Abstract
Background/Aim: In the present study we compared the glucose and glutamine requirements of cancer and normal cells to determine if the Warburg effect is cancer specific.
Materials and Methods: 143B human osteosarcoma, HT1080 human fibrosarcoma, HCT116 human colon cancer and normal Hs27 human fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with and without glucose; with and without glutamine; or with and without methionine. The EC50 of glucose, glutamine and methionine was compared in cancer and normal cells. Co-culture of Hs27 normal fibroblast with each cancer cell line was performed by using 12-well plates with and without glucose or methionine. Cell viability was determined with the WST-8 viability reagent, by phase-contrast microscopy or fluorescence microscopy.
Results: The EC50 of glucose for the three cancer cell lines ranged from 0.54 to 4.88 mM. The EC50 of glucose for Hs27 normal fibroblasts was 0.35 mM, which was not significantly lower than in HCT116 cells (p=0.2225). The EC50 for glutamine ranged from 0.15 to 0.54 mM for the cancer-cell lines and 0.24 mM for normal fibroblasts, which did not distinguish normal from cancer cells. For comparison the EC50 of cancer cells for methionine ranged from 3.8 μM to 21.4 μM while for normal fibroblasts the EC50 for methionine was 2.3 μM, which was significantly lower than in all the cancer cell lines (p<0.0167). In co-culture of cancer and normal fibroblasts, glucose-free or glutamine-free medium resulted in loss of cell viability by day 7 for both the cancer and normal cells. In contrast, in methionine-free medium, the normal fibroblasts were alive and healthy at day 7.
Conclusion: The Warburg effect of glucose and glutamine addiction is not cancer specific in comparison to methionine addiction (Hoffman effect), which is cancer specific, suggesting the Warburg effect is not a cancer paradigm.
- Glucose
- glutamine
- methionine
- requirement
- vulnerability
- co-culture
- cancer cells
- normal fibroblasts
- Warburg effect
- Hoffman effect
- cancer-specific
- paradigm
Introduction
Warburg’s discovery nearly 100 years ago of the high glycolysis rate of cancer cells, even in the presence of oxygen, suggesting an apparent addiction of glucose by cancer cells, came to be known as the Warburg effect (1). In the present century, numerous studies have been carried out on the Warburg effect as a paradigm for cancer research and treatment [reviewed in (2)]. There have been numerous publications from Craig B. Thompson’s group claiming glutamine dependence in cancer is a result of the Warburg effect and that glutamine dependence is a cancer-specific vulnerability (3-11).
However, no direct head-to-head experiments have been carried out on the glucose and glutamine requirement of cancer and normal cells under the exact same conditions.
In 1959, it was discovered by Sugimura et al. that in tumor-bearing rats fed chow depleted of methionine, tumor inhibition was greater compared to feeding chow depleted of other amino acids (12). In the early 1970s, a study found that cancer cells in culture were dependent on methionine and unlike normal cells were unable to grow when the methionine precursor homocysteine replaced methionine (13). Cancer cells were also first shown to be selectively sensitive to the enzyme methioninase (14). One of us (RMH) showed in 1976 that cancer cells were addicted to methionine (the Hoffman effect); the cancer cells, unlike normal cells, required an external source of methionine despite making greater-than-normal amounts of methionine from homocysteine (15). It was then shown that cancer-cell addiction to methionine was at least in part due to elevated transmethylation reactions in cancer cells (16-18).
Methionine addiction can be targeted by methionine restriction, including by using recombinant methioninase (rMETase), which selectively inhibits cancer cell growth and metastasis and sensitizes cancers to chemotherapy (19).
Positron-emission tomography (PET) imaging has exploited the Warburg effect using [18F]-fluorodeoxyglucose (20). Recently PET imaging used [11C]-methionine to exploit the methionine addiction of cancer (21, 22), indicating that methionine addiction of cancer (Hoffman effect) is stronger than the Warburg effect.
The present study compared the glucose, glutamine and methionine requirements of cancer and normal cells in monoculture and co-culture, where the conditions were exactly the same for cancer and normal cells, in order to determine whether the Warburg effect is cancer specific like the Hoffman effect.
Materials and Methods
Cell culture. The 143B human osteosarcoma cell line, the HT1080 human fibrosarcoma cell line, the HCT116 human colon cancer cell line and Hs27 normal human fibroblasts were obtained from the American Type Culture Collection (Manassas, VA, USA). Green fluorescent protein (GFP) -expressing HCT116 cells, and red fluorescent protein (RFP) -expressing 143B and HT1080 cells, were established as described elsewhere (23). The cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Thermo Fisher Scientific, Waltham, MA, USA) in an incubator at 37°C with 5% CO2.
Determination of the half-maximal effective concentration (EC50) of glucose, glutamine and methionine for cancer and normal cells. Each cell line was seeded (2.0×103 cells) in a 96-well plate with normal DMEM (100 μl/well) and incubated at 37°C overnight. Methionine-, cysteine-, glutamine-, or glucose-free DMEM, supplemented with 10% dialyzed fetal bovine serum, 1% penicillin/streptomycin was prepared. Methionine-restricted medium was prepared by adding L-cystine 2HCl (150 μM) and L-glutamine (4 mM Thermo Fisher Scientific, Waltham, MA, USA). Glutamine-restricted medium was prepared by adding L-cystine 2HCl (150 μM) and L-methionine (100 μM). Glucose-restricted medium contained methionine, glutamine and cystine as described above. Each cell line was treated with these media with different concentrations of methionine or glutamine or glucose at 37°C for 96 h. For determining the methionine requirement, concentrations of 0, 1, 2, 4, 8, 16, 32, 64 and 128 μM were added; for determining the glucose requirement, concentrations of 0, 0.125, 0.25, 0.5, 1, 2, 4, 8 and 16 mM were added; for determining the glutamine requirement, concentrations of 0, 0.03125, 0.0625, 0.125, 0.25, 0.5, 1, 2 and 4 mM were added. After the treatment period, the cell-viability reagent WST-8 (10 μl; Dojindo laboratories, Kumamoto, Japan) was added to each cell-culture well and the resulting absorbance at 450 nm was measured after 1 h. The effective concentration (EC50) values for glucose, glutamine and methionine for each cell line were estimated by nonlinear regression using a four-parameter logistic model (GraphPad Prism, version 10.6.1; GraphPad Software). Experiments were performed in triplicate.
Co-culture of cancer and normal cells to determine vulnerability to glucose depletion. Each cancer-cell line (5.0×104 cells) was seeded into 12-well plates together with Hs27 normal fibroblasts (5.0×104 cells). The day after seeding, after washing with phosphate-buffered saline (PBS), the medium in each well was replaced as follows: Control medium [with methionine and glucose (MET+ GLU+), DMEM, GlutaMAX™ supplement; Thermo Fisher Scientific, Waltham, MA, USA]. The methionine-restricted medium (MET− GLU+) and glucose-restricted medium (MET+ GLU−), composition was the same as in the EC50 experiment. At 4 and 7 days after the medium was replaced, the wells were washed with PBS twice. Experiments were performed in triplicate and analyzed by phase-contrast and fluorescence microscopy.
Microscopy. Phase-contrast microscopic and fluorescence microscopic images of green or red fluorescent protein were acquired from the same microscopic field with an Olympus IX71 microscope (Olympus corp. Tokyo, Japan) on days 0, 4 and 7.
Statistical analysis. Statistical analysis was performed using Welch’s t-test for co-culture experiments and using extra-sum of squares F-test for EC50 analysis (GraphPad Prism, version 10.6.1; GraphPad Software). Data are expressed as the mean±standard deviation (95% confidence interval) for the EC50 experiments. A value of p≤0.0167 was considered statistically significant for the EC50 experiments due to the Bonferroni correction (Hs27 vs. 143B, HT1080 and HCT116).
Results
EC50 of glucose for cancer and normal cells. The EC50 of Hs27 normal fibroblasts for glucose was 0.35 mM. For cancer cells, the EC50 was 2.89 mM for 143B osteosarcoma cells; 4.88 mM for HT1080 fibrosarcoma cells; 0.54 mM for HCT116 colon cancer cells. There were significant differences between the EC50 of glucose for 143B and HT1080 cancer cells compared to normal fibroblasts (p =0.0080, p<0.0001, respectively). However, there was no significant difference between the EC50 of glucose for Hs27 normal fibroblasts and the EC50 of HCT116 colon cancer cells for glucose (p=0.2225, Figure 1).
EC50 of glucose for cancer and normal cells. The EC50 of glucose for Hs27 normal fibroblasts was 0.35 mM; for 143B osteosarcoma was 2.89 mM; for HT1080 fibrosarcoma was 4.88 mM; for HCT116 colon cancer was 0.54 mM (p=0.0080, p<0.0001, p=0.2225, respectively compared to Hs27 normal fibroblasts). Thus, there was no significant difference between the glucose EC50 for Hs27 normal fibroblasts and HCT116 colon-cancer cells. GFP: Green fluorescent protein; RFP: red fluorescent protein. The EC50 is half the concentration that results in maximum cell viability. Please note that glucose concentrations on the x-axis are plotted on a logarithmic scale.
EC50 of glutamine for cancer and normal cells. The EC50 of glutamine was 0.15 mM for HCT116 colon-cancer; 0.54 mM for 143B osteosarcoma; 0.54 for HT1080 fibrosarcoma cells. The EC50 of glutamine for Hs27 normal fibroblasts was 0.24 mM. There were significant differences in the EC50 of glutamine between 143B and HT1080 cancer cells compared to Hs27 normal fibroblasts (p<0.0001 and p<0.0001, respectively). However, there was no significant difference in the glutamine EC50 between the HCT116 colon-cancer cells and Hs27 normal fibroblasts (p=0.0730, Figure 2).
EC50 of glutamine for cancer and normal cells. The EC50 of glutamine for Hs27 normal fibroblasts was 0.24 mM; for 143B osteosarcoma was 0.54 mM; for HT1080 fibrosarcoma was 0.54 mM; and for HCT116 colon cancer was 0.15 mM. The EC50 of glutamine for Hs27 human fibroblasts was within the range as cancer cell lines (0.15-0.54 mM). GFP: green fluorescent protein; RFP: red fluorescent protein. The EC50 is half the concentration that results in maximum cell viability. Please note that glutamine concentrations on the x-axis are plotted on a logarithmic scale.
EC50 of methionine for cancer and normal cells. The methionine EC50 for Hs27 normal fibroblasts (2.3 μM) was significantly lower than the EC50s for all cancer cell lines (15.7 μM for 143B osteosarcoma, p<0.0001; 21.4 μM for HT1080 fibrosarcoma, p<0.0001; 3.8 μM for HCT116 colon-cancer, p=0.0099. Figure 3).
EC50 of methionine for cancer and normal cells. The EC50 of methionine for Hs27 normal fibroblasts was 2.3 μM; for 143B osteosarcoma was 15.7 μM; for HT1080 fibrosarcoma was 21.4 μM; and for HCT116 colon cancer was 3.8 μM (p<0.0001, p<0.0001, p=0.0099, respectively compared to Hs27 normal fibroblasts). The normal fibroblasts had an EC50 for methionine significantly lower than any of the cancer cells. GFP: Green fluorescent protein; RFP: red fluorescent protein. The EC50 is half the concentration that results in maximum cell viability. Please note that methionine concentrations on the x-axis are plotted on a logarithmic scale.
Co-culture of cancer and normal cells to determine the cancer-specific vulnerability to glucose or methionine depletion. In media with both methionine (MET+) and glucose (GLU+), all cancer cell lines and normal fibroblasts grew very well, with cancer cells dominating the culture. In the MET+ GLU− condition, neither the cancer cells nor normal cells remained viable by day 7. This result can be seen by the disappearance of the cancer cells using both fluorescence and phase-contrast microscopy. Phase-contrast microscopy distinguishes normal and cancer cell by their very different morphology. Under MET− GLU+ condition, all cancer cells lost viability by day 7 as seen by loss of fluorescence. However, the normal fibroblasts survived under the MET− GLU+ condition as seen by phase-contrast microscopy. Representative images are presented in Figure 4, Figure 5 and Figure 6.
Co-culture of red fluorescent protein (RFP)-expressing 143B osteosarcoma cells and Hs27 normal fibroblasts to determine the cancer-specific vulnerability to glucose restriction or methionine restriction. In the medium with both methionine and glucose (MET+ GLU+), normal and cancer cells grew very well, and cancer cells dominated the culture. In the medium without methionine but with glucose (MET− GLU+), the Hs27 normal fibroblasts were alive and healthy at day 7. In contrast, the 143B osteosarcoma cells were not viable at day 7, as visualized by phase-contrast microscopy and by red fluorescent protein. Cell viability in the medium with methionine but without glucose (MET+ GLU−) was lost by day 7 for both 143B and Hs27 normal fibroblasts. Experiments were repeated three times. Representative images are shown. Scale bar=100 μm.
Co-culture of red fluorescent protein (RFP)-expressing HT1080 fibrosarcoma cells and Hs27 normal fibroblasts to determine the cancer-specific vulnerability of glucose restriction or methionine restriction. In medium with both methionine and glucose (MET+ GLU+), normal and cancer cells grew very well, and the cancer cells dominated the culture. In medium without methionine but with glucose (MET− GLU+), the Hs27 normal fibroblasts were alive and healthy at day 7. In contrast, the HT1080 fibrosarcoma cells were not viable at day 7 as visualized by phase-contrast microscopy and by RFP fluorescence. Cell viability in the medium with methionine but without glucose (MET+ GLU−) was lost by day 7 for both the HT1080 fibrosarcoma and Hs27 normal fibroblasts. Experiments were repeated three times. Representative images are shown. Scale bar=100 μm.
Co-culture of green fluorescent protein (GFP)-expressing HCT116 colon-cancer cells and Hs27 normal fibroblasts to determine the cancer-specific vulnerability of glucose restriction or methionine restriction. In medium with both methionine and glucose (MET+ GLU+), normal and cancer cells grew very well, and the cancer cells dominated the culture. In medium without methionine but with glucose (MET− GLU+), the Hs27 normal fibroblasts were alive and healthy at day 7. In contrast, the HCT116 colon-cancer cells were not viable at day 7 visualized by phase-contrast microscopy and by GFP fluorescence. Cell viability in the medium with methionine but without glucose (MET+ GLU−) was lost by day 7 for both the HCT-116 colon-cancer cells and Hs27 normal fibroblasts. Experiments were repeated three times. Representative images are shown. Scale bar=100 μm.
Discussion
For the past 100 years, the Warburg effect of glucose addiction (1-11) has been the paradigm of cancer-specific metabolism. In the present century, glutamine dependence has become part of the paradigm of cancer-specific metabolism related to the Warburg effect (3-9). In the cancer clinic, positron-emission tomography with [18F]-fluorodeoxyglucose glucose (FDG-PET) can be used to detect cancer due to the Warburg effect (20).
The results of the present study show that cancer cells and normal cells are not distinguished by their EC50 for glucose or glutamine, nor by their vulnerability to glucose restriction. In contrast, normal and cancer cells are distinguished by the higher EC50 of the cancer cells for methionine. The co-culture experiments show that the cancer cells, and not the normal cells, are vulnerable to methionine restriction. In contrast both normal and cancer cells are vulnerable to glucose restriction. Our previous study showed in co-culture of cancer and normal cells that glutamine dependence is not a cancer-specific vulnerability (24). However, methionine restriction is a cancer-specific vulnerability, unlike glucose or glutamine restriction. A limit of the present and previous study is that they were performed in vitro.
The present results are consistent with previous studies in which [11C]-methionine-PET imaging showed stronger, more specific signals than [18F]-FDG-PET of the same patients (21, 22).
The present results show the Hoffman effect of methionine addiction is cancer-specific and the Warburg effect of glucose/glutamine addiction is not. Currently, cancer therapeutics based on the Warburg effect are being developed that block various steps of glycolysis (25). The present results suggest that methionine addiction may be a better therapeutic target since it is cancer-specific unlike glucose and glutamine dependence. Early studies suggest the potential clinical promise of methionine restriction (26, 27). Although the Warburg effect is the current paradigm for cancer (1-11, 25), the present results suggest a paradigm shift to the Hoffman effect (28-31).
Acknowledgements
This paper is dedicated to the memory of A.R. Moossa, MD; Professor Philip Miles; Sun Lee, MD; Richard W. Erbe, MD; Professor Milton Plesur; Professor Gordon H. Sato; Professor Li Jiaxi; Masaki Kitajima, MD; Shigeo Yagi, Ph.D.; Jack Geller, MD; Joseph R Bertino, MD; J.A.R. Mead, Ph.D.; Eugene P. Frenkel, MD; John Mendelsohn, MD; Professor I.J. Fidler; Professor Lev Bergelson; Professor Sheldon Penman; Professor John R. Raper; Professor Peter H. Duesberg; Professor J.D. Watson; and Joseph Leighton, MD. May their memory be a blessing.
The Robert M. Hoffman Foundation for Cancer Research provided funds for the present study.
Footnotes
Authors’ Contributions
YM and RMH designed the study. KM, TI, BMK, JSK, QH, SL, BM, YT and KN critically reviewed the manuscript. YM performed the experiments. YM was the major contributor to writing the manuscript, and RMH revised the manuscript. All authors read and approved the final manuscript.
Conflicts of Interest
The Authors declare no competing interests in relation to this study.
Artificial Intelligence (AI) Disclosure
No artificial intelligence (AI) tools, including large language models or machine learning software, were used in the preparation, analysis, or presentation of this manuscript.
- Received January 15, 2026.
- Revision received February 14, 2026.
- Accepted February 24, 2026.
- Copyright © 2026 The Author(s). Published by the International Institute of Anticancer Research.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.












