Class C ABC transporters and Saccharomyces cerevisiae vacuole fusion

Abstract

Membrane fusion is carried out by core machinery that is conserved throughout eukaryotes. This is comprised of Rab GTPases and their effectors, and SNARE proteins, which together are sufficient to drive the fusion of reconstituted proteoliposomes. However, an outer layer of factors that are specific to individual trafficking pathways in vivo regulates the spatial and temporal occurrence of fusion. The homotypic fusion of Saccharomyces cerevisiae vacuolar lysosomes utilizes a growing set of factors to regulate the fusion machinery that include members of the ATP binding cassette (ABC) transporter family. Yeast vacuoles have five class C ABC transporters that are known to transport a variety of toxins into the vacuole lumen as part of detoxifying the cell. We have found that ABCC transporters can also regulate vacuole fusion through novel mechanisms. For instance Ybt1 serves as negative regulator of fusion through its effects on vacuolar Ca²⁺ homeostasis. Additional studies showed that Ycf1 acts as a positive regulator by affecting the efficient recruitment of the SNARE Vam7. Finally, we discuss the potential interface between the translocation of lipids across the membrane bilayer, also known as lipid flipping, and the efficiency of fusion.

Keywords:

• SNARE
• PI3P
• Vam7
• Ycf1
• Ybt1
• Nft1
• Vmr1
• Bpt1
• Ca²⁺ homeostasis
• lipid flipping

Abbreviations:

• ABC, ATP binding cassette
• DAG, diacylglycerol
• HOPS, homotypic fusion and vacuole protein sorting complex
• MDR, multidrug resistance
• MSD, membrane spanning domain
• NBD, nucleotide binding domain
• PA, phosphatidic acid
• PC, phosphatidylcholine
• PE, phosphatidylethanolamine
• PI, phosphatidylinositol
• PI3P, phosphatidylinositol 3-phosphate
• PI(3, 5)P₂, phosphatidylinositol 3, 5-bisphosphate
• PS, phosphatidylserine
• PX, phox homology
• SNARE, soluble N-ethylmaleimide-sensitive factor attachment protein receptors

Membrane Fusion

Eukaryotic cellular homeostasis requires the trafficking of membrane-bound cargo throughout the cell using mechanisms that are conserved from yeast to man as previously reviewed^1-3 (Fig. 1). Each transport pathway culminates in the fusion of donor and acceptor membranes allowing the transfer of cargo. Membrane fusion has been dissected into experimentally defined stages that begin with the ATP-dependent disruption of inactive cis-SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) complexes on each membrane. The AAA+ protein NSF/Sec18 and its adaptor protein α-SNAP/Sec17 execute this function in what is termed the priming stage.⁴ Primed SNAREs from two membranes form parallel four helical bundles in trans through their SNARE motifs containing a critical central polar glutamine (Q), or arginine (R) that interact in the ionic zero-layer.⁵ Each SNARE bundle is composed of 1 R-SNARE and 3 Q-SNARE coils. Saccharomyces cerevisiae vacuole fusion depends on the R-SNARE Nyv1 and the Q-SNAREs Vam3, Vti1 and Vam7. Vam7 is the only vacuolar SNARE lacking a membrane anchor. It associates with the membrane via its N-terminal phox homology (PX) domain that interacts with the lipid phosphatidylinositol 3-phosphate (PI3P).⁶

Figure 1. Stages of vacuole fusion and formation of the vertex ring. (A) Vacuole fusion undergoes experimentally defined stages. 1. Priming–dispersed vacuoles harboring inactive cis-SNARE complexes are activated by the AAA+ ATPase Sec18 and its adaptor protein Sec17; 2. Tethering – association of vacuoles through the activity of Ypt7 and the HOPS complex; 3. Docking – Vacuoles become tightly apposed leading to the formation of the vertex ring (orange) and trans-SNARE paring; 5 – The docking to fusion transition may also go through a hemifusion intermediate where the outer leaflets mix while the inner leaflets remain intact and prevent content mixing. 6. – Fusion occurs at the vertex ring leading the merger of both membranes (green) and content mixing. Fusion also leads to the internalization of the boundary membrane (yellow and blue). (B) The vertex ring is enriched with the Q-SNAREs Vti1, Vam3, and Vam7, and the R-SNARE Nyv1. The vertex ring domain is also enriched with Ypt7, HOPS and regulatory lipids including PI3P, PI(4,5)P₂, DAG and ergosterol. (Adapted from reference 2).

The next stage of fusion is the tethering reaction driven by Rab GTPases and their effector molecules. Yeast vacuole tethering requires the Rab Ypt7 and the HOPS (homotypic fusion and vacuole protein sorting) complex.^7,8 During the docking stage, trans-SNARE complexes form between vesicles and trigger the release of luminal Ca²⁺ stores.^9,10 During docking the interacting membranes become deformed and can be divided into three morphologically distinct domains.^11-13 Docked membranes become tightly apposed forming 2 flattened discs termed the boundary domain. The edge of the boundary where the two vacuoles come into contact is termed the vertex ring domain and is characterized by its sharp positive curvature where the membranes bend at the domain transition. The vertex ring is the site of fusion where it is enriched with SNAREs, Ypt7, HOPS and a group of regulatory lipids comprising of phosphoinositides (PI), diacylglycerol (DAG) and ergosterol (Fig. 1B). The remaining membrane domain termed the outer edge is free of contacts with other vesicles and retains its spherical shape. Fusion can occur through a hemifusion intermediate at the vertex ring where the outer leaflets of docked vacuoles fuse while the inner leaflets remain intact.^14-17 Thus, the positive curvature of vertex domains during docking changes to negatively curved domains upon hemifusion that contributes to destabilization of the bilayers.^18,19 Ultimately, the inner leaflets fuse and luminal contents mix.

The core fusion machinery has been identified and is sufficient to drive the fusion of reconstituted proteoliposomes.^20,21 However, living biological systems require multiple layers of regulation to respond to changing conditions to prevent spurious fusion. Several regulators have been identified that lie outside of the core fusion machinery including the class-1 casein kinase Yck3,^22-24 the phosphatidic acid (PA) phosphatase Pah1,²⁵ the PI 3-kinase Vps34,²⁶ the Na⁺/H⁺ exchanger Nhx1,²⁷ phospholipase C,²⁸ Rho GTPases,²⁹ and actin.^30,31

In the search for new regulators of vacuole fusion we have now turned to class C ABC transporters as inspired by the work of others. Recent work has demonstrated that ABC transporters, which are well characterized as proteins that detoxify the cell of metals, bile acids and other toxins, can also regulate vacuole homeostasis. For instance the yeast cadmium transporter Ycf1 physically interacts with vacuolar factors that regulate PI3P turnover as well as actin remodeling.³² As mentioned above, both PI3P production and actin dynamics are required for vacuole fusion. This served an impetus to examine the role of Ycf1 and the other class C ABC transporters on vacuole fusion. Here we review our findings of novel roles for ABCC transporters in regulating vacuole fusion.

ABC Transporters

ATP binding cassette (ABC) transporters operate in all living cells from bacteria to human and serve a variety of functions. Chiefly, they function as transporters that use the energy of ATP hydrolysis to move substrates across a membrane.³³ However, ABC proteins can also function as ion channels, channel regulators, receptors, proteases, as well as environmental sensors.³⁴ A variety of molecules are recognized as ABC substrates including ions, anticancer drugs, antibiotics, peptides, and phospholipids.^35,36 ABC transporters can also function as importers and exporters. In bacteria the ABC pumps LmrA and HlyB export toxins out of the bacterium conferring resistance.^37,38 As importers bacterial ABC transporters are employed to uptake nutrients as exemplified by the maltose importer MalFGK2 from Escherichia coli and the histidine permease HisQMP2 from Salmonella typhimurium.^39,40

The ABC transporter family shares a common topology consisting of 2 membrane-spanning domains (MSD) and two nucleotide-binding domains (NBD).⁴¹ The subunits may be formed from separate polypeptides, as found in numerous bacterial transporters. Most eukaryotic ABC transporters are expressed as single polypeptides. Each MSD spans the membrane multiple times to form the substrate translocation channel. The class C transporters are further divided into the short and long subgroups with the latter containing a third MSD N-terminal extension.⁴¹ Some of the predicted membrane-spanning α-helices may not be crucial for substrate translocation but may serve other functions such as membrane insertion or transporter regulation.⁴²

The dysfunction of ABC transporters is implicated in numerous human diseases such as cystic fibrosis, adrenoleukodystrophy, familial hyperinsulinemic hypoglycemia of infancy, and Stargardt disease.^43,44 A separate class of ABC transporters, termed P-glycoproteins, is implicated in multidrug resistance (MDR) and overexpression can lead to the resistance of chemotherapeutic agents.⁴⁵ Other important human ABC transporters are TAP1 and TAP2 (transporter associated with antigen processing), which translocate peptides from the cytosol to the endoplasmic reticulum for the presentation of antigens at the cell surface via MHC class-1 molecules.⁴⁶

The S. cerevisiae genome contains 30 ABC protein genes that are implicated in a variety of cellular functions including drug resistance, pheromone secretion, stress response, and cellular detoxification.^41,47,48 Based on phylogenetic analysis, yeast ABC proteins have been classified into six subfamilies: MDR, PDR; MRP/CFTR, ALDp, YEF3, and RLI subfamily⁴⁷ and later reclassified as ABCA-ABCG to facilitate correlation of experimental findings between yeast and human ABC transporters.⁴¹ The human ABCA subfamily is absent in yeast and there are 2 genes that do not fit within the current classification system.

The best-characterized members of the multidrug resistance related protein (MRP/CFTR or ABCC) family are Yor1 and Ycf1^49,50 (Table 1). Overexpression of Yor1 confers resistance to oligomycin, reveromycin A and organic anions.^51,52 Yor1 has also been implicated in the translocation of fluorescent phosphatidylethanolamine (PE) across the lipid bilayer.⁵³ Ycf1 plays a role in the detoxification of the cytosol by transporting metal-containing peptides and metal ions such as cadmium and arsenic as glutathione-S-conjugates into the vacuole.^50,54 Bpt1 was classified as a bilirubin translocator and also plays a similar role as Ycf1.⁵⁵ Ybt1 was reported to translocate bile acids into vacuoles.⁵⁶ The vacuole also contains the lesser-known ABC proteins Vmr1, Nft1 and the putative transporter YOL075c.⁴¹

Table 1. Location and function of yeast ABCC transporters.

Name	Location	Function	References
Yor1	PM^1	Multidrug transporter	^51 ^49
Ycf1	V	Glutathione S-conjugate transporter; transports toxic metals including cadmium, mercury and aresentie; transports Ade2 pigment; transports bile pigments and free bilirubin; Vam7 recruitment to vacuoles; positive regulator of vacuole fusion	^50,54,55,83
Ybt1	V	Bile and Ade2 transport; phosphatidylcholine translocation; regulates calcium transport; negative regulator of vacuole fusion	^56-58^,^102
Bpt1	V	Glutathione S-conjugate transporter; transports cadmium and Ade2; positive regulator of vacuole fusion	^55^,^83,103
Vmr1	V	Multidrug resistance; Glutathione S-conjugate transporter; Resistance to cadmium and other toxic metals	^53,104
Nft1	V	Unknown	^105

¹PM, plasma membrane; V, vacuole.

Ybt1 Acts as a Negative Regulator of Membrane Fusion

The role of Ybt1 was originally described as transporting bile acids from the cytoplasm to the lumen of yeast vacuoles.⁵⁶ It was later revealed Ybt1 translocates the lipid phosphatidylcholine (PC) across the membrane bilayer as part of choline recycling.⁵⁷ However, its role in vacuole fusion had not been explored. Deletion of YBT1 leads to a marked increase in vacuole homotypic fusion.⁵⁸ Vacuole fusion can be augmented by multiple means including deleting the type 1 casein kinase Yck3,⁵⁹ altering osmolarity,⁶⁰ and increasing the number of SNAREs per vacuole.¹⁷ Vacuoles purified from ybt1Δ yeast contained wild type levels of SNAREs suggesting that the augmented fusion was not caused by increased copies of SNAREs. It remains unclear whether deleting YBT1 affects Yck3 function or osmoregulation. The increased fusion was also not due to changes in any of the key fusion regulators including Ypt7 and HOPS. The changes in fusion did not alter sensitivities to characterized fusion inhibitors (e.g. antibodies against SNAREs) demonstrating that the increased fusogenicity was on pathway. Together these data suggest that the difference in fusion might be due to changes in the efficacy of the fusion machinery. Further experiments showed that the formation of vertex ring microdomains and trans-SNARE complexes were indistinguishable between ybt1Δ and wild type vacuoles, indicating that the alteration occurred after docking. The formation of trans-SNARE complexes triggers the release of vacuolar Ca²⁺ stores.¹⁰ In the absence of Ybt1 the kinetics of Ca²⁺ release was strikingly delayed relative to wild type vacuoles, suggesting that Ybt1 normally attenuates fusion by affecting the release of Ca²⁺ after docking. This is in accord with a report showing that elevated concentrations of extraluminal Ca²⁺ potently inhibited vacuole fusion.⁶¹

Others have shown that lysis can occur during the fusion process, which could contribute to the detected release of Ca²⁺.⁶² In their study, the authors overexpressed soluble GFP in the lumen of yeast vacuoles that lacked the protease Pep4 but contained the inactive alkaline phosphatase zymogen pro-Pho8. These were incubated with vacuole containing Pep4 but lacking pro-Pho8. Upon fusion, Pep4 activates Pho8 and alkaline phosphatase activity serves as a measure of fusion. Interestingly, they detected the release of GFP from the vacuole lumen when fusion was inhibited by various means. Released GFP was measured in the membrane and supernatant fractions after fractionation by centrifugation. They detected a substantial release of GFP in all conditions including those that potently inhibited fusion, as detected by Pho8 activity. Consequently they concluded that lysis readily occurs during incubation. Thus, this could contribute to the detection of extraluminal Ca²⁺. However, we do not believe that spurious lysis is the source of the detected Ca²⁺ signal. In the lysis experiments, released GFP was detected when fusion was inhibited by numerous reagents including the Ypt7 GAP Gyp1 as well as antibodies against SNAREs and the priming machinery. In sharp contrast, Ca²⁺ release is not seen when fusion is inhibited by Gyp1 or antibodies targeting Sec17 and SNAREs. This is in keeping with work by Merz and Wickner showing the link between SNARE pair formation and Ca²⁺ efflux.¹⁰

S. cerevisiae contains multiple Ca²⁺ transporters including the channels Yvc1, Cch1 and Mid1, the pump Pmc1, and the exchanger Vcx1.⁶³ The release of vacuolar Ca²⁺ during osmotic shock occurs through Yvc1 and requires the production of PI(3,5)P₂ by the PI3P 5-kinase Fab1.⁶⁴ However, the channel responsible for Ca²⁺ efflux upon trans-SNARE pairing remains unidentified.¹⁰ The role of Ybt1 in Ca²⁺ efflux is unclear. Ybt1 interacts with the Ca²⁺ ATPase Pmc1, which transports the cation into the vacuole lumen.⁶⁵ When inactive, resting Pmc1 is found in complex with free Nyv1. The Nyv1-Pmc1 interaction is subsequently disrupted when Nyv1 is competed away by its cognate Q-SNAREs.⁶⁶ The dissociation of Nyv1-Pmc1 interactions leads to Pmc1 activation, thus linking the fusion machinery to Ca²⁺ homeostasis. To date it is unknown whether Ybt1 interacts with the Nyv1-Pmc1 complex or with Pmc1 alone. It is possible that the absence of Ybt1 from a putative trimeric complex could destabilize the Pmc1-Nyv1 interaction and lead to facile trans-SNARE pairing and more efficient fusion. Others have found that Pmc1 and Yvc1 are both required for the uptake of extracellular Ca²⁺ during glucose induced Ca²⁺ signaling,⁶⁷ suggesting that the intake and efflux of vacuolar Ca²⁺ is interdependent. Yvc1 is a homolog of the TRPC family, which can be regulated through mechanical and osmotic stress.^68,69 Thus, changes in curvature and lipid composition during vertex ring formation could change the physical environment in which Yvc1 functions. As stated above, the formation of trans-SNARE complexes during vacuole docking leads to pronounced changes in membrane curvature and the lipid make up of the vertex ring microdomain. This coincides with the release of luminal Ca²⁺. Because Ybt1 is a PC flippase, it is possible that ybt1Δ vacuoles accumulate PC on the outer leaflet altering membrane curvature and mechanical signaling to Yvc1 or other unidentified Ca²⁺ channels. Thus, the augmented fusion of ybt1Δ vacuoles could be due to inefficient channel activation due to the changes in the physical properties of the vacuole membrane.

There are several possible mechanisms by which Ca²⁺ efflux might contribute to the regulation of vacuole fusion. As previously mentioned, the docking stage of the fusion pathway requires the formation of the vertex microdomain that is enriched in the proteins and lipids that drive fusion and disrupting the formation or stability of vertex microdomains alters fusion.^11-13 It is likely that altering the spatiotemporal release of Ca²⁺ could neutralize the electrostatic repulsion of tightly packed anionic lipids at the site of fusion, therefore facilitating the tight apposition of the two vesicles and altering the energy barrier for fusion.^70,71 Others have shown that Ca²⁺ can affect the rate of fusion. Using reconstituted proteoliposomes Diao et al. found that fusion could occur through a hemifusion intermediate during slow fusion, or fuse both leaflets simultaneously during fast fusion.¹⁶ Additionally hemifusion could occur in the absence of added Ca²⁺ and that its addition increased the amount of hemifused vesicles and underwent slow fusion. Interestingly, vesicles that only had point contacts (without hemifusion) rapidly fused when Ca²⁺ was added. It must be noted that these experiments were performed with synaptic SNAREs and the Ca²⁺ responsive regulator synaptotagmin 1. The yeast vacuole lacks a known synaptotagmin homologue making it difficult to draw direct comparisons between the 2 sets of data. That said it remains possible that delaying the Ca²⁺ efflux by ybt1Δ vacuoles could allow for a build-up in the number of hemifused and point-contacted vesicles relative to wild type conditions leading to an overall increase in fusion. The mechanisms for the differential effects of Ca²⁺ on docked and hemifused vesicles remain unclear. Furthermore, the physical requirements that determine slow and fast fusion are unknown. Yet, one could speculate that changes in local microviscosity and membrane tension may effect the final path of fusion. The concentration of extraluminal Ca²⁺ could control the level of electrostatic repulsions between membranes. Thus, the early release of Ca²⁺ could stabilize the interactions between membranes and promote the hemifusion path. In contrast the delay in Ca²⁺ efflux would result in a greater tension and trigger the fast fusion path. Clearly, much more work will be required to further investigate these possibilities.

In addition to effecting clustering of anionic lipids, Ca²⁺ can also effect the interactions of proteins with the membrane and interactions between proteins including SNAREs. Zilly et al. showed that sub-micromolar Ca²⁺ concentrations induce the clustering of plasma membrane SNAREs by neutralizing negatively charged amino acid side chains.⁷² Albeit, the role of charged amino acids in clustering was not directly tested. Moreover, they did not take into account the role of Ca²⁺ on the lipid bilayer. It was also found that ≥10 μM Ca²⁺ inhibits SNARE pairing. This illustrates that the spatial and temporal flux of Ca²⁺ could act as a positive and negative regulator of membrane fusion. This is consistent with data showing that vacuole fusion can be inhibited by either Ca²⁺ removal or an excess of extraluminal Ca²⁺.⁶¹ Thus, it is not unlikely that low levels of extraluminal Ca²⁺ promotes the stable interactions of SNARE proteins. This could promote full zippering of the trans-SNARE complexes, triggering the release Ca²⁺ that may reach inhibitory local concentrations and stop the fusion of lagging vacuoles. While the in vitro detection of Ca²⁺ efflux only measures global concentrations, it is possible that local concentrations of Ca²⁺ at the site of efflux may reach micromolar levels thus inhibiting fusion. Therefore, a delay in Ca²⁺ efflux in ybt1Δ vacuoles could promote the build-up of fusion-ready membranes relative to wild type vesicles.

Ycf1 and Fusion

The best-characterized yeast ABCC family member is Ycf1, which regulates the transport of cadmium, mercury and other toxins as glutathione conjugates into the vacuole lumen to detoxify the cytoplasm.^50,54,73 Paumi et al. used an integrated split-ubiquitin membrane yeast two-hybrid (iMYTH) analysis to discover novel Ycf1 binding partners.³² Of interest, Ycf1 was found to physically interact with the PI3P 5-kinase Fab1, which phosphorylates PI3P to produce PI(3,5)P₂.⁷⁴ This suggested that Ycf1 might play a critical role in vacuole homeostasis through binding Fab1. PI3P is an essential factor in the endolysosomal pathway where it interacts with various proteins to support vesicular transport.⁷⁵ On vacuoles PI3P interacts with the soluble SNARE Vam7 via its PX domain as well as the HOPS complex and the Ypt7 nucleotide exchange factor Mon1-Ccz1.^6,24,76,77 The functional role of PI3P can be inactivated by either specific phosphatases such as Ymr1/MTM family members or the kinase activity of Fab1/PIKfyve.^78-81 Fab1 activity leads to the production of PI(3,5)P₂, which itself can be modified by the phosphatase activity of (Fig. 4) to regenerate PI3P.⁸²

Deletion of Ycf1 caused a significant reduction in fusion efficiency that was linked to the exclusion of Vam7 from isolated ycf1Δ vacuoles, whereas other SNAREs, HOPS subunits and Ypt7 were uneffected.⁸³ The attenuated fusion was restored to wild type levels by the addition of recombinant Vam7, indicating that it was able to readily interact with other SNAREs, HOPS and lipids on the membrane to stimulate fusion. Because Vam7 associates with PI3P and Ycf1 might effect the PI3P levels during the fusion reaction we next asked whether the effect of deleting YCF1 on Vam7 recruitment was due to altered PI3P levels. When examining the levels of PI3P on wild type and ycf1Δ vacuoles, we found that the total levels of the lipid were identical in both strains. Together, these data suggested that the physical interactions between Ycf1 and Fab1 are not linked to the steady state levels of PI3P during fusion or the recruitment of Vam7 to vacuoles. Interestingly, deleting YCF1 has a significant effect on the localization of PI3P. In the absence of Ycf1, PI3P localization to vertex microdomains was elevated relative to wild type vacuoles. It remains unclear if the increase in PI3P enrichment at vertices could alter fusion or the recruitment of Vam7 to the membrane. It is possible that the increase in vertex localized PI3P is a side effect of an unidentified negative effects of deleting YCF1.

One possibility for the change in PI3P localization on ycf1Δ vacuoles could be linked to actin dynamics. Previous studies have shown that vacuolar actin undergoes remodeling during the fusion reaction.^30,31 Early in the pathway filamentous actin depolymerizes which could facilitate the lateral movement of proteins to the vertex microdomain. Late in the pathway globular actin polymerizes, where it could potentially serve as a molecular fence to stabilize protein complexes into small domains such as the vertex ring.^84,85 These possibilities are mirrored by the distribution of PI3P and actin dynamics. Using latrunculin to depolymerize actin leads to the accumulation of PI3P at vertex sites.¹³ Conversely, the use of jasplakinolide to stabilize actin filaments sharply reduces PI3P localization at vertices. Jasplakinolide has also been shown to block the vertex enrichment of SNAREs and HOPS.¹² Together, these data illustrate that the state of actin polymerization effects the lateral movement of PI3P, SNAREs and HOPS and their accumulation at the vertex microdomain.

The polymerization of actin requires numerous factors including the activity of the Rho1 GTPase.^86,87 Rho1 interacts with Ycf1 to protect cells from oxidative stress.⁸⁸ Rho1 is activated by its nucleotide exchange factor Tus1,⁸⁹ which physically interacts with Ycf1.³² This suggests that deleting YCF1 could indirectly effect actin dynamics by excluding Tus1 from the vacuole. The increase in vertex localized PI3P on ycf1Δ vacuoles is consistent with the notion that actin polymerization might be attenuated on ycf1Δ vacuoles.

Separately we found that the ATPase activity of Ycf1 was important for vacuole fusion. This was discovered using a point mutation in the NBD1 of Ycf1 that abolished ATPase activity. Complementation of ycf1Δ cells with Ycf1^K669M did not rescue the fusion defect, yet Vam7 levels were not reduced, indicating that Ycf1 performs a secondary ATPase-dependent mechanism to support vacuole fusion. This might appear to be inconsistent with data showing that the in vitro fusion of ycf1Δ vacuole was rescued by adding recombinant Vam7. However, it should be noted that adding recombinant Vam7 to in vitro fusion reactions can bypass blocks in fusion that occur prior to docking and trans-SNARE pairing.^10,17,90,91 Thus, it is possible that Ycf1 serves a second function during fusion that is linked to transport activity prior to docking. Because the in vitro fusion assay is devoid of toxic levels of Cd²⁺ and other soluble substrates, or glutathione S-transferase for that matter, we must suppose that Ycf1 carries out a novel ATPase dependent function. As mentioned above, ABC transporters can have many functions including the translocation of phospholipids across the membrane bilayer. This activity would likely alter the membrane asymmetry and curvature needed for optimal fusion efficiency. This in not unprecedented for a vacuolar ABCC as Ybt1 is known to translocate PC across the vacuole bilayer.^57,92 Therefore, it is possible that Ycf1 translocates a different lipid across the bilayer as part of regulating fusion efficiency.

ABCC Transporters and Membrane Asymmetry

During vesicular trafficking, membranes can be stretched, compressed, made to bulge out during budding, or invaginate to form intraluminal vesicles.⁹³ The curvature of membranes is altered where shape changes occur and might require adjusting the number of lipids on either side of the bilayer to accommodate protein and lipid packing. Membrane distortion can either dissipate compression or focus tension where it is needed in order to regulate mechanisms such as activating mechanosensitive proteins or triggering fusion and fission events.^64,94 Some of these adjustments are mediated by the translocation of lipids between leaflets by proteins termed flippases and floppases with specificities for different lipid classes.^92,95,96 The combined function of multiple flip-/floppases can result in establishing an asymmetric membrane with each leaflet containing a unique lipid profile as observed at the plasma membrane. The cytoplasmic surface of the eukaryotic plasma membrane is enriched in PI, PA, PE, and phosphatidylserine (PS) while the extracytoplasmic leaflet contains primarily PC and sphingomyelin.^97,98 The thermodynamic barrier prevents passive movement of phospholipids between leaflets, thus any changes must occur through the action of energy-dependent carriers.⁹⁹ Importantly, the translocation of lipids can cause changes in membrane shape that effects fission or fusion.⁹⁴ Some of the described flip/floppases belong to the ABC transporter family. For example the human multi-drug resistance associated protein (hMRP1) has been shown to transport PS analogs in human erythrocytes.^100,101 In yeast the ABCC transporter Yor1 translocates PE at the plasma membrane.⁵³ Together this indicates a potential role for other ABC transporters in maintaining lipid asymmetry and effecting membrane curvature.

Thus far Ybt1 has been demonstrated to translocate PC across the vacuole bilayer.⁵⁷ As described above, the deletion of Ybt1 caused an increase in fusion that was linked to a delay in Ca²⁺ efflux. Although it remains uncertain whether the flipping of PC is related to Ca²⁺ transport, it is appealing to think that changing the asymmetry and curvature of the bilayer could have direct physical effects on the vacuolar Ca²⁺ channel activated by trans-SNARE complex formation. The presence of up to 5 additional ABCC transporters on the vacuole opens up the possibility that multiple lipids could be translocated under various conditions to maintain vacuole homeostasis. Regarding fusion, changes in lipid asymmetry could effect the assembly and dynamics of the vertex microdomain. It is already known that disruption of the vertex microdomain by binding or modifying various regulatory lipids inhibits the organization and interaction of SNAREs as well as the binding of HOPS, Mon1 and Vam7 to the vacuole membrane.^13,24,76 Although it remains to be seen if lipid flippases directly effect the fusion machinery there is no doubt that ABCC proteins effect various aspects of vacuole function.

Concluding Remarks and Future Directions

Vacuole fusion is carried out by core machinery composed of a Rab GTP, 4 SNAREs, and the HOPS tethering complex. Like all membrane fusion events, the homotypic fusion of vacuoles relies on a complex circuitry of regulators be they kinases, phosphatases, lipases or ion transporters. While exploring the outer shells of fusion regulation we can now include members of the ABCC family. In addition to their critical roles in the detoxification of the cell cadmium and other toxic metals, we now know that Ycf1 can also regulate vacuole fusion through the recruitment of the soluble SNARE Vam7 to the vacuole. We have yet to discover how this occurs thus future studies will investigate whether the 2 proteins directly interact, or whether it occurs via an indirect mechanism. Similarly we now know that the function of Ybt1 is not limited to the transport of bile, but includes the regulation of Ca²⁺ transport during fusion. Future studies will focus on dissecting the interactions of Ybt1 with the other regulators of membrane fusion and how those interactions might be linked to Ca²⁺ efflux. In addition to elucidating the mechanisms of Ycf1 and Ybt1 we will continue in examining the potential roles of the other ABCC proteins in the regulation of this vacuole homeostasis. Discovering how this unexpected family of transporters influences vacuole fusion and homeostasis will be a challenge for years to come.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank members of the Fratti Lab for the critical reading of the manuscript.

Additional information

Funding

This work was supported by a grant from the National Institutes of Health (GM101132) to RAF.