GPI Anchor (Molecular Biology)

A large number of proteins on the surfaces of eukaryotic cells are covalently linked to a glycosylphosphatidylinositol (GPI) molecule (Table 1). GPI molecules consist of a glycan chain that, contains ethanolamine, glucosamine, and mannose residues linked to the inositol ring of phosphatidylinositol (Fig. 1). The ethanolamine in the GPI molecule is amide-linked to the carboxyl-terminal amino acid residue of the protein after it is translocated across the membrane of the endoplasmic reticulum. The GPI molecule is embedded in the lipid bilayer and is the membrane anchor responsible for retaining the protein at the cell surface. GPI-anchored proteins can be released from the membrane by highly specific phospholipases that cleave the GPI molecule. The complex structure of the GPI molecule (Fig. 1), compared to the other lipids used as membrane anchors, suggests that it has additional functions.

Figure 1. Structure of the GPI anchor. The C-terminal residue of a protein is attached via an amide linkage to the amino group of an ethanolamine phosphate in the GPI molecule (1). A conserved glycan backbone connects the ethanolamine phosphate to the phosphatidylinositol moiety. For simplicity, the sugar residues in the glycan are shown as hexagons. The exact linkages between the sugars and with the phosphatidylinositol and ethanolamine phosphate moieties (i.e., EtNP-6Mana1-2Mana1-6Mana1-4GlcNa1-6myoinositol), are highly conserved between protozoa and mammals. In contrast, the side chains (shown as X, Y, and Z) that are attached to the mannose residues in the glycan backbone are highly variable and can contain a variety of other sugars and additional ethanolamine phosphate groups. The hydrophobic part of the phosphatidylinositol (which will be embedded in the lipid bilayer) is shown as a 1-alkyl, 2-acyl structure, although 1,2-diacylglycerol, 1-alkylglycerol, or ceramide are also found. The hydrocarbon chain length and degree of unsaturation are also highly variable among proteins (the example shown here is for human folate-binding protein). The inositol ring can also become acylated (not shown), which makes the GPI anchor insensitive to degradation by PI-phospholipase C (PI-PLC). The sites of action of two different phospholipases that cleave GPI anchors are also shown. PI-PLC is a phospholipase C that can also cleave phosphatidylinositol and is produced by several different pathogenic bacteria [see Phospholipases C]. Although PI-PLC is not involved in physiological release processes, it is widely used in identifying and experimentally manipulating GPI-anchored proteins on cell surfaces. GPI-PLD is a GPI-specific phospholipase D and cannot hydrolyze phosphatidylinositol. GPI-PLD is abundant in mammalian plasma and may be involved in physiologically releasing GPI-anchored proteins from cell surfaces. Ins = myoinositol, GlcN = glucosamine, Man = mannose, EtNP = ethanolamine phosphate.

Table 1. Examples of GPI-Anchored Proteins

Biosynthesis of GPI-anchored proteins takes place in two main stages: (1) The GPI molecule is synthesized by sequential addition of the glycan chain components to a phosphatidylinositol molecule at the cytoplasmic surface of the endoplasmic reticulum. Then, the GPI molecule is translocated across the membrane of the endoplasmic reticulum to the lumenal surface (1-3). (2) The protein is translocated, cotranslationally, into the endoplasmic reticulum. The C-terminal region of the protein contains a special type of "GPI signal peptide" that is recognized and removed by an enzyme (believed to be a transamidase) located on the lumenal surface of the endoplasmic reticulum (4). The same enzyme catalyzes coupling of the terminal ethanolamine of the GPI molecule to the newly exposed carboxyl group in an amide linkage. The attachment process is completed within a few minutes of translation/translocation, and then the GPI-anchored protein is transported through the Golgi apparatus to the cell surface by the conventional vesicle transport mechanisms used by other plasma membrane proteins. It should be emphasized that relatively little direct biochemical information is available about the attachment process. Efficient GPI attachment may require precise spatial and temporal coordination of the transamidase with the special GPI signal peptide of the protein and translocation of the GPI precursor. Probably for that reason, reconstitution of GPI-anchoring in vitro from purified components presents a formidable technical challenge.

Unlike the other types of lipid anchor, the residue to which the GPI becomes attached (designated site w; see later) is quite variable and relatively difficult to identify from just the amino acid sequence of the protein. Nevertheless, rules for predicting the existence and the location of GPI anchors in a novel protein have been devised on the basis of two sources of information from about twenty naturally occurring, GPI-anchored proteins: (1) comparison of the protein sequence predicted from its gene sequence with the identity of the C-terminal residue in the mature protein, and (2) extensive mutational analysis of the C-terminal region of a limited number of model proteins (4). This region can be divided into three parts:

1. Hydrophobic C-terminus. A hydrophobic C-terminus is essential for GPI-anchoring, although the minimum number of hydrophobic residues required for GPI-anchoring is quite variable among proteins. In most known GPI-anchored proteins, the hydrophobic region is about 15 to 20 residues long, but it can be as short as eight residues. At first sight this region looks like a hydrophobic polypeptide anchor (see Membrane Anchors), and it may in fact serve this temporary function for the short interval between completion of translation/translocation and attachment of the GPI molecule. Polypeptide anchors are generally a few residues longer, however, and usually terminate in relatively hydrophilic residues. Furthermore, mutations that make this region more like a polypeptide anchor usually result in poor GPI anchoring.

2. Hydrophilic region: A hydrophilic region of approximately five to seven residues is located between the hydrophobic region and the attachment site. This region exhibits no strong preference for particular amino acid residues.

3. Cleavage-GPI attachment site: There is no strong consensus sequence, but there are definite amino acid preferences in the vicinity of the cleavage site: Site w (which becomes the C-terminal residue in the mature protein): Gly, Ala, Ser, Cys, Asp, and Asn preferred. Site w + 1: no preference. Site w + 2: Gly, Ala and Ser preferred. A statistical application of the w/w + 2 preferences predict the attachment site in naturally occurring proteins with approximately 80% accuracy. Apart from its predictive value, this information can be used for engineering novel GPI-anchored proteins. Thus, constructing chimeric polypeptides that contain the extracellular domain of a polypeptide-anchored protein and a GPI signal peptide has permitted expressing many proteins in a GPI-anchored form on the cell surface.

The proteins that use a GPI anchor are quite varied. More than 100 different GPI-anchored proteins have been identified (see Table 1for examples). There is, however, no strong correlation of GPI-anchoring with particular types of protein function. For example, cell surface enzymes and adhesion molecules are found in both GPI-anchored and polypeptide anchored forms. Furthermore, although the potential for reversible membrane binding provided by a lipid anchor is useful for regulating protein distribution inside a cell, it would permit continuous loss of proteins from the cell surface (for a general discussion of factors that can affect membrane affinity of lipid-anchored proteins see Membrane Anchors). It is not known why cell surface proteins use the complex GPI molecule instead of the relatively simple lipid anchors used by cytosolic proteins. Two biophysical properties of the GPI anchor, however, may help to reduce dissociation from the cell surface: (1) The GPI anchor usually contains two relatively long (i.e., >16-carbon) unbranched acyl or alkyl chains, which in combination would have relatively high affinity for the membrane (see Fig. 1). (2) Unlike other types of lipid anchors, there is a long and relatively flexible linking molecule, the glycan backbone (see Fig. 1) between the lipid in the bilayer and the protein. This type of linkage reduces the entropy cost of membrane binding because there are minimal restrictions on translational and rotational motion of the protein (see Membrane Anchors).

Although their functional advantages are uncertain, GPI anchors confer some unusual and well-studied properties on cell-surface proteins: (1) GPI anchors are hydrolyzed by specific phospholipases that remove the hydrophobic lipid group, resulting in a protein that is no longer anchored to the membrane (see Fig. 1). GPI-phospholipase D is a secreted enzyme present in mammalian tissues and plasma and could release proteins from the cell surface. The mechanisms by which this enzyme is regulated are unknown, however (5). (2) Under some conditions, GPI-anchored proteins associate preferentially with particular types of lipid molecule in the membrane lipid bilayer. It has been suggested that GPI-anchored proteins are clustered in particular regions at the cell surface (e.g., microdomains and caveolae), but the size, composition, and functional significance of the clusters are controversial (6). (3) Cross-linking GPI-anchored proteins on the cell surface produces an activating signal in some cell types (7). Because the lipid anchor does not cross the membrane, the mechanism by which the signal is transmitted, and its possible connection to microdomains is currently an area of considerable research interest. (4) In spite of the high membrane affinity of the anchor, GPI-anchored proteins can transfer from one cell to another both in vitro and in vivo (8, 9). The precise mechanism of this process and its physiological significance are unknown. Intercellular transfer is of considerable therapeutic interest, however, because it offers a relatively efficient method for inserting novel GPI-anchored proteins (natural or engineered) into the surface of a patient’s cells ex vivo (10).