The emerging significance of Vac8, a multi-purpose armadillo-repeat protein in yeast

ABSTRACT

Vac8 is the sole armadillo-repeat (ARM) protein in yeast. The function of Vac8 in the cytoplasm-to-vacuole targeting pathway has been known for a long time but its role in the phagophore assembly site localization and recruitment of autophagy-related protein complexes is slowly coming to light. Because Vac8 is also involved in formation of the nuclear-vacuole junction and vacuole inheritance, the protein needs to be a competent and wide-ranging mediator of cellular processes. In this article, we discuss two recent studies reporting on Vac8 and its binding partners. We describe Vac8 in the context of crystallized protein complexes as well as predicted models to reveal the versatility of Vac8 and its potential to become a subject of future autophagy research.

Abbreviation: ARM, armadillo repeat; Cvt, cytoplasm-to-vacuole targeting; IDPR, intrinsically disordered protein region NVJ, nucleus-vacuole junction; SEC, size-exclusion chromatography.

KEYWORDS:

• Atg13
• Atg14
• crystal structure
• Nvj1
• Vac17

The cytoplasm-to-vacuole targeting pathway in yeast relies on Vac8, as an earlier study showed that deletion of VAC8 abolishes the import of precursor aminopeptidase I (prApe1) to the vacuole [1]. Since then, it took twenty years to report that Vac8 is essential also for robust autophagy under starvation conditions. Vac8, via binding to Atg13, links the phagophore assembly site to the vacuolar membrane and, thereby, recruits the Atg1 initiation complex near the vacuole [2,3]. Vac8 interacts also with Atg11, to recruit cargo to the vacuole during selective autophagy, and with the Vps15-Vps34 dimer, to localize Atg14 to the vacuole independently of the autophagy machinery [4]. Vac8 is structurally a hybrid protein consisting of 12 armadillo repeats (amino acid residues 40–578) flanked at the N terminus by an intrinsically disordered protein region (IDPR; residues 1–40) and at the C terminus by a large disordered loop (residues 527–556) in armadillo-repeat 12 (ARM12; Figure 1A). The extended segment within the first 10 amino acid residues is very dynamic and harbors three cysteines (C4, C5, and C7) that are palmitoylated in vivo to anchor Vac8 in the vacuolar membrane [5]. The region 20–40 comprises a mobile H1 α-helix. The H1 is flexibly attached to ARM1 [6,7]. The function of the disordered loop in ARM12 is currently unknown.

Figure 1. Vac8 and its binding partners. (A) AlphaFold2 model of Vac8. C4, C5, and C7 are palmitoylated to anchor the protein in the vacuolar membrane. ARM1-ARM12, purple; IDPRs, gray. (B) Crystal structure of the Vac8[10–515]-Atg13[567–695] complex (PDB ID: 6KBM). The Atg13 rendered surface is shown in teal. (C) Crystal structure of the Vac8[10–515]-Nvj1[229–321] complex (PDB ID: 5XJG). The Nvj1 rendered surface is shown in lime green. (D) Crystal structure of the Vac8[10–515]-Vac17[290–344] complex (PDB ID: 7YCJ). The Vac17 rendered surface is shown in blue. (E) AlphaFold2 model of Atg14 from S. cerevisiae. The N-terminal domain folded into a coiled-coil, cyan green; C-terminal domain, teal.

Atg13 binds Vac8 via its C-terminal region (residues 660–685 in S. cerevisiae). Specifically, residues 660–671 of Atg13 loosely interact with ARM7-ARM10 of Vac8 and residues 672–685 tightly bind ARM2-ARM6. The ARM2-ARM10 architecture creates the inner groove of Vac8 where Atg13 attaches in an extended conformation [7] (Figure 1B). This interaction closely resembles the interaction of Vac8 with Nvj1, a nuclear membrane protein that, through binding to Vac8, brings the nuclear and vacuolar membrane together in a physical contact called the nucleus-vacuole junction (NVJ). The NVJ is important for piecemeal microautophagy of the nucleus/PMN [8]. Nvj1 attaches to Vac8 in an antiparallel fashion, as does Atg13. The residues 292–321 of Nvj1 bind in an extended conformation to the inner groove of Vac8 involving R276, R317, and R359 (Figure 1C). This cationic triad is important also for the binding of Vac8 to Atg13, but is less essential for the interaction of Vac8 with Vac17 [6]. Vac17 binds Vac8 during the process of vacuole inheritance when the Vac8-Vac17 complex connects with Myo2, a motor protein that “walks” along actin and, thereby, transports a portion of the vacuole in the mother cell to the daughter cell [5,9]. How can one protein handle all of these interactions? In this article, we discuss Vac8 based on two recent studies that provide new structural insights into the interactions between Vac8 and its binding partners.

One recent study [10] focuses on how Vac8 interacts with Vac17 when compared to Atg13 and Nvj1. Kim et al. show that Vac8 binds Vac17 in an antiparallel manner and in two separate interfaces. One interface encompasses residues 297–308 of Vac17 interacting in an extended conformation via hydrogen bonds with ARM4-ARM7 of Vac8. The other interface utilizes hydrophobic forces between a C-terminal helix of Vac17 (Hc; residues 332–340) and ARM1 and the H1 helix of Vac8 (Figure 1D). Mutations of Vac8 or Vac17 residues in the two interfaces abolish vacuole inheritance. In comparison to Atg13 and Nvj1 that bind Vac8 with a high affinity (K_d of 0.60 µM and 0.71 µM, respectively), Vac17 binds Vac8 relatively weakly (K_d of 2.53 µM). The interactions of Vac8 with its binding partners are mutually exclusive. In particular, when Vac8 is in the complex with Atg13 or Nvj1, Vac17 does not bind to these complexes. Conversely, Atg13 and Nvj1 do bind to the Vac8-Vac17 complex, but very weakly (K_d of 9–10 µM) [10]. From an earlier study, we know that Atg13 does not bind to the Vac8-Nvj1 complex and Nvj1 does not bind to the Vac8-Atg13 complex [6]. Despite these binding relations, Kim et al. conclude that Atg13, Nvj1, and Vac17 are unlikely to compete with each other for Vac8. This conclusion is justified by a previous finding that the endogenous level of Vac8 in the yeast cell is about an order of magnitude higher than the level of its binding partners [11]. Together, these findings mean that one Vac8 molecule can be involved only in one cellular process at a time, but all Vac8-mediated processes can run simultaneously and independently of one another.

In the earlier study probing the Vac8-Nvj1 interaction [6], size-exclusion chromatography (SEC) revealed that Vac8 in solution is a monomer, and so is its deletion mutant lacking H1. Interestingly, removal of the N- and C-terminal IDPRs in Vac8[40–515] makes the deletion construct insoluble, causing it to aggregate in solution, and making it impossible to elute outside of the void volume in SEC [6]. This result indicates that the IDPRs in Vac8 might be important for solubility of the protein. It is possible that these IDPRs in Vac8 function as entropic bristles keeping the protein soluble. Such a function has been found in other proteins with disordered tails [12,13].

In their study [10], Kim et al. used a partially deleted construct, Vac8[10-515] to reveal mechanistic details for how Vac8 binds partial Vac17[290–344] and remains monomeric. The crystal structure shows that Hc of Vac17 binds H1 and, thereby, stabilizes the interaction of H1 with the H2 and H3 helices of ARM1 in Vac8. This four-helical bundle (Figure 1D) allows for the masking of ARM1 by H1, and keeps Vac8 monomeric. The H1 flexibly attached to ARM1 is visible also in the crystalized Vac8[10-515]-Atg13[567–695] complex where Vac8[10-515] is monomeric [7]. The SEC analysis shows that Vac8-Nvj1[229–321], Vac8[1-532]-Nvj1[229–321], and Vac8[10-515]-Nvj1[229–321] complexes in solution harbor the monomeric Vac8 as well [6]. Therefore, H1 is proposed to be an important regulatory element that hinders self-association of Vac8 into a dimer [6,7,10]. Consistent with this proposition is the finding that deletion of H1 or shifting H1 away from ARM1 is associated with Vac8 oligomerization. Specifically, the crystalized Vac8[10-515]-Nvj1[229–321] complex with H1 shifted far away from ARM1 exhibits dimeric Vac8. The Vac8[40–515]-Nvj1[229–321] and Vac8∆H1-Nvj1[229–321] complexes in solution harbor the Vac8 dimer or a higher order oligomer, probably tetramer, detected by SEC [6]. Analogously, Vac8[40–515]-Atg13[567–695] in solution and Vac8∆H1-Atg13[567–695] in the crystal have dimeric Vac8 [7]. Kim et al. compare the crystal structure of Vac8[10-515]-Nvj1[229–321] with Vac8∆H1-Atg13[567–695], both harboring dimeric Vac8, to point out an arch-like overall shape for the former and a super-coil shape for the latter. These authors conclude that Vac8 dimers observed using partial Vac8 molecules in the crystal structures represent functional states of Vac8 with different binding partners. When compared to the Vac8[10-515]-Vac17[290–344] crystal structure, where Vac8 is a monomer, Kim et al. propose that Vac8 differentially navigates three independent cellular processes (vacuole inheritance, NVJ formation, and the Cvt pathway) and that the three crystal structures, two with dimeric Vac8 and one with monomeric Vac8, reveal the molecular basis for Vac8 regulation by its binding partners. Although such a postulation is very appealing, it needs to be emphasized that the crystal structures cannot visualize full-length Vac8 and that Vac8∆H1 is unable to restore the Cvt pathway in vac8∆ cells [7]; thus, Vac8∆H1 is nonfunctional in the Cvt pathway.

To ensure that the dimerization of Vac8 with Nvj1 or Atg13 is physiological and not forced oligomerization of partial Vac8 constructs, important questions remain to be elucidated. One such question is whether the Vac8-Atg13[567–695] complex in solution harbors monomeric Vac8 when tested by SEC. Another question is whether dimeric Vac8 in the crystalized Vac8[10-515]-Nvj1[229–321] complex can be a crystallographic artifact because SEC detects solely monomeric Vac8[10-515]-Nvj1[229–321] and, in the crystal, H1 of Vac8[10-515] replaces H3 in the incomplete ARM12 of neighboring molecules. Apparently, the crystal can force detachment of H1 from ARM1 and then translocation to partial ARM12 in other molecules. This effect is not possible with full-length monomeric Vac8 with an intact ARM12. Analogously, it remains unclear whether dimerization of Vac8 in the Vac8∆H1-Atg13[567–695] complex can simply be a result of missing H1, because H1 can hinder dimerization. Along these lines, one wonders if the Vac8∆H1-Nvj1[229–321] complex, where Vac8 lacking H1 carries the intact C terminus, adopts a super-coil shape, as does Vac8∆H1-Atg13[567–695]. Simply put, the arch and super-coil shape of the Vac8 dimer arise from different Vac8 constructs. Another question is how to reconcile the findings that Vac8[40–515] aggregates, Vac8∆H1 is monomeric, and both these constructs form dimers when they are in a complex with Nvj1[229–321] or Atg13[567–695] [6,7]. It seems plausible that the bound extended conformations of Nvj1[229–321] or Atg13[567–695] may function as solubilizing reagents preventing aggregation of Vac8[40–515] and, at the same time, they may facilitate dimerization of Vac8[40–515] or Vac8∆H1 via exposed hydrophobic residues that are stabilized in H2 and H3 of ARM1 by IDPRs of Nvj1 or Atg13. If this hypothesis is confirmed, then the unexpected and unique self-association of partial Vac8 lacking/misplacing H1 may not be physiological.

As mentioned above, in vivo, Vac8 interacts with the Vps15-Vps34 subcomplex to localize Atg14 to the vacuole [4]. The second recent study discussed in this article [14] showed that Vac8 recruits the class III phosphatidylinositol 3-kinase complex to the vacuole via direct binding to Atg14. Hitomi et al. found that the Atg14 C terminus interacts constitutively with Vac8. Specifically, deletion of the region 157–344 in Atg14 abolishes the protein interaction with Vac8 and prevents recruitment of the class III phosphatidylinositol 3-kinase complex. The Atg14-binding region on Vac8 has not been examined. In relation to the study by Kim et al., one wonders how Atg14 binds Vac8, and whether the Atg13 and Atg14 interactions with Vac8 are mutually exclusive, given that they are both constitutive. The disordered C termini of Atg13, Nvj1, or Vac17 attaching to the inner cavity of Vac8 [6,7,10] reveal the Vac8 binding pattern. The AlphaFold2 model of Atg14 from S. cerevisiae (Figure 1E) shows a disordered proline-enriched tail (residues 316–344) in the C-terminal domain that has been deleted in the study by Hitomi et al. This tail may suit the Vac8 binding pattern and may bind to the inner cavity of Vac8 in a manner similar to that of the Atg13[660–685] peptide, which is also enriched in proline. If confirmed, size exclusion chromatography of Vac8-Atg14[316–344] in solution may reveal the oligomeric state of Vac8 in this complex. It can be useful to note that the disordered tail in Atg14 from yeast is structurally homologous to the ATG14/Barkor Autophagosome Targeting Sequence (BATS; residues 413–492), a disordered region that harbors an amphipathic helix/AH targeting human ATG14 to the phagophore membrane [15]. Interestingly, the amphipathic helix in ATG14 is preceded by a proline-rich sequence segment.

In recent years, we obtained a better understanding of Vac8 interactions with its binding partners. Now we know that disordered extended conformations of the C-terminal tails in Atg13, Nvj1, and Vac17 mold, in a mutually exclusive manner and irrespective of their upstream domains, into the Vac8 inner cavity. Such a mechanism allows Vac8 to interact, in theory, with a large number of diverse proteins, as long as their Vac8-binding domain is disordered and extended. The many-to-one binding scenario, where many IDPRs bind to a groove of one globular protein is not unique to Vac8. Other proteins, such as YWHA/14-3-3 adopt it as well [16,17], which demonstrates the usefulness of this scenario throughout evolution. Given the abundance of protein disorder in the autophagy protein machinery, it is intriguing to wonder what other autophagy proteins may bind Vac8. The Vac8 protein is more important in yeast autophagy than we thought for many years. Perhaps, it is time to start looking for a Vac8-like ARM protein among the mammalian autophagy machinery.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the National Institute of General Medical Sciences (GM131919) and the Protein Folding Disease FastForwards Initiative, University of Michigan.