Taming the power of the immune system in the lab wasn’t easy. For a start, it was a mystery how we have so many different antibodies, millions at any one time. But if it’s one gene per protein, and we only had something like 30,000 genes, how could we have millions of different antibodies?
Understanding this problem was the key that unlocked one way to make antibodies against targets that we select. We can use them as tools in the lab, or to treat disease—therapeutic antibodies. Chances are you may know someone who has, or is, using them; if they have Crohn’s disease for example.
Just finding antibodies that stick to a target—some unique bump, crevice or corner on a virus, for example—isn’t enough. The next step is to find out if they do something useful. They may stick but not prevent infection, or not activate some receptor on a cell. Now a group from the Scripps Research Institute have found a way to do both, and they made something un-naturally useful too.
The key to how we can make so many different antibodies was the discovery that some of our genes can be shuffled about like cards in a deck. The croupiers are cellular tools: chopping up bits of DNA, mixing them and sticking them back together again. Making completely different new sequences.
The shuffled sequences code for the tips of antibodies—the parts that stick to foreign invaders in the body. The shuffling means we could make billions of different antibodies; without having to have billions of separate genes. And it all goes on inside an immune cell called a B-cell; as it matures it deals itself a hand of genes like no other B-cell, enabling it to make its own uniquely specific antibody.
Knowing how the genetics of the immune system work we can make ‘libraries’ of all the potential different genes for the antibody tips. Meaning we can take the job of making antibodies out of the hands of the body and into the lab.
The DNA for the binding tips are put into a virus that infects bacteria—a ‘bacteriophage’. The virus ‘displays’ the products of these genes as a protein on its surface. It’s a neat solution because it links the physical characteristics (the ‘phenotype’) of the virus to the genetics responsible for producing it (the ‘genotype’). The technique is called ‘phage display’. We can then search through this library of viruses and binding tip genes to find ones that stick to our target.
But we still need to see if they do anything. Normally this would mean taking the selected antibodies and screening them through a series of tests to see if they have the desired effect.
The team at Scripps made a change to normal phage display to get around this. First they decided on their target, a receptor for a protein called EPO, which when stimulated by EPO ups the production of red blood cells. They wanted a way to quickly find antibodies that would mimic EPO and stimulate its receptor, not just ones that would stick to it and do nothing.
They used phage display to find antibodies that could stick to their target. They took the genes for these antibodies and put them inside a different virus, one that infects human cells: a lentivirus. They poured this over lab grown human cells that have EPO receptors and grow in response to EPO. Infected cells started to produce the antibodies. Those that grew most had been infected with viruses whose antibodies could stimulate the EPO receptor.
When they looked closer, they saw that cells were being infected with two or three different lentiviruses, coding for 2 or 3 different antibodies. They grabbed the antibody sequences from the best growing cells and made them. When they tested each antibody on its own, it didn’t work very well. They tried them in combination too; it worked better, but again, not great. Why?
In nature the two arms of the Y-shaped antibody—the binding tips—are both exactly the same, sticking to the same target. The team realised that in cells infected by more than one virus, things weren’t behaving as normal. Something un-natural was going on. Antibodies were being put together with the binding tip from one virus, while the other arm had the binding tip from a different virus. The antibodies were bi-specific: binding to two different parts of the same target. Bi-specific antibodies don’t exist in nature.
The Scripps team checked to see if this was the case; mixing and matching binding tips from cells that multiplied. They found that one particular combination of binding tips was as good as real EPO at stimulating the receptor.
Making bi-specific antibodies in this way multiplies the number of possible antibodies beyond the billions we already have.
Since the technique also allows us to look for the effect we’re after before we know the identity of the antibody, it could also speed up the search for new drugs.
Suddenly the store cupboard of possible new therapeutic antibodies just got bigger.
Reference: Zhang, H., Wilson, I. a., & Lerner, R. a. (2012). Selection of antibodies that regulate phenotype from intracellular combinatorial antibody libraries. Proceedings of the National Academy of Sciences, 2012.
Antibody: Anna Tanczos, Wellcome Images
Shuffled cards: Todd Klassy