Lung squamous cell carcinoma exhibits a targetable glucose dependency unique among non-small cell lung cancers

ABSTRACT

Cancer cells consume high amounts of glucose for their cellular bioenergetic and anabolic requirements, relying on glucose to fuel their growth. We recently reported that lung squamous cell carcinoma, a major subtype of non-small cell lung cancer (NSCLC), exhibits remarkably elevated glucose transporter GLUT1 (encoded by SLC2A1) expression and glucose dependency, while another subtype of NSCLC, lung adenocarcinoma, shows significant glucose independence. Our findings highlight the metabolic heterogeneity of glucose metabolism among lung cancer subtypes, which can be exploited for targeted lung cancer therapies.

Otto Warburg's pioneering discovery of the metabolic preferences of cancer cells began an age of investigation which is now on the cutting edge of cancer research. For many years after the discovery of the “Warburg effect,” tumor metabolism was viewed as relatively homogenous among different cancers, displaying a uniformly elevated metabolic preference for glycolysis and lactate fermentation rather than mitochondrial oxidative phosphorylation to supply cellular energetic and anabolic needs. However, recent work has shed light on the understanding of cancer metabolism as a highly heterogeneous set of biochemical systems with different cancers exhibiting distinct metabolic characteristics.^1,2 As the realization of cancer-associated metabolic possibilities begins to expand, keen research is now focused on targeting tumor cell metabolism for the development of novel, targeted therapies which aim to reduce the incidences of tumor resistance, recurrence, and the toxic side effects commonly associated with conventional chemotherapies.³ To achieve effective metabolic targeting of tumor cells, significant progress must be made in identifying and exploiting metabolic markers unique to the individual cancer, which may highlight vulnerabilities and dependencies of the underlying malignancy.

As the leading cause of cancer-related deaths worldwide, lung cancer is responsible for more deaths than breast, prostate, and colon cancers combined.⁴ Lung cancer is largely divided into 2 subtypes, small cell and non-small cell lung cancer (NSCLC), with approximately 85% of lung cancers clinically presenting as NSCLC. Advances in understanding NSCLC have led to the development of effective molecularly targeted therapies. However, many of these targeted alterations are relatively common in lung adenocarcinoma (LUAD), a major subtype of NSCLC, but are rarely present in lung squamous cell carcinoma (LSQCC), the other major subtype of NSCLC, demonstrating that these NSCLC tumor subtypes exhibit very distinct genetic and phenotypic features.⁵ Furthermore, recent studies have led to the significant understanding of many metabolic alterations present in NSCLC, but do not distinguish one NSCLC tumor subtype from another.^1,2 We sought to identify a specific metabolic profile that could distinguish LSQCC apart from LUAD. Surprisingly, unbiased differential gene expression profiling between LSQCC and LUAD using The Cancer Genome Atlas (TCGA) revealed that expression of the glucose transporter GLUT1 and other key glycolytic enzymes are significantly enriched in LSQCC compared with normal lung tissues and LUAD.⁶ Further histopathological analysis of clinical LSQCC and LUAD patient samples, patient tissue microarrays, patient-derived xenografts, and a genetically engineered mouse model of NSCLC all converged on a unifying metabolic feature of LSQCC, that GLUT1 expression is vastly more elevated in LSQCC than in LUAD (Figure 1).

Figure 1.

Differential GLUT1 and p63 expression in lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LSQCC). Immunofluorescent staining for glucose transporter GLUT1 (Green) and p63 (red) in LUAD (left) and LSQCC (right) xenograft tumors. Nuclei were stained with DAPI (blue).

This discovery led us to hypothesize that elevated GLUT1 expression in LSQCC augments glucose uptake and glycolytic flux, and is associated with an increased dependency on glucose metabolism unique among NSCLC. Metabolic analysis in a panel of LSQCC and LUAD cell lines uncovered distinct metabolic phenotypes that could be grouped into highly glycolytic LSQCC cell lines that were susceptible to genetic GLUT1 knockdown, and relatively glucose-independent LUAD cell lines that displayed little sensitivity to the loss of GLUT1. In addition, glycolytic inhibition using the reversible GLUT1 inhibitor WZB117 elicited a reduction in glucose uptake and extracellular acidification rate (ECAR) that was associated with increased apoptosis and cell death only in LSQCC cell lines. We further explored the functional contribution of GLUT1 to in vivo glucose uptake and metabolism by observing ¹⁸Fluoro-2-deoxy-glucose (¹⁸F-FDG) uptake in the Kras^G12D; Lkb1-null (KL) NSCLC mouse model, which develops a heterogeneous combination of LSQCC and LUAD tumors within the same animal. Our study as well as previous clinical studies have identified higher ¹⁸F-FDG uptake in LSQCC subtypes of NSCLC using positron emission tomography (PET) scanning, and we found that even within mice bearing a combination of NSCLC tumors, ¹⁸F-FDG uptake was able to distinguish LSQCC from LUAD.^6,7 In vivo pharmacological inhibition of glycolysis using the non-metabolizable glucose analog 2-deoxy-D-glucose or WZB117 was sufficient to suppress tumor growth in LSQCC but not in LUAD xenografts.

Multiple oncogenic drivers associated with elevated glycolysis in cancers, such as KRAS activating mutations, are rarely exhibited in LSQCC patients, yet the dramatically high glycolytic capacity of LSQCC suggests that other physiologic, microenvironmental, or oncogenic processes must be driving elevated glucose metabolism.⁸ Previous studies and our analysis of TCGA data revealed that LSQCC patients exhibit a frequent genomic amplification of the catalytic subunit of the PI3K complex (PIK3CA) that results in the highest PIK3CA expression among all TCGA cohorts.^6,9 Oncogenic PI3K/Akt/mTOR signaling is known to increase glycolytic capacity through multiple mechanisms including increased stabilization of HIF-1α under non-hypoxic conditions,¹⁰ therefore we hypothesized that aberrant PI3K-mediated translational stabilization of HIF-1α contributes to elevated GLUT1 in LSQCC. Consistently, we found a strong association between PI3K/AKT/mTOR signaling and GLUT1 expression using TCGA proteomic data and histopathological analysis of LSQCC tumors in xenografts and KL mice. Notably, genetically depleting HIF-1α dramatically diminished GLUT1 expression in LSQCC cell lines. However, PI3K/AKT/mTOR signaling is highly activated in a vast variety of human cancers including LUAD, yet GLUT1 levels in these tumors are significantly lower than LSQCC suggesting that additional LSQCC-specific mechanisms regulating GLUT1 expression are involved. Further study is required to better elucidate oncogenic as well as microenvironmental cues contributing to the unique GLUT1 expression and glycolytic metabolism in LSQCC.

Our study provides the rationale for a novel LSQCC-specific metabolic targeting strategy, but more importantly, emphasizes the need to consider the unique metabolic characteristics of individual cancer subtypes beyond the generalized “Warburg effect.”

Acknowledgments

The authors apologize to colleagues whose work could not be cited due to space limitations. Our original work is partially supported by the National Cancer Institute (NCI) of the National Institutes of Health (NIH), R21CA208746 (J.K.) and American Lung Association, LCD-400239 (J.K.).