Assessing the Impact of Alternative Splicing on Domain Interactions in the Human Proteome

Abstract

We have constructed a database of alternatively spliced protein forms (ASP), consisting of 13,384 protein isoform sequences of 4422 human genes (www.bioinformatics.ucla.edu/ASP). We identified fifty protein domain types that were selectively removed by alternative splicing at much higher frequencies than average (p-value < 0.01). These include many well-known protein-interaction domains (e.g., KRAB; ankyrin repeats; Kelch) including some that have been previously shown to be regulated functionally by alternative splicing (e.g., collagen domain). We present a number of novel examples (Kruppel transcription factors; Pbx2; Enc1) from the ASP database, illustrating how this pattern of alternative splicing changes the structure of a biological pathway, by redirecting protein interaction networks at key switch points. Our bioinformatics analysis indicates that a major impact of alternative splicing is removal of protein-protein interaction domains that mediate key linkages in protein interaction networks. ASP expands the available dataset of human alternatively spliced protein forms from 1989 human genes (SwissProt release 42) to 5413 (nonredundant set, ASP + SwissProt), a nearly 3-fold increase. ASP will enhance the existing pool of protein sequences that are searched by mass spectroscopy software during the identification of peptide fragments.

The Coupling of Alternative Splicing and Nonsense-Mediated mRNA Decay

“…Many propose that it is a major mechanism underlying proteome expansion, 3 but alternative splicing can also modulate the function or activity of a gene, for instance by adding or removing exons encoding protein domains or by altering the stability of the transcript or resulting protein. [4][5][6] In the last few years, it has become clear that many alternative splice forms previously thought to encode truncated proteins are actually targets of NMD (Fig. 1).…”

The evolving roles of alternative splicing

“…and how they are affected (added, removed, altered) can provide important and easily interpretable biological conclusions. Taking this approach, Resch et al [9] identified a set of functional domains that are removed more often than would be expected given the frequencies of each domain and of alternative splicing, including KRAB, ankyrin repeat and tubulin-binding domains. 56% of the members of this functionally diverse set fall into the higher level category of protein-interaction domains, yet this is no more than would be expected, given that this category is very common (64% of Swiss-Prot domains studied).…”

“…These classic biochemical investigations have been followed by hundreds more that use a wide variety of methods to discover and detail the implications of alternative splicing for specific genes. As mentioned above, Resch et al [9] demonstrate how appropriately designed, broad studies can lead to very specific and interpretable results. For example, they showed that, for the genes encoding the Kruppel family of transcription factors, alternative splicing preferentially removes the protein-interaction domain(s) and not the DNA-binding domains.…”

“…Thus, the full-length isoforms are predicted to bind DNA and be responsive to interaction partners. Shorter isoforms that lack the protein-interaction domain(s) may bind DNA, but would be insensitive to interaction partners [9].In many cases, alternative isoforms differ by small alterations of functional elements or domains, not by complete removal of functional elements [20]. In these cases, consideration of protein isoforms as simply the sum of their parts may not be sufficient to predict the biology of the resultant isoforms.…”

“…Thus, the full-length isoforms are predicted to bind DNA and be responsive to interaction partners. Shorter isoforms that lack the protein-interaction domain(s) may bind DNA, but would be insensitive to interaction partners [9].…”