Transmembrane

This subsection of the ‘Sequence annotation (Features)’ section describes the extent of a membrane-spanning region of the protein. It denotes the presence of both alpha-helical transmembrane regions and the membrane spanning regions of beta-barrel transmembrane proteins.

We annotate transmembrane domains, when:
1. the transmembrane domains have been experimentally determined; there is some experimental evidence for the number, location and range of transmembrane domains; the structure of a protein family has been reviewed.
2. a protein is related to a well-characterized family known to contain transmembranes. In some families, such a G-protein coupled receptors, ion channels, transmembrane domains can be propagated by similarity, when there is some experimental evidence for the number, location and topology of transmembrane domains.
3. transmembrane domains are predicted by different prediction tools, provided their presence is consistent with the function of the protein and is detected by several prediction tools.

For multi-pass membrane proteins, we do not number transmembrane regions except in cases when an accepted numbering system exists, such as for seven-transmembrane proteins, ion channels, bacteriorhodopsin and others. In these cases we use the numbering system applied to the specific protein family concerned.

1. Annotation of experimentally proven transmembrane regions

Experimentally proven transmembrane regions are generally annotated using the qualifier ‘Potential’ due to the inherent difficulties in determining their precise boundaries experimentally.
Example: Q96BI3

However, when the experimental technique used allows the unambiguous assignment of the boundary to a particular position (X-ray crystallography etc.), then the ‘Potential’ qualifier is absent.
Example: P31224

In such cases, the positions of the transmembrane regions can be propagated ‘By similarity’ to closely related homologs.
Example: P24177

2. Annotation of predicted transmembrane regions

We also annotate transmembrane regions which are predicted by the application of the predictive tools TMHMM, Memsat, Phobius and the hydrophobic moment plot method of Eisenberg and coworkers. Note that these tools predict only alpha-helical membrane spanning regions: the positions of membrane spanning beta-sheet regions are annotated strictly according to experimental information.

For predicted alpha-helical transmembrane regions at least two methods must return a positive prediction in order for a region to be annotated as transmembrane in UniProtKB/Swiss-Prot. When predicted N-terminal signal peptides and transmembrane domains overlap, then the Phobius prediction is used to discriminate between the two. In all cases predicted transmembrane domains are annotated as ‘Potential’.

See also: Non-experimental qualifiers

3. Membrane protein topologies in UniProtKB/Swiss-Prot

a) Single-pass transmembrane proteins

We define 4 types of single-pass transmembrane proteins in UniProtKB/Swiss-Prot. Each of these 4 topologies is specified according to a controlled vocabulary in the ‘Subcellular location’ subsection of the ‘General annotation (Comments)’ section.

• Type I: (N-terminus out): characterized by the feature key: ‘Signal peptide’ in the ‘Sequence annotation (Features)’ section.
  Example: P26442

• Type II (N-terminus in): characterized by the description: ‘Signal-anchor for type II membrane protein’ in the ‘Sequence annotation (Features)’ section. The transmembrane domain is located close to the N-terminus of the protein and functions as an anchor.
  Example: P08195

• Type III (N-terminus out): characterized by the description: ‘Signal-anchor for type III membrane protein’ in the ‘Sequence annotation (Features)’ section. The transmembrane domain is located close to the N-terminus of the protein and functions as an anchor.
  Example: O70601

• Type IV (N-terminus in): the transmembrane domain is located close to the C-terminus of the protein and functions as an anchor.
  Example: Q9HDC5

b) Multi-pass transmembrane proteins

Proteins with 2 or more transmembrane domains are considered as integral multi-pass membrane proteins and are not classified further in UniProtKB/Swiss-Prot. Here again the topology is specified in the ‘Subcellular location’ subsection of the ‘General annotation (Comments)’ section.

Many multi-pass membrane proteins do not have a signal peptide. Therefore when predicted N-terminal signal peptides and transmembrane regions overlap, and Phobius cannot be used to discriminate between the two, we annotate the disputed region by default as ‘transmembrane’. In the following example, the protein is predicted to contain a signal peptide which is annotated as transmembrane region.
Example: Q8RKH1

There are, however, examples of well-characterized proteins that have both a signal peptide and multiple transmembrane regions.
Example: P30988

c) Beta-barrel transmembrane proteins

We do not predict the positions of the membrane spanning domains of beta-barrel transmembrane proteins, so sequence annotations for transmembrane regions may be absent. In this case, information about the structure and topology of these proteins may be indicated in the ‘Domain’ subsection of the ‘General annotation (Comments)’ section.
Example: P04840

Related keyword: Transmembrane