I recommend that you take a look at this:
Jeong KJ, Jang SH, Velmurugan N. (2011) Recombinant antibodies: engineering and production in yeast and bacterial hosts. Biotechnol J. 6:16-27.
This fairly recent view discusses the progress and prospects for producing functional antibodies in yeast and bacteria. As pointed out by @Nick in the comment to the OP, the main stumbling block in both cases is the engineering that would be necessary to ensure correct glycosylation in the heterologous expression systems. This quotation from the cited review sums up the situation:
The N-linked glycan structure in the Fc region of antibodies is indispensible for binding to receptors (FcγRs), but glycosylation has been one of the major limitations in choosing possible production hosts. Eukaryotic yeast as well as bacteria, including E. coli, have not been used as the main hosts for the production of recombinant antibodies (full-length IgG format) for use in humans because of glycosylation problems.
They do however go on to discuss some interesting developments such as the expression of a Campylobacter N-glycosylation pathway in E.coli, as well as strategies for the production of functional antibody fragments.