[niels_olson]

Hi, I do research using the same methods: immunohistochemistry. This was done with rabbit polyclonal antibodies. The key being polyclonal. We don't know what those antibodies are really staining. The difference, however, is intriguing.

[moyix]

Could you elaborate on what polyclonal means here? Does it mean that they might not have detected fungus after all, but some other thing that would activate the same antibodies?

[dnautics]

Yeah, but that's true for monoclonals too, just a matter of degree.

A polyclonal antibody is an extraction of (usually rabbit) blood from an animal that has been insulted with the agent you'd like to detect. The extraction is enriched for antibodies. They are less reliable, because there could also be other antibodies in there. Even the antibodies that recognize what you're looking for could be a population of antibodies, each recognizing a different part (think the blind men and the elephant parable). Presumably, you have tested that a positive control of what you're trying to look for works.

A monoclonal is different. You insult a mouse with a sample of what you'd like to detect, and then resect the spleen, which contains b-cells, cells that produce the antibody. You then fuse the b-cells with a special strain of cancer cell that will immortalize the b-cell (these are called 'hybridomas'). After sorting one cell to a well, you then test to see if the cells in the well have produced an antibody that detects what you're looking for. Therefore, you can be confident that a monoclonal antibody contains only one antibody, that recognizes only one antigen.

You don't necessarily know what the antigen is that is recognized (that's called the epitope). The epitope could be, say, a piece of protein on the outside of the fungus or a different piece of protein on the outside of the fungus. It's even conceivable that even for a monoclonal there are side-reactions against things that are naturally in the sample.

And: Epitope retrieval techniques are really complicated. So figuring out exactly what epitope is recognized is not trivial.

[twic]

It might be worth explaining a bit more about how antibodies are made. There are these cells called B lymphocytes, whose job is to produce antibodies. When they form, they randomise parts of the antibody genes:

https://en.wikipedia.org/wiki/V%28D%29J_recombination

As a result, every B lymphocyte makes one particular antibody that is different to every other B lymphocyte, and which recognises a different molecular pattern to every other B lymphocyte. There's a process in their development which weeds out cells which recognises patterns in your own body, leaving a population of cells which (hopefully) recognise every possible pattern which is not part of your own body.

B lymphocytes then wander round the body, looking for stuff which binds to their antibody; if they find it, they know it must be an invader, and they raise the alarm, triggering an immune response (subject to checks and balances from other parts of the immune system). As part of that, the cell which raised the alarm will proliferate, making millions of cells producing identical antibodies, which bind to the invader and mark it for destruction by other cells of the immune system.

If an invader has more than one molecular pattern (which it will - every patch on the surface of a protein is a pattern, and a bacterium will have all sorts of proteins and other things on its surface), then a corresponding number of B lymphocytes should recognise it, and proliferate in response.

So, if you insult a rabbit with your protein, you will activate all the B lymphocytes which recognise patterns on it. If you purify antibodies from its bloodstream, you will get antibodies recognising all those patterns - and any other antibodies which happen to be in the bloodstream at the time. If you isolate a single B lymphocyte its bloodstream, make it immortal, culture it, and purify the antibodies from that, you will get a single kind of antibody, recognising a single molecular pattern.

In software terms, a monoclonal is a bit like detecting spam by grepping for a single spam-specific word, and a polyclonal is a bit like doing an n-gram comparison with a corpus of known spam.

[nonbel]

>"you can be confident that a monoclonal antibody contains only one antibody, that recognizes only one antigen"

Not sure if you really meant that since it is kind of contradicted by some later statements, but all antibodies are promiscuous. Thinking you can map one antibody to one epitope is extremely dangerous. It is a matter of quantity (affinity/avidity), not quality.

[twic]

Exactly. When i did immunocytochemistry, i had more confidence in polyclonals than monoclonals. If you get a strong signal with a polyclonal, then you know that many different antibodies are all binding the same thing, so it's very likely it's your protein. If you get a strong signal with a monoclonal, that's just one antibody - so it might be your protein, or it might be some other protein that, by chance, your antibody happens to bind tightly.