Quantification of microenvironmental metabolites in murine cancer models reveals determinants of tumor nutrient availability

Mark R Sullivan, Laura V Danai, Caroline A Lewis, Sze Ham Chan, Dan Y Gui, Tenzin Kunchok, Emily A Dennstedt, Matthew G Vander Heiden, Alexander Muir

Preprint posted on January 03, 2019

Cancer metabolism: what can we learn from analysing metabolites in tumour infiltrating fluid rather than in blood plasma?

Selected by Maria Rafaeva

Categories: cancer biology, physiology


Many studies in recent years have expanded our knowledge of the metabolic landscape in cancer. Cancer cells rewire metabolism to be beneficial in changing conditions such as in hypoxia and new microenvironment of the metastatic niches. They also establish metabolic communication with other cell types such as fibroblasts and immune infiltrating cells (1). Inevitably, this complexity at the cellular level leads to the question, how different factors, especially metabolic therapies, can affect the organism as a whole and contribute to the metabolic tumour microenvironment. In order to address this question, there is an urgent need for developing methods that allow the study of cancer metabolism in vivo (e.g. metabolomics, intravital imaging) (2). Interestingly, metabolites are often screened in the blood plasma for diagnostic purposes, but assessment of tumour interstitial fluid (TIF), which reveals local tumour nutrient exchange levels, was so far rarely done due to technical limitations (3).

Key findings

In this preprint Mark Sullivan et al. establish a new metabolomics approach in order to compare metabolite levels between tumour interstitial fluid and plasma from murine cancer models. Precision of such analysis was previously limited due to the presence of different matrix effects in these biological liquids. The authors overcome this problem by using internal standards for every labelled sample (12C to known amounts of added 13C).

By hierarchical clustering and principal component analysis they show in a mouse model of pancreatic ductal adenocarcinoma (PDAC) that TIF composition significantly differs from plasma, which represents local changes caused by the tumour microenvironment. Changes in the major metabolites are in line with previous findings: depletion of glucose in TIF matches its enhanced consumption by cancer cells, and depletion of particular amino acids can be caused by enhanced immune cell infiltration.

Excited by the abilities provided by the new method, the authors next sought to assess the role of several biological factors on TIF composition by manipulating one factor at a time. In particular, they assessed complex factors (as tumour size), cancer cell extrinsic (as anatomical location, mouse diet) and intrinsic factors (as tissue of origin, tumour genetics).

They found that two categories of tumour size (small vs large) did not possess significant difference in TIF composition. Anatomical location of a tumour, such as orthotopic vs subcutaneous, turned out to be crucial for defining the TIF metabolic profile. Different mouse diet (standard chow with whole protein vs defined diet with purified amino acids) lead to different TIF metabolic profiles as well. However, changes in TIF largely correlated with the changes in plasma levels, which highlights a systemic response of the whole organism to diet.

Interestingly, tumours comprised of cancer cells with different tissue of origin (lung vs pancreas, same genetic drivers) produced different TIF metabolic profiles. This, at least, proves previous findings that these cell types have different branched-chain amino acid metabolism. Finally, genetic knockout of tumour suppressor Keap1, expected to promote metabolic changes in lung adenocarcinoma, was not sufficient for changing TIF composition.

What I like

This study provides a great tool to address questions in tumour metabolism in a physiological context by analysing TIF metabolites. When plasma levels of certain metabolites can be used as a diagnostic tool, TIF composition can provide more precise information about tumour nutrition.

Certainly, the scope of the paper allows only to start answering certain questions about changes in TIF profiles. Nonetheless, authors use elegant set ups to pinpoint if intrinsic or extrinsic factors in cancer have a role in defining tumour metabolism as a whole.

Future directions

As the next steps to this study, it is important to support findings in the PDAC model with similar analyses of TIFs in other cancer models and cells carrying other genetic alterations. Hypotheses explaining certain metabolite changes in TIFs can be confirmed by analysing tumour composition (e.g. presence of difference cell types), performing in parallel in vitro assays with extracted cancer cells, etc. In addition, it will be exciting to study TIFs of mice treated with metabolic therapy (e.g. including different delivery mechanisms) as it can provide more insight about in vivo effects of the drugs. Finally, TIFs from tumours with different metastatic potential can be compared to see if some metabolic changes of the tumour microenvironment can explain this behaviour. All this would complement interesting findings in this preprint.

Questions to authors

  • Would a cancer model with controllable tumour size upon harvest be a better model for size-dependent TIFs profiling (e.g. xenografts)?
  • Why do some of the tumours not yield any TIF? Is it just because they have a small size?
  • Could we experimentally separate consumption and secretion effects of a tumour in TIF?
  • How different you think can be tumour nutrition in case of studying cancer metabolism in immunocompetent vs immunocompromised mice?


  1. Wolpaw AJ, Dang CV. “Exploiting Metabolic Vulnerabilities of Cancer with Precision and Accuracy” Trends Cell Biol.2018; 28(3)201-212
  2. Muir A, Danai LV, Vander Heiden MG. “Microenvironmental regulation of cancer cell metabolism: implications for experimental design and translational studies” Disease Models & Mechanisms.2018; 11
  3. Wiig H, Tenstad O, Iversen PO, Kalluri R, Bjerkvig R. “Interstitial fluid: the overlooked component of the tumour microenvironment?” Fibrogenesis Tissue Repair. 2010; 3 12

Tags: cancer, metabolism, metabolomics, nutrients, tumour microenvironment

Posted on: 11th February 2019 , updated on: 3rd March 2019

Read preprint (No Ratings Yet)

  • Author's response

    Mark Sullivan, Matt Vander Heiden & Alexander Muir shared

    Thank you very much for discussing our preprint. We are very excited about the techniques, results, and implications. We are also happy that you found our work interesting.

    First, we wholeheartedly agree with the suggested future directions, and work is underway to address these important questions. Based on the differences between lung and pancreatic TIF composition (even when the tumors were driven by the same oncogenic drivers), we think each type of tumor will likely have different TIF composition, and different microenvironmental metabolic constraints. As TIF from more diverse models of cancer are analyzed, it will be interesting to see the commonalities and differences between different tumors and how those differences affect tumor metabolism. We also agree that it will be important to determine how altering the cellular composition of tumors regulates TIF composition. In particular, pancreatic tumors are well known to contain extensive numbers of stromal cells (Hwang et al., 2016) (e.g. fibroblasts and macrophages), which likely play key roles in determining TIF composition. Studying how different cell types in pancreatic tumors contribute to TIF composition is one way to address how stromal cells impact tumor metabolism. Lastly, examining TIF composition in primary and metastatic tumors will also be helpful to learn how primary and metastatic tumors may be differently constrained by the microenvironment.

    Below are our response’s to your questions:

    1. We agree that better controlling for factors such as tumor size would decrease the sample-to-sample variability in our measurements. One variable that we realize could impact our measurements is circadian rhythm. For this initial study, we did not control when we euthanized animals and collected plasma and TIF, and circulating metabolites are known to change throughout the day (Dallmann et al., 2012). In future studies, we will control for this variable. However, for TIF analysis, we still favor using the autochthonous models instead of subcutaneous grafts, despite their increased variability in size upon harvest. We found significant differences in TIF composition between subcutaneous allografts of pancreatic cancer cells and autochthonous tumors arising in the pancreas. Thus, we think that where the tumor is located is key to determining TIF composition. Thus, while xenografts allow better experimental control, these tumor models may not fully recapitulate the TIF composition of actual primary tumors. For example, we hypothesize that the lack of stromal cells in subcutaneous models of pancreatic tumors (Hwang et al., 2016) may contribute to the difference in TIF composition.
    2. We do not have a good understanding of why some tumors yield TIF and others do not. Size is one contributing factor. Smaller tumors yield less TIF, and are more likely to yield none. Other tumor factors also influence interstitial volume, and thus the amount of TIF to isolate. These factors include the degree of vascularization and stromal architecture, and it would be interesting to know if tumors that do not yield TIF are different from those that do in terms of interstitial volume, vascularization or stroma.
    3. This is an important question, however determining cancer consumption and secretion of certain metabolites based on TIF and plasma concentrations alone is likely not possible. Historically, a number of groups have analyzed metabolite concentrations in blood that enters and then exits tumors (Gullino et al., 1967; Kallinowski et al., 1988; Sauer et al., 1982). Taking these measurements allows for direct determination of metabolite consumption and secretion by tumors, and may provide even more insight when combined with TIF metabolite measurements. We hope in the future to revisit these experiments with modern mass spectrometry methods to analyze more metabolites than was possible when these original studies were performed.
    4. Because non-cancer cell types in the tumor, including immune cells, can interact with cancer cells in the tumor, the use of immunocompromised mice could affect the results of cancer metabolism studies. Anatomical location influences TIF nutrient levels, and it is possible that immune cells are contributing to differences in TIF based on anatomical site. So, eliminating large numbers of immune cells that could act as stromal determinants of TIF nutrient levels could very well change the nutrient microenvironment and cancer metabolism. We have not tested this possibility directly, but think it would be good to test in the future, and to keep in mind as a caveat when using immunocompromised mice as a model for cancer metabolism studies.

    Thank you again for your interest and thoughts on our work!


    Dallmann, R., Viola, A.U., Tarokh, L., Cajochen, C., and Brown, S.A. (2012). The human circadian metabolome. Proc Natl Acad Sci U S A 109, 2625-2629.

    Gullino, P.M., Grantham, F.H., and Courtney, A.H. (1967). Glucose consumption by transplanted tumors in vivo. Cancer Res 27, 1031-1040.

    Hwang, C.I., Boj, S.F., Clevers, H., and Tuveson, D.A. (2016). Preclinical models of pancreatic ductal adenocarcinoma. J Pathol 238, 197-204.

    Kallinowski, F., Vaupel, P., Runkel, S., Berg, G., Fortmeyer, H.P., Baessler, K.H., Wagner, K., Mueller-Klieser, W., and Walenta, S. (1988). Glucose uptake, lactate release, ketone body turnover, metabolic micromilieu, and pH distributions in human breast cancer xenografts in nude rats. Cancer Res 48, 7264-7272.

    Sauer, L.A., Stayman, J.W., 3rd, and Dauchy, R.T. (1982). Amino acid, glucose, and lactic acid utilization in vivo by rat tumors. Cancer Res 42, 4090-4097.

    Have your say

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.

    Sign up to customise the site to your preferences and to receive alerts

    Register here