UNIVERSITY PARK, Pa. — By now most of us know we’re not alone. From one perspective, in fact, our bodies are merely the host for a teeming biological horde. Many aspects of our lives — not only the presence or absence of certain diseases, but conditions like obesity, sleep patterns, even mood — may be determined, to a surprising extent, by the microbes living inside of us. 

Although the concept of the microbiome, the sum total of our resident bacteria, viruses, and fungi, has only recently been popularized, “It’s not a new idea that microbes have influence in the body,” Andrew Patterson said. “But it’s only now that technology is allowing us to see how microbes exert that influence, and to measure it.”

Patterson, Tombros Early Career Professor and professor of molecular toxicology at Penn State, is using one of the newer and more promising of these technologies, called metabolomics, to learn about the microbiome of the human gut. 

Metabolomics is the measurement of all the chemical products of metabolism present in a given biological sample, typically blood, urine or another bodily fluid. These “metabolites,” teased apart and painstakingly identified, provide a window into cellular processes, and can therefore be important indicators of disease, or of an individual’s response to a drug, an environmental toxicant, or even the chemical compounds in our diet.

Researchers in the field speak of the "unique chemical fingerprints" cellular processes leave behind. To use another metaphor, a metabolomic sample is like the transcript of a conversation — a precise account of the complex molecular signaling between resident microbes and host cells that ends in a physiological response.

Patterson relates the concept in still more common terms. “Today, you go to the doctor, you give blood, and they screen for maybe a dozen things,” he said. “We’re just doing that on an orders-of-magnitude larger scale, looking at as much as we possibly can.”

Birth of a technology

True metabolomics has its roots in the 1970s, when advances in mass spectrometry and other analytic techniques began to allow for increasingly sensitive measurement and separation of complex chemical samples. Powerful bioinformatics tools developed over the next few decades then enabled pioneers in the field to start building databases of metabolites. The first draft of the human metabolome, consisting of some 2,500 metabolites, was completed at the University of Alberta in 2007, and similar efforts were undertaken for other plant and animal species. 

By 2009, Jeffrey Peters, now distinguished professor of molecular toxicology and carcinogenesis and deputy director of the Penn State Cancer Institute, had recognized in the emerging technology an opportunity to expand interdisciplinary research at Penn State.

“It is becoming increasingly clear that integration of genomics, transcriptomics, proteomics, and metabolomics will be required to allow for the most comprehensive approaches for disease prevention and treatment,” Peters wrote in an early draft proposal. A timely investment in metabolomics, he argued, could make Penn State a leader in the field.  

Peter Hudson, then director of the Huck Institutes for the Life Sciences, was an indispensable advocate.

“Peter was able to get support from the various colleges, Agricultural Sciences, Eberly [College of Science], Health and Human Development, Medicine — they all got behind the idea,” Peters recalled. In Hudson’s vision, the proposed Metabolomics Core Facility would be a natural fit for the sort of interdisciplinary collaboration that Penn State is known for; it would be staffed by leading experts and outfitted with a full range of state-of-the-art instrumentation. 

Patterson was identified as the right person to lead the initiative: An alumnus of Penn State, he was then employed as a postdoc at the Laboratory of Metabolism at the National Cancer Institute, one of the top metabolism laboratories in the country. 

One of the projects Patterson had embarked on at NCI illustrates both the promise of metabolomics and some of its challenges. He, section head Frank Gonzalez, Curtis Harris, chief of the Laboratory of Human Carcinogenesis, and the rest of their team were looking for a noninvasive way of detecting lung cancer, the leading cause of cancer deaths in both men and women worldwide. They decided to analyze urine samples from a cohort of over 1,000 patients, spanning differences in race, ethnicity, smoking status, stage of cancer, and other factors, such as diet and medications. It was a needle-in-a-haystack approach — exactly the sort of thing metabolomics affords.

“We went in blindly, looking to see if we could find a metabolite that all these people had in common,” Patterson said. “You’re just generating a huge amount of data and then seeing what shakes out of it.” The number of variables present complicated matters, making it difficult to separate any signal they might find from the surrounding noise. 

“Another problem, and this is true of metabolomics in general, is that the existing databases are still relatively small,” Patterson said. The largest, the Human Metabolome Database, now has over 114,000 metabolites, but there are untold numbers that have yet to be identified. 

Despite these obstacles, when the mass spectrometry data came back, the researchers were able to tease out a compound of amino acid and sugar that appeared to be shared across all their samples. Creatine riboside, they called it. When they looked through the databases for a match, however, there was none to be found. Theirs was an entirely new metabolite.

Biomarkers and bile acids

It took over a year to confirm what they had discovered. Patterson had by then moved to Penn State, and he and Philip Smith, an Army mass spectrometry expert who had been hired as director of the brand-new Metabolomics Core Facility, set about recreating the compound in the lab. They tried any number of routes, mixing different amounts of creatine and ribose together in different forms, at different temperatures, then running the results through their mass spec and NMR machines. “We’re not synthetic chemists, by any stretch,” Smith said with a wry smile. 

Finally, they hit on the right sequence of steps. A chemist back at NCI was then able to refine their results and synthesize a pure version of the molecule.

“Once we had that,” Patterson said, “we could come back and compare it with all our samples, and we found out that indeed we had what we thought we had.” 

Creatine riboside has since been found in tumor tissue as well as in urine, and also associated with several other types of cancer. “It may be not just a good biomarker for early stage lung cancer, but a general marker of cancer,” Patterson said. He is working with Peters and others at the Penn State Cancer Institute to set up a clinical trial, screening lung-cancer patients at Hershey Medical Center for the presence of the metabolite.

In the meantime, Patterson has shifted his emphasis away from cancer. He’s now using metabolomics to learn how the presence or absence of certain types of bacteria in a person’s digestive tract can affect processes associated with drug metabolism, obesity, diabetes, and fatty liver disease.  

“Our broad hypothesis is that these bacteria modify our host cells in order to promote an environment that is good for themselves,” he said. One way they do this is by making ligands, small signaling molecules that bind to specific cellular receptors and hijack the cell’s activity. Among the favorite targets for these disruptions are bile acids. 

There are over 150 types of bile acid in the human gut, Patterson explains. Released by the liver, they play an important role in digestive processes, particularly in dissolving fat. Recently, however, it has become clear that some bile acids also act as signaling molecules or are toxic to bacteria. Some bacteria, in turn, have developed ways to protect themselves from harm by modifying bile acid exposure. For the human or animal host, said Patterson, “this modification can have significant effects on lipid and glucose metabolism."

In one instance, a colleague studying tempol, an antioxidant compound commonly used to reduce the side effects of radiation therapy, noticed that mice given the drug did not gain weight like other mice, even when fed a high-fat diet. After trying unsuccessfully to determine why this was so, he turned to Patterson and his colleagues at NCI. 

Gut samples showed that giving mice tempol resulted in a sharp drop in Lactobacillus, a type of bacteria that is capable of modifying bile acids. As it turns out, reducing Lactobacillus allowed one specific bile acid, tauro-beta muricholic acid, to increase. The team’s analysis showed that this increase inhibited the activity of the farnesoid X receptor (FXR), which regulates the metabolism of bile acids, fats and glucose in the body. Without FXR, the mice’s digestion of fat was significantly altered, and obesity was curtailed.

Building on this finding, Patterson and colleagues at Hershey and NCI have developed a pill modeled after tauro-beta muricholic acid, intended to target FXR directly. In mouse trials, the drug has proved effective in lessening weight gain and insulin resistance in the face of a high-fat diet. Patterson has formed a company, Heliome Biotech, to commercialize the discovery.  

A whole new perspective 

With the establishment of the Metabolomics Core Facility in University Park, Penn State has indeed carved out a niche in this emerging field. 

“We’ve been able to attract a lot of collaboration, from here and elsewhere, because we have a really unique set-up that allows us to do a lot of different things,” Patterson said. “We have an expertise in the microbiome space that few others have right now.” 

The facility is administered jointly by the Huck Institutes for the Life Sciences and the Penn State Cancer Institute, reflecting high hopes for the potential of metabolomics as a tool for personalized medicine. Because it provides direct evidence of cellular processes, Patterson said, “It gives us a whole new perspective from which to look at health and disease.” 

There are significant challenges, however, to the technology’s wider use. One of them is cost. “It’s an extremely expensive undertaking,” Patterson acknowledged. “It costs millions of dollars to outfit these labs, because you need different platforms to be able to measure the metabolites in as comprehensive a way as possible.” 

Even if a sample gets a good, thorough analysis, he added, a single snapshot is of limited use for evaluating physiological processes. “The key to understanding what’s going on is to be able to measure changes over time,” he said. Doing that would require a massive increase in routine clinical screenings, and raise issues of consistency and reproducibility that have yet to be addressed.

Not to mention capacity. “You’re talking about millions of data points for each sample,” Patterson said, “Then combine that with the data available from genomics, transcriptomics and proteomics. The biggest challenge will be how to integrate all of that in a way our human brains can make sense of.” But if and when it can be done, he added, “the possible applications of metabolomics are endless. 

“Right now we’re using it to identify biomarkers and to search out metabolic pathways. But it can also help us to better understand how the body responds to daily exposures to drugs and environmental chemicals.” 

One day, Patterson hopes, he’ll be able to use metabolomics to predict how a patient will respond to a given drug, allowing treatment choices to be tailored to the individual and the individual’s microbiome.

“That’s the thing that probably excites me the most,” he said. “We’re finding that drugs and chemicals that were thought to work only through our host cells actually very profoundly influence the microbes we carry as well.

“The understanding that toxicology is dictated not only by our genetics but also by the microbes residing inside us is a completely untapped area."

This story first appeared in the Fall 2019 issue of Research/Penn State magazine.