Classification of
Membrane Protein Type
|
Cell membranes are crucial to the life of a cell. A cell is enclosed by the plasma membrane (cell envelope), which defines its boundaries, and maintains the essential differences between the cytosol and the extracellular environment. Inside the cell there are various organelles such as the endoplasmic reticulum, Golgi apparatus, mitochondria, and other membrane-bound organelles. The characteristic differences between the contents of the cytosol and each of these organelles are maintained by their respective membranes (subcell envelopes). Although the basic structure of biological membranes is provided by the lipid bilayer, most of the specific functions are carried out by the membrane proteins. Membrane proteins consist of transmembrane proteins and anchored membrane proteins. The former contains one or more hydrophobic segments, and hence is relatively easily discriminated from nonmembrane proteins. The latter has a consensus sequence motif at either the N- or C-terminus, and hence can be recognized to some extent. For example, anchored membrane proteins are usually either isoprenylated at the C-terminus with the consensus sequence motif of CAAX, or myristylated at the N-terminus with the motif of GXXXS/T, or palmitylated at a specific Cys-residue of the N-terminal region. Another type of anchored membrane protein is of GPI-anchored proteins which are modified through glycosylphosphatidylinositol (GPI) at the C-terminus with a unique sequence feature, such as a hydrophobic tail. The way that a membrane-bound protein is associated with the lipid bilayer usually reflects the function of the protein. For example, only transmembrane proteins can function on both sides of the bilayer or transport molecules across it. By contrast, proteins that function on only one side of the lipid bilayer are often associated exclusively with either the lipid monolayer or a protein domain on that side. Also, associated with different locations, membrane proteins usually have different biological functions. Proteins associated with the cell plasma membrane act as sensors of external signals, transferring information across the membrane and allowing the cell to change its behavior in response to environmental cues. The ion gradients across membranes, which can be used to synthesize ATP, to drive the transmembrane movement of selected solutes, or to produce and transmit electrical signals in nerve and muscle cells, are established by the activities of specialized membrane proteins. Therefore, the determination of function for new membrane proteins can be expedited significantly if we can find an effective scheme and algorithm to predict their types and subcellular locations. Especially nowadays, the number of protein sequences entering into public data banks is rapidly increasing; it would be both time-consuming and costly to rely on completely experiments for the solution of these problems. Furthermore, the establishment of such an algorithm can also help prioritize genes and proteins to be identified by genomics efforts as potential molecular targets for drug design. The issue is, however, given the sequence of a membrane protein, can we predict its inherent attributes in a cell? In other words, is it associated with the cell membrane (i.e., plasma membrane) or with the membrane of a specific organelle inside the cell? How is it embedded in, or bound to, a membrane? Is it an inner or outer membrane protein? The present study was devoted to these problems.
In the literature, the definitions for the category of membrane proteins and their types are not unique. In this article, the membrane proteins are categorized into six types.
|

1. Input the primary sequence of an compartment-unknown protein.
2. Use the input sequence as a query to search the SWISSPROT protein database using PSI-BLAST program.
3. Find a number of homologous proteins from SWISSPROT and generate position specific scoring matrix from the multiple alignment of these sequences.
4. Extract four different feature vectors and predict by four SVM classifiers.
5. Fuse the outputs from four SVM classifiers.
6. Output the predicted membrane protein type.
Any suggestions or comments are welcome. Please send them to Howard Leung.