Targeting glutamine metabolism slows soft tissue sarcoma growth

Tumour cells frequently utilize glutamine to meet bioenergetic and biosynthetic demands of rapid cell growth. However, glutamine dependence can be highly variable between in vitro and in vivo settings, based on surrounding microenvironments and complex adaptive responses to glutamine deprivation. Soft tissue sarcomas (STSs) are mesenchymal tumours where cytotoxic chemotherapy remains the primary approach for metastatic or unresectable disease. Therefore, it is critical to identify alternate therapies to improve patient outcomes. Using autochthonous STS murine models and unbiased metabolomics, we demonstrate that glutamine metabolism supports sarcomagenesis. STS subtypes expressing elevated glutaminase (GLS) levels are highly sensitive to glutamine starvation. In contrast to previous studies, treatment of autochthonous tumour-bearing animals with Telaglenastat (CB-839), an orally bioavailable GLS inhibitor, successfully inhibits undifferentiated pleomorphic sarcoma (UPS) tumour growth. We reveal glutamine metabolism as critical for sarcomagenesis, with CB-839 exhibiting potent therapeutic potential.

In the manuscript titled "Enhanced Glutamine Metabolism Promotes Soft Tissue Sarcoma Progression" Leet et al show that murine UPS and human fibrosarcoma and leiomyosarcoma cell proliferation depends on glutamine derived glutamate to support carbon metabolism, aspartate production, and nucleotide biosynthesis.Importantly this dependency can be targeted therapeutically with the glutaminase inhibitor CB-839, currently in clinical trials for other tumor types.While previous studies in PDAC and lung cancer human lines and mouse models have shown that disparate results in regard to CB-839 sensitivity and glutamine dependence in vitro and in vivo, this study shows a promising effect of GLS inhibition in some sarcoma sub-types in vitro and in vivo in xenografts and in an autochthonous mouse model of UPS.
The manuscript is well written the study is timely and relevant for sarcoma and cancer metabolism research.The study includes both in vitro and in vivo analyses and characterization of mechanisms that underlie dependence on glutamine metabolism in murine and human sarcoma lines.Importantly, the authors show that inhibiting GLS using CB-839 has anti-tumor activity in xenografts but also in an autochthonous mouse model of UPS which should be a better predictor of cancer metabolic dependencies when compared with subcutaneously transplanted tumors.The study is well performed and it is relevant since it identifies a potential new treatment, that could be used in combination with other current drugs for the treatment of some sub-types of soft-tissue sarcoma, a disease that remains extremely challenging to treat.Some points should be address and clarified to improve the quality of this study: Major Points 1.As sarcomas are an extremely heterogenous group of tumors understanding which sarcoma subtypes could benefit from GLS inhibition for treatment is important.Based on GLS levels in patient samples and analysis of three human cell lines (2 sensitive: fibrosarcoma, leiomyosarcoma; 1 resistant: liposarcoma), the authors conclude that sub-types such as UPS and leiomyosarcoma may be sensitive to GLS inhibition while "liposarcomas may rely on other metabolic pathways for their energetic and anabolic purposes".However, I think this is the weakest aspect of this study for the following reasons: -The authors use GLS levels (mRNA and protein), to predict which sarcoma sub-types would, in principal be sensitive to GLS inhibition.However, are GLS levels a good predictor of CB-839 sensitivity?Although it is reasonable to target GLS in GLS-high tumors, it does not appear to be a consensus in the literature for biomarkers that could predict sensitivity to CB-839.See for example (PMID: 31088535), in which CB-839 resistant breast cancer lines and xenografts exhibit high GLS mRNA and protein levels.This caveat should be discussed.
-The number of sarcoma types and human lines tested for glutamine starvation and CB-839 sensitivity is too low to accurately predict sensitive sarcoma sub-types or correlate sensitivity with GLS levels.Only 3 sarcoma human cell lines were analyzed, all from different subtypes.The one 1 resistant line (liposarcoma) leads the author to suggest liposarcoma may be an exception in regards to glutamine dependence, but one single cell line may just reflect variability within sub-types.Additionally, to say in the abstract that "Multiple STS subtypes expressing elevated glutaminase (GLS) levels are highly sensitive to glutamine starvation" does not fully reflect what was tested.The same laboratory has previously used another human liposarcoma cell line in their studies (LPS246, is this line one resistant too?), and at least one more leiomyosarcoma cell line available at ATCC.Those, and if possible additional ones, should be tested for glutamine starvation and CB-839 treatment in vitro (alongside with GLS protein levels).
Given the challenges in studying this disease (e.g.very few human cell lines and available mouse models) a comprehensive analysis may be particularly hard to accomplish.If there is variability across the few lines available, the authors should simply acknowledge that more studies in additional sarcoma models will be required to more comprehensively understand these differences.
2. One of the most interesting results in this study is that mouse models harboring similar genetic alterations (KP mice, Mutant Kras and Trp53 -deleted tumors) show different sensitivity to CB-839, suggesting that "tumour microenvironment, surrounding tissue, and cell of origin influence metabolic responses in these models".Nevertheless, it role for KRAS in re-wiring metabolism in PDAC has been described.Would it be possible that sarcomas with RAS mutations show greater dependency on glutamine metabolism (which could also reflect different GLS levels across sub-types)?Do the 2 sensitive human cell lines harbor Ras mutations?There are other sarcoma datasets available that provide both mutational analysis and mRNA expression analysis (TCGA sarcoma data set).
3. In PDAC, the ineffective results of CB-839 in vivo were associated with highly adaptive metabolic networks in these cells following sustained exposure of the GLS inhibitor (also observed in vitro).It would be of interest to performed long-term proliferation assays in vitro with CB-839 (longer than 72h) in the sarcoma lines to determine if they remain sensitive or are equally able to re-wire metabolic dependencies which could also inform about the requirement for combination therapies.

Minor points:
-The authors re-use images from their previous publication in Nature communications (reference 41) with Figure S1A and S1F here corresponding to Figure 1B and 1D in ref 41.While the authors clearly state this model was previously described and in the figure legends that the images were adapted from the previous publication, it seems unnecessary to re-include the same images.Furthermore, while the representative tumors images and the same between studies, the graphs depicting tumor weight suggest they correspond to different cohorts (please see Fig1B in this study and Fig 1B in ref. 41).
-Page 11: "However, in marked contrast to previous in vivo studies, CB-839 administration to KP and KPH2 animals substantially inhibited tumour growth, as calculated from the difference in the muscular compartment of tumor-bearing limbs (red) relative to control limbs (green) (Fig. 7A-C; 7D-F)."Which in vivo studies are the authors referring here?There is no reference for this statement and is unclear whether the authors are refereeing to studies using sarcoma lines or other types of tumors (eg.PDAC).
-How did the authors choose a dose of 2uM for CB-839 for all lines analyzed?Were IC50 values determined, or it was based on previous studies.
-In the discussion: "Although human STSs show significant heterogeneity of GLS expression (Supplementary Fig. 4D-E), we noted a striking enrichment in UPS (Fig. 4F-H)" There is currently no Figure 4H (both graphs are depicted as Figure 4F).
-In some human cancer tissues, increased levels of GLS1 are associated with a higher disease stage and poor prognosis PMID:25844758.Did the authors see a similar trend in samples analyzed in Fig 4E (if clinical data is available).
The glutamine labeling to aspartate in Fig. 2h appears quite low, especially relative to glutamine labeling to alanine.Could the authors comment on this discrepancy (is there aspartate in the medium?), or put the numbers in context?
The labelling experiments the reviewer refers to use 15 N-labeled glutamine in media lacking aspartate.While both aspartate and alanine derive their nitrogen from glutamine/glutamate through transamination, they rely on separate precursors: aspartate on oxaloacetate and alanine on pyruvate.Since these labelling experiments were not performed under glucose-free conditions, it is likely that high enrichment in alanine is due to available pyruvate pools which would drive alanine production.This has now been clarified and highlighted in the main text (see p. 8, first paragraph).

2.
In Figs.2d, e, the authors note that the contribution of glutamine to the TCA cycle (as demonstrated by citrate) is similar, but the contribution to aspartate is increased.Because aspartate is generated from glutamine carbons via the TCA cycle, presumably glutamine contributes more to at least a portion of the TCA cycle in STS lines.Could the authors clarify this statement or include more TCA cycle metabolites in their analysis?
Due to the transamination reaction, enrichment of aspartate from U-13 C-glutamine can be from labelled oxaloacetate as well as labelled glutamate.This may be why enrichment in aspartate from U-13 C-glutamine is higher in the sarcoma lines, despite no changes in TCA cycle intermediates.Furthermore, it may be difficult to compare the enrichment of aspartate to TCA cycle metabolites since aspartate in this case is an "end-product", while the TCA cycle intermediates continue to cycle (see Supplementary Fig. 2E).Previously, we only showed enrichment into citrate and have now included other TCA cycle metabolites as the reviewer has suggested (see Fig. 2D; p. 7, fourth paragraph).In addition, aspartate may be elevated despite no significant differences in TCA cycle metabolites in sarcoma cells because of increased glutamate-oxaloacetate transaminase 2 (GOT2) expression.GOT1/2 act to generate aspartate from the TCA cycle (see Supplementary Fig. 2E, 2G; p. 8, first paragraph).