Many cell-surface proteins are anchored to the membrane by glycosylphosphatidylinositol (GPI), and it has remained unclear how these GPI-anchored proteins (GPI-APs) are internalized by cells. But now, in Developmental Cell, Mayor and colleagues describe a new endocytic pathway that delivers these proteins to recycling endosomes (REs).

Whether recycling GPI-APs are delivered to the Golgi or the RE has been a point of debate, so the authors monitored the endocytic destinations of various GPI-APs — the folate receptor, decay-accelerating factor and GPI–green fluorescent protein (GFP) — in two cell types. They detected all endocytosed GPI-APs in the RE, but not in the Golgi, in both cases.

Although internalized GPI-APs colocalize with transferrin — an endocytosed marker of the RE — a fraction of these GPI-APs consistently remains segregated from transferrin and the Golgi. To find out what this means, Mayor and co-workers compared the early steps in GPI-AP endocytosis with markers of the clathrin-mediated endocytosis pathway, and found that immediately after internalization, a large fraction of GPI-APs is present in compartments that are distinct from transferrin-receptor-containing endosomes. The authors named these tubular–vesicular compartments 'GPI-AP-enriched early endosomal compartments' (GEECs).

Mayor and colleagues showed that GEECs do not colocalize with compartments marked by GFP–rab5 and GFP–rab4 — markers that colocalize with transferrin in sorting endosomes. This further indicates that GEECs are distinct from the classical clathrin-pit-derived sorting endosomes. The authors also found that GEECs constitute an important pinocytic pathway in the cells studied.

Internalization into GEECs specifically requires the GPI anchor, as the authors found that transmembrane isoforms of GPI-APs and a transferrin receptor lacking its cytoplasmic tail were excluded from GEECs. They also showed that GEEC uptake of both GPI-APs and the fluid phase is independent of dynamin 2 and caveolae — other markers of well-established routes of endocytosis.

As the Rho family of GTPases are known to have a regulatory function in endocytic processes, Mayor and co-workers investigated their effect on GEEC formation. They found that inactivation of cdc42 — but not RhoA or Rac1 — inhibits the GEEC pathway, which reduces fluid-phase uptake and results in a redirection of GPI-APs to the clathrin-mediated endocytic pathway.

Endocytosis of GPI-APs seems to be important for their function and pathology — for example, the GPI-anchored prion protein is converted to its infectious form after internalization. The discovery by Mayor and colleagues of a new cdc42-regulated, clathrin- and caveolae-independent, recycling route for these proteins is therefore likely to be of considerable biological and immunological importance.