Geranylgeranyl generosity: a new prenyl‐transferase gives a fat to a SNARE protein

Protein prenylation, a well‐defined protein consensus motifs direct modification by one of three prenyl‐transferases, has been an area of fairly settled science for 20 or 30 years. Protein prenylation, the specific prenyl modification (farnesyl or geranylgeranyl), as well as the prenyl‐transferases involved can be inferred by protein sequence. Two new papers now upset this settled wisdom with the discovery of a fourth prenyl‐transferase, namely geranylgeranyl‐transferase‐III (GGTase‐III) (Kuchay et al, 2019; Shirakawa et al, 2020).

The identification of a novel prenyl‐transferase, GGTase‐III, adds new insight into protein prenylation and the importance of prenyl modification of its substrate Ykt6 for Golgi complex maintenance.

Prior to GGTase‐III discovery, the known members of the prenyl‐transferase family were farnesyl‐transferase (FTase), geranylgeranyl‐transferase‐I (GGTase‐I), and Rab protein geranylgeranyl‐transferase (RabGGTase or GGTase‐II). Each of them is an α/β heterodimer, with the α subunit largely devoted to binding the protein substrate and the β subunit contributing active site residues as well as a hydrophobic cavity to hold the prenyl‐pyrophosphate reactant, either farnesyl‐pyrophosphate or geranylgeranyl‐pyrophosphate (GGPP) (Lane & Beese, 2006; Leung et al, 2006). FTase and GGTase‐I both act on C‐terminal‐CaaX sites, where the cysteinyl modification site (C) locates four residues from the C terminus, followed by two aliphatic residues (aa), and finally, the indeterminate residue X. The identity of the X residue determines whether the protein will be farnesylated by FTase or geranylgeranylated by GGTase‐I. FTase and GGTase‐I use the same α subunit but have different β subunits. RabGGTase and the newly identified GGTase‐III also share a subunit; they utilize the same β subunit, i.e. RABGGTB, but have a unique α subunits. As the name indicates, RabGGTase is devoted to the geranylgeranylation of Rab proteins, the G protein regulators of vesicular trafficking. Rab proteins are dually prenylated on two C‐terminal or C‐proximal cysteines; examples of such C‐terminal sequences include ‐CC, ‐CCxx, and ‐CxC.

The GGTase‐III discovery finds its origins in the genome project, where a new protein with prenyl‐transferase homology, namely prenyl‐transferase alpha subunit repeat‐containing 1 (PTAR1), was identified. The two new studies show that PTAR1 is indeed a prenyl‐transferase α subunit and that it combines with RabGGTase β subunit to yield the new GGTase‐III (Kuchay et al, 2019; Shirakawa et al, 2020). Both groups provide strong biochemical support for geranylgeranylation as well as in‐depth X‐ray structural analysis of the new enzyme.

To identify likely substrates, Shirakawa et al used purified GGTase‐III to transfer a biotinylated GGPP analogue to HeLa cell extract proteins, allowing potential substrates to be pulled down. This analysis identified just one potential substrate, the atypical SNARE protein Ykt6, which functions both in intra‐Golgi trafficking and in autophagosome–lysosome fusion. Ykt6 is atypical in a couple of respects. Differing from other SNAREs, which use C‐terminal transmembrane domain to anchor to resident membranes, Ykt6 instead relies on two C‐terminal lipid modifications. The Ykt6 C‐terminal sequence ‐CCAIM is reported to be both farnesylated by FTase and then subsequently palmitoylated (Fukasawa et al, 2004), palmitoylation being another lipid modification that is used for tethering proteins to membranes (Chamberlain & Shipston, 2015). A second unique feature of Ykt6 is its N‐terminal longin domain, a globular domain that regulates function by folding back over the SNARE domain to form the “closed” conformation, which is inactive and untethered to membranes (Fukasawa et al, 2004). The farnesyl modification, which is the first lipid to be added, stabilizes Ykt6 into the closed conformation, while subsequent palmitoylation was thought to provide the trigger that springs Ykt6 into its active “open” conformation—exposing the two C‐terminal lipids for membrane anchoring and the SNARE domain for interaction with cognate SNARE partners to drive membrane fusion (Fig. 1). Shirakawa et al, however, now provide convincing evidence that this second lipidation is geranylgeranylation, not palmitoylation. Indeed, using deoxycholate‐based gel electrophoresis that nicely distinguishes the different Ykt6 prenylation states, they find Ykt6 to be exclusively expressed as the dually farnesylated and geranylgeranylated form in wild‐type cells.

Figure 1. Suggested models of Ykt6 C‐terminal lipidation regulating both membrane insertion and activation of its membrane fusion actions.

Left: Key elements of the prior closed/open model (Fukasawa et al, 2004) are depicted. Right: A revised model, based on the new results by Shirakawa et al (2020), is shown.

Having palmitoylation be the Ykt6 conformational switch is attractive, in that reversible palmitoylation is known to regulate the membrane localization of numerous signaling proteins to and from membranes (Chamberlain & Shipston, 2015). Indeed, many palmitoylation‐regulated proteins show C‐terminal sequences similar to Ykt6, namely a C‐terminal‐CaaX that gets prenylated plus additional proximal cysteine(s) that get palmitoylated, with notable examples including N‐ and H‐Ras (Goodwin et al, 2005; Rocks et al, 2005). Shirakawa et al now point out that the prior evidence supporting Ykt6 palmitoylation is not particularly strong, while their new data supporting protein geranylgeranylation are compelling. Ykt6 dual prenylation proceeds via a multi‐step pathway (Fig. 1). Newly synthesized Ykt6 is first farnesylated by FTase. Next, as for other ‐CaaX‐ending proteins, the Ykt6 C terminus is further processed by two endoplasmic reticulum‐localized activities: The C‐terminal aaX tripeptide is clipped off by the RCE1 peptidase, and then, the newly exposed C‐terminal carboxylic acid is methylated by the ICMT methyl‐transferase. In a final step, GGTase‐III geranylgeranylates the adjacent cysteine.

Can Ykt6 geranylgeranylation be reconciled with the closed/open Ykt6 model? Consistent with prior results (Fukasawa et al, 2004), Shirakawa et al find that the initial, singly prenylated, farnesylated Ykt6 adopts the closed conformation. Their evidence further suggests that this closed conformation is maintained even following subsequent geranylgeranylation. Indeed, their modeling indicates that the longin domain hydrophobic cavity can accommodate both prenyl moieties within the closed configuration. Thus, geranylgeranylation appears not to be the conformational trigger, indicating that other yet‐to‐be‐discovered regulatory mechanisms “open” Ykt6 for action.

This short News & Views focuses on just one aspect of this paper (Shirakawa et al, 2020). Many more insights await the reader of the actual paper.

The EMBO Journal (2020) 39: e104744