We recently demonstrated that IFI 16, a human member of a family of interferon-inducible nuclear proteins, can function as a potent repressor of transcription. All members of this family are found in the nucleus and contain 1 or 2 copies of a conserved 200 amino acid repeat domain. IFI 16 migrates on SDS-PAGE as three distinct protein species (IFI 16A, 16B, 16C) clustered at 85-95 kDa, and we therefore set out to determine the molecular mechanisms underpinning the production of these different isoforms. In the present report, we have used thermal cycling amplification of reverse-transcribed mRNA (RT-PCR) and Southern blotting of genomic DNA to show that the three protein isoforms result from translation of three separate mRNA species produced by differential mRNA splicing. This differential splicing gives rise to variability in the central ("hinge") domain of the molecule which separates the two 200 amino acid repeats. The longest mRNA (approximately 2.7 kb) encodes an open reading frame of 2355 bp and generates the IFI 16A isoform of 785 amino acids. It contains sequences from 11 exons, including a newly identified exon (7a) which appears to have arisen by tandem duplication of exon 7. The second isoform (IFI 16B, corresponding to the form reported previously) is the most abundantly expressed, and results from deletion of exon 7a (168 bp) to encode a protein of 729 amino acids. The smallest mRNA encodes the IFI 16C isoform (2019 bp), has deleted both exon 7 and exon 7a, and shortens the protein by a further 56 amino acids. Culture of IFI 16-expressing cells with tunicamycin and incubation of cellular lysates with endoglycosidase H suggested that neither IFI 16A nor IFI 16B is glycosylated; however, some IFI 16C molecules showed a minor degree of complex carbohydrate addition. Furthermore, immunoprecipitation and Western blotting indicated that all three IFI 16 isoforms are phosphorylated on serine and threonine residues, but not on tyrosine. Thus, the three IFI 16 protein isoforms arise due to alternative RNA splicing and not due to differential glycosylation or phosphorylation. Finally, IFI 16 isoforms can homo- and heterodimerize, and we have mapped the dimerization domain to the amino terminus which contains an imperfect leucine zipper domain.