Another longin SNARE for autophagosome-lysosome fusion—how does Ykt6 work?

ABSTRACT

Formation of the autolysosome involves SNARE-mediated autophagosome-lysosome fusion, which is mediated by a combination of the Qa SNARE STX17 (syntaxin 17), the Qbc SNARE SNAP29 and the R-SNAREs VAMP7/8. 2 very recent reports have now implicated another R-SNARE with a longin domain, YKT6, in this fusion process. Interestingly, these reports painted two different pictures of YKT6’s involvement. Studies in HeLa cells indicated that YKT6, acting independently of STX17, could form a separate SNARE complex with SNAP29 and another Qa SNARE to mediate autophagosome-lysosome fusion. Conversely, work in Drosophila larvae fat cells showed that while Ykt6 could form a SNARE complex with Snap29 and Syx17/Stx17, it is readily outcompeted by lysosomal Vamp7 in this regard. Moreover, its activity in autophagosome-lysosome fusion is not impaired by mutation of the supposedly critical ionic zero-layer residue from R to Q. In this regard, YKT6 may therefore act in a noncanonical way to regulate fusion. Here, we ponder on the fresh mechanistic perspectives on the final membrane fusion step of macroautophagy/autophagy offered by these new findings. Further, we propose another possible mechanism as to how YKT6 might act, which may provide some reconciliation to the differences observed.

Abbreviations: LD: longin domain

KEYWORDS:

• Autophagosome
• autophagy
• lysosome
• SNARE
• Syntaxin 17 (STX17)
• YKT6

Introduction – SNAREs in autophagosome-lysosome fusion

Macroautophagy (hereafter autophagy) is a cellular homeostasis process conserved in eukaryotes. During the process, cytoplasmic contents, including whole organelles such as the mitochondria, are packaged into vesicular enclosures known as autophagosomes, with the latter’s contents subsequently degraded and recycled via fusion with the lysosomes [1–3]. The autophagy pathway has a largely unique set of machineries and components, but some of which overlap with those mediating the eukaryotic vesicular membrane transport machinery. Such components include RAB GTPases, membrane tethering molecules and complexes [4–8] and those required for membrane fusion, such as NAPA/α-SNAP (NSF attachment protein alpha) [9] and SNAP receptors (SNAREs). In fact, multiple SNAREs and SNARE-mediated fusion events have been documented to be involved, either directly or indirectly, in the various steps of autophagosome biogenesis and maturation [6,10–18].

SNAREs are a family of largely membrane tail-anchored proteins with either 1 or 2 α-helical SNARE domains [19], which mediate intracellular membrane fusion processes. Commonly, SNAREs from a vesicle membrane (v-SNAREs) interact in trans with SNAREs on the target membrane (t-SNAREs) to form a functional SNARE complex through their SNARE domains. The complexing SNAREs constitute a SNAREpin [20] complex that could ultimately ‘zip-up’ to drive the fusion of 2 negatively charged membrane bilayers. A fusion productive SNARE complex typically constitutes a 4-helix bundle, with their SNARE domains aligned in a parallel manner to each other. Buried in the largely hydrophobic core of this 4-helix bundle is a unique ionic layer comprized of 3 glutamines (Q) and 1 arginine (R), each contributed by an α-helical SNARE domain of the complexing SNAREs. The presence of either Q or R in this ‘ionic zero layer’ position is uniquely conserved across members of different SNARE subfamilies [21]. SNAREs could thus be classified as either Q-SNAREs (with different subtypes Qa, Qb, and Qc) or R-SNAREs in this manner, and most known functional SNARE complexes bear the ‘3Q + 1R’ stoichiometry. Among the R SNAREs, the longins [22,23] bearing a conserved PFN (profilin)-like fold known as a longin domain (LD) (for example VAMP7/TI-VAMP, SEC22 and YKT6), could be distinguished from the brevins (for example VAMP1 through VAMP5) that have shorter N-termini without an LD.

The final step of membrane alteration in autophagy involves the formation of the autolysosome via the fusion of mature autophagosome with lysosomes with the aid of specific SNAREs [24]. The autophagosome gains fusion competence only after it matures, as lysosomes should not fuse with pre-autophagosomes or phagophores. The v-SNAREs responsible for autophagosome-lysosome fusion are therefore targeted to the autophagosome outer membrane post biogenesis. Seminal work published earlier from the Mizushima and Juhász laboratories have shown that an autophagosomal Qa SNARE, STX17, together with its cognate SNARE partners, the Qbc SNARE SNAP29 and either one of the R-SNAREs VAMP7 or VAMP8, forms a SNARE complex that mediate this fusion step in mammalian [6] and in Drosophila larvae fat cells [25], respectively. STX17 has a unique C-terminal tandem transmembrane domain structure containing glycine zipper-like motifs that could form a hairpin, and this structure is essential for its selective targeting to the outer autophagosomal membrane. Importantly, STX17 also engages subunits of the homotypic fusion and vacuole protein sorting (HOPS) complex, a membrane tethering complex previously known for its role in the endocytic pathway [26], thus implicating a role for HOPS in the membrane tethering process that precedes autophagosome-lysosome fusion [5,7]. Interestingly, oligomeric forms of ATG14, better known for its role in the localization of the autophagy-specific phosphatidylinositol 3-kinase complex to initiate autophagosome formation, also binds to STX17 and promotes membrane tethering that enhances autophagosome fusion to endolysosomes [27]. How the mature autophagosome acquires lysosome fusion competence is not completely clear yet. However, STX17’s interactions with IRGM (immunity related GTPase M) and the mammalian paralog of Atg8/MAP1LC3/LC3 (microtubule-associated protein 1 light chain 3) was recently shown to mediate the SNARE’s recruitment onto the autophagosome membranes [28].

The Mizushima and Juhász laboratories have now provided new evidence for the involvement of another SNARE, YKT6, in autophagosome-lysosome fusion. Like VAMP7, YKT6 is a longin-type R-SNARE. However, YKT6 is rather promiscuous in terms of its SNARE partners, and has been implicated in a number of different membrane trafficking processes.

YKT6, a jack of many trades in vesicular membrane trafficking

Ykt6 is an unusual SNARE molecule, both in terms of its structure as well as its perceived multifunctional roles. YKT6 was first characterized as an essential gene in the yeast Saccharomyces cerevisiae, whose product is prenylated and functions in endoplasmic reticulum (ER)-Golgi transport [29]. This SNARE is conserved from yeast to humans. YKT6 is likewise also found to play a role in ER-Golgi transport, forming a functional SNARE complex with BET1, STX5 (syntaxin 5) and GOSR1/GS28 [30]. Subsequently, YKT6 was also shown to play a role in early Golgi transport [31], plasma membrane fusion [32] and endosome-trans-Golgi network (TGN) retrograde transport [33] in mammalian cells, as well as TGN fusion in Arabidopsis [34]. In yeast, Ykt6 is also involved in vacuole fusion [35,36] and cargo transport to the vacuole [37,38]. Interestingly, it has also been implicated in processes that include exosome secretion [39,40] in mammals and Drosophila [39,40], and the biogenesis of glycosomes in the parasite Trypanosoma brucei [41].

The functional versatility of YKT6 could be explained by its rather unique features and post-translational modifications [38,42–44]. Unlike the other longins such as SEC22 and VAMP7, YKT6 lacks a transmembrane domain and its membrane association is dependent on its C-terminal lipid anchors. The C terminus of YKT6 comprize a sequence that includes 2 cysteines (-CCAIM), and the sequence is within the context of a lipid modification signal motif (CAAX) [45]. The more distal cysteine is constitutively lipidated by farnesylation, whereas the other is palmitoylated in a reversible manner [45]. Structural studies suggest that upon farnesylation, the molecule’s SNARE domain forms α-helices that fold back onto the longin domain, forming a hairpin-like, closed conformation [42,46]. In this conformation, the lipid group is not exposed, and the closed form of YKT6 therefore remains cytoplasmic without being membrane associated [38]. Palmitoylation at the proximal cysteine results in an open conformation of YKT6 that can then become membrane anchored (see Figure 1(d)). This reversible palmitoylation-mediated translocation between the cytoplasm and membranes apparently enhances YKT6’s general mobility and its potential engagement of different SNARE partners to participate in different membrane fusion processes.

Figure 1. A schematic illustration of the mode of YKT6 function in autophagosome (AP)-lysosome (L) fusion. (a) The model of Matsui et al. [47] for HeLa cells, in which STX17 and YKT6 contributes to fusion in separate SNARE complexes I (comprized of autophagosomal STX17, SNAP29, and lysosomal VAMP7) and II (comprized of autophagosomal Ykt6, Snap29 and lysosomal Stx7). (b) The model of Takáts et al. [48] in Drosophila cells, in which Ykt6 is recruited onto lysosomes and regulates fusion in a noncanonical manner by engaging Snap29 and HOPS, but is ultimately displaced from the final SNARE fusion complex by Vamp7. (c) Another hypothetical model for Ykt6 action, in which YKT6 is recruited onto autophagosomes (instead of lysosomes) prior to STX17 translocation to autophagosomes. YKT6 on autophagosomes could mediate fusion on its own by engaging SNAP29 and a lysosomal Qa SNARE (such as STX7), albeit at a low basal level (dotted arrow). However, when STX17 is recruited onto autophagosomes during autophagy induction, it would take over as the main SNARE that drives fusion, when STX17-SNAP29 forms a SNARE complex in trans with lysosomal VAMP7 (perhaps with YKT6 displaced by VAMP7). (d) A schematic diagram of the domain structure of YKT6, which harbors a C-terminal sequence of -CCAIM. When the more distal Cys residue is farnesylated (green), the molecule’s SNARE domain folds back onto the N-terminal longin domain, forming the ‘closed’ form of YKT6 that remains cytoplasmic. Palmitoylation (blue) at the other Cys results in an ‘open’ conformation of YKT6 that becomes tail anchored to a membrane.

YKT6’s involvement in autophagosome-lysosome fusion

Evidence for YKT6’s involvement in autophagosome-lysosome fusion comes from 2 different experimental models. Matsui et al. [47] have generated STX17 knockout (KO) human HeLa cells using CRISPR-Cas9 genome editing. The authors found that these cells retained some degree of autophagic flux, as indicated by the accumulation of the phosphatidylethanolamine-conjugated form of LC3 (LC3-II), after treatment with the lysosomal vacuolar-type H⁺-ATPase inhibitor bafilomycin A₁. The latter also caused accumulation of partially digested autolysosomes as visualized by electron microscopy (EM). SNARE molecules that are associated with unfused autophagosomes in cells expressing a dominant-negative mutant form of STX17 (which lacks its N-terminal domain) were sought for by the authors. YKT6 is one SNARE found to exhibit such autophagosome localization. GFP-tagged YKT6 colocalizes with LC3 even in STX17 KO cells and the N-terminal longin domain is required for its targeting to autophagosomes. Autophagosome-lysosome fusion that is partially retained in STX17 KO cells is further blocked by YKT6 silencing, and this defect cannot be rescued by STX17 overexpression. YKT6 overexpression also cannot rescue the autophagy flux phenotype of STX17 KO cells, suggesting that STX17 and YKT6 act independently in HeLa cells in terms of autophagic flux.

Matsui et al. further confirmed that, as an R-SNARE, YKT6 can associate with the Qbc SNARE SNAP29, and together these can associate with multiple Qa SNAREs to form SNARE complexes stable enough for their constituents to be co-immunoprecipitated. Given that YKT6 appears to function independently of STX17 in autophagy, the authors identified the lysosome-localized STX7 (syntaxin 7) as a potential Qa SNARE partner for YKT6 and SNAP29 in HeLa cells. A SNARE complex based on YKT6, rather than STX17, can thus account for the remaining autophagic flux in the absence of STX17. Notably, in this case, the autophagosome bears an R-SNARE instead of a Q-SNARE in comparison with the fusion complex based on STX17.

In another report based on a Drosophila model, Takáts et al. [48] have investigated the SNARE requirements of autophagy in L3-stage larvae fat cells. The authors found autophagy flux impairment in RNAi lines targeting fly Ykt6, which is recapitulated by Ykt6 missense and nonsense mutants, and the phenotype resembles deficiencies seen in strains with loss of Syx17, Snap29, Vamp7 or HOPS function. Autophagosomes that are unable to fuse with lysosomes accumulate with Ykt6 deficiency, but not with loss of Syx5/Stx5, indicating that Ykt6 has a direct role in autophagosome processing that is independent of its role in the early secretory pathway. HA-tagged Ykt6 in transgenic lines appears largely cytoplasmic in well-fed larvae, but can be found to label Cp1/cathepsin L-positive lysosomes and fly Atg8. The presence of Ykt6 on lysosomes or autolysosomes is further confirmed by EM.

Possible interactions between Drosophila Ykt6 and the other known SNAREs in autophagosome-lysosome fusion were investigated with tagged proteins in the Dmel2 fly cell line. Ykt6 was found to co-immunoprecipitate with Syx17 only in the presence of Snap29, and these 3 SNAREs can likely form a ternary complex. Interestingly, this complex appears to be less stable than the Syx17-Snap29-Vamp7 complex, as Vamp7 can readily displaced Ykt6 from the Syx17-containing complex. A similar observation was made with the recombinant SNARE motifs of Syx17, Vamp7, Ykt6 and Snap29 in vitro. Structural modelling also suggest that Vamp7 will form a tighter 4-helix bundle with Snap29 and Syx17 than Ykt6. Most importantly, epistasis analysis showed that while overexpression of Ykt6 fails to rescue the autophagy defect resulting from the silencing of Vamp7, overexpression of Vamp7 can restore the defects caused by Ykt6 silencing. Therefore, Vamp7 appears to act downstream of Ykt6 in a functional pathway.

Two other findings of Takáts et al. in the Drosophila model are noteworthy. First, the authors showed that a mutation of Ykt6 in the ionic zero layer arginine (R) residue to glutamine (Q) does not abolish its function with regard to autophagy, unlike the palmitoylation or farnesylation site mutants. Second, Ykt6, like Syx17, binds to the HOPS complex, with its longin domain and SNARE domain engaging different subunits. Therefore, the authors proposed that Ykt6, at least in the Drosophila model explored, has a noncanonical, perhaps regulatory role in autophagosome-lysosome fusion.

Models for YKT6’s mechanism of action in autophagosome-lysosome fusion

The recent findings on YKT6’s involvement in autophagosome-lysosome fusion as discussed above, provided some fresh perspectives to our understanding of this conserved cellular process of critical importance, but also raised new questions. Both reports agree that YKT6 has a role to play in autophagosome-lysosome fusion. However, there are fundamental differences between the findings of these 2 reports, particularly pertaining to the mode of action of YKT6 in the fusion process. Matsui et al. found human YKT6 in HeLa cells acting independently of the STX17-containing SNARE complex in mediating fusion, likely functioning through another SNARE complex of its own, particularly in the complete absence of STX17 in the STX17 KO cells. However, Takáts et al.’s results indicated that Drosophila’s Ykt6 in larvae fat cells functions in a noncanonical manner; it appears to act upstream of the fusion step requiring the Syx17-based SNARE complex, and perhaps regulating the formation and action of the latter. In this regard, it has been shown that some SNAREs may have a regulatory function in SNARE complex formation that fine-tunes membrane fusion specificity [49]. What could be the reason(s) for these differences? While these may be simply due to inherent differences in species and cell type, can these differences be reconciled?

An important question that has remained unclear from the reports is how cytoplasmic YKT6 is eventually recruited to its site of action upon induction of autophagy. While Matsui et al. [47] found that human YKT6 localizes to autophagosomes in starved HeLa cells (colocalization and comigration in a sedimentation gradient with LC3 and STX17), Takáts et al. [48] emphasized the localization of Drosophila Ykt6 to lysosomes and autolysosomes (the latter is apparently supported by immunogold labeling of Ykt6 at the limiting membranes of autolysosomes). It is unclear whether the latter simply reflects a difficulty in morphologically distinguishing unfused autophagosomes from autolysosomes in larvae tissue sections, or perhaps more interestingly, in how YKT6 is differentially recruited to the 2 fusing membranes. It is theoretically possible for YKT6 to be recruited to lysosomes where it acts as an R-SNARE (as VAMP7 does), and for this to occur more readily in some cell types relative to others. However, it is also possible that YKT6, like STX17, is recruited to the mature autophagosome membrane instead, and its interaction with subunits of the HOPS complex would aid membrane tethering prior to fusion. The membrane recruitment process and the mechanism of YKT6 in the different cell types are critically important points that warrant further focus.

Another unanswered question pertains to YKT6’s exact role in autophagosome-lysosome fusion under physiological conditions. It remains possible that in human cells, YKT6 serves a minor yet non-redundant role in fusion for a small fraction of autophagosomes, and this role becomes greatly amplified as cells without STX17 ‘adapt’ or ‘switch’ to using YKT6 as a main fusion-driving SNARE. Both sets of results have in fact indicated that the Stx17-based SNARE complex is of primary importance and has a non-redundant function in autophagosome-lysosome fusion. Phenotypes resulting from STX17 deficiency cannot be effectively complemented by YKT6 even when the latter is overexpressed, but Vamp7 overexpression in fly larvae can potentially rescue Ykt6 deficiency defects. Furthermore, the readiness by which Vamp7 displaces Ykt6 from its cognate SNARE complex in Drosophila cells perhaps indicates that, at least in these cells, Ykt6’s role in fusion could be limited while there is sufficient Vamp7 around. It is also unclear if the R-to-Q mutant of Ykt6, although genetically able to complement autophagy defects due to loss of the endogenous Ykt6 in Drosophila cells, can form a SNARE complex that is functional in the actual membrane fusion process.

Therefore, during autophagy induction, YKT6 could serve either as a redundant/backup fusion-driving SNARE, or more as a regulatory factor. As suggested by Takáts et al’s model (see their Figure 6) [48], YKT6 is recruited to lysosomes and co-engages HOPS and SNAP29 with STX17 on the autophagosome membrane (see also Figure 1, model B). Conversely, we postulate that YKT6 may actually be recruited to autophagosomes rather than lysosomes. In this scenario, YKT6’s ability to engage both cytoplasmic HOPS and SNAP29 may then serve to prime autophagosomes for tethering and to become fusion ready. In this case, the bulk of the fusion process occurs only when STX17 is eventually translocated or recruited onto autophagosome membranes, with the encountering of lysosomal VAMP7 displacing YKT6 (see Figure 1, model C). The STX17-mediated process could therefore become inefficient without priming by YKT6 (thus the autophagy flux phenotype), but autophagy could ultimately still occur, as STX17 itself would still engage both HOPS and SNAP29. However, when STX17 is not translocated to the autophagosomes, fusion may still be completed, albeit at a low basal level, if a lysosomal Qa SNARE (such as STX7) is engaged. This YKT6-mediated fusion process could become markedly enhanced (perhaps as a form of adaptation) in cells depleted of STX17, as in the case of STX17 KO HeLa cells.

Our model offers a reconciliation of the differences in the observations made by Matsui et al. [47] and Takáts et al. [48] and their interpretations of YKT6’s modes of action. Importantly, in this model STX17 recruitment to autophagosomes could represent a regulatory step at the level of autophagosome clearance, which could serve to elevate fusion activity during autophagy induction. Whether this is indeed the case would of course require further investigations. Evidence for the model would include demonstration of membrane fusion competence with in vitro reconstitution involving YKT6, SNAP29 and STX7. In this regard, 2 very recent reports on in vitro reconstitution of autophagosme-vacuole fusion are worth noting. It was shown that the yeast Ykt6 is an autophagosomal SNARE required for autophagosome-vacuole fusion in Saccharomyces cerevisiae [50,51]. There are, however, also other possible models for YKT6 action in autophagosome-lysosomal fusion. As pointed out by a reviewer of this manuscript, YKT6 may also function downstream of STX17 recruitment at the autophagosome to stabilize or ‘shield’ the STX17 complex from nonproductive engagements with other molecules (or itself) until VAMP7 is engaged. A better understanding of the temporal order of YKT6’s recruitment to autophagosomes in comparison to STX17 under basal conditions and following autophagy induction would help to resolve the issue.

Acknowledgments

BLT is supported by the NUS Graduate School for Integrative Sciences and Engineering (NGS). The authors are grateful to the reviewers, whose constructive comments and criticisms greatly improved the manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.