Functional premature polyadenylation signals and aberrant splicing within a recombinant protein coding sequence limit expression

Recombinant glycoproteins can be produced at high levels in permanently transfected mammalian cells using expression vectors with strong viral promoters. CHO-K1 cell lines developed to produce the recombinant complement activator blocking protein, CAB-2 (a fusion of membrane co-factor protein, MCP, and decay accelerating factor, DAF), showed unexpectedly low expression. Northern blot analysis revealed that in addition to the expected 2300 base CAB-2 mRNA species, these cell lines expressed 790 and 1500 base mRNA species accounting for ~50% and ~10% of the total CAB-2 mRNA, respectively. RT-PCR studies established that the 1500 base species resulted from aberrant splicing from within the DAF region of the CAB-2 coding sequence to a site within the 3' untranslated region. 3' RACE analysis confirmed that the 790 base species resulted from premature polyadenylation at an AATAAA site within the MCP coding region of CAB-2. Another prematurely polyadenylated species, not observed on Northern blots, was observed in the DAF region by 3' RACE. Analysis of human tissues and cell lines revealed that these internal polyadenylation signals in native MCP and DAF coding regions also generated prematurely polyadenylated mRNAs. Genetic modification of these functional RNA processing elements within the CAB-2 gene eliminated the aberrant mRNA species and significantly increased recombinant CAB-2 expression. These results illustrate that protein expression can be limited by aberrant mRNA processing and demonstrate the importance of identifying and eliminating these mRNA processing signals from within coding DNA to maximize recombinant protein expression.