
  2. Aier I, Semwal R, Sharma A, Varadwaj PK: A systematic assessment of statistics, risk factors, and underlying features involved in pancreatic cancer. Cancer Epidemiol 2019, 58:104-110.
  3. Löhr JM: Pancreatic cancer should be treated as a medical emergency. Bmj 2014, 349:g5261.
  4. Karamitopoulou E: Tumour microenvironment of pancreatic cancer: immune landscape is dictated by molecular and histopathological features. Br J Cancer 2019, 121(1):5-14.
  5. Löhr M, Schmidt C, Ringel J, Kluth M, Muller P, Nizze H, Jesnowski R: Transforming growth factor-beta1 induces desmoplasia in an experimental model of human pancreatic carcinoma. Cancer Res 2001, 61(2):550-555.
  6. Longati P, Jia X, Eimer J, Wagman A, Witt MR, Rehnmark S, Verbeke C, Toftgård R, Löhr M, Heuchel RL: 3D pancreatic carcinoma spheroids induce a matrix-rich, chemoresistant phenotype offering a better model for drug testing. BMC Cancer 2013, 13:95.
  7. Heuchel RL, Nania S, Liu JP, Löhr JM: High-throughput drug screening model using 3D cultured human pancreatic ductal adenocarcinoma cells. Cancer Research 2016, 76:2.
  8. Löhr JM, Karimi M, Omazic B, Kartalis N, Verbeke CS, Berkenstam A, Frödin JE: A phase I dose escalation trial of AXP107-11, a novel multi-component crystalline form of genistein, in combination with gemcitabine in chemotherapy-naive patients with unresectable pancreatic cancer. Pancreatology 2016, 16(4):640-645.
  9. Jesnowski R, Fürst D, Ringel J, Chen Y, Schrödel A, Kleeff J, Kolb A, Schareck WD, Löhr M: Immortalization of pancreatic stellate cells as an in vitro model of pancreatic fibrosis: deactivation is induced by matrigel and N-acetylcysteine. Lab Invest 2005, 85(10):1276-1291.
  10. Arvidsson G, Henriksson J, Sander B, Wright AP: Mixed-species RNAseq analysis of human lymphoma cells adhering to mouse stromal cells identifies a core gene set that is also differentially expressed in the lymph node microenvironment of mantle cell lymphoma and chronic lymphocytic leukemia patients. Haematologica 2018, 103(4):666-678.
  11. Norberg KJ, Liu X, Fernández Moro C, Strell C, Nania S, Blümel M, Balboni A, Bozóky B, Heuchel RL, Löhr JM: A novel pancreatic tumour and stellate cell 3D co-culture spheroid model. BMC cancer 2020, 20(1):475-475.
  12. Collisson EA, Bailey P, Chang DK, Biankin AV: Molecular subtypes of pancreatic cancer. Nat Rev Gastroenterol Hepatol 2019, 16(4):207-220.
  13. Öhlund D, Handly-Santana A, Biffi G, Elyada E, Almeida AS, Ponz-Sarvise M, Corbo V, Oni TE, Hearn SA, Lee EJ et al: Distinct populations of inflammatory fibroblasts and myofibroblasts in pancreatic cancer. J Exp Med 2017, 214(3):579-596.
  14. Liu X, Gündel B, Li X, Liu J, Wright A, Löhr M, Arvidsson G, Heuchel R: 3D heterospecies spheroids of pancreatic stroma and cancer cells demonstrate key phenotypes of pancreatic ductal adenocarcinoma. Transl Oncol 2021, 14(7):101107.
  15. Tiriac H, Belleau P, Engle DD, Plenker D, Deschênes A, Somerville TDD, Froeling FEM, Burkhart RA, Denroche RE, Jang GH et al: Organoid Profiling Identifies Common Responders to Chemotherapy in Pancreatic Cancer. Cancer Discov 2018, 8(9):1112-1129.
  16. Boj SF, Hwang CI, Baker LA, Chio, II, Engle DD, Corbo V, Jager M, Ponz-Sarvise M, Tiriac H, Spector MS et al: Organoid models of human and mouse ductal pancreatic cancer. Cell 2015, 160(1-2):324-338.

Rainer Heuchel

PI, Karolinska Institutet

Preclinical research (3d-culture approaches and GEMMS) in pancreatic ductal adenocarcinoma and acute/chronic pancreatitis. Focus on epithelial/stroma interactions/crosstalk.