TRAPP structures reveal the big picture

Two papers in this issue provide new structural insights into the “TRAnsport Protein Particle” (TRAPP) complexes, which play crucial roles in Golgi function. Both papers focus on TRAPPIII, which activates the Rab protein Ypt1 in yeast or the homologous Rab1 in metazoans. The structures illuminate how TRAPPIII specifically recognizes its Rab protein substrate. Joiner et al (2021) also describe a membrane‐anchoring mechanism for yeast TRAPPIII, while Galindo et al (2021) characterize the large subunits that define metazoan TRAPPIII.

Structures of the TRAPPIII complexes from yeast and metazoans clarify how TRAPPIII is recruited to Golgi membranes and how it activates the Ypt1/Rab1 GTPase.

Golgi‐associated TRAPP complexes are intricate and fascinating. The TRAPP core, sometimes referred to as TRAPPI, is a heteromeric peripheral membrane protein composed of small subunits. Additional large subunits bind the TRAPP core to form TRAPPII, while a different set of additional large subunits bind the TRAPP core to form TRAPPIII (Lipatova & Segev, 2019). TRAPPI is present in the cell only at low levels and may or may not act on its own, but the functions of TRAPPII and TRAPPIII are well‐established. Both of these complexes act as guanine nucleotide exchange factors (GEFs) for Rab GTPases, thereby recruiting the GTPases to Golgi membranes. TRAPPII acts late in the Golgi to activate Ypt31/Ypt32 in yeast or the homologous Rab11 in metazoans (Thomas & Fromme, 2016; Riedel et al, 2018; Lipatova & Segev, 2019). TRAPPIII acts earlier in the Golgi to activate Ypt1 in yeast or the homologous Rab1 in metazoans (Riedel et al, 2018; Thomas et al, 2018; Lipatova & Segev, 2019). These GTPases help to coordinate the maturation of Golgi compartments, and Ypt1/Rab1 also functions in autophagy. The challenge is to elucidate how TRAPP complexes are recruited to membranes and how they activate their cognate GTPases.

Previous structural studies have clarified the detailed organization of the TRAPP core, the basic arrangement of the additional large subunits, and the mechanisms of nucleotide exchange (Kim et al, 2006; Cai et al, 2008; Lipatova & Segev, 2019). Yet, those data did not resolve key questions about how the large subunits confer the unique properties of TRAPPII and TRAPPIII. The new papers address this deficiency by characterizing TRAPPIII complexes from Saccharomyces cerevisiae and Drosophila.

Joiner et al (2021) focus on yeast TRAPPIII. The work was directed by Chris Fromme, whose group has performed beautiful biochemical and cell biological analyses of yeast TRAPP complexes. Those efforts helped to establish the specific GEF activities of TRAPPII (Thomas & Fromme, 2016) and TRAPPIII (Thomas et al, 2018), and demonstrated that the length of the C‐terminal hypervariable domain (HVD) of a Rab protein creates a steric constraint that enables TRAPPII to activate Ypt31/32 with its relatively long HVD but not Ypt1 with its relatively short HVD (Thomas et al, 2019). Yeast TRAPPIII is formed by addition to the TRAPP core of a single protein called Trs85. Joiner et al (2021) now present a 3.7 Å cryo‐EM structure of TRAPPIII bound to Ypt1, and they complement the structural data with functional tests of Trs85.

Yeast TRAPPIII is a narrow rod. The new structure reveals that Trs85 contains an α‐solenoid that binds to one end of the TRAPP core (Fig 1A) through a mostly conserved interaction surface. Mutational analysis confirmed the importance of this interaction for Trs85 assembly into TRAPPIII and for Ypt1 activation. Joiner et al (2021) propose that Trs85 helps to anchor TRAPPIII to membranes, based on two lines of evidence. First, Trs85 is needed for stable association of purified TRAPPIII with liposomes. Second, Trs85 contains a conserved sequence that is not resolved in the structure but is predicted to form an amphipathic α‐helix, and as judged by mutational analysis, this putative α‐helix mediates stable membrane binding of TRAPPIII and promotes Ypt1 activation in vivo and in vitro. The same putative α‐helix is required for Trs85 function in autophagy, suggesting a shared mechanism of membrane recruitment. The implication is that by binding TRAPPIII to membranes, Trs85 ensures efficient activation of Ypt1.

Figure 1. Cryo‐EM structures of yeast and metazoan TRAPPIII bound to Ypt1/Rab1.

(A) TRAPPIII from Saccharomyces cerevisiae has an elongated heptameric core, with the large TRAPPIII‐specific subunit Trs85 bound at one end. A putative amphipathic α‐helix in Trs85 seems to anchor TRAPPIII to the membrane. Trs85 corresponds to the N‐terminal half of metazoan TRAPPC8. The yeast TRAPPIII structure was determined with bound Ypt1. A portion of the Ypt1 HVD was visible, and the orange projection depicts the approximate location of the HVD, which is anchored in the membrane by C‐terminal prenyl groups (red). (B) TRAPPIII from Drosophila has an elongated octameric core, with the large TRAPPIII‐specific subunit TRAPPC8 bound at one end and the large TRAPPIII‐specific subunit TRAPPC11 bound at the other end. These two large subunits form “arms” that are linked at a vertex by TRAPPC12 and TRAPPC13. The metazoan structure was determined with bound Rab1, but the HVD was not visible. At the top, a best guess is depicted for the orientation of metazoan TRAPPIII relative to the membrane. At the bottom, the structure has been rotated to look down on metazoan TRAPPIII, which is roughly triangular.

Another exciting aspect of the yeast TRAPPIII structure is identification of a binding site for the HVD of Ypt1. HVDs help to confer specificity for the recognition of Rab proteins by their GEFs. Although HVDs are traditionally considered to be unstructured, part of the Ypt1 HVD is visible at low resolution bound to a TRAPP core subunit. The HVD terminates in a prenylated membrane anchor, so knowledge of the HVD location, coupled with the presence of a positively charged surface that could interact with anionic lipids, suggests an orientation of TRAPPIII on the membrane surface (Fig 1A).

The experimental tractability of yeast TRAPPIII comes at a price because, in most other eukaryotes, TRAPPIII is more elaborate. The metazoan counterpart of Trs85 is a protein of about twice the size called TRAPPC8. Only the N‐terminal half of TRAPPC8 resembles Trs85. Moreover, metazoan TRAPPIII contains additional large subunits called TRAPPC11, TRAPPC12, and TRAPPC13 (Riedel et al, 2018). A similar arrangement is found in the fungus Aspergillus (Pinar et al, 2019). Thus, a full analysis of TRAPPIII requires the use of other model organisms.

Galindo et al (2021) tackle this task by characterizing Drosophila TRAPPIII. The work was directed by Sean Munro, whose group has made major strides in developing Drosophila as a model for studying the metazoan secretory pathway. They previously established that metazoan TRAPPIII activates Rab1, while metazoan TRAPPII mainly activates Rab11 (Riedel et al, 2018). Galindo et al (2021) now present cryo‐EM structures of Drosophila TRAPPIII at resolutions of 4.3–5.5 Å.

This work is a significant contribution because there are no crystal structures for the large subunits of metazoan TRAPPIII, and there was little information about how these subunits are arranged in the complex. The new structures reveal that metazoan TRAPPIII is roughly triangle‐shaped, with TRAPPC8 and TRAPPC11 forming arms that attach to the ends of the catalytic core (Fig 1B). These arms are linked at a vertex of the triangle by TRAPPC12 and TRAPPC13 (Fig 1B). Although the membrane orientation of metazoan TRAPPIII is still somewhat speculative, the triangle probably lies flat against the membrane surface (Fig 1B).

An additional finding from Galindo et al (2021) sheds light on substrate binding by TRAPPIII. It was known that the Rab HVDs are only partly responsible for specific recognition by the TRAPP complexes (Thomas et al, 2019). Now, the C‐terminal half of TRAPPC8 is seen to contact Rab1, an interaction that presumably helps TRAPPIII to be selective. This mechanism is not available in yeast, suggesting that multiple interactions can cooperate to ensure recognition of the correct Rab protein. Galindo et al (2021) speculate about another capability that may be restricted to organisms with large TRAPPIII arms, namely regulated access to the Rab‐binding site. Cryo‐EM provides hints that the TRAPPIII arms are flexible, perhaps enabling them to occlude the Rab1‐binding site in response to regulatory input.

A paired reading of these papers affirms the value of combining data from different model systems and also offers a perspective on the current state of cryo‐EM structure determination. This method is enabling rapid advances, but for demanding structures such as the TRAPP complexes, the path is not yet easy. Both groups performed gymnastics to extract suitable images despite biased orientations of the TRAPP particles, and both groups relied on cross‐linking mass spectrometry to verify the structures. Moreover, the analysis built on prior crystallography of core subunits and subcomplexes. This type of structural biology remains a highly skilled endeavor.

There is still much to learn about the TRAPP complexes. A preliminary analysis of Drosophila TRAPPII by Galindo et al (2021) suggests that its overall structure resembles that of TRAPPIII. Why are TRAPPII and TRAPPIII so big and complicated? Part of the answer may be that the TRAPP complexes have multiple functions. One class of functions enables TRAPP complexes to associate with the Golgi at the appropriate times—e.g., in yeast, TRAPPIII is recruited in a Trs85‐dependent manner at an intermediate point in Golgi maturation (Thomas et al, 2018), and then TRAPPII is recruited in a process that involves binding to the Arf1 GTPase (Thomas & Fromme, 2016). Moreover, TRAPP complexes have been proposed to have tethering activity, particularly for ER‐derived COPII vesicles (Cai et al, 2007). Yeast TRAPPII also recruits a GTPase‐activating protein for Ypt6, which acts between Ypt1 and Ypt31/Ypt32 (Brunet et al, 2016). These observations suggest that we should think of the TRAPP complexes not merely as Rab GEFs, but rather as central components of a circuit that controls Golgi operation and dynamics.

The EMBO Journal (2021) 40: e108537.