Glycosylphosphatidylinositol (GPI)-anchored proteins (APs) are anchored at the outer leaflet of the plasma membrane (PM) bilayer by covalent linkage to a typical glycolipid and expressed in all eukaryotic organisms so far studied. Lipolytic release from PMs into extracellular compartments and intercellular transfer are regarded as the main (patho)physiological roles exerted by GPI-APs.
In eukaryotic cells, a specific class of surface proteins is anchored at the outer leaflet of the phospholipid bilayer of plasma membranes (PMs) via a glycosylphosphatidylinositol (GPI) glycolipid moiety which encompasses about 150, i.e., 0.5–0.8%, of the translated proteins in mammals 
(for a review dealing with the structure and biogenesis of GPI-APs, see 
). One of the major characteristics of GPI-anchored proteins (GPI-APs) is their release from the PMs through a small set of phospholipases of unique substrate but different cleavage specificity (for a review, see 
) rather than a large panel of proteases, each with a different substrate and unique cleavage specificity, as is required for transmembrane proteins (for a review, see 
). An additional feature is their intercellular transfer, which relies on the complete GPI anchor remaining attached and special carrier mechanisms and structures, among them extracellular vesicles (EVs) and micelle-like complexes. Both attributes alone or in concert may be regarded as the main driving forces for the development and conservation during evolution of GPI-APs vs. transmembrane proteins. The intercellular transfer of GPI-APs with the different structures and molecular mechanisms involved and its putative (patho)physiological implications, which may become cause for the reconsideration of Darwin’s theory of intercellular flow of hereditary information 
, published as Pangenesis theory more than 150 years ago 
, are addressed here.
2. Some Thoughts about the Evolution of GPI-AP Transfer
In the yeast Saccharomyces cerevisiae
, a considerable portion of its GPI-APs (around 60) become transferred and coupled to the cell wall and consequently is engaged in cell wall integrity and assembly 
, among them Cwp2p 
and Tip1p 
. At variance, other yeast GPI-APs remain anchored at the PMs 
or reside at both locations, among them Gas1p 
and Gce1p 
. During the transfer of GPI-APs to the cell wall, the glucosamine moiety of the glycan core together with the PI building block is eliminated and concomitantly the protein moiety becomes coupled to the ß-1,6-glucan of the cell wall via the three mannose residues remaining left at the glycan core 
in the course of a transglucosylation reaction, which involves Cwg6/GPI-3 
. Interestingly, the decision between targeting to the cell wall or the PMs critically depends on the amino acid sequence upstream of the GPI attachment (ω) site 
. Different (positive or negative) signals for cell wall targeting 
or targeting by default 
have been proposed so far. Importantly, more recent studies have shown that the type of inositolphosphorylceramide in the GPI lipid portion is involved in the retention of the GPI-APs at the PMs 
. Finally, Cwh43p and the genetically related Ted1p, which encode proteins engaged in the elimination of EtN-P from the second mannose residue of the glycan core 
, as well as Dcw1 and Dfg5p, putative mannosidases 
, may operate as components of a sorting machinery which is localized at or near the PMs and functions by recognizing the amino acid sequence upstream of the ω-site for the differential transfer of a specific class of GPI-APs from the PMs to the cell wall.
On the basis of the apparent dual localization of GPI-APs in yeast with some of them finally residing at the PMs in the course of trafficking along the typical secretory pathway 
, and others continuing their journey—following their release from the PMs—across the periplasmic space to the cell wall and finally being covalently coupled to its ß-glucan by transglucosylation, it is tempting to speculate about the evolution of the anchorage of membrane proteins by GPI. GPI-APs may have been introduced by yeast operating at both the PMs and the cell surface or cell wall. It is conceivable that, for transglucosylation to occur, the full-length GPI-APs, equipped with the complete glycan core, have to be presented in micelle-like (lyso)phospholipid-containing GPI-AP complexes or EVs to the cell wall ß-glucan and the associated enzymic apparatus. The latter manages to transfer the GPI-APs-ß-glucan intermediates from the PMs across the periplasmic space to the cell wall. Alternatively, protrusions of the PMs, which harbor the envisaged cell wall GPI-APs and come into close contact to those sites of the cell wall, may be involved in the expansion and growth of the cell wall.
The expression of micelle-like GPI-AP complexes or EVs in yeast has not been studied so far. However, previously soluble versions of a subset of full-length GPI-APs, among them Gce1, from the PMs into the periplasmic space of Saccharomyces cerevisiae
under conditions of glucose repression have been reported 
. This may be compatible with the signal-induced transfer of full-length GPI-APs with the final destination cell wall in micelle-like complexes or EVs across the aqueous periplasmic milieu, prior to their transglucosylation to the ß-glucan. With the absence of cell walls in higher eukaryotic cells such as in mammalian organisms, the mechanism of the—spontaneous or signal-induced—release of full-length GPI-APs into micelle-like complexes or EVs could have acquired new functions. GPI-APs may succeed in the “direct”, “vertical”, and/or “horizontal” transfer between neighboring tissue cells with accompanying (patho)physiological consequences.
Taken together, it is concluded that GPI-APs may not have lost their functional diversity during evolution but rather have replaced one major role, their involvement in the biogenesis of and protein anchorage at the cell wall of unicellular fungi, with another one of no less importance: GPI-APs enable the intercellular transfer of proteins in multicellular eukaryotic organisms. Nevertheless, this apparent switch in function does not necessarily exclude the—seemingly remote—possibility that in fungi full-length GPI-APs in micelle-like complexes, which have been released into the periplasmic space, manage to pass the (porous network of the) cell wall without transglucosylation. Thereafter, those may reach the environmental medium and undergo transfer to the PMs of neighboring cells, with the accompanying effects on the phenotype of the latter. This transfer of GPI-APs could be interpreted as exchange of information.
3. Some Thoughts about PIGs, Mediators of Insulin Action, and GPI-AP Transfer
As discussed above, anabolic effects, i.e., stimulation of glycogen and lipid synthesis, are elicited in acceptor cells such as human adipocytes or ELCs upon transfer of full-length GPI-APs from donor cells with the concerted action of lipid-containing micelle-like GPI-AP complexes and lipolytically cleaved GPI-APs (PIG-proteins), which are produced in response to physiological stimuli such as hormones (e.g., insulin) or therapeutic agents such as antidiabetic SUs of the third generation (e.g., glimepiride). This raises the question about the relationship of the molecular mechanism underlying this insulin-mimetic activity and the insulin-mimetic activity exerted by the so-called soluble mediators of insulin action (PIGs). The latter were identified almost four decades ago in the incubation media of a multitude of insulin target cells, such as adipocytes, but have resisted purification to homogeneity and unambiguous structural elucidation so far. The previous model of their generation encompassed the two-fold—lipolytic and proteolytic—cleavage within the GPI anchor and at the carboxy-terminus of the protein moiety, respectively, of GPI-APs and, as the underlying molecular mechanisms of their insulin-mimetic action, their specific transport across the target cell PMs and allosteric regulation of the activity of key metabolic enzymes in the cytoplasm in insulin-like fashion (e.g., 
; for a review, see 
). However, this model has remained a matter of intense dispute for decades, as held true for the corresponding data basis 
As summarized above, strongly argue that the insulin-independent stimulation of glucose and lipid metabolism by PIG-proteins, generated in response to physiological and pharmacological stimuli, as well as by synthetic PIGs, added to the incubation medium, in concert with full-length GPI-APs, which interact with GPI-binding proteins, relies on extracellular sites and mechanisms of action. Thus, almost 40 years of research have finally led to a shift in the thinking about the mode of the insulin-mimetic action of PIGs from intracellular to extracellular. This shift coincides with (i) the dissociation of full-length GPI-APs from GPI-binding proteins into the incubation medium of insulin target cells in vitro and the circulation or interstitial spaces of treated rodents in vivo, respectively, and (ii) the subsequent transfer to and insertion into the outer leaflet of the PMs of insulin target cells of the full-length GPI-APs. This extracellular mode of action of PIGs which—by nature—is susceptible towards a multitude of exogenous—experimental, (patho)physiological and environmental—factors may explain the prominent problems with the (high) statistical variance and, in part, even (limited until missing) reproducibility of some of the published findings regarding the insulin-mimetic activity in vitro and in vivo of both isolated and synthetic PIGs (this issue has been discussed extensively in ref. 
Furthermore, the extracellular mode of action of PIGs may pave the path for the development of novel anti-diabetic agents. Previous efforts to use the structure of PIGs for the design of insulin-mimetic small molecules with oral bioavailability were severely hampered by the apparently erroneous assumption of the need for cell permeability. Furthermore, it turned out to become difficult to convert the PIG glycan core into a non-carbohydrate structure in order to gain stability during and to increase the efficacy of the gastrointestinal passage for oral absorption of putative insulin-mimetic drugs, which rely on this mode of action. In fact, systematic variation of the individual carbohydrate components of the PIG glycan core as well as of their glycosidic linkages on the basis of experimentally established structure-activity relationship led to the design of PIGs which exerted almost full insulin activity in vitro (e.g., stimulation of glucose transport in primary rat adipocytes, see 
). However, those efforts failed in the generation of non-carbohydrate PIG-mimetics with insulin-mimetic activity comparable to that of the authentic PIGs as well as in the reduction of the size of the latter in order to reduce the expenditure of their chemical synthesis.
The elucidation of the extracellular site of action of the PIG-proteins which involves the displacement of full-length GPI-APs from serum GPI-binding proteins, among them—but not restricted to—albumin, and the parallel introduction of an innovative chip-based and microfluidic SAW biosensor for its assaying (even in the high-throughput mode) may create novel opportunities of drug discovery research for the pharmaceutical industry, which is engaged in the therapy of metabolic diseases, especially type 2 diabetes and obesity. In particular, in the future efforts to identify serum GPI-binding proteins (different from albumin) could lead to the characterization of those which are equipped with an allosteric site of regulation. It may be of relevance that the activity of GPLD1 has already been demonstrated to be positively controlled by divalent cations 
. The use of allosterically controlled GPI-binding proteins in screening efforts may lead to the design of non-carbohydrate small-molecule insulin-mimetic drugs of lower EC50
compared to those which rely on the competitive displacement of full-length GPI-APs.