Dissecting the mechanism of atlastin-mediated homotypic membrane fusion at the single-molecule level

Shi, Lijun; Yang, Chenguang; Zhang, Mingyuan; Li, Kangning; Wang, Keying; Jiao, Li; Liu, Ruming; Wang, Yunyun; Li, Ming; Wang, Yong; Ma, Lu; Hu, Shuxin; Bian, Xin

doi:10.1038/s41467-024-46919-z

Download PDF

Article
Open access
Published: 20 March 2024

Dissecting the mechanism of atlastin-mediated homotypic membrane fusion at the single-molecule level

Nature Communications volume 15, Article number: 2488 (2024) Cite this article

1008 Accesses
1 Altmetric
Metrics details

Subjects

Abstract

Homotypic membrane fusion of the endoplasmic reticulum (ER) is mediated by dynamin-like GTPase atlastin (ATL). This fundamental process relies on GTP-dependent domain rearrangements in the N-terminal region of ATL (ATL_cyto), including the GTPase domain and three-helix bundle (3HB). However, its conformational dynamics during the GTPase cycle remain elusive. Here, we combine single-molecule FRET imaging and molecular dynamics simulations to address this conundrum. Different from the prevailing model, ATL_cyto can form a loose crossover dimer upon GTP binding, which is tightened by GTP hydrolysis for membrane fusion. Furthermore, the α-helical motif between the 3HB and transmembrane domain, which is embedded in the surface of the lipid bilayer and self-associates in the crossover dimer, is required for ATL function. To recycle the proteins, Pi release, which disassembles the dimer, activates frequent relative movements between the GTPase domain and 3HB, and subsequent GDP dissociation alters the conformational preference of the ATL_cyto monomer for entering the next reaction cycle. Finally, we found that two disease-causing mutations affect human ATL1 activity by destabilizing GTP binding-induced loose crossover dimer formation and the membrane-embedded helix, respectively. These results provide insights into ATL-mediated homotypic membrane fusion and the pathological mechanisms of related disease.

Motion of VAPB molecules reveals ER–mitochondria contact site subdomains

Article Open access 24 January 2024

In situ architecture of the ER–mitochondria encounter structure

Article 10 May 2023

Insights into membrane association of the SMP domain of extended synaptotagmin

Article Open access 17 March 2023

Introduction

In eukaryotes, the endoplasmic reticulum (ER) spreads throughout the cell as a dynamic and continuous network of sheets and tubules with three-way junctions. This unique structure is formed by homotypic fusion between ER membranes^1,2, which is catalyzed by dynamin-like integral membrane protein atlastin (ATL) in metazoans, Sey1p in yeast, and Root Hair Defective 3 (RHD3) in plants^{3,4,5,6,7,8,9,10,11,12}. Because the ER network is critical to cellular function, the specific isoforms of ATLs are associated with a number of cellular processes, including the bone morphogenetic protein signaling pathway^13,14,15,16, lipid droplet size regulation¹⁷, nuclear envelope assembly¹⁸, ER-phagy^19,20, COPII formation²¹, autophagosome formation²², and flavivirus replication^23,24. Mutations in human ATL1 are implicated in hereditary spastic paraplegia (HSP) and hereditary sensory neuropathy type I (HSN1)^25,26. The observation of abnormally long, non-branched ER tubules in mammalian cells expressing ATL1 with disease mutations suggests that the disease is caused by defects in ER homotypic fusion^6,10,27,28.

The membrane fusion events within a cell are classified into two types: heterotypic and homotypic fusion. Heterotypic membrane fusion has been extensively studied and shown to be driven by the folding energy from conformational changes in SNARE complexes, which are disassembled by NSF and SNAPs for recycling after fusion²⁹. However, the detailed processes underlying homotypic membrane fusion mediated by ATL or other related dynamin-like proteins, such as mitochondrial outer membrane fusion protein mitofusin³⁰, remain elusive.

ATL comprises an N-terminal cytosolic region, including a GTPase domain and a middle domain, two transmembrane domains (TMDs), and a C-terminal tail. The crystal structures of the N-terminal cytosolic region of ATL (ATL_cyto) and its ortholog in yeast have been determined for different nucleotide-bound states^{31,32,33,34,35,36}. Two different conformations of human ATL1_cyto have been observed in the presence of GDP^31,33. Though biochemical experiments have demonstrated that GDP-bound ATL1 is a monomer or weak dimer, both of the conformations revealed that a crystallographic dimer formed between the two GTPase domains. Moreover, in one conformation, the middle domains that fold into three-helix bundles (3HBs) point in opposite directions (Fig. 1a, Form 1 extended dimer), whereas, in the other, the linker regions between the GTPase domains and 3HBs cross one another and the 3HBs are in a parallel orientation (Fig. 1a, Form 2 loose crossover dimer). Interestingly, these two conformations partly use the same residues involved in the interface between the GTPase domain and 3HB, though it is an intramolecular interaction in Form 1 dimer and intermolecular interaction in Form 2 crossover dimer³¹. A third conformation of ATL1_cyto dimer (Fig. 1a, Form 3 tight crossover dimer) was obtained in the presence of non-hydrolyzable GTP analog GMppNHp or the presence of GDP/AlF₄⁻³² that mimics the transition state of GTP hydrolysis^37,38. Similar to Form 2, the 3HBs in Form 3 are parallel but much closer to one another. ATL-mediated membrane fusion has been suggested to be driven by the GTP-dependent movement of its 3HB relative to the GTPase domain^{31,33,39,40,41}. However, structural and biochemical studies of human ATLs and Drosophila ATL (dmATL) have led to conflicting membrane fusion mechanisms.

**Fig. 1: Conformations of ATL1_cyto dimers in the presence of GTPγS and GDP/AlF₄⁻ revealed by intramolecular smFRET assays.**

Three different models have been proposed to understand the homotypic membrane fusion process. In one model, ATL molecules bind GTP and generate Form 1 dimers across different membranes, which is followed by a large conformational change from Form 1 to Form 3, triggered by GTP hydrolysis to pull the two membranes towards each other⁴². The second model, however, suggests that GTP binding is enough to induce the tight Form 3 crossover dimer formation for fusion and GTP hydrolysis is only to initiate dimer dissociation for a new round of fusion⁴³. In the third model, GTP hydrolysis is proposed to occur within ATL monomers^32,44, but this model is not supported by biochemical analyses of ATL⁴⁵ and contradicts the results from protein-induced fluorescence enhancement (PIFE) experiments⁴³. In addition to the controversy regarding domain rearrangements upon GTP binding and GTP hydrolysis, the ATL monomer conformations regulated by the sequential release of Pi and GDP⁴⁶ to start a new round of fusion are still poorly understood. These conundrums are due to a lack of an approach to comprehensively analyze the structural dynamics of ATL in each step of the GTPase cycle.

Single-molecule techniques have provided a unique opportunity to dissect the fundamental mechanisms of protein function by directly observing and accurately detecting the behaviors of individual molecules. Here, we develop three different experimental systems to apply intramolecular and intermolecular single-molecule Förster resonance energy transfer (smFRET) imaging to observe the conformational dynamics and pathogenic mutations of individual ATL1_cyto molecules in solution in the absence or presence of different nucleotides. Our data provide evidence that ATL1_cyto can adopt a loose (Form 2-like) crossover dimer conformation upon GTP binding, a surprising observation given that the current models in the literature only predict a Form 1 or Form 3 dimer conformation for GTP-bound ATL1_cyto^42,43. The loose association of 3HBs is tightened in the transition state of GTP hydrolysis. The physiological importance of GTP hydrolysis-induced tightening of the 3HB-3HB interface is supported by the single-molecule analysis of ATL1_cyto with a disease mutation. In addition, based on the results of single-molecule FRET imaging and molecular dynamics (MD) simulations, we found that the α-helical motif between the 3HB and TMD, which is embedded in the surface of the lipid bilayer and self-associates in the ATL dimer, is required for membrane fusion. Finally, we show that the sequential release of Pi and GDP regulates the protein recycling process in a stepwise manner. Pi release, which disassembles the dimer, activates flexibility in the GTPase domain-3HB region with a preference for the Form 1 conformation, and the subsequent GDP dissociation alters the conformational preference of monomeric ATL1_cyto towards the Form 2 conformation to enter the next GTPase cycle. These results provide insights into the mechanisms of dynamin-like protein-mediated homotypic membrane fusion and pathological mechanisms in HSP.

Results

Dimerization of ATL1_cyto induced by GTP binding and hydrolysis

The domain rearrangements of individual ATL1_cyto molecules in solution in the absence or presence of different nucleotides were monitored using smFRET. We generated three constructs of ATL1_cyto (residues 1–447) in which all native cysteines were substituted with alanines and T51, K400, or the T51/K400 pair was mutated to cysteine [termed Cysless ATL1_cyto-T51C (ATL1_cyto-T), Cysless ATL1_cyto-K400C (ATL1_cyto-K), and Cysless ATL1_cyto-T51C/K400C (ATL1_cyto-TK), Supplementary Figs. 1–3]. K400 was previously used as a labeling site for bulk FRET assays⁴⁴. We also added an Avi tag followed by a His tag to the carboxyl terminus of ATL1_cyto-K and a C-terminal His tag to ATL1_cyto-T and ATL1_cyto-TK (Supplementary Fig. 2 and 3). ATL1_cyto-TK had similar GTPase activities as wild-type (WT) ATL1_cyto (Supplementary Fig. 4). As the GTPase activity of ATL relies on intramolecular and intermolecular domain interactions^{31,32,33,39,40}, these results indicate that the mutations we introduced do not affect the GTP-dependent conformational dynamics of the protein.

We first performed photobleaching experiments using LD655-labeled ATL1_cyto-K to test the dimerization state of the protein at the single-molecule level. The C-terminal Avi tag of LD655-ATL1_cyto-K was biotinylated (Supplementary Fig. 5). The modified proteins were mixed with or without 1 mM GDP, GTPγS, GDP/AlF₄⁻ at a high protein concentration of ~2 μM, followed by dilution and immobilization of the protein ( ~ 0.3 nM) in a streptavidin-coated microfluidic chamber (Supplementary Fig. 6a, b). The samples were imaged using total internal reflection fluorescence (TIRF) microscopy. At this low protein concentration, the individual fluorescently labeled particles could be distinguished (Supplementary Fig. 6c), and almost all of them displayed one-step photobleaching in the absence or presence of GDP (Supplementary Fig. 6d, f). With GTPγS or GDP/AlF₄⁻, the fluorescence of most LD655-ATL1_cyto-K molecules was photobleached in two steps (Supplementary Fig. 6c, e, f). These data show that the ATL1_cyto molecules are monomeric in the absence or presence of GDP and form dimers in the presence of GTPγS or GDP/AlF₄⁻ under the conditions of single-molecule experiments, consistent with previous bulk FRET experiments^44,45.

To assess the conformation of a protomer in the ATL1_cyto dimer upon GTP binding or in the transition state of GTP hydrolysis using intramolecular smFRET, we labeled ATL1_cyto-TK with the LD555/LD655 fluorophore pair⁴⁷ (Fig. 1a) and biotinylated the C-terminal Avi-tagged ATL1_cyto-K (Supplementary Fig. 5). These two modified proteins were incubated together with 1 mM GTPγS or GDP/AlF₄⁻ to form ATL1_cyto-TK-ATL1_cyto-K dimer, which was subsequently diluted and immobilized in a streptavidin-coated microfluidic chamber through biotin-streptavidin binding (Fig. 1b). In this system, only the signals of fluorescently labeled ATL1_cyto-TK that dimerized with biotinylated ATL1_cyto-K-Avi in a nucleotide-dependent manner were recorded. The corresponding FRET efficiencies for individual LD555/LD655-labeled ATL1_cyto-TK molecules were recorded at an exposure time of 30 ms, and hundreds of smFRET trajectories exhibiting anti-correlated donor and acceptor fluorescence intensities were extracted (Fig. 1c and Supplementary Fig. 7).

According to the reported crystal structures of ATL1_cyto^31,32,33, labeled residues 51 and 400 in one ATL_cyto-TK molecule are close in Form 1 dimer and distant in Form 2 and Form 3 crossover dimers (Fig. 1a). Analyses of intramolecular smFRET data showed that ATL1_cyto-TK exhibited single low-FRET distributions centered at ~0.28 in the presence of GTPγS or GDP/AlF₄⁻ (Fig. 1d and Supplementary Table 1). In addition, the FRET values over time for almost all ATL1_cyto-TK molecules were stationary (Supplementary Fig. 7). A possible conformation corresponding to this low-FRET state is the Form 3 tight crossover dimer (Supplementary Tables 1 and 2), which was obtained in the GMppNHp- or GDP/AlF₄⁻-bound condition³². Notably, this low-FRET state is also likely to represent a conformation similar to Form 2 crossover dimer in which the 3HBs loosely associate with one another, as the intramolecular smFRET imaging may not be able to recognize the difference between the Form 2 and Form 3 conformations. Nevertheless, we barely observed a higher FRET state corresponding to the Form 1 conformation.

Next, we performed intermolecular smFRET imaging to confirm the crossover dimer conformation in the presence of GTPγS or GDP/AlF₄⁻. A mixture of LD555-ATL1_cyto-T and biotinylated LD655-ATL1_cyto-K with GTPγS or GDP/AlF₄⁻ was diluted and added to a streptavidin-coated microfluidic chamber (Supplementary Fig. 8a, b). The immobilized individual nucleotide-dependent ATL1_cyto-T-ATL1_cyto-K dimers were imaged over time by TIRF microscopy (Supplementary Fig. 8b). Consistent with the short distance between the two labeled sites in the Form 2 and Form 3 crossover dimers (Supplementary Fig. 8a and Table 2), the labeled ATL1_cyto-T-ATL1_cyto-K dimer only had a single high-FRET peak centered at ~0.80 in the presence of GTPγS or GDP/AlF₄⁻ (Supplementary Fig. 8c, d and Table 3). Thus, instead of generating a Form 1 extended dimer, GTP binding to ATL1_cyto induces crossover dimer formation.

GTP binding generates loose ATL1_cyto crossover dimer, which is tightened by GTP hydrolysis

To further explore the conformational preferences of ATL1_cyto dimer in the presence of GTPγS and GDP/AlF₄⁻, we mixed LD555-ATL1_cyto-K and biotinylated LD655-ATL1_cyto-K with 1 mM GTPγS or GDP/AlF₄⁻ and monitored the corresponding FRET values for the immobilized individual nucleotide-dependent ATL1_cyto-K dimers over time (Fig. 2a, b). Surprisingly, in the presence of GDP/AlF₄⁻, we observed a high-FRET state corresponding to Form 3 crossover dimer conformation, whereas GTPγS-loaded ATL1_cyto-K exhibited both a high-FRET distribution centered at ~0.66 and low-FRET peak centered at ~0.28 (Fig. 2c–e, Supplementary Fig. 9 and Table 4). Transitions between low- and high-FRET states were rare (Supplementary Fig. 9), but it is possible that the ATL1_cyto dimers undergo a low-frequency transition between low- and high-FRET states on a sub-millisecond timescale in the presence of GTPγS, and this fast transition could not be efficiently captured by our smFRET setup (30 ms per frame). When the ATL1_cyto molecules dimerize, the distance between labeled position 400 in two 3HBs is shortest in Form 3 (Fig. 2a and Supplementary Table 2). Given that the FRET efficiency negatively correlates with the distance between the donor and acceptor fluorophores and no Form 1 extended dimer was formed in the presence of GTPγS or GDP/AlF₄⁻ (Fig. 1a–d and Supplementary Fig. 8), these data demonstrate that the GTP-bound ATL1_cyto crossover dimers adopt two conformations corresponding to the loose crossover dimer and Form 3 tight crossover dimer, respectively, and the tight association of 3HBs is completely formed in the transition state of GTP hydrolysis.

**Fig. 2: Conformations of ATL1_cyto dimers in the presence of GTPγS and GDP/AlF₄⁻ revealed by intermolecular smFRET assays.**

Disease mutation S398Y in ATL1 affects the formation of loose crossover dimer

HSP has been pathogenically linked to defects in ER homotypic fusion^6,10,27,28. Most ATL1 mutations causing HSP localize at the intermolecular and intramolecular domain interfaces and exhibit weakened nucleotide binding and/or reduced propensity for dimerization³³. However, some mutations cause disease despite exhibiting WT levels of nucleotide-dependent dimerization when examined by commonly used biochemical and biophysical methods³³. One such mutation is S398Y, which localizes at the center of the second α-helix of 3HB (Fig. 3a and Supplementary Fig. 1).

**Fig. 3: Loose crossover dimer formation and the linker region between the 3HB and TMD play roles in ATL-mediated membrane fusion.**

To further understand the pathological mechanisms in HSP, we tested the activity of ATL1 with a disease mutation in vivo. Yeast cells lacking Sey1p and another ER-shaping protein, Yop1p (DKO), had a significant reduction in the intricate tubular ER network at the periphery of the cell due to a defect in ER membrane fusion, and the expression of WT ATL1 can rescue this abnormal cortical ER morphology (Supplementary Fig. 10)^3,6,48. However, in the DKO cells expressing disease mutant ATL1 S398Y, restoration of the tubular ER network to the WT levels was not observed (Supplementary Fig. 10). Next, we applied intermolecular smFRET experiments to the mutant ATL1_cyto-K carrying S398Y (Fig. 2b). In the presence of GDP/AlF₄⁻, the FRET trajectories and distributions of ATL1_cyto-K-S398Y were reminiscent of ATL1_cyto-K (Fig. 3b, c, Supplementary Fig. 11 and Table 4). However, the occupancy of the low-FRET state observed with GTPγS was decreased in the case of ATL1_cyto-K-S398Y (Fig. 3b–d, Supplementary Fig. 11 and Table 4). This change revealed that mutating Ser to Tyr at position 398 of ATL1 destabilizes the Form 2-like crossover dimer in the presence of GTPγS with differences in the relative free energy of ~0.8 k_BT (Fig. 3e). Analyses of this disease mutation imply that the formation of loose crossover dimer upon GTP binding is physiologically important.

The helical motif between the 3HB and TMD is essential for membrane fusion

Another HSP-causing ATL1 mutation, N440T, localizes to the linker region between the 3HB and TMDs (Fig. 3f and Supplementary Fig. 1). When we used intermolecular smFRET to examine ATL1_cyto-K carrying N440T in the presence of GTPγS or GDP/AlF₄⁻ (Fig. 2b), the FRET trajectories and distributions of ATL1_cyto-K-N440T were similar to ATL1_cyto-K (Fig. 3d, g, h, Supplementary Fig. 12, and Table 4). These results indicate that this disease mutation does not affect nucleotide-dependent dimerization of ATL. To gain insights into the mechanisms underlying disease pathogenesis, we inspected the reported crystal structures of GMppNHp-bound WT ATL1_cyto and ATL1_cyto-N440T³². Interestingly, a helix formed in the linker region of WT ATL1_cyto, but it became unstructured when mutating N440 to Thr (Fig. 3f).

To confirm the critical role of the helix in the linker region in the function of ATL, we introduced a proline substitution at position 443 of ATL1 and corresponding position 418 of dmATL, which is in the middle of the linker region. This mutation is expected to disrupt the structure of the α-helix (Fig. 3f and Supplementary Fig. 1). In membrane fusion assays^9,31, we reconstituted full-length dmATL into donor and acceptor proteoliposomes. The fusion of fluorescently labeled donor proteoliposomes with unlabeled acceptor proteoliposomes resulted in dilution and dequenching of the FRET pair, NBD-PE and Rhodamine-PE, showing increased NBD fluorescence. Though robust membrane fusion was achieved by WT dmATL in the presence of GTP, dmATL-H418P had significantly lower fusion activity (Fig. 3i). In DKO yeast cells, the N440T and H443P mutants did not have fenestrated ER restored to WT levels (Supplementary Fig. 10). In agreement with the results of the disease mutation experiments, this fusion defect was not caused by an impairment in crossover dimer formation, as the GTPγS- and GDP/AlF₄⁻-dependent intermolecular smFRET trajectories and distributions of ATL1_cyto-K-H443P were highly consistent with ATL1_cyto-K and ATL1_cyto-K-N440T (Fig. 3d, j, k, Supplementary Fig. 13 and Table 4).

To investigate the role of the helical motif linking the 3HB and TMD in ATL function, we studied the location of the helical motif in the membrane. We modeled the full-length ATL1 protein, including the TMDs and C-terminal tail, in the Apo monomer (Fig. 4a) and GDP/Pi-bound Form 3 tight crossover dimer conformations (Fig. 4b) using AlphaFold2 prediction with the AF2complex algorithm⁴⁹. The Form 3 conformation represents the post-fusion state of ATL1 dimer (see Discussion). We then used the OPM web server⁵⁰ to predict how these models of ATL1 interact with the membrane. The prediction suggested that the helical motif is located at the surface of the outer leaflet of the membrane in both forms (Fig. 4a, b). We also performed all-atom MD simulations of ATL1 monomer and Form 3 crossover dimer to determine how they behave in an ER-like membrane environment (Supplementary Table 5). The simulations showed that the helical motif inserted slightly deeper into the membrane (Fig. 4c, d). In particular, the hydrophobic residues oriented their side chains towards the nonpolar phase of the lipid bilayer, and the polar residues oriented their side chains towards the solvent (Fig. 4b). This orientation may help optimize the energetics of the ATL1-membrane interaction.

**Fig. 4: Interaction and membrane location of the helical motifs linking the 3HBs to the TMDs of ATL1 molecules in molecular dynamics simulations.**

Interestingly, in the Form 3 tight crossover dimer, strong interactions exist between the helical motifs from each protomer over the 1-μs MD simulation (Fig. 4b, e, Supplementary Fig. 14a and Supplementary Movie 1), resulting in extensive intermolecular contacts (Fig. 4f). Notably, we did not observe any interaction between the TM domains in the MD simulations, but this may depend on the accuracy of the AF2 prediction, which needs further investigation. We also performed all-atom MD simulations of the ATL1 dimer in Form 2 conformation and observed that the helical motifs in the dimer approached and interacted with each other during the 1-μs simulation (Supplementary Fig. 14b). However, the 3HBs did not move closer within this time scale, which confirms that the assembly of the 3HBs and the interaction between the helical motifs are independent, and this helical motif has strong potential to self-associate. To validate the experimental observations, additional simulations for the GDP/Pi-bound Form 3 dimer were initiated using the crystal structure of the ATL1_cyto-N440T³², in which the helical motif is unstructured. We observed a significant decrease in interactions between the helical motifs (Fig. 4g, h and Supplementary Fig. 14c).

Therefore, we propose that the membrane-embedded helical motifs between 3HBs and TMDs in the ATL crossover dimer interact with each other at the initial stage of lipid mixing to allow the energy provided during the tight association of 3HBs to be transferred all the way into the membrane to complete membrane fusion.

Conformational dynamics of ATL1_cyto regulated by Pi and GDP release

Finally, we used intramolecular smFRET imaging to analyze the behaviors of ATL1_cyto-TK with or without GDP (Fig. 1a), which represent the sequential release of Pi and GDP^27,44,45 for dimer disassembly and protein recycling^33,40,51. The ATL1_cyto-TK monomers labeled with LD555/LD655 in the absence or presence of 1 mM GDP were immobilized in a streptavidin-coated microfluidic chamber through an interaction between its C-terminal His tag and biotinylated anti-His antibody (Fig. 5a). Interestingly, the FRET distributions of individual ATL1_cyto-TK in the absence or presence of GDP were broad with an ultra-low-FRET peak centered at ~0.18, medium-FRET peak centered at ~0.41, high-FRET peak centered at ~0.63, and ultra-high-FRET peak centered at ~0.83 (Fig. 5b, c and Supplementary Table 6).

**Fig. 5: Conformational dynamics of ATL1_cyto-TK monomers in the absence or presence of GDP revealed by intramolecular smFRET assays.**

To probe the kinetics of the intramolecular conformational transitions, we fitted the individual intramolecular smFRET trajectories of the ATL1_cyto-TK monomers in the absence or presence of GDP using a four-state Hidden Markov Model process (Fig. 5b). The FRET efficiencies before and after each transition were plotted as a transition density plot (TDP) to indicate the frequencies of conformational transitions (Fig. 5d). In contrast to the GTPγS- and GDP/AlF₄⁻-bound ATL1_cyto dimers, which have a relatively stable GTPase domain-3HB region (Fig. 1c and Supplementary Fig. 7), frequent relative movements between the GTPase domain and 3HB of monomeric ATL1_cyto were observed in Apo and GDP-bound conditions, reflecting a weakened interaction between the GTPase domain and 3HB (Fig. 5d). In the presence of GDP, transitions between medium- and high-FRET states and between high- and ultra-high-FRET states were predominantly observed, whereas the Apo form had frequent transitions between medium- and ultra-low-FRET states (Fig. 5d).

As expected, the conformations of Apo and GDP-bound ATL1_cyto-TK were not identical. The ATL1_cyto-TK monomer in the presence of GDP preferred the ultra-high-FRET Form 1 conformation (Fig. 5c, Supplementary Tables 2 and 6), whereas nucleotide-free ATL1_cyto-TK had a dominant occupancy in the ultra-low-FRET state (Fig. 5c), which represents the Form 2 or Form 3 conformation of ATL1_cyto-TK as a monomer (Supplementary Tables 2 and 6). This change was the result of a dramatically decreased transition rate from the ultra-low-FRET state ( ~ 0.18) toward the medium-FRET state ( ~ 0.41) (Fig. 5e and Supplementary Fig. 15). The energy barriers for the transition towards a high-FRET state ( ~ 0.63) and subsequent ultra-high-FRET state ( ~ 0.83) were also elevated. These analyses result in a qualitative model of a GDP release-altered conformational preference of ATL1_cyto (Fig. 5f).

Collectively, these results demonstrate that the GTPase domain-3HB region of ATL becomes flexible after GTP hydrolysis and the sequential release of Pi and GDP generates two different conformational preferences of the ATL monomer in a stepwise manner to recycle the protein for a new round of fusion.

Discussion

In this study, we measured smFRET efficiency to monitor the conformations of ATL1_cyto and its pathogenic mutations in the absence or presence of different nucleotides, despite some discrepancies between the theoretical dye-dye distances calculated by the FRETpredict tool (Supplementary table 2)^52,53 and the experimental dye-dye distances based on the smFRET efficiency (see Methods for calculation, Supplementary tables 1, 3, 4, and 6), which could be attributed to uncertainties in the Förster radius R₀ and limited MD sampling. Our results partly confirm the prior mechanistic model for ATL function in which the loading of GTP induces the crossover dimerization of ATL⁴³. However, our proposed model in this study also differs from previous studies. First, we show that the energy input from GTP binding is not enough to complete the tightest (Form 3) crossover dimer formation, as a proportion of ATL1_cyto molecules in the presence of GTPγS interact as in the Form 2-like conformation, where their 3HBs loosely associate with one another (Fig. 6, Step I). Given that only the N-terminal cytosolic region of ATL1 in solution was used here, more membrane-anchored full-length ATL molecules are assumed to have a Form 2-like crossover dimer conformation upon GTP binding due to the electrostatic repulsive force between the two apposed membranes, which increases the difficulty for tight association of 3HBs. Consistent with this proposed model, GTP binding to ATL has been reported to be insufficient to induce the formation of membrane tethers^45,54. Another important finding was that, although GTP hydrolysis did not trigger a large conformational change in ATL1_cyto (from Form 1 to Form 3) as suggested previously⁴², it did tighten the 3HB-3HB interface, allowing multiple ATL molecules on each membrane to cooperate to promote tethering and fusion between two bilayers (Fig. 6, Step II and II’).

**Fig. 6: Model of a successful membrane fusion event mediated by ATL.**

After the membranes are fused by ATL, Pi and GDP are released in a sequential manner to disassemble ATL dimers for new rounds of fusion. Another unanswered question prior to this study was the domain rearrangements of ATL coupled with this process. On the basis of our smFRET results, the sequential release of Pi and GDP regulates the conformational changes in ATL in a stepwise manner for protein recycling. Unlike the membrane fission-specific dynamin superfamily GTPases, such as MxA, in which the flexibility in the GTPase domain-BSE region is released upon GTP binding⁵⁵, the frequent relative movement between the GTPase domain and 3HB of ATL was activated by Pi release with a preference for the Form 1 conformation (Fig. 6, Step III). The activation of flexibility in the GTPase domain-3HB region, which reflects the weakened interaction between the GTPase domain of one protomer and 3HB of the other promotor, is expected to initiate disassembly of the ATL dimer, and the relatively small interfacial area in the Form 1 dimer may facilitate this process. We also found that the GTPase domain-3HB region of ATL1 remains flexible in the Apo state, but GDP release leads to alternation of the conformational preference towards a Form 2 or Form 3 conformation (Fig. 6, Step IV), which may lower the difficulty of GTP binding-induced Form 2-like crossover dimer formation between ATLs sitting in apposed membranes (Fig. 6, Step I).

Single-molecule approaches also offer a solution to better understanding the pathogenetic mechanisms that are sometimes hidden. The presence of Ser at position 398, where the corresponding residues in human ATL2/3 and other species are Ala, is a unique feature of human ATL1 (Supplementary Fig. 1). Its mutation to Tyr causes HSP but has no detectable defect in nucleotide-dependent dimerization when examined by commonly used biochemical and biophysical methods³³, though the corresponding mutation in dmATL results in impaired dimer formation²⁷. Via a combination of smFRET analyses and imaging experiments using yeast cells, our results indicated that ATL1 molecules carrying S398Y have a defect in the generation of Form 2-like dimer upon GTP binding and are not able to efficiently mediate ER membrane fusion in vivo, implying the physiological importance of loose crossover dimer formation (Fig. 6, Step I). Based on a previous report that most attempts to induce fusion through trans-dimerization of ATL molecules are futile^45,54 and that a successful membrane fusion event requires the cooperation of ATL oligomers formed through their TMDs in each membrane⁴⁸, we propose that GTP binding generates several ATL1 molecules with similar loose crossover dimer conformations, allowing them to undergo conformational changes synchronously triggered by GTP hydrolysis for fusion (Fig. 6, Step II and II’). However, if the tight crossover dimers of ATL1 are generated directly from ATL1 molecules in the Apo state with a variety of conformations due to the flexibility in the GTPase domain-3HB region, the driving force for efficient membrane fusion may not be enough.

Another question raised by the finding of GTP-dependent conformational changes in ATL was how the energy released from tight crossover dimer formation is transferred to the membrane. The role of the linker region between the 3HB and TMD in this process was inspired by the crystal structures of ATL1_cyto carrying disease mutation N440T, in which the helix of the linker region is disrupted³². Our single-molecule FRET imaging assays, combined with MD simulations, further showed that this helical linker region is buried in the lipid bilayer and self-associates in the ATL dimer but is not involved in the nucleotide-dependent assembly of 3HBs. We speculate that the helix formed in the linker region is structurally linked to the energy supply for membrane fusion (Fig. 6, Step II’). A similar mechanism of SNARE-mediated heterotypic membrane fusion has been reported, in which the helix of the linker domain between the SNARE motif and TMD not only provides additional energy^56,57,58, but also interacts with membranes to facilitate membrane fusion⁵⁹. Interestingly, ATL-mediated homotypic membrane fusion also requires a membrane-destabilizing motif, which is the C-terminal amphipathic helix of ATL^48,51,60.

In summary, our results suggest a comprehensive model for ATL-mediated membrane fusion (Fig. 6) and contribute to understanding the pathogenesis of HSP. The overall mechanisms may be shared by other dynamin-like proteins that act as homotypic fusogens, and the smFRET experiments and MD simulations we performed here can be adapted to further explore their differences.

Methods

Reagents

Reagents were obtained from the following sources: Ni Sepharose 6 Fast Flow (Cytiva, 17531802), Glutathione Sepharose 4 Fast Flow (Cytiva, 17513201), tris(2-carboxyethyl)phosphine (Thermo, 75259), EDTA-free protease inhibitor cocktail (Roche, 11873580001), PreScission Protease (Absin, abs01243), Anapoe-X-100 (Anatrace, 9002-93-1), ATP (Roche, 11140965001), MgOAc (Sigma-Aldrich, M0631), L-glutamic acid potassium (Sigma-Aldrich, G1501), biotinylated anti-His antibody (Bioss Antibodies, bs-0287R-bio), streptavidin (Sangon Biotech, A610492), LD555-MAL (Lumidyne, 04), LD655-MAL (Lumidyne, 10), GDP (Sigma-Aldrich, G7127), GTPγS (Roche, 10220647001), GMppNHp (Sigma-Aldrich, G0635), mPEG-SVA-5000 (Laysan Bio, 170-106), Biotin-PEG-SVA-5000 (Laysan Bio, 170-124), benzoic acid (Sigma-Aldrich, 242381), protocatechuate 3,4-dioxygenase (Sigma-Aldrich, P8279), and n-dodecyl-β-D-maltoside (Avanti Polar Lipids, 850520). All lipids were obtained from Avanti Polar Lipids: 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC, 850457); 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (POPS, 840034); 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD-PE, 810144); and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (Rhodamine-PE, 810150).

Strains and plasmids

Yeast strains used in this study were BY4741 (MATa his3Δ1 leu2Δ2met15Δ ura3Δ) and JHY4 (BY4741 sey1Δ::kanMX4 yop1Δ::HIS3MX6)⁶. The region encoding the N-terminal cytosolic region of human ATL1 (a.a. 1–447) was cloned into the pET-28a vector using the NcoI and NotI sites. The region encoding the full-length human ATL1 (a.a. 1-558) was cloned into the pRS316 vector along with the endogenous promoter and terminator of SEY1 using the NotI and XhoI sites. The region encoding the full-length dmATL (a.a. 1-541) was cloned into the pGEX-6p-1 vector using the BamHI and XhoI sites. Point mutations, HA tags and Avi tags were introduced by site-directed mutagenesis.

Protein expression and purification

ATL1_cyto and its mutants were expressed in Rosetta (DE3) (Biomed) E. coli cells and purified as described previously³¹. Briefly, cells were harvested and lysed by sonication in purification buffer (25 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM MgCl₂, 1X complete EDTA-free protease inhibitor cocktail, 0.5 mM TCEP). The lysates were clarified by centrifugation at 200,000 × g for 1 h at 4 °C using a TYPE 45 Ti rotor. The protein was isolated using an Ni-NTA column and further purified by size exclusion chromatography using Superdex 200 Increase 10/300 GL in purification buffer without protease inhibitor cocktail and TCEP. Fractions containing ATL1_cyto were pooled and concentrated.

Full-length dmATL and its mutant were expressed in Rosetta (DE3) (Biomed) E. coli cells as described previously³¹. Briefly, cells were harvested by centrifugation and lysed by sonication in purification buffer containing 10% (v/v) glycerol. Membranes were pelleted by centrifugation at 200,000 × g for 1 h at 4 °C using an SW 41 rotor (Beckman Coulter), followed by solubilization in 2.5% (w/v) Anapoe X-100 and 10% (v/v) glycerol in purification buffer for 1 h at 4 °C. The extract was clarified by centrifugation at 200,000 × g for 1 h at 4 °C using an SW 41 rotor and the protein purified from the supernatant by glutathione Sepharose beads. After washing with 0.1% (w/v) Anapoe X-100 and 10% (v/v) glycerol in purification buffer, the GST tag was cleaved by PreScission Protease on a column overnight at 4 °C. The untagged protein was washed with 0.1% (w/v) Anapoe X-100 and 10% (v/v) glycerol in purification buffer.

Protein biotinylation

For biotinylation, a mixture of 100 μL 100 mM ATP, 100 μL 100 mM MgOAc, 100 μL 500 μM d-biotin, 10 μL 100 μM purified BirA biotin ligase, and 700 μL 50 μM protein was incubated in biotinylation buffer (25 mM HEPES, pH 7.4, 0.5 mM TCEP, 200 mM L-glutamic acid potassium) overnight at 4 °C. Biotinylated protein was incubated with 15 μM streptavidin for 20 min on ice, followed by SDS-PAGE analysis to measure the efficiency of biotinylation. Excess free d-biotin was removed by 10 K MWCO centrifugal filters (Amicon).

Fluorescence microscopy

Yeast cells were grown to an optical density at 600 nm (OD600) of 1.2-1.5, pelleted, and resuspended in 10% glycerol solution. Resuspended cells were imaged using a super-resolution microscope (DeltaVision OMX SR, Cytiva) with a 60X oil-immersion objective. Sec63p-EGFP was excited by a 488 nm laser. All images were captured by a scientific Complementary Metal Oxide Semiconductor (sCMOS) camera.

Fluorophore labeling

For intramolecular smFRET assays, ATL1_cyto-TK was incubated with LD555 and LD655 at a ratio of 1:1.2:1.2 in labeling buffer (25 mM HEPES, pH 7.4, 150 mM KCl, 5 mM MgCl₂) for 5 h at 4 °C. For intermolecular smFRET experiments, ATL1_cyto-K was incubated with LD555 or biotinylated ATL1_cyto-K with LD655, each at a ratio of 1:1.2 in labeling buffer for 5 h at 4 °C. Excess free fluorophores were removed by Zeba 7 K MWCO spin desalting columns (Thermo Scientific). TCEP was added to the dye-labeled proteins at a final concentration of 0.5 mM.

Single-molecule FRET imaging and data acquisition

The slide surface of the chamber used in smFRET experiments was modified by the mixture of mPEG-SVA-5000 and Biotin-PEG-SVA-5000 at a ratio of 100:1 in modification buffer (0.1 M NaHCO₃ and 0.6 M K₂SO₄) overnight. Proteins ( ~ 2 μM) were incubated with 1 mM GDP, GTPγS, or GDP/AlF₄⁻ on ice for 1 h, followed by dilution in labeling buffer containing 10 μg/mL streptavidin and 1 mM nucleotide in the chamber to immobilize the dimers ( ~ 0.3 nM). For the strategy shown in Fig. 5a, 10 μg/mL streptavidin, 10 μg/mL biotinylated anti-His antibody and 1 mM nucleotide were added to the chamber with dye-labeled ATL1_cyto-TK and incubated for 5 min to immobilize the monomers ( ~ 1 nM). Buffer containing 2.5 mM benzoic acid, 2 mM protocatechuate 3,4-dioxygenase (PCD) and 1 mM nucleotide was infused into the chamber for oxygen-scavenging before measuring fluorescence measurements.

The imaging was performed using a TIRF microscope (Nikon, Ti2) with a single-frequency 532 nm laser as the excitation source for smFRET experiments and 640 nm laser as the excitation source for photo-bleaching experiments. The microscope was equipped with a high numerical aperture oil immersion objective (Nikon, 100X, N.A. 1.49, oil immersion) to provide high-resolution imaging. To distinguish between the fluorescence signals of the donor and acceptor channels, a 630 nm dichroic mirror was employed. The fluorescence signals were recorded at a frame interval of 30 ms for smFRET experiments and 60 ms for photo-bleaching experiments using an EMCCD camera (Andor, IX897). The raw data were saved in 16-bit TIFF format and further processed for analysis.

smFRET data analysis

The FRET trajectories were extracted from the recorded movies, and the FRET efficiency was calculated using the equation FRET = I_LD655/(γI_LD555 + I_LD655), where I_LD555 and I_LD655 represent the LD555 and LD655 fluorescence intensities, respectively, and γ is the correlation coefficient correcting the difference between the quantum yields and the detection efficiencies of LD555 and LD655. In detail, the parameter γ is equal to the ratio of the reduction in acceptor intensity (F_A) to the increase in donor intensity (F_D) when acceptor was photobleached (γ = F_A/F_D). We counted the proportions of hundreds of molecules as a normal distribution in each experiment and regarded the central value of the normal distribution as the final γ value (Supplementary Table 7). The FRET efficiency data from all trajectories were compiled into histograms. These histograms were then fitted into the sum of GaussAmp distributions. The occupancy of each state was estimated by the area under each GaussAmp curve and further used to evaluate the free energy difference between state i and state j using the equation ΔG = -k_BTln(A_i/A_j)^61,62, where A_i and A_j are the occupancies of state i and state j, respectively, k_B represents the Boltzmann constant, and T is the thermodynamic temperature. We employed a MATLAB script based on the Hidden Markov Model algorithm^63,64 to fit the FRET efficiency trajectories to analyze the dynamics of intramolecular conformational states. The TDPs were derived from the conformational transitions of four-state Hidden Markov Model fitting. The experimental dye-dye distance (R) based on smFRET efficiency (E) was calculated by the following equation: E = 1/[1 + (R/R₀)⁶], in which the estimated Förster radius R₀ was 62 Å⁴⁷.

Lipid-mixing assays

The lipid-mixing assays were performed as described previously³¹. Briefly, lipid mixtures of donor liposomes (82:15:1.5:1.5 mole percent of POPC:POPS:NBD-PE:Rhodamine-PE) and acceptor liposomes (85:15 mole percent of POPC:POPS) were dried with N₂ to form a film. Purification buffer containing 10% (v/v) glycerol was added to the tube to rehydrate the lipid film, and the suspension underwent 10 freeze-thaw cycles in liquid nitrogen and 37 °C water bath. The liposomes were extruded 21 times through polycarbonate filters with a pore size of 100 nm (Avanti Polar Lipids). The mixture of purified dmATL with liposomes at a protein to lipid ratio of 1:1000 was incubated at 4 °C for 1 h with gentle shaking. The detergent was removed by bio-beads SM-2 resin (Bio-Rad) to reconstitute the proteins into liposomes.

The in vitro lipid-mixing assays were performed in 100 μL volumes. The final lipid concentration was 0.5 mM with donor and acceptor proteoliposomes containing dmATL added at a 1:1 ratio. Reactions were initiated by the addition of 1 mM GTP to the mixtures in a 96-well plate (Corning). The fluorescence intensity of NBD was monitored at 37 °C using an excitation wavelength of 460 nm and emission wavelength of 538 nm every 1 min over 30 min in a Cytation 5 Imaging Reader (BioTek). All data were corrected by subtracting the baseline values obtained at 0 min. The data were expressed as a percentage of the total NBD fluorescence determined by adding 10 μL of 2.5% n-dodecyl-β-D-maltoside to the reactions after 30 min.

GTPase activity assays

The GTPase assays were performed using the Enzchek Phosphate Assay Kit (Invitrogen) in 100 μL volumes with 5 μL of 20X reaction buffer (1 M Tris, pH 7.5, 20 mM MgCl₂, 2 mM sodium azide), 200 mM 2-amino-6-mercapto-7-methylpurine riboside, 1 unit/mL purine nucleoside phosphorylase (PNP), and ATL1_cyto or its mutants. Reactions were initiated by the addition of 1 mM GTP in a 96-well plate (Corning). The absorbance at 360 nm was measured at 37 °C every 1 min over 30 min in a Cytation 5 Imaging Reader (BioTek). Data were expressed as absolute absorbance values subtracting the the absorbance at 0 min.

Molecular dynamics simulations

We modeled the full-length GDP/Pi-bound ATL1 dimer in Form 3 using AlphaFold2 prediction with the AF2complex algorithm⁴⁹. Based on the AF2 prediction and crystal structure of ATL1 bound to GDP (PDB ID: 3Q5E), we further modeled the full-length ATL1 monomer. Next, we inserted these full-length models into a lipid bilayer composed of 80% POPC and 20% POPE, and surrounded them with a water box containing 0.15 M NaCl. The box sizes were 13.5 nm × 13.5 nm × 13.5 nm for ATL1 monomer and 12.1 nm × 12.1 nm × 16.9 nm for Form 3 ATL1 dimer, resulting in ~232,000 and ~250,000 total atoms, respectively. We used the OPM webserver⁵⁰ to align the TM region with the lipid bilayer and the CHARMM-GUI webserver⁶⁵ to build the systems. We then minimized the energy using the steepest descent algorithm and equilibrated the systems in six steps with gradually reduced position constraints. We ran production simulations in NPT conditions using the CHARMM36m force field⁶⁶ for the protein, lipid, GDP, and Pi molecules, and TIP3P for the water molecules. We maintained the temperature at 310 K with a Nosé–Hoover thermostat (1 ps coupling constant) and the pressure at 1.0 bar with a Parrinello–Rahman barostat (5 ps time coupling constant). We used a cut-off of 1.2 nm for van der Waals and short-range electrostatic interactions, with a switch function at 1.0 nm. We used the particle mesh Ewald method with a 0.12 nm mesh spacing for long-range electrostatic interactions. The same methodology was employed to construct models for the N440T mutant of the GDP/Pi-bound Form 3 ATL1 dimer and GDP-bound Form 2 ATL1 dimer. Notably, the N440T mutant model was based on the crystal structure (PDB 4IDQ). We performed two independent simulations, each lasting 1000 ns, for all models using Gromacs 2021.5⁶⁷ on GPUs. Trajectories were analyzed by PLUMED⁶⁸, mdanalysis⁶⁹, and mdtraj⁷⁰. Theoretical FRET fluorophore distances were calculated from the MD trajectories using FRETpredict⁵² and subsequently fitted with a normal distribution using scipy⁵³.

Data availability

All underlying data are available from the corresponding authors upon request. The source data underlying Figs. 1d, 2d, 2e, 3c, d, h, i, k, 5c, e and Supplementary Figs. 4, 6f, 8d, 10b, 15 are provided as a Source Data file. The crystal structures of human ATL1_cyto and its mutants with different nucleotides have previously been deposited in PDB: 3QOF, 3QNU, 4IDN, 4IDP, 3Q5E, 4IDQ. The molecular dynamics simulations data are available at https://github.com/yongwangCPH/papers/tree/main/2024/ATL1. Source data are provided with this paper.

Code availability

The custom MATLAB code is available at https://github.com/yangchenguang-1994/HMM-FRET.

References

Friedman, J. R. & Voeltz, G. K. The ER in 3D: a multifunctional dynamic membrane network. Trends Cell Biol. 21, 709–717 (2011).
Article CAS PubMed PubMed Central Google Scholar
Hu, J., Prinz, W. A. & Rapoport, T. A. Weaving the web of ER tubules. Cell 147, 1226–1231 (2011).
Article CAS PubMed PubMed Central Google Scholar
Anwar, K. et al. The dynamin-like GTPase Sey1p mediates homotypic ER fusion in S. cerevisiae. J. Cell Biol. 197, 209–217 (2012).
Article CAS PubMed PubMed Central Google Scholar
Chen, J., Stefano, G., Brandizzi, F. & Zheng, H. Arabidopsis RHD3 mediates the generation of the tubular ER network and is required for Golgi distribution and motility in plant cells. J. Cell Sci. 124, 2241–2252 (2011).
Article CAS PubMed Google Scholar
Crosby, D. et al. Reconstitution of human atlastin fusion activity reveals autoinhibition by the C terminus. J. Cell Biol. 221, e202107070 (2022).
Article CAS PubMed Google Scholar
Hu, J. et al. A class of dynamin-like GTPases involved in the generation of the tubular ER network. Cell 138, 549–561 (2009).
Article CAS PubMed PubMed Central Google Scholar
Jang, E. et al. Human atlastins are sufficient to drive the fusion of liposomes with a physiological lipid composition. J. Cell Biol. 222, e202109090 (2023).
Article CAS PubMed PubMed Central Google Scholar
Liu, X. et al. Atlastin-1 regulates morphology and function of endoplasmic reticulum in dendrites. Nat. Commun. 10, 568 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Orso, G. et al. Homotypic fusion of ER membranes requires the dynamin-like GTPase atlastin. Nature 460, 978–983 (2009).
Article ADS CAS PubMed Google Scholar
Rismanchi, N., Soderblom, C., Stadler, J., Zhu, P. P. & Blackstone, C. Atlastin GTPases are required for Golgi apparatus and ER morphogenesis. Hum. Mol. Genet 17, 1591–1604 (2008).
Article CAS PubMed PubMed Central Google Scholar
Zhang, M. et al. ROOT HAIR DEFECTIVE3 family of dynamin-like GTPases mediates homotypic endoplasmic reticulum fusion and is essential for Arabidopsis development. Plant Physiol. 163, 713–720 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Bryce, S. et al. Human atlastin-3 is a constitutive ER membrane fusion catalyst. J. Cell Biol. 222, e202211021 (2023).
Article CAS PubMed PubMed Central Google Scholar
Fassier, C. et al. Zebrafish atlastin controls motility and spinal motor axon architecture via inhibition of the BMP pathway. Nat. Neurosci. 13, 1380–1387 (2010).
Article CAS PubMed Google Scholar
Summerville, J. B. et al. The effects of ER morphology on synaptic structure and function in Drosophila melanogaster. J. Cell Sci. 129, 1635–1648 (2016).
CAS PubMed PubMed Central Google Scholar
Zhao, G. et al. Mammalian knock out cells reveal prominent roles for atlastin GTPases in ER network morphology. Exp. Cell Res 349, 32–44 (2016).
Article CAS PubMed Google Scholar
Zhao, J. & Hedera, P. Hereditary spastic paraplegia-causing mutations in atlastin-1 interfere with BMPRII trafficking. Mol. Cell Neurosci. 52, 87–96 (2013).
Article CAS PubMed Google Scholar
Klemm, R. W. et al. A conserved role for atlastin GTPases in regulating lipid droplet size. Cell Rep. 3, 1465–1475 (2013).
Article CAS PubMed PubMed Central Google Scholar
Wang, S., Romano, F. B., Field, C. M., Mitchison, T. J. & Rapoport, T. A. Multiple mechanisms determine ER network morphology during the cell cycle in Xenopus egg extracts. J. Cell Biol. 203, 801–814 (2013).
Article CAS PubMed PubMed Central Google Scholar
Chen, Q. et al. ATL3 Is a Tubular ER-Phagy Receptor for GABARAP-Mediated Selective Autophagy. Curr. Biol. 29, 846–855.e846 (2019).
Article CAS PubMed Google Scholar
Liang, J. R., Lingeman, E., Ahmed, S. & Corn, J. E. Correction: Atlastins remodel the endoplasmic reticulum for selective autophagy. J. Cell Biol. 217, 4049–4050 (2018).
Article PubMed PubMed Central Google Scholar
Niu, L. et al. Atlastin-mediated membrane tethering is critical for cargo mobility and exit from the endoplasmic reticulum. Proc. Natl Acad. Sci. USA 116, 14029–14038 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Liu, N., Zhao, H., Zhao, Y. G., Hu, J. & Zhang, H. Atlastin 2/3 regulate ER targeting of the ULK1 complex to initiate autophagy. J. Cell Biol. 220, e202012091 (2021).
Article CAS PubMed PubMed Central Google Scholar
Monel, B. et al. Atlastin Endoplasmic Reticulum-Shaping Proteins Facilitate Zika Virus Replication. J. Virol. 93, e01047–19 (2019).
Article CAS PubMed PubMed Central Google Scholar
Neufeldt, C. J. et al. ER-shaping atlastin proteins act as central hubs to promote flavivirus replication and virion assembly. Nat. Microbiol 4, 2416–2429 (2019).
Article PubMed PubMed Central Google Scholar
Guelly, C. et al. Targeted high-throughput sequencing identifies mutations in atlastin-1 as a cause of hereditary sensory neuropathy type I. Am. J. Hum. Genet 88, 99–105 (2011).
Article CAS PubMed PubMed Central Google Scholar
Zhao, X. et al. Mutations in a newly identified GTPase gene cause autosomal dominant hereditary spastic paraplegia. Nat. Genet 29, 326–331 (2001).
Article CAS PubMed Google Scholar
Ulengin, I., Park, J. J. & Lee, T. H. ER network formation and membrane fusion by atlastin1/SPG3A disease variants. Mol. Biol. Cell 26, 1616–1628 (2015).
Article CAS PubMed PubMed Central Google Scholar
Montagna, A., Vajente, N., Pendin, D. & Daga, A. In vivo Analysis of CRISPR/Cas9 Induced Atlastin Pathological Mutations in Drosophila. Front Neurosci. 14, 547746 (2020).
Article PubMed PubMed Central Google Scholar
Rizo, J. Molecular Mechanisms Underlying Neurotransmitter Release. Annu Rev. Biophys. 51, 377–408 (2022).
Article CAS PubMed PubMed Central Google Scholar
Gao, S. & Hu, J. Mitochondrial Fusion: The Machineries In and Out. Trends Cell Biol. 31, 62–74 (2021).
Article CAS PubMed Google Scholar
Bian, X. et al. Structures of the atlastin GTPase provide insight into homotypic fusion of endoplasmic reticulum membranes. Proc. Natl Acad. Sci. USA 108, 3976–3981 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Byrnes, L. J. et al. Structural basis for conformational switching and GTP loading of the large G protein atlastin. EMBO J. 32, 369–384 (2013).
Article CAS PubMed PubMed Central Google Scholar
Byrnes, L. J. & Sondermann, H. Structural basis for the nucleotide-dependent dimerization of the large G protein atlastin-1/SPG3A. Proc. Natl Acad. Sci. USA 108, 2216–2221 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Kelly, C. M., Byrnes, L. J., Neela, N., Sondermann, H. & O’Donnell, J. P. The hypervariable region of atlastin-1 is a site for intrinsic and extrinsic regulation. J. Cell Biol. 220, e202104128 (2021).
Article CAS PubMed PubMed Central Google Scholar
Wu, F., Hu, X., Bian, X., Liu, X. & Hu, J. Comparison of human and Drosophila atlastin GTPases. Protein Cell 6, 139–146 (2015).
Article CAS PubMed Google Scholar
Yan, L. et al. Structures of the yeast dynamin-like GTPase Sey1p provide insight into homotypic ER fusion. J. Cell Biol. 210, 961–972 (2015).
Article CAS PubMed PubMed Central Google Scholar
Ahmadian, M. R., Mittal, R., Hall, A. & Wittinghofer, A. Aluminum fluoride associates with the small guanine nucleotide binding proteins. FEBS Lett. 408, 315–318 (1997).
Article CAS PubMed Google Scholar
Praefcke, G. J., Geyer, M., Schwemmle, M., Robert Kalbitzer, H. & Herrmann, C. Nucleotide-binding characteristics of human guanylate-binding protein 1 (hGBP1) and identification of the third GTP-binding motif. J. Mol. Biol. 292, 321–332 (1999).
Article CAS PubMed Google Scholar
Winsor, J., Hackney, D. D. & Lee, T. H. The crossover conformational shift of the GTPase atlastin provides the energy driving ER fusion. J. Cell Biol. 216, 1321–1335 (2017).
Article CAS PubMed PubMed Central Google Scholar
Morin-Leisk, J. et al. An intramolecular salt bridge drives the soluble domain of GTP-bound atlastin into the postfusion conformation. J. Cell Biol. 195, 605–615 (2011).
Article CAS PubMed PubMed Central Google Scholar
Pendin, D. et al. GTP-dependent packing of a three-helix bundle is required for atlastin-mediated fusion. Proc. Natl Acad. Sci. USA 108, 16283–16288 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Hu, J. & Rapoport, T. A. Fusion of the endoplasmic reticulum by membrane-bound GTPases. Semin Cell Dev. Biol. 60, 105–111 (2016).
Article CAS PubMed Google Scholar
Winsor, J., Machi, U., Han, Q., Hackney, D. D. & Lee, T. H. GTP hydrolysis promotes disassembly of the atlastin crossover dimer during ER fusion. J. Cell Biol. 217, 4184–4198 (2018).
Article CAS PubMed PubMed Central Google Scholar
O’Donnell, J. P. et al. Timing and Reset Mechanism of GTP Hydrolysis-Driven Conformational Changes of Atlastin. Structure 25, 997–1010.e1014 (2017).
Article PubMed PubMed Central Google Scholar
Liu, T. Y. et al. Cis and trans interactions between atlastin molecules during membrane fusion. Proc. Natl Acad. Sci. USA 112, E1851–E1860 (2015).
CAS PubMed PubMed Central Google Scholar
Sprang, S. R. G protein mechanisms: insights from structural analysis. Annu Rev. Biochem 66, 639–678 (1997).
Article CAS PubMed Google Scholar
Girodat, D., Pati, A. K., Terry, D. S., Blanchard, S. C. & Sanbonmatsu, K. Y. Quantitative comparison between sub-millisecond time resolution single-molecule FRET measurements and 10-second molecular simulations of a biosensor protein. PLoS Comput Biol. 16, e1008293 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Liu, T. Y. et al. Lipid interaction of the C terminus and association of the transmembrane segments facilitate atlastin-mediated homotypic endoplasmic reticulum fusion. Proc. Natl Acad. Sci. USA 109, E2146–E2154 (2012).
Article CAS PubMed PubMed Central Google Scholar
Gao, M., Nakajima, An, D., Parks, J. M. & Skolnick, J. AF2Complex predicts direct physical interactions in multimeric proteins with deep learning. Nat. Commun. 13, 1744 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Lomize, M. A., Pogozheva, I. D., Joo, H., Mosberg, H. I. & Lomize, A. L. OPM database and PPM web server: resources for positioning of proteins in membranes. Nucleic Acids Res. 40, D370–D376 (2012).
Article CAS PubMed Google Scholar
Moss, T. J., Andreazza, C., Verma, A., Daga, A. & McNew, J. A. Membrane fusion by the GTPase atlastin requires a conserved C-terminal cytoplasmic tail and dimerization through the middle domain. Proc. Natl Acad. Sci. USA 108, 11133–11138 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Montepietra, D. et al. FRETpredict: a Python package for FRET efficiency predictions using rotamer libraries. Commun. Biol. 7, 298 (2024).
Virtanen, P. et al. SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat. Methods 17, 261–272 (2020).
Article CAS PubMed PubMed Central Google Scholar
Saini, S. G., Liu, C., Zhang, P. & Lee, T. H. Membrane tethering by the atlastin GTPase depends on GTP hydrolysis but not on forming the cross-over configuration. Mol. Biol. Cell 25, 3942–3953 (2014).
Article PubMed PubMed Central Google Scholar
Chen, Y. et al. Conformational dynamics of dynamin-like MxA revealed by single-molecule FRET. Nat. Commun. 8, 15744 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Gao, Y. et al. Single reconstituted neuronal SNARE complexes zipper in three distinct stages. Science 337, 1340–1343 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Ma, L. et al. Munc18-1-regulated stage-wise SNARE assembly underlying synaptic exocytosis. Elife 4, e09580 (2015).
Article PubMed PubMed Central Google Scholar
Stein, A., Weber, G., Wahl, M. C. & Jahn, R. Helical extension of the neuronal SNARE complex into the membrane. Nature 460, 525–528 (2009).
Article ADS CAS PubMed PubMed Central Google Scholar
Rathore, S. S. et al. Intracellular Vesicle Fusion Requires a Membrane-Destabilizing Peptide Located at the Juxtamembrane Region of the v-SNARE. Cell Rep. 29, 4583–4592.e4583 (2019).
Article CAS PubMed PubMed Central Google Scholar
Faust, J. E. et al. The Atlastin C-terminal tail is an amphipathic helix that perturbs the bilayer structure during endoplasmic reticulum homotypic fusion. J. Biol. Chem. 290, 4772–4783 (2015).
Article CAS PubMed PubMed Central Google Scholar
Fersht, A. R., Matouschek, A. & Serrano, L. The folding of an enzyme. I. Theory of protein engineering analysis of stability and pathway of protein folding. J. Mol. Biol. 224, 771–782 (1992).
Article CAS PubMed Google Scholar
Munro, J. B., Vaiana, A., Sanbonmatsu, K. Y. & Blanchard, S. C. A new view of protein synthesis: mapping the free energy landscape of the ribosome using single-molecule FRET. Biopolymers 89, 565–577 (2008).
Article CAS PubMed PubMed Central Google Scholar
Joo, C. & Ha, T. Single-molecule FRET with total internal reflection microscopy. Cold Spring Harb. Protoc. 2012, pdb.top072058 (2012).
Article PubMed Google Scholar
McKinney, S. A., Joo, C. & Ha, T. Analysis of single-molecule FRET trajectories using hidden Markov modeling. Biophys. J. 91, 1941–1951 (2006).
Article ADS CAS PubMed PubMed Central Google Scholar
Hsu, P. C. et al. CHARMM-GUI Martini Maker for modeling and simulation of complex bacterial membranes with lipopolysaccharides. J. Comput Chem. 38, 2354–2363 (2017).
Article CAS PubMed PubMed Central Google Scholar
Huang, J. et al. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nat. Methods 14, 71–73 (2017).
Article CAS PubMed Google Scholar
Abraham, M. J. et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1-2, 19–25 (2015).
Article ADS Google Scholar
consortium, P. Promoting transparency and reproducibility in enhanced molecular simulations. Nat. Methods 16, 670–673 (2019).
Article Google Scholar
Michaud-Agrawal, N., Denning, E. J., Woolf, T. B. & Beckstein, O. Software News and Updates MDAnalysis: A Toolkit for the Analysis of Molecular Dynamics Simulations. J. Comput. Chem. 32, 2319–2327 (2011).
Article CAS PubMed PubMed Central Google Scholar
McGibbon, R. T. et al. MDTraj: A Modern Open Library for the Analysis of Molecular Dynamics Trajectories. Biophys. J. 109, 1528–1532 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank Yongli Zhang for critical reading of the manuscript. X.B. is supported by the National Natural Science Foundation of China (32371287 and 92154001), the Fundamental Research Funds for the Central Universities (63223043 and 63233053), and the Talent Training Project at Nankai University (035-BB042112). L.M. is supported by the National Natural Science Foundation of China (32271274 and 31770812). Yong W. is supported by the National Key R&D Program of China (2021YFF1200404), the National Natural Science Foundation of China (32371300), and the Fundamental Research Funds for Central Universities (K20220228) and acknowledges access to computational resources from the Information Technology Center and State Key Lab of Computer-Aided Design (CAD) & Computer Graphics (CG) of Zhejiang University. M.L. is supported by National Key R&D Program of China (2019YFA0709304) and National Natural Science Foundation of China (12090051 and T2221001).

Author information

These authors contributed equally: Lijun Shi, Chenguang Yang.

Authors and Affiliations

State Key Laboratory of Medicinal Chemical Biology, College of Life Sciences, Frontiers Science Center for Cell Responses, Nankai University, Tianjin, 300071, China
Lijun Shi, Kangning Li, Yunyun Wang & Xin Bian
National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing, 100190, China
Chenguang Yang, Ming Li, Lu Ma & Shuxin Hu
University of Chinese Academy of Sciences, Beijing, 100049, China
Chenguang Yang
College of Life Sciences, Zhejiang University, Hangzhou, 310027, China
Mingyuan Zhang, Keying Wang & Yong Wang
College of Life Sciences, Nankai University, Tianjin, 300071, China
Li Jiao & Ruming Liu
The Provincial International Science and Technology Cooperation Base on Engineering Biology, International Campus of Zhejiang University, Haining, 314400, China
Yong Wang

Authors

Lijun Shi
View author publications
You can also search for this author in PubMed Google Scholar
Chenguang Yang
View author publications
You can also search for this author in PubMed Google Scholar
Mingyuan Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Kangning Li
View author publications
You can also search for this author in PubMed Google Scholar
Keying Wang
View author publications
You can also search for this author in PubMed Google Scholar
Li Jiao
View author publications
You can also search for this author in PubMed Google Scholar
Ruming Liu
View author publications
You can also search for this author in PubMed Google Scholar
Yunyun Wang
View author publications
You can also search for this author in PubMed Google Scholar
Ming Li
View author publications
You can also search for this author in PubMed Google Scholar
Yong Wang
View author publications
You can also search for this author in PubMed Google Scholar
Lu Ma
View author publications
You can also search for this author in PubMed Google Scholar
Shuxin Hu
View author publications
You can also search for this author in PubMed Google Scholar
Xin Bian
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

All authors participated in designing the experiments. L.S., K.L., Yunyun W. and X.B. purified the proteins. L.S., C.Y. and K.L. performed smFRET assays. K.L., Yunyun W. and X.B. performed biochemical characterizations of the proteins. L.S., J.L. and R.L. performed functional studies of the proteins in yeast cells. Yong W., M.Z. and K.W. performed MD simulations. L.S., C.Y., M.L., L.M., S.H., M.Z., Yong W. and X.B. analyzed and interpreted data. L.S., C.Y., L.M., S.H. and X.B. initiated the project. L.S., C.Y., L.M., S.H., Yong W. and X.B. wrote the manuscript.

Corresponding authors

Correspondence to Yong Wang, Lu Ma, Shuxin Hu or Xin Bian.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Youngsoo Jun and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Movie 1

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Shi, L., Yang, C., Zhang, M. et al. Dissecting the mechanism of atlastin-mediated homotypic membrane fusion at the single-molecule level. Nat Commun 15, 2488 (2024). https://doi.org/10.1038/s41467-024-46919-z

Download citation

Received: 27 July 2023
Accepted: 13 March 2024
Published: 20 March 2024
DOI: https://doi.org/10.1038/s41467-024-46919-z

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.