We present evidence that mammalian cells have the ability to generate functional mRNA molecules by trans‐splicing. Rat cells, transformed by an early SV40 DNA fragment (Bst/Bam) synthesize a truncated T antigen (T1 antigen), although the cells do not have a direct sequence homology for the T1 antigen at the DNA level. The Bst/Bam DNA fragment encodes exclusively for the second SV40 T antigen exon (aa 83‐708) and contains the entire small t antigen intron. To synthesize the corresponding mRNA (T1 mRNA), the cells utilize a cryptic 5′ splice site within the second exon (codons for aa 131/132) as donor site and the upstream small t antigen 3′ splice site as the acceptor site. Since these sites are in an inverted order on the pre‐mRNA, two Bst/Bam transcripts are required to generate one T1 mRNA molecule. HeLa cell nuclear extracts also performed the trans‐splicing reaction in vitro.

Bacterially‐expressed fusion proteins containing the DNA‐(region C) or hormone‐binding (region E) domains of the chicken progesterone receptor (cPR) fused to the C terminus of Escherichia coli beta‐galactosidase were analysed for the specificity of interaction with natural and synthetic hormone‐responsive elements (HREs) and progestins, respectively. The purified fusion protein containing the progestin‐binding domain bound progesterone with an apparent Kd of 1.0‐1.5 nM and was specifically photocross‐linked with the synthetic progestin R5020 in crude bacterial lysates. Labelling of intact bacterial cells with [3H]R5020 revealed that the majority, if not all, of the bacterially produced hormone‐binding domain was active. No differences in the binding to a synthetic palindromic glucocorticoid/progestin‐responsive element (GRE/PRE) were found when the bacterially produced cPR DNA‐binding domain was compared in methylation interference assays with the full‐length chicken progesterone receptor form A expressed in eukaryotic cells. The study of dissociation kinetics, however, revealed differences in the half‐life of the complexes formed between the palindromic GRE/PRE and either the receptor form A or the fusion protein containing the cPR DNA‐binding domain. DNase I protection experiments demonstrated that the bacterially produced region C of the cPR generated specific ‘footprints’ on the mouse mammary tumour virus long terminal repeat (MMTV‐LTR) which were nearly identical to those previously reported for the rat glucocorticoid receptor.

The early SV40 BstXI-BamHI (Bst/Bam) DNA fragment encodes exclusively for the second exon of the large T-antigen and contains the intact small t-antigen intron. Rat cells transformed by the p14T, a construct that carries the Bst/Bam DNA fragment as a tail-to-head tandem duplication, synthesize a truncated T-antigen (T1-antigen) without having a direct equivalent at the DNA level. Formation of the T1-mRNA occurs by means of two distinct mechanisms: alternative-tandem-cis-splicing and trans-splicing. To generate the T1-mRNA the cells utilize a cryptic 5' splice site, located within the second exon of the large T-antigen and the regular small t-antigen 3' splice site. Since these splice sites are in an inverted order two Bst/Bam transcripts are required to generate one T1-mRNA molecule. For alternative-tandem-cis-splicing the cells utilize a 4.4 kb pre-mRNA that contains the sequence of the entire Bst/Bam tandem repeat. The proximal Bst/Bam segment provides the 5' donor splice site and the distal segment the 3' acceptor site. This requires that the pre-mRNA not be cleaved after the RNA polymerase II has passed the polyadenylation signal of the proximal Bst/Bam DNA segment. Synthesis of the 4.4 kb pre-mRNA was demonstrable by RT-PCR but not by Northern blot analysis. For trans-splicing, the cells utilize two separate pre-mRNA molecules. One transcript provides the cryptic 5' splice donor site and the other the 3' splice acceptor site. To demonstrate this a three base pair deletion was introduced into the proximal Bst/Bam segment of the p14T DNA (p14Tdelta-3) as a marker, destroying the recognition site for Pf/MI restriction enzyme. This deletion allowed the differentiation between the proximal and distal Bst/Bam segment. RT-PCR analysis and DNA sequencing confirmed that the p14Tdelta-3 transformed cells generate the T1-mRNA by intra- and inter-molecular RNA splicing.

Spliceosome-mediated RNA trans-splicing enables correction or labeling of pre-mRNA, but therapeutic applications are hampered by issues related to the activity and target specificity of trans-splicing RNA (tsRNA). We employed computational RNA structure design to improve both on-target activity and specificity of tsRNA in a herpes simplex virus thymidine kinase/ganciclovir suicide gene therapy approach targeting alpha fetoprotein (AFP), a marker of hepatocellular carcinoma (HCC) or human papillomavirus type 16 (HPV-16) pre-mRNA. While unstructured, mismatched target binding domains significantly improved 3′ exon replacement (3’ER), 5′ exon replacement (5’ER) correlated with the thermodynamic stability of the tsRNA 3′ end. Alternative on-target trans-splicing was found to be a prevalent event. The specificity of trans-splicing with the intended target splice site was improved 10-fold by designing tsRNA that harbors secondary target binding domains shielding alternative on-target and blinding off-target splicing events. Such rationally designed suicide RNAs efficiently triggered death of HPV-16-transduced or hepatoblastoma-derived human tissue culture cells without evidence for off-target cell killing. Highest cell death activities were observed with novel dual-targeting tsRNAs programmed for trans-splicing toward AFP and a second HCC pre-mRNA biomarker. Our observations suggest trans-splicing represents a promising approach to suicide gene therapy.