Nothing in a flamingo’s genetic code says that it should be pink. Instead, by snacking on the carotenoid-rich algae and shrimp in its environment, the flamingo takes on its characteristic rosy hue. Like flamingos, humans are also products of our environment. While genetics determine much of human biology, factors like the food we eat, the medicine we take, and the diseases we develop give a more complete picture of human health.

The many factors that contribute to health in small and large ways may seem infinite, but metabolites — the small molecules produced as byproducts when cells break down food, drugs, or other chemicals — give scientists a window into the molecular processes that drive human health. For example, metabolites produced by healthy cells often differ from those released by cancer cells. Differences in individual metabolism can make certain drugs effective in some people but ineffective in others.

Taking a systems-level approach, Stefano Tiziani, a chemical biologist at the University of Texas at Austin, studies the metabolome in the context of cancer to find more effective cancer treatments and to identify people who may or may not benefit from a particular chemotherapeutic treatment.

How can understanding the metabolome lead to the development of better cancer drugs?

One incredible example of how understanding the metabolome led to a successful chemotherapeutic therapy is in amino acid depletion therapy for kids with acute lymphoblastic leukemia (ALL). One of the drug treatments is the enzyme asparaginase, which simply depletes the concentration of the metabolites asparagine and glutamine. Due to a change in their metabolism, all cancer cells are under a large pressure to uptake these metabolites, so depleting them is part of the successful chemotherapeutic treatment.

In my research, we want to understand how the metabolic response of a person’s cells correlates with their response to chemotherapeutic treatment. By doing this, we can potentially identify those patients who don't respond to treatment and why they don't respond. From the metabolomic readout, we can come up with a new treatment strategy for those patients. We also use metabolomics to identify chemotherapy drugs that might be more effective in combination than when used alone.

How do you use metabolomics to identify new cancer drug combination treatments?

We recently developed a new method to test the synergistic combination of two drugs for prostate cancer by assessing the metabolome (1). “Synergistic” is like the word “significant” in that it is a statistical term. We can’t just say that two drugs work well together; we need to quantify it. To do that, we developed a new algorithm called principal component analysis-based Euclidean distance synergy quantification (PEDS). 

We realized that when the majority of people apply any -omics analysis to drug synergy questions, they determine the synergistic combination based on one variable such as cell survivability or apoptosis assays. That doesn't really take advantage of the systems biology platform. Our PEDS algorithm can take hundreds of variables and assess the overall metabolism to determine if two drugs or natural compounds either synergize or antagonize. With this information, we can assess how this drug combination acts at the molecular level. We can identify biomarkers associated with the mitochondria such as those involved in oxidative phosphorylation, the TCA cycle, and glycolysis, for example. With this metabolic information, we can characterize how cells respond to treatment and evaluate whether the patient would successfully respond to treatment or not.

What was your reaction when you identified two drugs that synergized well for prostate cancer treatment?

The two drugs — the glutaminase inhibitor CB-839 and docetaxel, an approved cancer drug — came out synergistically in a completely unbiased way. They worked very well in vitro in the 3D prostate cancer model and in mice, which we were very pleased to see. The beauty of this combination was that it came out of screening about 300 compounds, not a thousand compounds, because I run a lab, not a core facility.

We’re partnering with a number of hospitals and medical research institutes to test and analyze data from clinical trials to generate new hypotheses for potential drug combinations. We are already working on leukemia and glioblastoma, and we have other drugs that we are working on. If we can identify a new combination, hopefully it will benefit the patient and be more selective and less toxic than current cancer treatments.

How much more complicated would it be to look for synergy between three drugs rather than two?

I think it’s feasible. The main challenge will be optimizing the drug concentrations and ratios to use. When we want to move from cell lines to animal models, we need know the appropriate concentrations and ratios of drugs. Our paper gives a good foundation, so we would translate the algorithm from a two-drug combination to a three-drug combination to an "n" combination. The number of possibilities would increase dramatically.

What excites you most about developing these metabolomic screens and algorithms?

One of the most exciting parts about metabolomics is that we can collaborate with so many people — researchers who work at the genomic, proteomic, and epigenetic levels. We can use metabolomics not only for drug screening but for phenotypic screening. For example, after someone drinks a cup of coffee or eats something high in omega three fatty acids, metabolomic screening can identify how their metabolism switches. Think about it. One day when we take a blood test, the doctor will evaluate not only 10 to 50 biomarkers, but potentially thousands. Can you imagine how much more information will come from a simple test? We can potentially use these metabolic biomarkers to predict, for example, the risk of developing a disease and start preventative treatments sooner.

Reference

Lu, X. et al. Metabolomics-based phenotypic screens for evaluation of drug synergy via direct-infusion mass spectrometry. iScience  25, 104221 (2022).