Convocatoria de proyectos de la I.E.R Bojacá-Chía al programa Ondas  Fecha : Lunes 24 de Enero de

In the following we will describe some interesting functional implications of alternative splicing events observed in our MS dataset confirms the soundness of the identified splicing products.

12.2 Results 129

We will describe some very interesting functional mechanisms of changes in localization sig- nals, transmembrane helices and the domain composition of multi-domain proteins which are discussed in the biological literature.

Terminal modifications of protein domains

A large fraction of isoforms (81%) have modifications at the terminal ends of the reference protein which may contribute to changes in localization or degradation signals. Using SignalP and TargetP annotations [46] from InterProScan [126], we find 117 SignalP and 160 TargetP peptide matches. 25% and 31%, respectively, of those are found to be altered by alternative splicing potentially accounting for changes in the cellular localization or faster degradation.

One example is the Human Deoxyuridine Triphosphate Nucleotidohydrolase gene (P33316) which is important for DNA replication. Two distinct isoforms DUT-N and DUT-M originating from the same gene are located in the nucleus or targeted to the Mitochondria respectively [91] depending on the unique N-terminal signal which can be altered by alternative splicing. The existence of both isoforms is confirmed by unique mass spectrometry peptides in our dataset. Another example is the BACH (Human Brain Acyl-CoA Hydrolase) protein (O00154) in Human, where different isoforms (including the isoform 3 confirmed by mass spectrometry data) [180] are shown to be located in the cytoplasm (and potentially targeted to the nucleus) or transported to the Mitochondria.

Known isoforms with alterations of signal peptides in our dataset contain isoforms impor- tant for brain development and neuronal activity. While the Vol-L and Vol-S isoforms of ITA3 (O44386) from Drosophila [60] take part in Integrin mediated short-term memory learning, the isoforms of the Gamma-aminobutyric acid receptor subunit beta-3 (P28472) are involved in the synaptic inhibition of neuronal activity [84]. The detailed functions of those signal peptide alter- ations are unknown in both cases.

Alternative splicing of transmembrane regions

Similar to changes in the cellular localization of a protein via alternative splicing of target or signal peptides, the removal or the inclusion of trans-membrane domains can lead to membrane- bound or secreted / cytoplasmic isoforms. According to TMHMM [88] annotations 79 proteins in our dataset contain transmembrane domains from which 33 (41%) appear to be alternatively spliced. Among the isoforms discovered in our dataset are several interesting cases also described in the biological literature.

ADAM23 (O75077), a metalloprotease and disintegrin protein which plays an important role in brain development gives rise to three isoforms which differ in their C-terminal transmembrane domain. While the native isoform and isoform 2 have different transmembrane domains, isoform 3 lacks this domain and is found to be secreted. Isoform 3 (confirmed in MS-data) can then com- pete with the native, membrane-bound variant for binding Integrin. Different ADAM23 isoforms may therefore alter cell-cell or cell-matrix interaction during brain development [152]. Another interesting case is TNR25 (Q93038) in Human where several isoforms are produced. TNR25 is involved in receptor signaling inducing apoptosis. Different isoforms are found to be integrated

130 12. Alternative splicing and proteome complexity in mass spectrometry data

into the membrane or secreted from the cell. The existence of the secreted isoform number 8 can be specifically confirmed by MS data. The secreted forms seem to bind extracellular ligands and block receptor signaling before the signal is transmitted into the cell. Therefore they are sug- gested to set the balance between TNR25 mediated apoptosis and cell survival [140]. Both cases represent examples for an interesting functional principle of splicing resulting from secreted or membrane-bound isoforms. While membrane bound isoforms are responsible for the interaction of the cell with its environment, secreted isoforms are able to block extra-cellular signals before they reach the cell by acting as regulatory signal sinks.

Changing domain composition by alternative splicing events

In order to analyze the frequency of splicing events altering the domain composition of multi- domain proteins on a sequence basis we used PFAM patterns. Out of the 2002 patterns annotated for the 733 reference proteins, 370 (18%) are found to be altered significantly (more than 75% of the residues of the domain affected) by alternative splicing events. Unfortunately the dataset is too small to identify patterns being spliced significantly more often than expected e.g. via the LOD analysis used in Chapter 8.

Nevertheless, two interesting examples for changing the function of a protein via changing its domain composition are the Human proteins ZN396 (Q96N95) and ZSCA1 (Q8NBB4). Both consist of a SCAN-box domain as well as threeC2H2-Zinc finger motifs and members of this family often act as transcriptional repressors. In both cases isoforms are confirmed by MS dis- playing the deletion of all three zinc-finger domains leaving only the SCAN domain. Isolated SCAN domains have been shown to act as selective oligomerization domains which may mod- ulate the formation of functional SCAN domain zinc-finger transcription factors [138] and the isoform of ZN396 has been shown to e.g. hetero-associate with the native variant [176]. The fact that two of those events are confirmed in MS-data may point to a large importance of this mechanism to regulate the transcriptional landscape of the human genome via changes in the domain composition of multi-domain transcription factors.