How do the various assays differentiate normal antibodies from autoantibodies?
- Thread starter Jaybee00
- Start date
Jonathan Edwards
Senior Member (Voting Rights)
Assays differentiate antibodies to foreign antigens and antibodies to self antigens (autoantibodies) simply by starting with the relevant antigen and measuring what sticks to it. Autoantibodies are normal antibodies in any other sense. We all have autoantibodies that don't happen to do much. They are not different from any other antibodies except in that they happen to stick to self antigens in dishes.
The problem with assays is not so much error as to what they mean. If antibodies stick to an antigen in an assay that does not necessarily mean they will stick in vivo. There is a rough affinity threshold that determines that an antibody at least might do something pathological. I forget what it is but a dissociation constant of 10 ^-8 sticks in my mind. ELISA and similar assays usiing washing steps to try to mimic this affinity threshold but it is all very arbitrary.
There are all sorts of problems of non-specificity like antibodies sticking to plastic - which why you 'block' the plastic with an irrelevant protein like casein. There are also complicated issues about whether or not the antigen stuck to a plate displays the face (epitope) that antibodies bind to in vivo. There are also issues relating to functional antibodies. Autoantibodies to a protein may only matter if they bind to a site that triggers something like phosphorylation, as in a receptor signalling. Antibdoies that bind the some antigen on plastic by binding at a different place may be irrelevant.
There is a lifetime of work involved in trying to gto sort out all the interpretation problems and for most diseases we give up and accept that what you measure in an. assay is a very crude indirect measure of any antibody relevant in vivo.
The problem with assays is not so much error as to what they mean. If antibodies stick to an antigen in an assay that does not necessarily mean they will stick in vivo. There is a rough affinity threshold that determines that an antibody at least might do something pathological. I forget what it is but a dissociation constant of 10 ^-8 sticks in my mind. ELISA and similar assays usiing washing steps to try to mimic this affinity threshold but it is all very arbitrary.
There are all sorts of problems of non-specificity like antibodies sticking to plastic - which why you 'block' the plastic with an irrelevant protein like casein. There are also complicated issues about whether or not the antigen stuck to a plate displays the face (epitope) that antibodies bind to in vivo. There are also issues relating to functional antibodies. Autoantibodies to a protein may only matter if they bind to a site that triggers something like phosphorylation, as in a receptor signalling. Antibdoies that bind the some antigen on plastic by binding at a different place may be irrelevant.
There is a lifetime of work involved in trying to gto sort out all the interpretation problems and for most diseases we give up and accept that what you measure in an. assay is a very crude indirect measure of any antibody relevant in vivo.
Jonathan Edwards
Senior Member (Voting Rights)
Maybe they aren't there. I think jnmaciuch and I would agree that they probably aren't.
If they are there then they might still be hard to find for very complicated stereochemical reasons. That said, most autoantibodies we now know about can be identified either with tissue staining techniques or electrophoresis of self proteins and showing binding with a 'Western' blot. A few autoantibodies were only identified in the 1990s - like the antineutrophil cytoplasmic antibodies - but most had already been known about even if the protein had not been identified. Antinuclear antibodies turned out to be antibodies to about 50 different proteins, each associated with a slightly different pattern of disease, but we knew they were there in the 1960s.
Don't you have to know what antibody you're looking for? Surely an assay can't contain every single human protein, so it doesn't seem like failing to find autoantibodies with these assays would be evidence that there aren't any clinically meaningful ones.
Jonathan Edwards
Senior Member (Voting Rights)
If you use staining of tissues you are likely to pick up antibodies to any protein in those tissues. People have screened pretty much every tissue looking for autoantibodies. SO these assays contain all tissue fixed proteins.
If the protein is circulating then it should run out on an SDS-PAGE gel on electrophoresis and show up as a smudge on a nitrocellulose Western blot. So that assay contains all soluble proteins.
That might not be good enough but in recent years commercial outfits have developed chips with thousands of proteins on them - pretty much all proteins expressed in all tissues. I don't know the detail but I imagine you have a tiny dot of each protein in an array and scan the chip after applying patient serum and signal reagent with a LASER. Jo Cambridge has done this for ME/CFS. There are some differences from normals but not at the levels we normally associated with pathogenic antibodies.
So, no, you don't really need to know what protein you are looking for.
Utsikt
Senior Member (Voting Rights)
Do we know how that affinity threshold compares to the effective thresholds during the sorting stages in bone marrow etc.?
Are they caused by the same physics, so the thresholds are similar, or are there some kind of buffers of sort to weed out almost auto-reactive as well?
jnmaciuch
Senior Member (Voting Rights)
In addition to what Jonathan said, you can also do BCR sequencing like was done in one of the studies from Audrey Ryback. It won’t tell you what antigen the antibody is for (unless it’s similar to a sequenced antibody for a known antigen), but that would give you evidence of clonal expansion. (In layman’s terms, if a B cell receptor binds to an antigen and bypasses self tolerance, it will start cloning itself and you’ll end up with a large proportion of very similar sequences in your data).
There might be some unusual instances where something doesn’t show up using either method. Null results on both….its hard to make a case that there’s something to be found.
Jonathan Edwards
Senior Member (Voting Rights)
Nobody knows very much about this, I suspect. There are probably data that narrow things down to an order of magnitude or two for Kd but one might expect the system to be pretty precise on this and as you say it would make sense to be over-cautious.
The 'physics' both for B cell selection and autoantibody effect involves engagemnt of complement in some situations, but not others. It is vastly complex. Which is maybe why it isn't so surprising that the system has leaky spots called autoimmune diseases.
And there are situations where antibody can cause harm just through concentration, without it necessarily being important what it binds to. An example is Waldenstrom's macroglobulinaemia where high IgM levels mediate venulitis. MS lesions may be similar, with high local IgG mediating demyelination.
It is all very complicated.