As stated here it is possible to produce fragments of antibodies in bacteria and harvest them (from the medium, I guess, but I don't have access to the full article). As it is possible to design antibodies, would it be possible and feasible to produce important antibodies large scale through bacteria like it is done for example with insulin?
I'll preface this by saying that I'm talking about antibodies of mammalian/human origin, consisting of two identical heavy and two identical light chains disulfide-bonded into the classic antibody structure:
It is possible to produce antibody-like molecules known as single-chain antibodies, including single-chain variable fragments (scFv) in bacteria and viruses, where it is also known as phage display. These hosts can also produce a modified Fab fragment that is engineered as a single polypeptide chain, and multiple other single-chain antibody-like molecules as shown below. In the past, this has primarily been done in E. coli, as they are a common tool used in protein expression, and have many advantages as an experimental system in general. However, single-chain antibody production in E. coli is hampered by the fact that these exogenous proteins tend to form inclusion bodies instead of being secreted into the medium, where they can be collected and purified much more easily, without the need for lysing the host cells. The paper you linked to describes a method for producing scFvs in the Gram-positive bacterium Bacillus megaterium, which is becoming more common for the production of recombinant proteins.
So, it is possible to produce antibody-like molecules in bacteria. However, unlike insulin, full-length antibodies are made up of multiple polypeptide chains, and are both glycosylated and disulfide-bonded. While there is evidence that bacteria can accomplish both of these biochemical tasks, the processes seem to require additional chaperone, trafficking, and regulatory proteins that bacteria do not possess, and as far as I know scientists have so far not been successful in producing full "proper" antibody molecules in prokaryotes.
However, there are many mammalian/human expression systems for producing fully functional recombinant antibodies in very high yield, and a number of FDA-approved monoclonal antibody drugs are now on the market, with many more in various stages of development and clinical trials.
@MattDMo hit the highlights, but I wanted to comment on the creation of libraries of anbitodies for screening in bacatera. But there has been interest in nanobodies - the antibodies of camelids (llamas and camels) have only one chain, which makes the idea of heterologous expression in bacteria even simpler.