Øystein Fluge and Olav Mella interviewed on BBC Radio

[Jonathan Edwards]

Professor Edwards, mind if I ask, what are some of the challenges involved in trying to identify an autoantibody in this situation?

Is it necessary to have a good idea of the target of the autoantibody from the start, or does that not factor in?

I guess they'd need to do some sort of pooling of sera from the rituximab responders. Is obtaining enough of a potential autoantibody for analysis difficult?

I'm basically wondering why it's taken so long to find the pathogenic autoantibody(ies) in ME/CFS, if one exists - which I guess it likely will if the study results continue to hold up. Thank you.

Yes, you have to make the correct guess about roughly what protein the antibody is against and find some sort of tissue sample or extract that will contain that protein in a form that has not been damaged by processing and the antibody will bind to. There are about 10,000 basic proteins that an antibody might be to and there might be as many as 100,000 'conformations' since proteins can take various forms. If we include post translational modifications like citrullination (in RA there are antibodies to this) the number goes up again and if we consider individual epitope shapes that antibodies could bind to we are up in the hundreds of millions I suspect.

In very simple terms, the difficulty of finding an autoantibody is like the difficulty in making a key that will open a lock that is on your safe and you have lost the key to. How long would that take you?!!! Nearly all autoantibodies have been discovered by chance. When the tissue immunofluorescence technique was discovered lots of autoantibodies were discovered because all you had to do was take a dozen sections of different tissues and look to see if patient antibodies bound to any particular part of any tissue. The problem is that not all autoantibodies will show up in this sort of preparation. An anti-cytokine antibody would not show up because there isn't enough cytokine there in the tissue to show up - etc etc.

Pooling sera is no use. It will just dilute and confuse things. You have to test each one individually. Generally speaking you do not extract or analyse the antibody, you just know it is there in your serum because it binds to a section or a blot and you can then light it up with a fluorescent tag for antibody or a silver stain (like developing a photo) once you have washed away all the irrelevant antibody in the serum.

Autoantibodies are still being discovered all the time despite people having been looking for fifty years or more. They may have been missed because nobody has looked at sections of the right tissue - this applies to the new anti-brain antibodies and also the anti-neutrophil antibodies. They may also be missed because they bind to proteins that do not show up in tissues, being in too small amounts or too evenly distributed and which are damaged when transferred to Western blots.

We also have the problem that auto-antibodies may bind to proteins in the body in a way that you cannot get to work in the lab for all sorts of reasons to do with changes in the biochemical environment. It could also be that the binding in the body is weaker than what you need in the lab to get a fixed staining result. We can only find antibodies that will stay stuck while you wash off the irrelevant antibody for about an hour.

So all in all the absence of a known autoantibody does not mean there is none - or that there is an easy way to find it.

 

[Sasha]

@Jonathan Edwards - is there any prospect of high-throughput techniques for this, like there is for what I shall vaguely refer to as 'other stuff'?

 

[Jonathan Edwards]

@Jonathan Edwards - is there any prospect of high-throughput techniques for this, like there is for what I shall vaguely refer to as 'other stuff'?

Old fashioned immunofluorescence on tissue sections from the 1960s is pretty much the paradigmatic 'high throughput' technique in that you are testing for antibodies to a thousand proteins at once. I am afraid that high tech language sometimes obscures simpler realities! The problem with a lot of modern 'high throughput' techniques is that they are one, two or three steps removed from the physiological context of whole tissue - and that is probably the main reason why one might miss things or get artefactual results. The neurologists who have found the new anti-neuronal antibodies have done so by going back to tissue sections.

 

[Sasha]

Old fashioned immunofluorescence on tissue sections from the 1960s is pretty much the paradigmatic 'high throughput' technique in that you are testing for antibodies to a thousand proteins at once. I am afraid that high tech language sometimes obscures simpler realities! The problem with a lot of modern 'high throughput' techniques is that they are one, two or three steps removed from the physiological context of whole tissue - and that is probably the main reason why one might miss things or get artefactual results. The neurologists who have found the new anti-neuronal antibodies have done so by going back to tissue sections.

So if you wanted to do 'high throughput' it would be the case of getting a thousand lab technicians to fumble about with bits of blotting paper?

 

[Jonathan Edwards]

So if you wanted to do 'high throughput' it would be the case of getting a thousand lab technicians to fumble about with bits of blotting paper?

No, you only need one technician. You just need to get them to use a method that is very like the real situation in the body. The nitrocellulose blotting paper of Western blots is actually not very good because you have to chemically reduce all the proteins in order to run them on the electrophoretic gel first. That burns off quite a lot of recognition sites.

What you actually need is a very intelligent lab director like Angela Vincent or Mady Hornig who is still actually involved in supervising the work and a technician who is not afraid to get an unexpected result and note it down.

 

[Sasha]

No, you only need one technician. You just need to get them to use a method that is very like the real situation in the body. The nitrocellulose blotting paper of Western blots is actually not very good because you have to chemically reduce all the proteins in order to run them on the electrophoretic gel first. That burns off quite a lot of recognition sites.

What you actually need is a very intelligent lab director like Angela Vincent or Mady Hornig who is still actually involved in supervising the work and a technician who is not afraid to get an unexpected result and note it down.

But you mentioned that we're looking for one thing in maybe hundreds of millions. If we don't have a clue what it is (or not enough of a clue to narrow that down), does it make sense to just run 100 patients' samples through and test for these hundreds of millions of things and see what's in common?

Sorry, I don't think I'm appreciating the subtleties! But I'm wondering if there's a 'just shove it all through' version of this.

 

[Jonathan Edwards]

But you mentioned that we're looking for one thing in maybe hundreds of millions. If we don't have a clue what it is (or not enough of a clue to narrow that down), does it make sense to just run 100 patients' samples through and test for these hundreds of millions of things and see what's in common?

Sorry, I don't think I'm appreciating the subtleties! But I'm wondering if there's a 'just shove it all through' version of this.

No, a tissue section is as shove it all through as you can possibly get. It has all the possibilities for that tissue laid out on it. You might need to look at sections of fifty different tissues but that is not really the problem and you can get five tissues on one slide anyway. Ten slides would have all the hundreds of millions of shapes that are relevant - unless these shapes are not localised enough or get damaged in section drying. If you use a 'soup' in a modern high throughput system you lose an localisation and probably damage the proteins more. In order to pick up things like anti-cytokine antibodies we need much lower throughput systems with spots of highly concentrated localised cytokine put on a chip to be flooded with patient serum. You can make a chip with 100 cytokines on it but in comparison to tissue that is very low throughput.

 

[Sasha]

No, a tissue section is as shove it all through as you can possibly get. It has all the possibilities for that tissue laid out on it. You might need to look at sections of fifty different tissues but that is not really the problem and you can get five tissues on one slide anyway. Ten slides would have all the hundreds of millions of shapes that are relevant - unless these shapes are not localised enough or get damaged in section drying. If you use a 'soup' in a modern high throughput system you lose an localisation and probably damage the proteins more. In order to pick up things like anti-cytokine antibodies we need much lower throughput systems with spots of highly concentrated localised cytokine put on a chip to be flooded with patient serum. You can make a chip with 100 cytokines on it but in comparison to tissue that is very low throughput.

I still don't think I've really got my head around this but I think you're saying that labs aren't testing for all the millions of possibilities? Would it make sense for them to try to do so?

 

[deleder2k]

No, a tissue section is as shove it all through as you can possibly get. It has all the possibilities for that tissue laid out on it. You might need to look at sections of fifty different tissues but that is not really the problem and you can get five tissues on one slide anyway. Ten slides would have all the hundreds of millions of shapes that are relevant - unless these shapes are not localised enough or get damaged in section drying. If you use a 'soup' in a modern high throughput system you lose an localisation and probably damage the proteins more. In order to pick up things like anti-cytokine antibodies we need much lower throughput systems with spots of highly concentrated localised cytokine put on a chip to be flooded with patient serum. You can make a chip with 100 cytokines on it but in comparison to tissue that is very low throughput.

It seems that Øystein Fluge and Olav Mella thinks that the body may have created an antibody against some sort form of NOS. If that is indeed correct, would that easier to locate it, or doesnt that matter? It sounds like finding an antibody is finding a needle in a haystack.

 

[Jonathan Edwards]

I still don't think I've really got my head around this but I think you're saying that labs aren't testing for all the millions of possibilities? Would it make sense for them to try to do so?

Old fashioned tissue sections will test for all the millions of possibilities but they still may not show up. In the past labs may not have looked at sections from enough tissues but that is less likely now. Just at present ME sera are being tested on central nervous system tissues in ways that may not have been tried before and that might come up with something but even if it does not we may still miss autoantibodies that are relevant.

Try this analogy. You want to find out if you have a key that will open one of a million locks. So you find a box that has all the million locks in (a set of slides with all fifty tissues) and you make a million copies of your key (your serum with millions of copies of antibody molecules) and try them all out. It is easy because the locks are magnetic and each one attracts one of your key copies into its keyhole.

This may work but it might be that in real life the locks are fitted so that the key will only work if you push it in hard and jiggle it a bit, or you have to turn it twice to open or something like that. So although you are testing al possibilities your system may not discover a fit. In tissue sections you can only see binding if the protein is localised to specific places like membranes or organelles. If the protein is everywhere you just get brighter background levels.

 

[Jonathan Edwards]

It seems that Øystein Fluge and Olav Mella thinks that the body may have created an antibody against some sort form of NOS. If that is indeed correct, would that easier to locate it, or doesnt that matter? It sounds like finding an antibody is finding a needle in a haystack.

There is a specific difficulty in finding autoantibodies to proteins in blood vessels which is that blood vessels often bind antibodies in a non-specific way. So if you see a blood vessel pattern on a tissue section stained with serum you tend to ignore it. There are ways of picking up genuine signals but you need someone very experienced and also very clear thinking to notice these things.

 

[Sasha]

Thanks, @Jonathan Edwards - that's a very helpful analogy.

 

[Joolz]

@Jonathan Edwards: How would you explain immunological (flu-like) symptoms by the Fluge/Mella theory?

 

[msf]

I have posted this before, but it seems relevant to the present discussion of anti-NOS antibodies, so I thought I would post it again.

http://www.ncbi.nlm.nih.gov/pubmed/25617880

The paper suggests that the nitrosative auto-antibodies aren't related to the severity of the disease, but that oxidative auto-antibodies were (if I understood it correctly). This might be because they only included measures of illness that were are affected by oxidative auto-antibodies. Or perhaps the nitrosative auto-antibodies are like the cytokines that Lipkin et al found are increased in ME patients, that is a symptom rather than a cause? The paper suggests that bacterial translocation might be causing this auto-immunity, it would be interesting if they were able to show this, but I'm not sure how you would be able to do this, even if the levels of auto-antibodies and CD14 correlated it wouldn't prove causation.

 

[Jonathan Edwards]

@Jonathan Edwards: How would you explain immunological (flu-like) symptoms by the Fluge/Mella theory?

We are probably all dealing with viruses all the time but most of the time the immune system works silently without generating any symptoms. If the threshold for cellular activation is changed by some mediator that affects blood flow and metabolic regulation then the silent activities of the immune system might become intrusive. That is the sort of idea I think they are considering. Mediators like NO and steroids affect almost every cellular activity. B cells have steroid receptors. They probably respond to NO. I am not sure that I am convinced that NO activity is the right signal for ME but I cannot exclude it.

 

[Jonathan Edwards]

I have posted this before, but it seems relevant to the present discussion of anti-NOS antibodies, so I thought I would post it again.

http://www.ncbi.nlm.nih.gov/pubmed/25617880

The paper suggests that the nitrosative auto-antibodies aren't related to the severity of the disease, but that oxidative auto-antibodies were (if I understood it correctly). This might be because they only included measures of illness that were are affected by oxidative auto-antibodies. Or perhaps the nitrosative auto-antibodies are like the cytokines that Lipkin et al found are increased in ME patients, that is a symptom rather than a cause? The paper suggests that bacterial translocation might be causing this auto-immunity, it would be interesting if they were able to show this, but I'm not sure how you would be able to do this, even if the levels of auto-antibodies and CD14 correlated it wouldn't prove causation.

Right, nobody has been able to prove bacteria cause autoimmunity and the old molecular mimicry theory has very little going for it. I am not too impressed by the abstract of this paper, which contains no actual data - always a bad sign to my mind.

 

[msf]

Yes, it was published in the letters section of a medical journal for some reason. I posted it as a possible area for future research, not because it provides good evidence for what it is claiming.

As you said, no one has been able to prove bacteria cause autoimmunity, but I don't think they are proposing a molecular mimicry theory here - at least that isn't how I read it. I thought they were proposing that a oxidative and nitrosative response to LPS was causing the creation of auto-antibodies to these epitopes, and that when NAIOS were given to patients this reduced the levels of oxidative and nitrosative epitopes and thereby attenuated the autoimmune response to them.

 

[starlily88]

Øystein Fluge and Olav Mella interviewed on BBC Radio (Naked Scientists) about Rituximab and their B-cell theory

http://www.bbc.co.uk/programmes/p02ldh8b

Interview from 39:22

I hit your URL - but did not hear Fluge/Mella. How can I hear them? Thanks (new on here)

 

[deleder2k]

I hit your URL - but did not hear Fluge/Mella. How can I hear them? Thanks (new on here)

Press play and search for 39:22

 

[Snow Leopard]

It seems that Øystein Fluge and Olav Mella thinks that the body may have created an antibody against some sort form of NOS. If that is indeed correct, would that easier to locate it, or doesnt that matter? It sounds like finding an antibody is finding a needle in a haystack.

In a word. No. It is the excess production of NO that is associated with hypotension. The hypothesis is that there is an antibody directed towardsn extracellular signalling (either an endocrine factor, or a receptor) that leads to dysregulation of one of the NO synthases.

There is actually potential for such perpetuating loops within the cell. Eg MAPK pathways both stimulate NO synthesis, and can be stimulated by the presence of NO.

NO is also involved in hypoxic metabolism (and can lower oxygen consumption inhibit enzymes like aconitase etc), calcium channel activity, the list goes on.

In terms of ME, there are a few hypotheses related to this by Martin Pall in various published papers.

But all this comes back to the question, what is the extracellular trigger? Is it antibody mediated, and if so, what is the antibody directed against?

I do believe there are clues out there - eg the lack of a strong inflammatory response suggests there is no build up of immune complexes outside of circulation and this significantly narrows the possibilities. I think anti-cytokine/growth factor antibodies are unlikely for example.

I have some ideas, but they are probably wrong, so...