Vesicle transport, cilium formation, and membrane specialization: The origins of a sensory organelle

Almost all mammalian cells can possess a single cilium, called a primary cilium. On polarized epithelial cells, this cilium extends from the apical surface into the extracellular space. Long underappreciated, the cilium has recently received a great deal of attention because it has become clear that ciliary defects contribute to important human diseases such as polycystic kidney disease and retinal degeneration. However, despite the near ubiquity of cilia and their importance to human disease, little is known about the regulation of ciliogenesis. In this issue of PNAS, Vieira et al. reveal that FAPP2 is a new player in ciliogenesis (1).

FAPP2 and the related protein, FAPP1, are resident trans-Golgi network (TGN) proteins previously implicated in vesicle trafficking to the apical membrane (2). FAPPs are defined by an N-terminal pleckstrin homology (PH) domain that binds to phosphatidylinositol-4-phosphate [PI(4)P] (3), the phosphoinositide most prevalent in the Golgi complex. Phosphoinositides have long been known to function in regulated transport and specifically in transport from the Golgi complex (4).

Vieira et al. (1) examine the function of FAPPs using RNAi-mediated knockdown in Madin–Darby canine kidney (MDCK) cells. Normally, MDCK cells grown on filters or in Matrigel (an extract of the extracellular matrix produced by the Engelbreth–Holm–Swarm tumor) form a well polarized epithelial monolayer, complete with primary cilia. Depletion of FAPP2, but not FAPP1, abrogates cilium formation.

Because vesicle trafficking has not previously been linked to vertebrate ciliogenesis, a central question raised by this work is how FAPP2 contributes to cilium formation. Because the primary cilium is an apical structure in epithelial cells, one obvious explanation would be that FAPP2 depletion alters some fundamental aspect of cell architecture, such as organelle structure or apical–basal polarity. Indeed, Fan et al. (5) have demonstrated that interfering with the function of known regulators of apical–basal polarization blocks cilium formation. However, this does not seem to be the case with FAPP2. Distribution of standard markers of apical and basolateral polarity is not disrupted with diminished FAPP2 and neither is formation of the tight junction or morphology of the TGN (2). Subtle changes in apical–basal polarity are apparent in FAPP2-depleted cells, as demonstrated by the altered distribution of galectin-3, a marker of the subapical domain.

If apical–basal polarity is not grossly perturbed, how might FAPP2 function in ciliogenesis? Apical transport uses a number of different mechanisms, but apical transport of a variety of different proteins, including YFP-GL-GPI, A-VSVG-GP, and HA, is diminished in FAPP-depleted cells (2). Interestingly, ultrastructural analysis reveals that in FAPP2-depleted cells, vesicles accumulate between the apical membrane and the centrioles (1). Proteins involved in vesicle transport can participate in multiple events, including cargo selection, vesicle release from the TGN, recruitment of transport machinery, docking with the target membrane, and membrane fusion itself. That FAPP2 depletion causes vesicles to accumulate under the plasma membrane begins to suggest that formation and release of vesicles destined for the apical membrane are not abolished, but that the later steps of vesicle docking or fusion are affected. It would be interesting to know a bit more about these subapical vesicles, including whether they are decorated with FAPP2 or carry apical cargo, as expected.

Little is known about the regulation of ciliogenesis.

The centrioles that form the core of the basal body, the foundation of the cilium, are normally closely associated with vesicles and the TGN (see figure 3 in ref. 6). Vesicles carrying transmembrane proteins destined for the cilium are thought to fuse to the plasma membrane adjacent to the basal body (7, 8). Therefore, it might be expected that disruption of apical transport could prevent the delivery of ciliary transmembrane proteins. One possible explanation for the role of FAPP2 in ciliogenesis is that it functions in the apical delivery of a transmembrane protein essential for ciliogenesis.

Another possibility is suggested by investigations into a previously unrevealed function of FAPP2. In addition to its PH domain, FAPP2 has a C-terminal domain homologous to Glycolipid Transfer Protein, raising the prospect that FAPP2 binds glycolipids. To investigate this possibility, Vieira et al. (1) use the amphiphilic dye Laurdan and the distribution of Forssman glycolipid, a glycolipid predominant on the apical and ciliary membranes (9, 10). In FAPP2-knockdown cells, Forssman glycolipid is redistributed to the basolateral domain. Similarly, Laurdan, a dye that incorporates into and reflects the degree of organization of lipids in the membrane bilayer (11, 12), shows an altered distribution of its generalized polarization function (a property of its emission spectrum) in FAPP2-knockdown cells. Together, these results indicate that FAPP2 functions not only in protein trafficking but also in lipid distribution.

It is certainly possible that FAPP2 is essential for organizing lipids in such a way that promotes ciliogenesis. Examination of the cilium with Laurdan staining reveals that the base of the cilium is a region of particularly high lipid ordering (1). Perhaps this lipid ordering is essential for either the docking of the basal body to the plasma membrane or the extension of the ciliary axoneme, both essential steps in ciliogenesis. In addition, the biophysical properties of lipid ordering might contribute to the structural and functional properties of the cilium. One possible role for this region, suggested by Vieira et al. (1), is that it could act as a barrier to the diffusion of proteins into or out of the cilium. It will be exciting to see the results of tests of this idea.

Further close examination of the apical membrane reveals that two apical glycosylphosphatidylinositol (GPI)-linked proteins are excluded from a region surrounding the cilium (1). This striking result indicates that a circle of apical membrane around the base of the primary cilium is fundamentally different from the rest of the apical membrane. This difference is further revealed by the finding that this domain contains galectin-3, previously known as a marker of the subapical domain. Thus, instead of a homogenous region, the apical surface of epithelial cells comprises at least three distinct regions: the majority of the apical membrane surrounding a periciliary domain of ≈2 μm² and, within that, the cilium proper. Each of these regions has a distinct protein composition and perhaps a distinct lipid composition (Fig. 1).

Fig. 1.

Previous studies have shown that different plasma membrane domains of polarized epithelial cells have distinct protein compositions. For example, Polycystin-1 (PC1), Polycystin-2 (PC2), and Fibrocystin are components of the ciliary membrane (light blue), whereas gp58 and E-cadherin are components of the basolateral membrane (dark blue). Vieira et al. (1) reveal that there are previously unrecognized domains. The base of the cilium is a region of high lipid order (orange), and the apical membrane can be divided into a periciliary domain expressing galectin-3 (green) and the remainder marked by apical GPI-linked proteins (red). Also shown are the Golgi complex (black); the centrioles of the basal body (gold); and the TGN (pink), the site of FAPP2 localization.

Key questions remain regarding the function of this periciliary domain. Is this periciliary domain the same as the domain of ordered lipids at the base of the cilium? Is this domain essential for centriole docking to the membrane or some other step in ciliogenesis? Is FAPP2 required for formation of the periciliary domain or for delivery of vesicles to this domain? How are some apical proteins excluded from this domain?

In addition to raising many interesting questions, the presence of a distinct periciliary domain suggests a possible solution to an unresolved question about the trafficking of transmembrane proteins to cilia. In Caenorhabditis elegans, mutation of a subunit of the AP-1 clathrin adaptor complex prevents the trafficking of transmembrane proteins to primary cilia (13, 14). AP-1 had previously been known to function in sorting to endosomes and the basolateral membrane, not to the cilium (15). Given that the periciliary domain is marked by galectin-3, perhaps it has some properties in common with the subapical domain. If so, these shared properties may allow AP-1 to target vesicles to the periciliary domain, explaining why a protein complex not previously implicated in apical transport directs the movement of proteins to the cilium.

The presence of this distinct patch of apical membrane may also have implications for the evolutionary origins of the cilium. A wide diversity of eukaryotic organisms form cilia, including metazoans, Chlamydomonas, Tetrahymena, Trypanosoma, Leishmania, Giardia, and Paramecia. This phylogenetic breadth suggests that the last common eukaryotic ancestor possessed a cilium, and that organisms lacking cilia, such as fungi and some plants, have subsequently lost them (16, 17).

Ciliated organisms possess conserved proteins involved in ciliogenesis, such as those that participate in intraflagellar transport (IFT), the active movement of proteins up and down the cilium (18). IFT proteins show both overall structural and sequence similarity to components of coat protein I (COPI) and clathrin coats (19). This homology raises the possibility that IFT has evolved from a vesicle coat complex involved in trafficking. Because IFT particles associate with a region around the centrioles of the basal body, the proto-IFT complex may have participated in trafficking from the TGN to a specialized membrane domain adjacent to the centrioles. Certainly, the identification of the periciliary domain as a distinct membrane between the basal body and cilium supports such a scenario.

Membrane specialization could confer advantages such as compartmentalization of sensation or organization of motility, both of which are important ciliary functions in existing organisms. Jékely and Arendt (19) suggest that expansion of this specialized membrane patch and coordination with centriole-based microtubules may have changed this domain into the first recognizable cilium. It is interesting to speculate that this inconspicuous and previously unrecognized periciliary domain might be the descendant of an ancient instance of cell polarization. It is clear that despite over a century of work on the “difficult problem of the structure of cilia” (20), this area will be the source of other fascinating insights for quite some time to come.