This is a loose transcription of Chris Armstrong’s presentation at Open Medicine Foundation’s Community Symposium, “Molecular Basis of ME/CFS” in August 2017. Some additional notes have been added.
Thank you very much everyone. I’m going to be talking to you today about ME, Metabolism and I, which will be largely focussing on metabolomics, and the studies that we have done in ME/CFS.
I’ll briefly provide an overview of metabolomics. Essentially metabolites are small organic molecules that exist in many different parts of your body, in many different biofluids. They engage with the external environment, and are altered by the external environment. They initiate genes which will create proteins, which will then alter the metabolites themselves. So metabolites sit within a large dynamic of different things within the cellular process, and that’s why they give us so much information about what’s going on in the cells themselves.
On the right of this slide, there are some graphs, which provide a representation of metabolomics and (probably more well-known fields) proteomics and genomics. In response to an external stress of any kind, there will be rapid changes in the metabolomics information. During the same time period, there will be slow change in the proteomics and, in the genomics, there will be very minimal change. A stressor can include things like like eating, waking up or exercise. Metabolomics will give us information about all those different things, and it is constantly changing. This is one of the reasons why we are looking at metabolomics for ME/CFS, and why we started researching it about 6 or 7 years ago. Given how fluctuating the symptoms are in ME/CFS patients, we wanted to see how sensitive the metabolomics are. We wanted to see whether we could reliably find metabolites that are strongly related to symptoms in people with ME/CFS, and at the same time look at the underlying pattern of the illness.
Today, I am going to describe the metabolomics work which we have done in ME/CFS. We initially did a very preliminary metabolomics study in 2012. From there we did a much larger study, which we separated into two separate papers. Today, for the first time I will describe the results of these studies together, and then I will describe what we think these results mean.
Because there are such large differences between males and females when it comes to metabolite changes, we looked at was only a female cohort. We had 35 ME/CFS patients, versus 24 what we call “non-ME/CFS patients”. We say “non-ME/CFS” because “healthy” is such a hard thing to define. We just wanted people who weren’t diagnosed with ME/CFS, and were relatively healthy. We looked at urine, blood and faecal samples, which was all collected within a 24 hour period. We also took symptom questionnaires from the patients at the time. From the faecal samples, we also looked at the microbes, getting fresh quantitation of the microbes that were inhabiting the colon at the time. Then we did metabolomics analysis on the urine, blood and faecal samples, and put this altogether in a glorious, large data matrix, so we could really assess that information both as a whole, as well as individual parts.
Before I describe our results, I wanted to show this physiological disorder slide because, after I’ve given talks in the past, people have asked me “Is this information of a physiological disorder?”. These are probably the graphs that indicate to me the best (or the most simply) that ME/CFS is a physiological disorder. I’m not going to explain how the graphs work, but essentially a computer will use algorithms to divide data based on patterns that it observes, and will separate different samples within it. We don’t actually tell the algorithm whether the data points are from people with ME/CFS or controls. After the fact, we label which ones are controls and which are ME/CFS, and you can see that the computer already does a good job of separating them out. Even in the top graphs, that’s just information straight out of the machine. We can just put blood, urine and faecal samples straight into the NMR, take the spectrum out, chop it up into little pieces and put it into this algorithm and it can tell you that there is a defined difference between ME/CFS and controls.
Now, if you’re looking at something like diabetes, where you have a strong biomarker in glucose, you would get quite a strong change, quite significant from controls, because glucose plays such an important role in diabetes. So what these graphs also tell us is that there’s no real biomarker that we can see from our samples yet, however you can see the broad change over the metabolism, and that there is a definable difference.
Looking at our research, these were the changes in the metabolites that we saw from blood, urine and faecal samples. Of the 29 metabolites that we looked at, 7 were significantly altered within the blood. Eight of the 30 metabolites in the urine were statistically altered. And 4 of the 24 metabolites in the faecal samples were significantly altered.
What do we think is occurring in these patients? Given this information that we got from this metabolite data, we wanted to come with a broad mechanism that could describe what is occurring. We also had the microbe data, and we found significant differences in the bacteria between the two groups. We found an increase in certain bacteria that are a bit more scavenging in nature, the clostridium. And we have, in the past, also seen a lot of bacterial overgrowths.
An overview of what happens in a healthy individual. First, you would eat some food, which would typically contain complex carbohydrates, sugars, fat and protein. The sugar would be absorbed from the small intestine into the blood and then into the cell. The complex carbohydrates would make their way into the colon, to be digested into short chain fatty acids (SCFA), which are important for intestinal health. The fat and protein would be digested into fatty acids and amino acids in the small intestine, and would also be absorbed into the cell itself. Those fatty acids and amino acids are important for creating other cellular proteins and cell fats within the cell, as well as for making enzymes and bile acids which then feed back into the small intestine to be used to digest other fats and proteins, making a nice little cycle. The orange dots across the top of the slide represent bacteria in the gut.
There are three main pathways for energy production in the cells. The first involves fatty acids being used by the mitochondria (represented by the green oval on the slide) to produce energy. This pathway also uses oxygen, and so is called an aerobic pathway. The second pathway involves glucose being converted to acetyl CoA and being used by the mitochondria with lactic acid production as a byproduct. The third pathway involves glucose but not involving the mitochondria. The final pathway does not involve oxygen, and so is called an anaerobic pathway.
The tables in the bottom right hand side of the graph show the relative values for how much energy per molecule is produced per molecule of sugar used. You can see, per molecule, not using oxygen is quite inefficient, but given that this pathway is much faster in nature, over time it will produce a lot of energy.
AMPK (adenosine monophosphate-activated protein kinase) is important to note. It is a major protein metabolic switch within cells.
Contrast this with a low energy system, like ME/CFS.
First off, AMPK detects that there are low nutrients and energy. It turns on like a switch and, once the switch is on, AMPK prioritises the use of mitochondria for efficient energy production, and diverts resources away from other processes, because it sense the low energy environment. The body wants to make energy in the most efficient way per molecule that it can. So, fats and amino acids, which would usually be used for cell proteins and cell fats, are diverted to make energy. This process requires oxygen and the end-products of glycolysis will also be used to make energy this way. When nutrient deprivation continues for a longer period then glucose becomes important for fuelling muscles and the brain, in this instance the end-products of glycolysis are increasingly used to form glucose via gluconeogenesis instead of being used to make ATP by mitochondria.
We think that there is a long term low energy adaptation that happens over time. Because resources are being diverted away from other processes, and towards energy production, there are less cell proteins and cell fats being created which, in turn, means less enzymes and bile acids being created, which means two things: (1) that the body is less able to manage digestion within the small intestine, and (2) fewer fats and proteins feeding back into the blood (which negatively impacts energy metabolism). With the intestines less able to digest fats and proteins, this means that they aren’t getting digested when they should be, and that they stay in the intestines a lot longer, and provide substrates for more bacteria to grow there. This results in changes in the gut microbiome, either with new bacteria, changes in bacteria or more growth of bacteria (represented by the red dots on the slide).
These bacteria have the ability to digest fats and protein, which they use to feed off. This may why we see an overgrowth and a change in these bacteria in ME/CFS. They can also digest these fats into smaller amino acids, fatty acids and into more short-chain fatty acids (via anaerobic metabolism in the colon). Short-chain fatty acids are good for intestinal health, but once there are too many of them, they will overflow into the blood. Whilst they can be used by mitochondria for energy, they also can switch on AMPK as well. So it creates a kind of positive feedback loop, where AMPK can take protein away from enzymes. This is good for the bacteria, because they get more substrate (protein and fats), and so they create the sort of environment that keeps that going.
We want to know whether cellular proteins are being used in this way, and this is something we definitely would like to study in the future.
So I have covered ME and metabolism. The “I” in the title of my presentation is actually for the individual, and this brings me to future studies. As I mentioned at the start, with metabolomics the fluctuations in response to external stressors are quite rapid. So many factors can influence metabolomics, especially when combined with all the genetic and other factors that can vary between the individuals. If grouping data from many individuals together, the individual variations can cloud the results. It seems more pertinent to research the individual over time, and look at how they change from their worst to their best days (and the days in between), and see whether we can trace changes in the metabolism that explain those changes in their symptoms.
I’d like to acknowledge the funders of our research: our major funders – The Mason Foundation and Solve /MECFS Initiative. And thanks to Open Medicine Foundation for inviting me; Bioscreen for looking into the bacteria; CFS Discovery, which is the clinic from which we get our samples, which is extremely valuable, and the clinician who works with the patients, Don Lewis.
Melbourne Bioanalytics is our new website, to provide information prior to publication and to keep the community involved in the research.