While it is debated whether all members of the BLTP family mediate lipid transfer in one direction only, most of the bridge-like proteins themselves appear to span membrane contact sites in consistent directions, co-ordinated by interactions with adaptors that are specific to the N and C termini of these proteins. This is most clearly borne out in the interactions of the single VPS13 protein in yeast, where N-terminal ER-to-C-terminal organelle contacts are formed by competitive recruitment of the C-terminus to organelle-specific adaptors (Bean et al., 2018). Even there, the list of competing interactors is likely to be further extended, as yeast VPS13 is also detected at junctions between the yeast vacuole and mitochondria(Gonzalez Montoro et al., 2018), suggesting a different N-terminal adaptor, or a tripartite junction including ER. Nevertheless, this suggests that the orientation of the BLTP at a given junction is important. Interestingly, one recent preprint (Hanna et al., 2025) suggested that some BLTPs may re-orient in response to signalling. In the scenario, it is assumed that the change in orientation determines whether the BLTP3A bridge forms between lysosomes and vesicles (at rest) or between the ER and lysosomes (under lysosomal damage conditions), via a rapid change in N and C terminal interacting partners. However, it is still unknown whether the direction (N-terminus to C-terminus) of lipid transfer changes. Previous in vitro experiments with the closely related BLTP3B/SHIP164 (Hanna et al., 2022) have demonstrated that recombinant BLTP3B protein is capable of dimerising in a head-to-tail orientation, but it is unclear whether this is a conformation found in vivo. If it were, lipids would translocate topologically ‘backward’ over one-half of the junction, and lipid transfer would not be an intrinsic property of the mechanism enabling lipid passage through the RBG hydrophobic groove.

While the example of BLTP3A is the most striking to date, it is apparent that the selection of membranes that the BLTPs transfer to and from is largely determined by the N and C terminal interacting proteins of these family, as the BLTPs do not display strong intrinsic preferences for target junctions in the absence of other factors recruiting them to membranes. As described in yeast, overexpression of proteins that can recruit VPS13 family proteins [the adaptor proteins mcp1 (John Peter et al., 2017), ypt35 (John Peter et al., 2017), and spo71 (Park et al., 2013)] compete (Adlakha et al., 2022; Bean et al., 2018) amongst each other to recruit the single yeast VPS13 protein to different ER-organelle membrane contact sites. This is somewhat distributed over VPS13 paralogues in other species, where each VPS13 has an evolutionarily distinct domain arrangement (Figure 2; Levine, 2022; Leterme et al., 2023). In humans, each of the four paralogues (VPS13A-D) have specific interactors which act together to recruit the family to target membranes (largely via specialised interactions to the VPS13 C-terminus, see Table 2), but even so, each human VPS13 is found at multiple organelle junctions.

Other proteins of the family appear to be less specialised, and it is unclear whether the distinction between the donor and acceptor ends of the protein (and therefore, the direction of lipid transport over these bridges), or the conditions for RBG proteins to be recruited to organelle junctions, is as highly regulated as it appears to be for the ATG2/VPS13 grouping when expressed at endogenous levels.

In some of these proteins, the selection of a ‘donor’ end of the protein may be fixed: BLTP1 has recently been identified with its N-terminal transmembrane helix in complex with two proteins (spigot/C1orf43, an ER-resident protein that cups the N-terminal chorein-like domain, and intake, a nematode-specific helical transmembrane protein which contacts both BLTP1(C. elegans LPD-3) and spigot) (Kang et al., 2024a), while BLTP2 sports its own N-terminal transmembrane helix. At the C-terminal, BLTP1 has a C-terminal amphiphilic helix and polybasic patch, which would be expected to interact with charged phosphoinositide lipid and perhaps dock to membranes (Wang et al., 2022), but would be far easier to dynamically dissolve and remodel (to perhaps select a new target membrane) than the interactions supported by the N-terminus. It is not clear whether the transport of its lipid cargo is in fact uni-directional (even if the orientation of the protein is fixed) as BLTP1 is suspected of being capable of transporting lipids between ER and plasma membrane bidirectionally (Toulmay et al., 2022; John Peter et al., 2022a) at its endogenous levels of expression.

Nevertheless, it would appear that at least some members of the BLTP family have specialisations that may favour unidirectional transport, particularly the VPS13/ATG2 grouping. In the case of VPS13 family proteins, donor/acceptor membranes are also largely determined by protein–protein interactions at the C-terminus of the protein (see Table 2). Interestingly, in at least some subcellular organelles, ATG2 and VPS13 proteins are partially redundant (Lei et al., 2022; Dabrowski et al., 2023), suggesting that despite refined mechanisms for targeting these proteins, which become relevant in human disease, the fundamental property of bulk lipid transfer is the most important facet of their function.

Generally, evidence shows that the chorein domains of BLTPs can bind and solubilise the major membrane phospholipids [phosphatidylethanolamine (PE), phosphatidylcholine (PC), phosphatidic acid (PA), phosphatidylserine (PS), phosphatidylinositol (PI), and phosphatidylglycerol (PG)] and are a poor selector for cholesterol (reviewed here; Hamai and Drin, 2024). However, it does not appear that the chorein domains are particularly selective for their cargoes, which appear to be representative of the major glycerophospholipid classes present in donor membranes. Studies looking at the affinity or transport of lipids show that in recombinant systems, BLTPs can transport PE, PC, PA, and PS and to some extent PI (Kumar et al., 2018; Hancock-Cerutti et al., 2022; Valverde et al., 2019), as well as fluorescent nitrobenzoxadiazole (NBD)-labelled lipids. Deletion of Atg2 in S. cerevisiae stops R18 (octadecyl rhodamine B) transfer (Hirata et al., 2017). R-18 is a single chain C-18 lipid with a hydrophilic fluorescent ‘headgroup’, illustrating the lack of headgroup selectivity of BLTPs. In chimeric assays, the chorein domains of ATG2 and VPS13 can be swapped (Osawa et al., 2019) and still retain function.

As the research currently stands, selectivity for lipids is more likely to be driven by the recruitment of the complex to the appropriate junction, and regulation by the properties of the two membranes driving a transport gradient. Assays in reconstituted systems show that BLTP3B rapidly retrieves a packet of lipid from the donor liposomes and then remains stable if there is no acceptor liposome to receive transferred lipid, suggesting that donor lipid is moving down an internal gradient within the RBG tube (Hanna et al., 2022), which drives extraction from the donor.

X-ray crystallography resolved the S. pombe ATG2 chorein domain in complex with PE (Osawa et al., 2019), where the majority of the interactions were with the acyl chains of the lipid, with what appeared to be a relatively loose selectivity for acyl chain length and little interaction with the lipid headgroup except a weak interaction with the PE phosphoryl group stabilised by an arginine residue in the chorein domain (see Figure 3A). It is also possible that some lipids are incorporated into the flow of lipids into chorein domains as ‘passengers’ and are not necessarily selected for. Molecular dynamics (Wang Y. et al., 2024) models favour an entry mechanism of spontaneous absorption of PC by the hydrophobic chorein domain interior, when the acyl chains of the absorbed lipid have been exposed to the cytosol by membrane bending (e.g., a highly curved membrane modelled in this case by a 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) micelle). This process appears to involve a number of conformational steps in the chorein domain: First, one of the two DOPC acyl chains spontaneously inserts into the hydrophobic chorein domain cavity while the charged headgroup interacts with a charge at the ‘mouth’ of the chorein domain scoop; second, the charged headgroup interacts with other slightly deeper positively charged residues to flip the polar head of PC into the cavity; finally, the second acyl chain trails behind and the lipid is transferred onwards, dragging a new acyl chain from the next lipid to enter behind it, potentially generating a continuous stream of oriented lipid. It is speculated that the energy required to drive these conformational changes in the chorein domain, particularly that needed to untangle one acyl chain from its hydrophobic association with other lipids in the micelle, may be supplied by interacting proteins that lower this energy barrier.

Current thinking suggests that BLTPs exploit membrane gradients to transfer bulk membrane phospholipids. Unless a gradient is built by the expenditure of energy (e.g., enzymatic activity), BLTPs would exploit passive gradients between membranes, as well as the properties of the transported lipids themselves. Some gradients are easily generated by local phenomena (e.g., placing the BLTP very close to a lipid synthase on the donor membrane (Lees and Reinisch, 2020) or a lipid hydrolase at the acceptor membrane, thus causing local depletion), or by deformation of the target membrane which may cause lipids to sort away from their site of arrival on a membrane. Liposome curvature has been shown to drive BLTP-mediated lipid transfer in vitro (Chowdhury et al., 2018), most likely because lipids are already partially exposed to the cytosol as the bending angle increases and thus are more easily solubilised by the chorein domain, and/or added more easily to the acceptor membrane as exposure of acyl chains on the bent target membrane favours the addition of further lipids.

The disturbance of acyl chains on the donor membrane is likely to facilitate lipid transfer. In models of the interaction of the liprin-like C terminal domain of VPS13 family proteins interacting with membranes, the liprin-like domain disturbs the cytosolic leaflet of the acceptor membrane bilayer, exposing the hydrophobic lipid chains to the ‘incoming’ lipid being donated by the BLTP, thus likely increasing the efficiency of their incorporation in the target membrane. Interestingly, several missense mutations of this liprin-like domain cause disease in human VPS13s, suggesting their relevance to the lipid transfer process (Dall’Armellina et al., 2023). Recent structural studies (Wang Y. et al., 2024) showed that in human ATG2A, this domain is extremely flexible with respect to its position relative to the RBG tube and assists in feeding lipids to the ATG9A scramblase in the target membrane.

Consistent with this mechanism being exploited by BLTPs in vivo, examination of subcellular organisation places lipid transfer via ATG2 to the very edge of the growing omegasome (Chowdhury et al., 2018), and BLTP2 is also found at the tips of tubular endosomes defined by Rab8 and Rab10 (Parolek and Burd, 2024), both where the curvature is highest.

Other properties, such as lipid packing, can also be sensed and/or rectified via BLTP-mediated transfer, suggesting that gradients of lipid composition or fluidity also drive BLTP function. This is illustrated in the case of the role of C. elegans lpd-3/BLTP1 countering the effect of cold stress on membrane fluidity (Wang et al., 2022), which is discussed below.

An important consideration is that the action of BLTPs is to drive bulk lipids to the cytosolic leaflet of a membrane bilayer, which then may or may not be redistributed by a scramblase. Eukaryotic membranes favour a less fluid, more tightly packed outer membrane leaflet, while the cytosolic face of the membrane bilayer (the target of BLTP-mediated transfer) is more loosely packed, thus allowing spontaneous membrane deformation (Bogdanov, 2023), and consequent segregating and sorting of membrane subdomains, which may influence a number of properties of the acceptor membrane. While the ER, the most common source of donor lipid for BLTP-mediated transfer, is both enriched in lipid synthases and relatively symmetric across inner and outer membrane leaflets, the acceptor organelles are successively more and more cholesterol-enriched (Maxfield and van Meer, 2010) and asymmetric [the outer leaflet being more rigid and the cytosolic leaflets of most organelles and both leaflets of the ER harbouring more flexible unsaturated lipids (Dingjan and Futerman, 2021)]. Thus, it is conceivable that the deposition of lipids will allow significant changes to membrane signalling due to localised changes in membrane properties.

How the nature of the donor membrane composition may impact proteins such as BLTP1, which appears to transfer lipids via its N-terminal helix and partner transmembrane protein C1orf43 (Kang et al., 2024a) from the PS-rich luminal leaflet of the ER at to the similarly PS-enriched cytosolic leaflet of target membranes (Kobayashi and Menon, 2018), while most other ER-resident lipids are symmetrically distributed across bilayers, is yet to be explored. It is intriguing, however, that both BLTP1 and C1orf43 are required for phagocytosis of L. pneumophila (Jeng et al., 2019) by macrophages and for phagocytosis of neuronal debris (Kang et al., 2024a), suggesting that continuous access to the ER lumen is necessary for function. Similarly, yeast Csf1/BLTP1 immunoprecipitated the ER lumenal enzyme Mdc4 (Toulmay et al., 2022), which depletes PE for the synthesis of GPI anchors. This may suggest that local depletion of PE in the lumenal bilayer is relevant to BLTP1 function.

Several models have been made of lipid interactions with members of the BLTP family, which all favour a continuous stream or bolus of lipid passing through the RBG portion of BLTPs (Li et al., 2020), following the hydrophobic track of residues down the centre of the RBG as predicted by mutagenesis studies (Valverde et al., 2019; Tan and Finkel, 2022). It is unclear whether, having entered the RBG ‘tube’, lipids are subject to any flow regulation as they pass from one side of the BLTP bridge to the other, but some indications exist that some kind of flow regulation or management may take place.

As lipids traverse the RBG repeats, modelling and cryo-electron microscopy find that BLTPs are suited to the co-ordinated deposition of multiple lipids at a time, which allows the lipids to orient and group themselves, thus minimising the energy needed to transfer lipids from one membrane to another. Recent cryo-electron microscopy studies (Kang et al., 2024a) and assays with NBD-PE in reconstituted systems (Hanna et al., 2022) show multiple lipids at a time are docked in the RBG groove.

Recent and elegant molecular dynamic (MD) simulations of lipid docking on open membrane-less models of the RBG portion of BLTPs (Srinivasan et al., 2024), shown in Figure 3B, show no intrinsic preference in terms of direction of entry for the open RBG repeats of ATG2 in MD simulations of free lipids in solution with BLTPs. Simulating sequential docking of multiple free lipids in yeast Atg2, Vps13, and Csf1/BLTP1, the authors show that these proteins can carry in the RBG repeat ‘tube’ at least 15, 49, and 53 phospholipids, respectively, where the charged lipid headgroup faces the solvent. Interestingly, this modelling also showed that the docked lipids were not equally distributed down the length of the RBG tube, instead finding bottlenecks in the hydrophobic track of Atg2 and Csf1/BLTP1 where lipids did not preferentially reside, which may provide an opportunity for transport down the length of the RBG tube to be regulated. These models, where only the open RBG tube is considered, would favour the bi-directional transport of lipids via BLTPs as the free lipids entered from both N and C terminal directions of the modelled open tubes. These bottlenecks in RBG repeats were also found in cryo-EM structures of ATG2A (Wang Y. et al., 2024), suggesting that the passage of lipids through BLTPs is not uninterrupted, and may require changes in conformation to allow passage of lipids. Several possible methods of regulation may exist that would modulate the diameter of the RBG tube during lipid transport, such as locally twisting or compressing RBG repeats, either by the many conserved flexible repeats and motifs that are scattered along the length of RBG proteins (Levine, 2022; Dall’Armellina et al., 2023) or by physical changes in the distance between membranes which may act as a kind of pump to drive lipids across the hydrophobic path.

Other models of BLTP family interaction with lipid in RBG bridges agree with a model of interrupted, and therefore regulatable, lipid transfer. A recent preprint (Kang et al., 2024a) showed that, in cryo-EM studies of C. elegans lpd-3/BLTP1 extracted with its native complex and cargo lipids, this protein was intimately twined via BLTP1’s N-terminal transmembrane helix and its chorein domain with two ER transmembrane proteins (intake, a C. elegans specific gene, and C1orf43/spigot) which allowed a continuous hydrophobic path from the ER lumen to the interior of the BLTP1 RBG repeats, suggesting BLTP1 is continuously anchored to its donor membrane and in contact with transportable lipid. In this study, Kang et al. were able to reveal some very interesting features of the interaction of phospholipid with RBG repeats. First, they noted that the distribution of phospholipids along the RBGs resolved in their structure was not homogenous: The narrower ~25 Å RBG tubes at the N-terminus of BLTP1 harboured a single co-ordinated row of phospholipids, whereas the broader (~40 Å) RBG repeats towards the middle of the protein were associated with 2–5 rows of lipid within the RBG repeat. Most interestingly, while the hydrophobic tails of these lipids faced the hydrophobic interior, the phospholipid headgroup was oriented towards the water in the half-open RBG repeat, by the presence of alternating acid and basic residues, which coordinated the lipids in rows. The more ionisable residues per RBG repeat, the more rows of lipids could be co-ordinated, such that in the wider RBG repeats, lipids were packed at a similar density to one leaflet of a membrane bilayer. The authors speculate that this allows the interior of the RBG repeat to conduct oriented clusters of lipids as if they were small patches of a membrane leaflet. The continuous anchoring of BLTP1 with membrane integral proteins at its N-terminus, and the accommodation of more lipids at the centre of the bridge than at its beginning would likely drive a gradient of movement away from the donor membrane. Interestingly, this was also observed in ATG2A, where, except for a bottleneck at the N terminal region, the hydrophobic track broadened along RBGs until the last segment, which contacts the acceptor membrane. This segment was unresolved in these structures, suggesting an internally generated lipid transfer gradient from the N-terminal to the C-terminal of the BLTP (Wang Y. et al., 2024).

Unfortunately, it was not possible to solve the structure of the entire BLTP1 complex ex vivo at the resolution needed to identify docked lipids over the entire length of the membrane bridge, so the nature of the interaction of BLTP1 with the plasma membrane was not observed in this system. Computational models of BLTP1 structure favour a widening of RBG repeats from the N- to C-terminus, until reaching a narrow neck at the C-terminal region of the protein (Wang et al., 2022). However, BLTP1 (Toulmay et al., 2022; John Peter et al., 2022a) like SHIP164/BLTP3B (Hanna et al., 2022) is of the subgroup of BLTPs for which bidirectional lipid transport is suspected, and this correlates with very little specialisation of the C terminal acceptor end of the BLTP bridge.

There are arguments, however, that the more specialised members of the BLTP family (ATG2, VPS13) for which the evidence for unidirectional lipid transport is stronger, may have adopted features that make lipid transport favour the N-terminal to C-terminal direction. Our own study modelling interactions of the specialised N- and C-terminal domains of VPS13A-D with donor and acceptor membranes (Dall’Armellina et al., 2023) showed that the liprin-like helical bundle, which is found in ATG2/VPS13 proteins, allows a narrowing of the extended RBG repeats until a single helix with a hydrophobic face covers the last distance between the RBG tube and where the liprin-like domain disturbs the cytosolic face of the target membrane (Figure 3C). We speculate that these features reduce the ability of solubilised lipids to ‘crawl’ back up the RBG tube and thus impose a direction of transport, in addition to the liprin-like domain facilitating lipid admixture by disturbing the acceptor membrane cytosolic leaflet.

Based on our modelling on VPS13s, along with work from others since, we propose two models to regulate the passage of lipids through RBG bridges: (1) a spring-like action of compressing and expanding RBG repeats (which potentially may be regulated by interaction with the many conserved cytosolic loops and motifs down the length of the BLTP proteins), and (2) a transient interaction model, where VPS13 RBGs are rigid, and lipids fall down the gradient between donor and acceptor membranes. In this model, the N-terminal chorein domain is weakly associated with donor membranes via long, flexible tethers to proteins such as VAPA/B, allowing transient ‘gulps’ of donor lipids. This could well be similar in other BLTPs, as current evidence suggests that BLTPs are strongly associated with either their donor (BLTP1-2) or their acceptor (VPS13) membranes, but not both—allowing transient interactions with one side of the intraorganelle bridge to drive pulses of lipid down the RBG tube.

The presence of BLTPs, despite their apparent lack of selectivity, also provides some advantages when dynamically manipulating membrane properties, as bulk transfer allows movement of lipids between organelle membranes on the scale of minutes (Wong et al., 2019; Reinisch and Prinz, 2021) thus allowing fast and presumably local response to changes in membranes: This is particularly seen by the action of VPS13C (Schrӧder et al., 2024; Wang X. et al., 2024) BLTP3A(Hanna et al., 2025) and ATG2 (Tan and Finkel, 2022) being recruited to damaged lysosomes for the process of lysosomal repair—a process that is accomplished on the scale of minutes and heavily involves a variety of other LTPs outside of the BLTP family.

In vitro studies using Förster resonance energy transfer (FRET)-based assays in liposomes have measured the transfer of a 2% mix of NBD-PE from donor liposomes at a surprisingly low rate of ~0.017 s-1 per molecule of the BLTP (Maeda et al., 2019). However, subsequent reanalysis of the composition of those liposomes, along with the fact that BLTPs unselectively transport phospholipids regardless of labelling, suggests that the maximal lipid flux per BLTP molecule in that same assay is 50 times higher, at 0.85 molecules s-1, (Hamai and Drin, 2024). This occurs even when there is not necessarily a very strong gradient between donor and acceptor liposomes, limiting the drive to traverse the BLTP. Assays where yeast Atg2-Atg18 are coupled with Atg9 scramblase show a doubling in the efficiency of in vitro NBD-PE transfer between liposomes (regardless if Atg9 was interacting with Atg2 on donor or acceptor liposomes), suggesting that transfer rates are dependent on whether both the inner and outer leaflets of membranes are accessible to lipids donated by BLTP complexes (Chumpen Ramirez et al., 2023).

In vivo measurements, however, put the lipid transfer rate much higher: ~200 lipids per second per molecule of Atg2 in yeast (Dabrowski et al., 2023). In all, this suggests that the rate at which lipids traverse the BLTP will be very much context-dependent, responding to local membrane geometry, membrane composition, and the role of BLTP-associated proteins in processing the lipids that are passing through.

Insights into the consequences of lipid transfer mediated by BLTPs have been gained through structure–function analyses and knockout studies across various organisms and cell types. Collectively, these findings suggest that the effects of BLTP-mediated lipid transfer depend on both the specific lipid being deposited and the manner in which the donated lipid is incorporated with the acceptor membrane. This process influences the geometry and biophysical properties of the target organelle, ultimately modulating cell signalling pathways and membrane sorting.

A key factor in determining how acceptor membranes are altered lies in the presence of accessory complex components associated with BLTPs. These proteins often function as part of larger macromolecular assemblies, interacting with accessory proteins such as Rabs and adaptor proteins or associating with membrane scramblases, which modulate the impact of the transferred lipid on the target membrane. The specific targeting and interaction partners at each BLTP-mediated intraorganellar contact appear to dictate the functional outcome of BLTP activity at a given membrane junction.

BLTPs are thought to deposit bulk membrane lipids onto their target membrane, but the context and method of deposition have different effects on the membrane properties of the target organelle. Experiments show BLTPs can drive membrane expansion in the case of VPS13/ATG2 proteins (Chowdhury et al., 2018; Da Costa et al., 2020) favouring models where these more specialised BLTPs do in fact drive unidirectional transport. Other members of the family such as BLTP1-2 have less specialisation at their C-termini but nevertheless drive changes in membrane properties such as membrane fluidity and consequent signalling.

One key distinction between the actions of BLTPs in their different cellular contexts is the presence or otherwise of associated scramblases as this imposes a geometry as to how the transported lipid will be distributed on the acceptor membrane. These transmembrane proteins facilitate donated lipids accessing the lumenal leaflet of the membrane bilayer, which has the effect of equilibrating the donated lipid across both bilayers of the membrane, which allows expansion of acceptor membranes without much bending. This is best understood for the VPS13/ATG2 grouping of BLTPs. It was noted that VPS13A associates with the scramblase XK (Ryoden et al., 2022) and shares a common disease phenotype (Park and Neiman, 2020) where PS exposure and PC internalisation of the outer face of the plasma membrane are compromised. In yeast, VPS13 forms a complex with the yeast-specific MCP1 protein, which, despite no sequence conservation with XK, also has scramblase activity (Adlakha et al., 2022). This is also the case for the similar ATG2 proteins that interact with the lipid scramblases ATG9a, TMEM41B, and VMP1(Ghanbarpour et al., 2021; Maeda et al., 2020). Cryo-EM suggests that human ATG2A directly feeds lipid to the lateral pores found in a single ATG9 unit in a target membrane, funnelling transferred lipid to the central ATG9A scramblase pore formed by ATG9A trimers (Wang Y. et al., 2024), and thus to the equilibrated membrane bilayer.

In these cases, the equilibration of the inner and outer leaflets of the target organelle will also drive transfer gradients by spreading the donated lipid over both the inner and outer leaflets of the membrane bilayer, by helping to maintain a concentration gradient between the donor membrane and the acceptor membrane cytosolic leaflet. This method of lipid deposition is supported by the established role of scramblase-associated ATG2 in depositing lipids for phagophore formation (Maeda et al., 2019; Chowdhury et al., 2018; Valverde et al., 2019; Osawa et al., 2019). Interestingly, viral replication of tomato bushy stunt virus in yeast and plants requires both Atg2 and its associated scramblase Atg9 (Kang et al., 2024b) showing the importance of this mechanism to ATG2 function.

In proteins such as BLTP3B/SHIP164, which are not known thus far to form complexes with a scramblase, overexpression of the BLTP and a targeting protein (in this case its interactor, syntaxin 6) leads to excessive tubulation of endosomes, possibly driven by the asymmetric delivery of lipids to the outer leaflet of the acceptor membrane, which, if not equilibrated, would need to deform and bend to accommodate the excess lipid (Otto et al., 2010). This change in membrane geometry leads to changes in the trafficking of endosomal cargoes such as cation-independent mannose 6 phosphate receptor and transferrin receptor. This also appears to be the case for the VPS13 family of proteins when these proteins are not interacting with scramblase, where VPS13-mediated transport then favours membrane distortion: Yeast VPS13 directs lipid to lysosomes to accomplish ESCRT-mediated inward budding and membrane sorting to form ILVs (Suzuki et al., 2024) or VPS13B allows the formation of extended tubular ERGIC to accommodate the trafficking and secretion of long extracellular proteins such as procollagen (Du et al., 2024). Similarly, Drosophila BLTP1, apart from other roles in neurons, also has a role in astrocytes, where its depletion interferes with the formation of phagosomes for clearance of neuronal debris(Kang et al., 2024a; Kang et al., 2023).

Although BLTPs transport lipids that are abundant in most membranes, the donor and acceptor membranes connected by BLTP bridges exhibit distinct characteristics, such as the different phosphatidylinositol phosphates (PIPs) that define organelles, and varying levels of cholesterol, which does not appear to be transported as cargo. They also differ in lipid composition (e.g., PC/PE/PS ratio), lipid saturation levels, and the level of asymmetry between the cytosolic and lumenal/extracellular leaflets of the bilayer [reviewed here (Kobayashi and Menon, 2018; Van Meer et al., 2008)].

The rapid admixture of lipids from other organellar membranes, such as the relatively unsaturated lipids found in fluid ER, the site of synthesis for most of the lipids known to be transported via BLTPs, is likely to change the properties of both donor and acceptor membranes. These changes may be to the geometry (e.g., cone-shaped lipids such as PE favour membrane bending, and unsaturated lipids favour membrane flexibility), or the capacity of the membrane to form ordered domains for signalling [as, e.g., is found at ER-membrane contacts (King et al., 2020)]. This has been proposed for BLTP2, which drives lipids from the ER to the plasma membrane to donate lipids to macropinosomes and tubular endosomes as they fuse to the plasma membrane (Dai et al., 2025). The absence of BLTP2 in this context leads to the accumulation of vacuolar structures that are contiguous with the plasma membrane and open to extracellular space, showing the importance of lipid deposition and changed local membrane properties via these proteins to assist membrane traffic, membrane deformation, and consequent signalling.

Nevertheless, there is a degree of subtlety as to how each BLTP is able to change membrane properties. In systems where some level of functional redundancy might be anticipated between members of the BLTP family such as BLTP2 and BLTP1—which both mediate transfer between ER and the plasma membrane—certain functional roles are specialised. These roles may be influenced by specific protein interactions with these BLTPs, as well as by the physical context of their membrane junctions and the local curvature induced by lipid transfer. Both BLTP1 and BLTP2 have important roles in homeoviscous adaptation (HVA), that is, maintaining membrane sorting and signalling properties under cold stress (Ernst et al., 2016). This network of adaptations is large, contextually dependent, and involves many different potential lipid modifications, or selections for properties [e.g., acyl chain length or saturation (Ernst et al., 2016; Ballweg et al., 2020)]. C. elegans LPD-3/BLTP1 is responsible for mitigating membrane rigidification (such as cold stress), by depleting PC from the ER membrane and blocking compensatory upregulation FAT-7 fatty acid desaturase, which uses an alternate path to increase membrane fluidity (by generating unsaturated acyl chain on resident lipids). Interestingly, the BLTP1 function can be bypassed by the addition of unsaturated phospholipids (lecithins) to the plasma membrane (Wang et al., 2022; Pandey et al., 2023). In yeast, CSF1/BLTP1 mutations cause a cold-sensitive phenotype (Tokai et al., 2000), which manifests as a lack of adaptive response (generation of shorter acyl chain lipids and increased acyl chain unsaturation), which could be in part reverted by treatment with the unsaturated lipid oleic acid (John Peter et al., 2022b).

BLTP2 regulates membrane fluidity by increasing PE levels at PM, facilitating cancer growth (Banerjee et al., 2024). Its yeast orthologue, Fmp27, exhibits a cold-dependent growth phenotype, resolved by supplementation with propanolamine, a lipid similar to PE that cannot be converted to PC, suggesting a specific requirement for BLTP2-delivered PE at these sites. While wild-type yeast acclimatise to cold stress by raising the plasma membrane PE:PC ratio, yeast lacking Fmp27/BLTP2 fail to increase plasma membrane PE (Banerjee et al., 2024). The authors also showed that a shift to cold temperature provoked a change in lipid unsaturation in both cell lines, which was abolished by the presence of abundant ethanolamine, a precursor to PE, indicating redundant pathways maintaining membrane fluidity in HVA (Banerjee et al., 2024).

PE is also a precursor of GPI anchors, and while mutation of yeast Fmp27/BLTP2 does not appreciably affect the synthesis of GPI anchors (Banerjee et al., 2024), yeast Csf1/BLTP1 mutants have disturbed GPI anchor biosynthesis (Toulmay et al., 2022). This highlights that although broadly similar, specificity is present between the effects of lipid transport by BLTPs and thus must be conveyed by either location or admixture of lipids.

However, the downstream consequences of altering membrane properties, even by manipulation of the highly abundant phospholipids that BLTPs are known to transfer, are considerable. Changing the balance of PE to PI, where PE is a curvature promotor and PI, because of hydrogen bonding between headgroups, favours uncurved membrane (Klose et al., 2012), induces membrane bending, lipid segregation, and consequent organisation of signalling complexes which affect minor membrane components and signalling lipids such as phosphoinositide lipids (PIPs). This is best illustrated in the downstream consequences of BLTP1 mutation in various model systems. In C. elegans and mouse fibroblasts, BLTP1 mutants show defects in the positioning and abundance of the low-abundance PIP lipid PI(3,4,5)P_3, as well as disorganised actin polymerisation (Wang et al., 2022; Khuong et al., 2010). In contrast, in Drosophila melanogaster, BLTP1 loss affects the plasma membrane positioning and abundance of PI(4,5)P₂, leading to defects in synaptic vesicle cycling (Verstreken et al., 2009; Khuong et al., 2010). It also appears that loss of this pathway triggers other strong signalling across membranes, leading to an insulin-dependent hyperactivation of mTOR signalling and consequent reduction in lifespan for LPD-3/BLTP1 mutants in C. elegans, which could be rescued by lecithin (Pandey et al., 2024), whereas insulin secretion is compromised in Drosophila Hobbit/BLTP2 mutants (Neuman and Bashirullah, 2018). A change in PIP signalling has also been observed in target membranes of cells depleted of VPS13 family proteins (Park et al., 2015), suggesting common effects of bulk phospholipid transfer by BLTPs on membrane signalling properties via PIP lipids.

It is unclear whether these effects on PIP lipid signalling are due to a direct defect in PIP lipid transfer via the RBG bridge, or the deficit in PIP lipids stems from downstream consequences of changes to membrane fluidity and therefore recruitment of specific PIP kinases and phosphatases, which preferentially segregate to lipid microdomains defined by acyl chain composition (Wang and Richards, 2012; Wenk et al., 2003).