Stanford Community Symposium 2018: Phair, Metabolic traps, Tryptophan trap

Questions about the metabolic trap, and some concerns​

Thanks for the feedback.

Is the genetic evidence robust?
I'll start with my main concern about the trap, the strength of evidence for a genetic problem. Phair started this work with an intriguing approach of looking for genes where every one of the 20 patients in the study had at least one (predicted) damaged copy and up came IDO2.

He then compared the frequency of damaging mutations in the 20 patients with frequency in a much larger reference population, see below:



The mutation he focused on was R248W, with a P value of 0.036, just inside the nominal P value of 0.05. However, he looked at multiple mutations, I am pretty sure more than the 5 shown here. And once you start making more than one comparison, you need to adjust P value downwards to correct for the resulting problem of false positives. 0.05 only applies to a single comparison.

And 0.036 would not survive any kind of statistical correction for multiple comparisons. That means that the original clue identifying IDO2 doesn't really hold up, which strikes me as a problem.

Understanding the biological mechanism that leads to ME/CFS

These issues may well have been covered elsewhere, in which case my questions are redundant.

Unfortunately, Phair did not have time at Stanford symposium to go into his rationale for how the metabolic trap leads to ME/CFS.


1. Serotonin
He briefly discussed that increased (or disturbed) serotonin levels, which he said would result from elevated tryptophan levels caused by the trap, could affect particular nuclei in the brain. These are the Raphé nuclei, which connect to do a whole bunch of other parts of the brain. And it’s possible to explain many of the symptoms of ME/CFS by invoking disturbances to these nuclei, it is not clear why it is the ME/CFS set of symptoms are not whole bunch of other things affected by them (including temperature).

But he only had a couple of minutes to explain things and I wondered if he'd expanded on this idea elsewhere?

Also, most of the body’s serotonin is in the gut, not in the brain. And @alex3619 has mentioned that IDO is mainly expressed in the immune cells and the liver. Again, has Phair discussed the serotonin issue in more detail elsewhere?

2. Kynurenine
Thanks to @Ravn for mentioning the kynurenine issue, where depleted kynurenine might lead to depleted levels of NAD. NAD plays a central role in mitochondrial energy production (as well as being an energy molecule itself, a bit like ATP).

3. Tryptophan
Intriguingly, Phair said that his model predicted that trap would last for eight weeks after just 10 days of high tryptophan, and after eight weeks there would be no return. Did he explain how this would happen? It wasn't clear to me from the evidence that was presented.

In particular, how would a trigger event such as a severe infection cause the high tryptophan levels needed to prime the trap?

Experimental evidence
I was impressed that Phair and colleagues had moved rapidly to try to find experimental evidence for the metabolic trap. He presented some very provisional data (for either 3 or 6 patients) supporting the model.

1. Relevant controls
Does anyone know how he chose the controls? Surely, proving the trap hypothesis requires controls that have a similar genetic make up patients. The whole idea is that patients have common mutations that put them at risk. But it is not until an unlikely trigger event, such as a prolonged infection, that the trap is sprung. So I think it's necessary to demonstrate a difference from healthy people with a similar level of damaged IDO2 genes – whose tryptophan levels should not have been caught by the metabolic trap. Maybe he is already using such controls: I seem to remember that controls were taken from family members, who presumably have a similar genetic make up. Again, any hard facts on this?

2. Link to serotonin and kynurenine
Data was shown for tryptophan and kynurenine, but if the biological effects are mediated result from changes in the levels of serotonin and NAD, we it would help to see data for those molecules as well. Maybe that's part of plan.
---

I am thinking of writing to Robert Phair, but I thought I'd ask here first, as I suspect that some of my points have already been addressed.

 

Kynurenine also has regulatory effects on T cells. We know ME patients have chronic T cell activation and we also know killing T cells with cyclophosphamide cures ME, for most of the patients at least. Metabolic trap and chronic immune activation may be a vicious cycle.

Maybe that is how an infection causes metabolic trap with immune activation and once the cells are trapped, low kynurenine levels cause continuous immune activation. Maybe, 4-6 months response time to cyclophosphamide is how long cells need to get out of the trap.

Just speculating...

 

Actually that is for IDO2. Its expressed in a few cell lines, including immune cells and liver. In immune cells it seems to be involved in regulating T cells. So its hard to see how this might lead to serotonin complications unless resident immune cells in the brain are activated.

I think the NAD connection has been discussed elsewhere, but I do not recall where.

I have some additional concerns, and possible answers, but I wont write about that now, I just got up.

 

I thought this initial investigation had no controls. He was primarily looking at the Trp/Kyn ratio.

 

Not sure anymore what I based it on but my interpretation was that they were looking for any mutation that could significantly inhibit IDO2 because it's the inhibition of IDO2 that sets the scene for the trap, not what specific mutation “broke” IDO2 in the first place, just the fact that it is “broken”. Mind you, all that is just my (mis?)interpretation.

Not that I'm aware of. Anybody else?

Not really. We've been speculating here in this thread amongst ourselves about that but nothing based on any actual info from Phair.

Likely a reflection that they haven't had time/resources to do more testing yet. But I think there was a suggestion - maybe during the panel discussion after? - that there would be tests (in Sweden?) with supplementing kynurenine and see if that helps. I'm not clear if that was “helps downstream symptoms caused by low kynurenine” or “helps undo the trap”. Maybe it was just “see what happens”.

Excellent idea. You have a lot of good questions that, as far as I know, haven't been answered yet. And if, energy permitting, you do manage to report back, even better: your writing is always so wonderfully clear and easy to follow.

 

Correct

But recently they've been saying that they find IDO2 in other places as well actually, including in the brain: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5638149/

I've found a potential answer to that question, though who knows if I'm correct. Since I found the info through contract work for Stanford, I have to sit on it until given permission to do otherwise.

As usual, I can't talk about much, guys -- though I'll alert Phair to this thread. He can choose what to discuss/address; that's the most ethical thing.

 

November statement from Robert Phair on his website....hard to interpret... running on a good track sounds good though...

Here's a link to my full talk on the IDO Metabolic Trap Hypothesis at the 2018 Stanford/OMF Community Symposium.

https://www.youtube.com/watch?v=uej1LXzRbnY

As Ron Davis says, following my talk, this hypothesis is NOT established fact. "That's what science is", he said, "constant disappointment." Then the audience laughs. But every scientist knows exactly what Ron means. Science is hard work. Just as important, though, is that those disappointments tell us how to proceed. Every experiment that proves a theory wrong gets us closer to the truth. I think we're running on a good track. Please stay tuned.

 

October statement from Phair


This is a diagram of the mathematical model used to produce the simulations I presented in my lecture at the 2018 OMF CFS Community Symposium at Stanford. Black arrows represent chemical reactions or transport. Colored rectangles represent molecules either inside or outside the cell. Red dotted lines terminated by a minus represent inhibition. Green dashed lines terminated by plus represent activation or catalysis. In my view (as of October 2018), there are many unlikely combinations of events that can drive a person into this trap. But eliminating these initial causes will not result in escape from the trap. In other words, the way out is not the reverse of the way in. In this respect the metabolic trap hypothesis is similar to the Naviaux hypothesis for CFS. The IDO metabolic trap is NOT conclusively established. We're working hard to convince ourselves that it's a real feature of ME/CFS and, if so, to find a way out of the trap. My work is funded by the Open Medicine Foundation

 

My concern here is about tissue distribution of IDO2, and what that might mean. There might not be enough of it to make much serotonin unless its really being pumped out, and the biggest organ for that is the liver. However all (this is not confirmed to be all) immune cells will be affected, and many of those reside in the brain. So the local concentrations of serotonin in the blood might be high, rather than total blood serotonin or serotonin concentration on average. It might also occur in bursts, with short very high levels, though I have no idea if this is the case it remains a possibility.

For a summary of a lot of the factors you might look here - https://www.genecards.org/cgi-bin/carddisp.pl?gene=IDO2
You can scroll down for tissue distribution. It is not clear what cell types in the brain contribute to IDO2 expression.

Note the African Sleeping Hypothesis ties into the Tryptophan Trap Hypothesis. I have not investigated this further. (Yet)

What impresses me is the fact that every single patient in the severe patient study has a problem with this gene. This is very unlikely to be by chance. The abnormal Trp/Kyn is also shown in six out of six patients. We need more data of course, with many more patients, but its looking likely as a contributing factor if not the sole cause.

Its also unlikely to be the only factor in ME. If maybe 40% of the population have an inactivating IDO2 mutation, and less than 1% have ME (possibly more like 0.2%) and only about 10% of patients (sometimes up to 20%) who have a severe infection get ME, then some other factors are involved. One possibility is infections or immune signals rising inside the brain, that is they cross the blood brain barrier. IDO1, for example, might have increased expression in inflammatory states.


I have several hypotheses about this trap. First we need to realise that as its intracellular, then not all cells expressing IDO2 may be affected.

Hypothesis 1 is that total numbers of cells affected may contribute to ME severity.

Hypothesis 2 is the pattern of distribution of affected cells might contribute to ME symptom expression.

 

Since some folks responded to Rapamune, a mTOR inhibitor, I found this interesting:

Downstream of IDO1 are three effector pathways that transduce the effects of IDO1 activity: general control over nonderepressible 2 (GCN2) is activated, mammalian target of rapamycin (mTOR) is inhibited, which is related to Trp deprivation, and the aryl hydrocarbon receptor (AhR) pathway is activated with Kyn as an endogenous AhR ligand.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6090955/

 

Thanks. Perhaps it would help if I summarise my questions here (modified after the helpful feedback, esp from @alex3619 and @Ravn).

1. Biology of causing ME/CFS

1.1 How do kynurenine, serotonin (and NAD?) leads to ME/CFS, particularly given the limited gene expression of IDO 2 (see below)?

1.2 How could a trigger, such as a severe infection, leads to initial high tryptophan levels and so spring the metabolic trap?

2. Model

2.1 The model predicted that after 10 days of high tryptophan levels the system would switch to the new high tryptophan state for 30 weeks. And after eight weeks of initial high tryptophan it would pass the point of no return. What happens, in biological terms, after 10 days to make the changes irreversible?

2.2 Would the model predicts that for most cell types in the body where IDO 2 is not expressed behave as if IDO 2 was damaged? If so, does that mean that almost all cell types would be affected? And would everyone would be at risk of these cell types falling into the metabolic trap, regardless of their ID0 2 status?

3. Experimental evidence

3.1 Are the controls matched for high levels of damage to copies of IDO 2, so that the evidence indicates the problem is due to the trap being srung, not simply due to damaged IDO 2?

3.2 Will the experiments also be measuring serotonin and NAD, if appropriate, to explain the symptoms of ME/CFS?

4 Statistics
Would it be appropriate to apply a statistical correction for the multiple comparisons made when looking at multiple individual predicted-damaging mutations?

Sorry, that's a lot of questions in the end.

--------

Tissue expression of IDO 2​

I agree with @Alex that tissue expression appears to be an important factor in the story. I think different databases give slightly different results, but reports are consistent that expression of IDO 2 is far more limited than for IDO 1. Here is data from the GT EX gene expression database:

IDO 1


IDO2:


Note that expression for whole blood, which will be based on all immune cells together, is very low. But it has been measured at significant levels in, for example, dendritic cells.

 

This is a nice summary. The only error is mine, not Simon's. My error was that I said there were, on average, 1.7 damaged copies out of 2. What the SIPS patients actually have is, on average, 1.7 (predicted) damaging mutations, and every patient has at least one. Since there are at least 5 mutations predicted to be damaging, I should have said 1.7 out of 5 or even 1.7 out of 10 since there are two copies of the IDO2 gene. Furthermore, it's always possible that a person with two damaging mutations has those mutations in the same copy (the one from Dad, say). This leaves the copy from Mom perfectly normal. So the person is effectively heterozygous for a broken IDO2. Depending on the relative expression of IDO1 and IDO2 in a given cell type, this person could still be at risk for falling into the metabolic trap. To me, the bottom line is that even SIPS patients who do not have the most common mutation (R248W) have one or more OTHER damaging mutations in the same gene. To me this was, and is, profoundly suggestive. If a person has only 2 damaging mutations, and those 2 include one on the copy from Dad and one on the copy from Mom, then the person is effectively homozygous for a broken IDO2. There's an interesting probability question here.

One more minor point is worth noting: the 10-25% figures are "attack rates" given by David Bell in his 1994 book. The range of attack rates is quite broad in CFS epidemics, perhaps 0.2% - 25%. If we are convinced there is a required genetic predisposition for CFS, then the high end of the attack rate scale seems to me to require a COMMON, not rare, mutation. CFS must then be seen as a high-prevalence, low penetrance disease with the low penetrance arising from an improbable set of simultaneous triggers.

 

Exactly. Though 'simultaneous' may not be required; rather:

• Trigger
• Second trigger before recovery from first trigger

....which may not require the triggers to occur simultaneously per se.

 

@Simon M Thanks for thinking about the metabolic trap so carefully. JaimeS sent me your questions, and as I read through them and some of the other thoughtful comments in this thread, I simply felt compelled to join s4me and be part of your conversation. Obviously doing so takes me away from the actual work, but as I said at the beginning of my Stanford Community Symposium talk, I'm keenly aware that the CFS patient community includes many experts on the disease, and I consider you all colleagues.

OK. This is my first post using Multiquote, so if I get it wrong, please be patient. Your question 1.1 is 4 questions in one. Let's start with expression. If we had expression data for all the enzymes in all the relevant cell types I'd be ecstatic. The data you show are on tissues, some many cell types are lumped together and that's just insufficient for our purposes. Also, the lower limit of detection (LOD) for these mRNAs isn't low enough. On many sites you can plot those expression data semi-log and I recommend you try that and see what you think.

What we need is single cell RNAseq (scRNAseq) and many labs are now doing this for individual cell types. There's even a portal for these data at the Broad Institute at MIT. This is all made possible by NGS. Not all the cell types we'd want have been done yet, but some antigen presenting cells (APCs) (dendritic cells) have been reported and both IDO1 and IDO2 are expressed in these cells. TDO2 is not. This is the expression pattern that must obtain if the IDO metabolic trap is to occur. We're going to try to differentiate stem cells into serotonergic neurons in culture and do two things: see if we can induce the trap experimentally, and also do scRNAseq. If we get to that point, I've got three ideas about how to get out of the trap that we can try experimentally.

Kynurenine (Kyn) is important as an immunoregulator. Kyn or Kyn metabolites control the function of Treg cells and thus also control the expansion of effector T cells. If an APC is in the trap and not making Kyn, then T cell expansion takes off and you get autoimmune symptoms. That's the immune side of the theory.

The CNS side of the theory is way more difficult to test since we cannot get serotonergic neurons from the midbrain Raphe nuclei of anyone. That's why we want to try making our own from stem cells. But the theory is, IF the enzymes of the kynurenine pathway are expressed in serotonergic neurons, then the IDO metabolic trap is feasible in those neurons. If a serotonergic neuron is in the trap, a bunch of potentially pathological consequences follow. First you have little or no kynurenine, and its downstream neuroactive metabolites. In neighboring microglia, you might get the same autoimmune cascade and see the brain inflammation that Jared Younger describes. In the serotonergic neurons you will further build up tryptophan (Trp) because the major Trp degradation pathway (the kynurenine pathway) is substrate inhibited. What happens to serotonin depends on which isoform of TPH2 the cell expresses. One isoform is, itself, substrate inhibited, so the neuron may produce far too little serotonin. The other isoform is normal Michaelis Menten so you'd get saturation and perhaps far too much serotonin. While there are only ~12,000 serotonergic neurons in brain, each one is thought to modulate the activity of perhaps 100,000 glutamatergic or GABA-ergic neurons in places like those in my Community Symposium slide. I'm not saying this is it; I'm saying there is a lot of explanatory power in the IDO metabolic trap hypothesis, and it's worth testing.

Finally, Ron Davis has said that the kynurenine pathway is important because it makes NAD+ and it will make less if it's in the metabolic trap. Ron and I disagree about this. I think everyone is eating so much nicotinate (niacin) that they will have no trouble making sufficient NAD+. You might find it instructive to read some of the history of a disease called Pellagra.

I'll work on your other questions as I have time. Thanks for posting them.

 

I've thought a good deal about this. Ron has too. One way is to simply consume too much Trp. The gut, the BBB and the neurons all use the same LAT1 transporter to transport Trp. In cells it has a uM Km on the way in and a mM Km on the way out. You will know many athletes who fell victim to CFS. I know several who were consuming a "ton" of whey protein when they fell ill. You might find it interesting to google "tryptophan poisoning." I know one CFS patient who attributes her CFS to Trp poisoning.

Another way to increase Trp is to inhibit IDO1 for a period of time. This works just as well as increasing extracellular Trp in simulations of my mechanistic computer model. There are lots of possible physiological inhibitors. If you and others on s4me have time to build a comprehensive list from the literature, I'd be right on it.

 

This is one of the ones I don't buy, though. It just seems as though this would be a really challenging feat to accomplish.

People developing ME do a great deal to try and compensate for symptoms. I tried whey also due to plummeting energy levels, and a lot of people I've spoken to were trying very strict diets and exercise regimens seemingly in response to worse and worse baseline function.

So I call chicken or egg on this one.

 

Alex, I agree with you that bistability is a natural way to think about CFS. And having grown up in non-linear systems theory in the 1960s before I became a physiologist in the 1970s, I too have been thinking about bistability in nonlinear biological systems for a long time. For CFS, I had the advantage of access to the SIPS WGS data and so it was natural to look for mutations that might uncover a bistability.

There is an old paper from Suzanne Vernon when she was still in Atlanta that represented a bistable model for CFS in the adrenocortical system. It was a collaboration with engineers at Georgia Tech. Also, there is an appendix to one of Sarah Myhill's papers written by a friend of hers who is a mathematician. He too, immediately thought of bistability when she told him about ME/CFS.

To me, the exciting thing about the IDO metabolic trap is that it represents a real, testable instance of such a bistability. It has some genetic support. It works theoretically. And if we can get the right cell types in culture, we can test it experimentally. If it's really a feature of ME/CFS, we can test ways to get cells out of the trap as a prelude to getting PEOPLE out of the trap.

 

Despite dietary niacin, my vitamin B3 levels were below 15% of the RDA. So it could be possible for some of us, but perhaps it's a gut issue preventing absorption?

 

In biological terms, what happens is that (after the point of no return) IDO1 is now inhibited by its own substrate, Trp. Importantly this can only happen in cells where the kynurenine pathway is the major degradation route for Trp and where TDO2 is not expressed. It becomes irreversible because it is no longer self-correcting. Instead, as Trp increases, IDO flux decreases, and as IDO flux decreases, Trp increases, and so on.