Rethinking the Roles of Cancer-Associated Fibroblasts in Pancreatic Cancer

Abstract

Bearing a dismal 5-year survival rate, pancreatic ductal adenocarcinoma (PDAC) is a challenging disease that features a unique fibroinflammatory tumor microenvironment. As major components of the PDAC tumor microenvironment, cancer-associated fibroblasts are still poorly understood and their contribution to the several hallmarks of PDAC, such as resistance to therapies, immunosuppression, and high incidence of metastasis, is likely underestimated. There have been encouraging advances in the understanding of these fascinating cells, but many controversies remain, leaving the field still actively exploring the full scope of their contributions in PDAC progression. Here we pose several important considerations regarding PDAC cancer-associated fibroblast functions. We posit that transcriptomic analyses be interpreted with caution, when aiming to uncover the functional contributions of these cells. Moreover, we propose that normalizing these functions, rather than eliminating them, will provide the opportunity to enhance therapeutic response. Finally, we propose that cancer-associated fibroblasts should not be studied in isolation, but in conjunction with its extracellular matrix, because their respective functions are coordinated and concordant.

Summary.

Despite the recent advances on the biology of cancer-associated fibroblasts in pancreatic cancer, there are still controversies concerning their roles. Here we discuss new perspectives on how to study their functions in pancreatic cancer.

Pancreatic ductal adenocarcinoma (PDAC) is the most common form of pancreatic cancer, and has an abysmal prognosis, with only ∼13% of patients surviving 5 years.¹ The lack of early detection markers and early metastatic dissemination are major reasons for the poor overall survival.2, 3, 4 Moreover, the complex and unique biology of this tumor also contributes to the poor prognosis. PDAC is characterized by extreme desmoplasia, which is the expansion and activation of fibroblasts and their extracellular matrix (ECM).^5,6 These cancer-associated fibroblasts (CAFs) are 1 of the most abundant cell types in the tumor,^7,8 yet their biology and their functions in PDAC are still greatly underappreciated. Over the past decade, efforts to understand CAF origins9, 10, 11, 12, 13 and heterogeneity^14,15 have been the chief areas of focus, with the advent of sophisticated mouse models, single-cell RNA sequencing, and big data approaches. CAF subtyping16, 17, 18 has helped to define some major transcriptional profiles that suggest what roles CAFs play in tumors.19, 20, 21 Nevertheless, the actual functional consequences of these subtypes, whether they are protumor or antitumor, or ultimately if they can be targeted for therapeutic benefit has yet to be robustly explored.

CAFs are a very challenging cell type to study because of their plastic and stem-like properties, the lack of cell-type specific markers, and context dependent roles.²² Moreover, CAFs are intimately linked to the ECM,²³ and thus, must be studied under more physiologically relevant conditions to accurately interpret their biology. Unfortunately, most of the studies exploring CAF biology have been conducted using 2-dimensional culture systems without appropriate consideration of the reciprocity between CAFs and their ECM. In this minireview with a perspective, we outline key concepts and suggest approaches to resolving some important challenges/controversies in the understanding of CAF biology, and where we envision the field going over the next decade.

CAF Subtyping: Functionality is Key

Molecular subtyping of CAFs (and other cells) has been a great tool for understanding CAF biology,14, 15, 16, 17, 18 but there is a trend for the field becoming too dogmatic with the use of markers to separate these cells into distinct “buckets.” In many recent studies, the great bulk of the data relies on the definition of transcriptional subclusters, as an attempt to define classifications that are suggestive of CAF functionality. But ultimately, what is the biologic meaning of a certain set of patients presenting many subclusters of CAFs? Are we to surmise each plays a distinct role in disease? Often, if one analyzes these clusters carefully, these multiple molecularly distinct subclusters collapse into a few key functional subtypes. This is because when studying cell subsets in the context of disease, there is a tendency to overlook the core functions of those cells. Activated fibroblasts and/or CAFs are by their nature highly secretory, producing ECM and a plethora of soluble factors, such as cytokines, chemokines, and growth factors,^22,24 a function independent of their subtype. Of course, the specific molecules they produce are likely influenced by context, in response to the integration of surrounding signals at any given time, but they still exert these core functions. We propose that critical differences between fibroblast populations are whether they support tumor development and progression or if they restrict these processes. In other words, are they protumor or antitumor? This is not meant to discourage or dismiss the importance of transcriptomics studies, but we suggest to rely on these studies as important tools to be used in conjunction with functional studies for a more complete and complimentary approach. Realigning our vision to define mechanisms that facilitate protumor and antitumor states will allow us to focus approaches where we can reprogram protumor CAFs to a more normalized, antitumor state, while limiting their protumor functions.

CAF + ECM as Functional Units

Especially in the case of highly fibrotic PDAC, CAFs exist and function in a 3-dimensional environment, surrounded by ECM that they largely produce themselves. Yet, historically, most in vitro studies of CAF function rely on isolated fibroblasts that are grown as a monolayer on plastic then manipulated genetically or pharmacologically, and finally analyzed for changes in their transcriptional or proteomic profiles over time. Clearly, this way of modeling CAF biology in vitro is far from physiologically relevant. CAF studies would benefit greatly from studying these cells in conjunction with their ECM. For example, CAF-derived ECMs activate naive fibroblasts, inducing expression of protumor proteins and pathways, such as p-FAK and internalized α5β1 integrin.²⁵ Moreover, the protumor protein NetrinG1 is dramatically upregulated in CAFs cultured in their own ECMs, compared with 2-dimensional growth on plastic, simulating what is observed in PDAC tissue from surgical samples.²⁶ In other 3-dimensional models, such as spheroids and organoids, the inclusion of microenvironmental cells into the systems, such as fibroblasts and immune cells, generate drug responses and physical architecture much more comparable with that of patient tumors.27, 28, 29 “Organ on a chip” models,^30,31 which also factor fluid pressure and shear forces, mimicking the high tumor interstitial pressure found in PDAC tumors, modulate signaling and cell interactions in a manner reminiscent of human tumors.

Thus, we consider CAF together with its native ECM (CAF + ECM) as a functional unit,²³ a concept coined by Dr. Edna Cukierman, will not only improve how we study CAF biology, but is mandatory to better understand the PDAC microenvironment as a whole. It is well known that the CAF + ECM unit imposes a physical barrier for treatment and penetration in PDAC³²; but more than physical, CAF + ECM is an entity that harbors reservoirs of secreted factors^33,34 in the tumor microenvironment (TME), including immunomodulatory factors,^35,36 metabolites,^37,38 and prosurvival factors,³⁴ making this unit a potential master controller of cancer progression. Additionally, the underlying fibrillar structure of collagens can also dictate disease progression^2,39 and cellular functions.^23,40,41 Consequently, understanding how the physical characteristics, composition, and stored factors impart the functionalities of CAF + ECM, will leverage ways to successfully modulate the TME to oppose PDAC progression.

Normalization, Not Elimination

Studies of mouse models of PDAC show great therapeutic promise in depleting the stroma, especially when targeting the hedgehog pathway⁴² or associated ECM.⁴³ However, these approaches have universally failed in the clinic,44, 45, 46, 47, 48 with follow-up studies demonstrating increased metastasis and immunosuppression in murine models.49, 50, 51 This is likely because CAFs are extremely plastic and respond to changing environment cues, overcoming the loss of functionality from the missing cell compartments. We propose that a better way to approach CAFs as a therapeutic target, therefore, would be to focus on strategies to normalize CAF function, while maintaining the CAF + ECM functional units. Our group recently demonstrated that inhibition of the synaptic protein NetrinG1 in CAFs limits protumor features of CAFs, whereas their ability to generate productive ECM was not compromised,²⁶ decreasing tumor burden in mice. LRRC15 has emerged as another potential CAF normalization target. Using diphtheria toxin depletion of LRRC15⁺ cells (mostly stromal mesenchymal cells/CAFs) in several tumor models, including PDAC, there is a reduction in fibroblast content and a shift toward a “universal fibroblast phenotype”⁵² that is potentially tumor restraining.⁵³ Another study shows that activating the vitamin D receptor reprograms pancreatic stellate cells in PDAC to a more quiescent normalized state, rendering chemotherapies more effective in mouse models.⁵⁴ These approaches suggest that efforts should be focused on mitigating the tumor-supportive functions to normalize the functionality of CAFs, rather than eliminating them entirely.

Understanding Functionality of CAFs for Better Therapeutics

Why do chemotherapy and radiotherapy often fail in patients with PDAC? The obvious answer is that by the time that these patients are diagnosed, there is already metastatic disease. Although this is certainly true, current therapies also fail to inhibit the progression of the primary disease. It is known that chemotherapy and radiotherapy often directly affect the TME in PDAC.55, 56, 57, 58, 59 Many reports show that there is an increase in transforming growth factor-β levels and in ECM stiffness, as well as enhanced fibrosis and activation of fibroblastic populations in preclinical and clinical settings,^60,61 which culminate in increased immunosuppression and more challenges to therapy. However, not all patients have a strong stromal response to chemotherapy/radiotherapy, and some studies have suggested that a lower stromal/tumor cell ratio to be a negative prognostic indicator.^59,62 Thus, the utility of measuring a bulk fibroblastic/stromal response to tumors is controversial and perhaps patient specific, and this lends further credence to dissect the intricacies behind fibroblastic function. The best approach would be to modulate the TME to predispose tumors to chemotherapies/immunotherapies, without intensifying the protumor features of the TME. One excellent example of this approach is targeting of focal adhesion kinase in PDAC. CAFs express high levels of pFAK³⁹⁷, and its genetic inactivation limits tumor burden and immunosuppression in PDAC mouse models.⁶³ Furthermore, focal adhesion kinase inhibitors in PDAC mouse models reprogram the stroma, reducing desmoplasia and immunosuppression, enhancing the efficacy of radiotherapy and immunotherapy. This combinatorial approach is currently being studied in a clinical trial.^64,65 With advances in the understanding of CAF + ECM biology, we will be better poised to target the TME therapeutically first, without destroying the CAF + ECM unit, resulting in the normalization of protumor functions while preserving the antitumor functions, and enhancing subsequent therapies. Can CAF metabolism be targeted?

CAFs are highly secretory, because they are constantly producing ECM components, growth factors, and chemokines and cytokines, and therefore rely on a unique metabolic machinery to enable these functions to be maintained, even under challenging metabolic conditions. In addition to overcoming the nutrient-poor TME that challenges their own survival, they also provide metabolic support to other cells in their TME, including cancer cells, by providing lipids, amino acids, exosomal content, nucleosides, carbohydrates, and by recycling metabolic waste products.^26,37,38,66, 67, 68, 69, 70, 71, 72, 73 Not surprisingly, there are many studies focusing on targeting different metabolic programs in CAFs, with successful functional outcomes initially, such as with glutaminase inhibitors.^74,75 Nevertheless, there are many redundant metabolic circuits that allow cells to rewire these circuits,⁷⁶ overcoming nutrient deficiencies induced by metabolic enzyme blockade.⁷⁷ Therefore, a more effective strategy, perhaps, would be to target critical metabolic pathways in CAFs that impart a secondary beneficial consequence, such as in the case of targeting glutamine synthetase. Although dual inhibition of glutamine synthetase in CAFs and glutaminase in tumor cells limits tumorigenesis by disrupting a key metabolic circuit (glutamine/glutamate cycle),⁶⁹ there is also a secondary effect on CAFs, where genetic or pharmacologic inhibition of glutamine synthetase leads to a large reduction in protumor cytokine secretion in PDAC CAFs.²⁶ Moreover, further disrupting glutamine/glutamate cycling by inhibiting the synaptic protein vesicular glutamate transporter 1 (VGlut1) in CAFs also reduces their protumor cytokine secretion. This is likely the unanticipated reason why the inhibition of a metabolic enzyme/circuit was successful in limiting tumorigenesis, because both loss of metabolic support of tumor cells and increased antitumor immunity were beneficial to the host. Tryptophan catabolism is another example of targeting metabolism that leads to increased antitumor immunity. Cytotoxic T cells rely on tryptophan for proper function,^78,79 and CAFs often express high levels of IDO-1, an enzyme responsible for depleting tryptophan in the TME.⁸⁰ Inhibition of IDO-1 leads to increased tryptophan levels and enhanced cytotoxic T-cell function.^81,82 These studies strongly suggest that we should consider the therapeutic efficacy of targeting CAF metabolism not only to identify resistance mechanisms to metabolic reprograming, but also to discover metabolic targets that result in limiting a secondary critical of CAF function (eg, production of protumor ECM or immunomodulatory factors).

Conclusions/Highlights/Future

There have been many advances in the understanding of CAF biology, but clearly there is still more to be deconvoluted. Sophisticated approaches, such as single-cell RNA sequencing, spatial transcriptomics, and the integration of big data provide a step toward uncovering new potential targets and particularities in the complex relationship of the different components of the TME and how they interact and adapt. However, simply identifying new dots is not the same as connecting them, which requires appropriately designed functional studies that fully appreciate and mimic the in vivo PDAC TME, as best as we possibly can.

Holistic approaches of the role of the TME that takes the CAF + ECM functional unit into account, will allow us to better identify approaches to reprogram CAF function to tip the balance away from protumor activities and toward antitumor functions. We propose that these approaches are much more likely to identify unique vulnerabilities that can be targeted to enhance therapies (Figure 1). Whether these vulnerabilities will involve the modulation of metabolic circuits, or other functional targets, it is likely important we avoid ablating CAFs and their ECM indiscriminately.

Figure 1.

Reassessing core concepts of CAF biology to improve PDAC therapy.Box 1: Although molecular profiling of CAF subtypes can provide a snapshot of CAF identity, the more relevant approach is to define specific functions of these fundamentally plastic cells as either protumor (dark brown cell) or antitumor (light brown cell) and to target those functions. Thus, defining the overall functional output of a CAF should be the experimental goal, to maximize therapeutic benefit. Box 2: Inhibition of protumor CAF functions (dark brown cell CAF), leading to CAF normalization (light brown cell CAF), will increase the efficacy of standard of care and targeted therapies (yellow bolt, green arrow), resulting in activation of antitumor immunity (green cells) and increased tumor cell elimination. Box 3: CAFs do not exist in isolation. They interact with each other, both physically and at a distance, and are intimately connected to an abundant ECM (blue lines) that is largely deposited by CAFs. Understanding how CAFs and their associated immediate ECM-rich environment act together as a functional unit will ultimately inform how we can normalize the stroma toward a tumor-constraining phenotype. Box 4: Elimination of either the fibroblast population as a whole, or the collagen-rich ECM, in pancreatic tumors has a detrimental therapeutic outcome in patients. Therefore, uncovering means to normalize the ECM (curly blue lines) and CAFs (light brown cells) will fortify the overall architecture of the pancreas while constraining tumor growth. Box 5: Targeting CAF metabolism alone has been ineffective because of multiple redundant and compensatory metabolic pathways in both CAFs and tumor cells. However, secondary consequences to disrupting particular metabolic circuits in CAFs, such as disrupting their immunosuppressive capacity, can increase intrinsic antitumor immunity, enhancing susceptibility to immune therapy. (Some of the images were used from biorender.com.)

In summary, we believe that consideration of these key factors will be the best approach for the success of targeting the TME in PDAC: functionality, modulation, and normalization.