Activation and substrate specificity of the human P4-ATPase ATP8B1

Dieudonné, Thibaud; Kümmerer, Felix; Laursen, Michelle Juknaviciute; Stock, Charlott; Flygaard, Rasmus Kock; Khalid, Syma; Lenoir, Guillaume; Lyons, Joseph A.; Lindorff-Larsen, Kresten; Nissen, Poul

doi:10.1038/s41467-023-42828-9

Download PDF

Article
Open access
Published: 18 November 2023

Activation and substrate specificity of the human P4-ATPase ATP8B1

Nature Communications volume 14, Article number: 7492 (2023) Cite this article

3427 Accesses
60 Altmetric
Metrics details

Subjects

Abstract

Asymmetric distribution of phospholipids in eukaryotic membranes is essential for cell integrity, signaling pathways, and vesicular trafficking. P4-ATPases, also known as flippases, participate in creating and maintaining this asymmetry through active transport of phospholipids from the exoplasmic to the cytosolic leaflet. Here, we present a total of nine cryo-electron microscopy structures of the human flippase ATP8B1-CDC50A complex at 2.4 to 3.1 Å overall resolution, along with functional and computational studies, addressing the autophosphorylation steps from ATP, substrate recognition and occlusion, as well as a phosphoinositide binding site. We find that the P4-ATPase transport site is occupied by water upon phosphorylation from ATP. Additionally, we identify two different autoinhibited states, a closed and an outward-open conformation. Furthermore, we identify and characterize the PI(3,4,5)P₃ binding site of ATP8B1 in an electropositive pocket between transmembrane segments 5, 7, 8, and 10. Our study also highlights the structural basis of a broad lipid specificity of ATP8B1 and adds phosphatidylinositol as a transport substrate for ATP8B1. We report a critical role of the sn-2 ester bond of glycerophospholipids in substrate recognition by ATP8B1 through conserved S403. These findings provide fundamental insights into ATP8B1 catalytic cycle and regulation, and substrate recognition in P4-ATPases.

Structural basis of the P4B ATPase lipid flippase activity

Article Open access 13 October 2021

Structure and autoregulation of a P4-ATPase lipid flippase

Article 26 June 2019

Autoinhibition and activation mechanisms of the eukaryotic lipid flippase Drs2p-Cdc50p

Article Open access 12 September 2019

Introduction

Lipid flippases actively transport lipids from the exoplasmic leaflet to the cytosolic leaflet of cellular membranes thereby creating and maintaining an asymmetric lipid distribution between the two leaflets¹. Transbilayer lipid asymmetry in the late secretory pathway is critical for eukaryotic cell functions such as membrane homeostasis, migration, cell signaling, blood coagulation, and morphogenesis². Most eukaryotic lipid flippases belong to the P4-ATPases subfamily of the P-type ATPase superfamily (with the sole known exception of ABCA4)^3,4. The P-type ATPases share a common architecture with three cytosolic domains that are involved in the autocatalyzed ATP hydrolysis: the nucleotide-binding (N) domain binds ATP, the phosphorylation (P) domain contains a conserved aspartate residue, which is phosphorylated during the transport cycle, and the actuator (A) domain contains a conserved glutamate, which is responsible for the subsequent dephosphorylation of the phosphoenzyme intermediate, that enables the protein to return to its initial state⁵ (Supplementary Fig. 1a). These three domains are linked to the membrane domain, which is typically formed by ten transmembrane segments (TM1–10). TM4 contains a conserved, proline-containing motif, which is specific to each P-type subgroup. For P4-ATPases it is P(I/V)S(L/M) and it constitutes part of the transport binding site and forms a hydrophobic gate preventing lipids from diffusing freely through the protein^6,7 (Supplementary Fig. 1a).

Substrate transport by P-type ATPases is driven by large conformational changes of the transporter switching between so-called E1 and E2 states that allow alternating access to each side of the membrane. These conformational changes are tightly coupled to the phosphorylation and dephosphorylation reactions performed by the cytosolic domains. The different steps of the P-type ATPase catalytic cycle are described by the Post-Albers scheme^8,9 (Supplementary Fig. 1b), with the transition from the E1 state to E1P state being the phosphorylation reaction, and the transition from the E2P state to E2 state being dephosphorylation. In P4-ATPases, substrate lipid recognition occurs in the outward-open E2P state followed by cognate substrate-induced occlusion triggering dephosphorylation^10,11,12. In addition, most P4-ATPases form a binary complex with a CDC50 subunit³, and this interaction is essential for the correct subcellular membrane targeting and flippase activity of the complex^13,14,15.

The human genome encodes 14 different P4-ATPases from ATP8, ATP9, ATP10, and ATP11 groups, each containing different isoforms³. ATP8B1 was originally identified as the causative gene for progressive familial intrahepatic cholestasis 1 (PFIC1), a rare but severe liver disease characterized by impaired bile flow¹⁶. Milder diseases caused by mutations of the ATP8B1 gene include benign recurrent intrahepatic cholestasis of type 1 (BRIC1) and intrahepatic cholestasis of pregnancy (ICP1). The function of ATP8B1 has been associated with the integrity of the canalicular membrane of hepatocytes¹⁷, but the exact mechanism is still unclear, in part obscured by discrepancies regarding the lipid substrate specificity. Furthermore, genetic variation at a site upstream of the ATP8B1 coding region has been identified in resilience to Alzheimer’s Disease¹⁸, and loss-of-function mutations of the closely related ATP8B4 represent Alzheimer’s risk¹⁹, but in both cases again through unknown mechanisms.

Initially, ATP8B1 was reported as a phosphatidylserine (PS)-specific flippase, but other studies have since proposed specificity to phosphatidylethanolamine (PE), phosphatidylcholine (PC) or cardiolipin (CL)^{20,21,22,23,24,25}. More recently, ATPase activity measurements using the purified ATP8B1-CDC50A and ATP8B1-CDC50B complexes also lead to contrasting conclusions^26,27. Previously, we reported that PC, PE and PS were able to stimulate the ATPase activity, but not CL or sphingomyelin (SM)²⁶, whereas Cheng et al., found that PS, but not PC could stimulate activity²⁷. Additionally, ATP8B1 activity is tightly regulated by its N- and C-terminal ends that closely interact with the cytosolic domains in an autoinhibited E2P state^26,27. Furthermore, ATP8B1 activity depends not only on a substrate lipid, but also on the presence of phosphoinositides, with a strong affinity for the triple-phosphorylated form, PI(3,4,5)P₃²⁶.

Here, we present nine different cryo-EM structures, functional studies, and molecular dynamics simulation data of the ATP8B1-CDC50A complex in conformations covering most of the transport cycle and different substrate complexes. From our data, we identify the presence of a water molecule associated with the E1 states of the cycle and located in the canonical transport site of P-type ATPase. We also report that the autoinhibited ATP8B1-CDC50A complex fluctuates between an outward-open and a closed conformation, identify a PI(3,4,5)P₃ binding site, new transport substrates, and reveal structural determinants of the broad glycerophospholipid specificity of ATP8B1.

Results

Cryo-EM structures of the ATP8B1-CDC50A complex along its catalytic cycle

We expressed the human ATP8B1-CDC50A complex in Saccharomyces cerevisiae. Isolated membranes were solubilized with n-Dodecyl β-D-maltoside (DDM) and purified by affinity purification via a BAD-tag (biotin acceptor domain)²⁸. DDM was exchanged for Lauryl Maltose Neopentyl Glycol (LMNG) prior to further purification by size-exclusion chromatography. We used two different constructs for our studies: full-length ATP8B1 (WT-ATP8B1) and a variant containing an engineered 3C protease site allowing autoinhibition of ATP8B1 to be relieved by proteolytic cleavage of the C-terminal end (ΔCter-ATP8B1) (Supplementary Fig. 2). The ΔCter-ATP8B1 construct was incubated with nucleotides (ADP or the non-hydrolyzable ATP analog adenine-(βγ-methylene)-triphosphate (AMPPCP)) and/or AlF_x to stabilize forms similar to E1-ATP, E1P-ADP and E1P conformations, corresponding to autophosphorylation from ATP and subsequent ADP release events (Fig. 1 and Supplementary Figs. 3–16, Supplementary Table 1). Furthermore, the phosphate mimics BeF_x and VO₃^- (metavanadate) were used in presence of transport substrate to stabilize the E2P outward-open and the E2┅Pi lipid-occluded conformation of the complex (Fig. 1 and Supplementary Figs. 3–16, Supplementary Table 1). Finally, the full-length ATP8B1 preincubated with ATP revealed two off-cycle autoinhibited states (Fig. 1 and Supplementary Figs. 3–16).

**Fig. 1: Cryo-EM maps of the ATP8B1-CDC50A flippase complex.**

A equivalent binding site of P-type ATPases is occupied by water in the E1 half-cycle of ATP8B1

The cryo-EM structure of the ATP8B1-CDC50A complex in the presence of AMPPCP mimics an ATP-bound E1 state and reveals the ATP binding pocket of ATP8B1 (Fig. 2a, b). As expected, AMPPCP is bound similarly to other P-type ATPases. The adenosine base is stacked by a π-π-interaction with F596. Additionally, D554 is directly engaged in the coordination network of the ß-phosphate, mediated by R652. Indeed, the R652 rotamer state is constrained by interactions with D554 and D734 (Fig. 2b). Noteworthy, the D554N mutation has been identified in patients suffering from PFIC1²⁹. Interestingly, extra density was also observed at an interacting distance to the ß-phosphate of AMPPCP, suggesting the presence of an additional cation or a water molecule (see Supplementary Discussion). The E1P-ADP state mimicked by ADP and AlF_x is similar to previously published structures of P4-ATPases^7,12,30. However, closer inspection of all E1 states along with the outward-open E2P states (autoinhibited and active conformations, see further below) reveals the presence of a density at the canonical, phosphorylation-coupled transport site of P-type ATPases between TM4, TM5 and TM6 (Fig. 2c). Due to the geometry of the site with no more than four ligands, we modeled the site as a water molecule. Interestingly, the water molecule is at interacting distance to S403 from the PISL motif of TM4, the backbone carbonyl group of two residues from TM6 (N989 and T993) and the highly conserved K957 of TM5, which is also described as essential in ATP11C³⁰.

**Fig. 2: ATP binding and the presence of a water molecule in the core of the lipid transport site.**

Given the high conservation of these residues among all P4-ATPases, we compared the cryo-EM densities of previously published P4-ATPases obtained at a sufficient resolution to observe a coordinated water molecule (Supplementary Fig. 17). In almost all maps, a density could be observed at this site, consistent with a water molecule. The water molecule site partially overlaps with transition metal and cation sites of P1 and P2-ATPases (Supplementary Fig. 18). Furthermore, the recent cryo-EM structure of human ATP13A2 (a P5-ATPase), in the E1-ATP conformation also revealed water molecules in the canonical P-type ATPase binding site³¹ (Supplementary Fig. 18).

Autoinhibited ATP8B1-CDC50A can adopt an open and a closed conformation

Previously, two independent structures of the ATP8B1-CDC50A complex in the E2P_{autoinhibited} state were reported^26,27. Surprisingly, they are not identical, especially in the N-terminal part of the transmembrane region (Supplementary Fig. 19a), where TM1 and TM2 have a different orientation either in a “close” or “open” conformation. Here, we collected cryo-EM data of the full-length ATP8B1-CDC50A complex preincubated with ATP and the activating lipid PI(3,4,5)P₃. Upon data processing two different conformations were observed (Fig. 3). The first one, representing ~90% of the particles, was in a “closed” conformation virtually identical to previous studies of autoinhibited forms^7,26,32 (Figs. 1 and 3 and Supplementary Fig. 19b). In this conformation, TM1 and TM2 are tightly packed to the rest of the transmembrane helix bundle, thus masking the substrate recognition PISL motif of TM4 and closing the exoplasmic lipid entry (Fig. 3 and Supplementary Fig. 19b). However, approx. 10% of the particles adopted an “open” conformation corresponding to the structure reported by Cheng et al.,²⁷. Here, TM1-2 are rotated and the lipid-binding groove is open as observed for non-autoinhibited P4-ATPases^30,32,33 (Fig. 3, Supplementary Fig. 20). In both cases, we still observe clear densities for the N- and C-terminal tails, interacting with the A, P and N domains in a similar manner (Fig. 3).

**Fig. 3: ATP8B1 autoinhibition mechanism and lipid groove opening.**

To better understand whether the TM1-2 orientation of this E2P_{autoinihibited} “open” state is different from the active form of ATP8B1 we collected cryo-EM data of the C-terminally truncated ATP8B1 trapped in an E2P conformation with BeF_x in presence of a POPC lipid substrate (Figs. 1 and 3). In the resulting E2P_active conformation, the transmembrane bundle displays the same organization as the E2P_{autoinhibited} “open” conformation with the lipid groove being open and filled with PC (Fig. 3 and Supplementary Fig. 19b). Interestingly, in the C-terminally truncated E2P_active conformation, no density associated with the N-terminal tail could be observed in the cryo-EM map. This suggests that the N-terminal tail only interacts with the core protein in the presence of the autoinhibitory C-terminal tail. Given the previously described strong synergistic effect of the N-terminus on C-terminal tail inhibition²⁶, our structural data suggest that the N-terminal tail stabilizes the C-terminal tail in the bound form, thus reinforcing autoinhibition.

PI(3,4,5)P₃ binding site

As previously shown, C-terminally truncated ATP8B1 exhibited strong stimulation by different phosphoinositides in the presence of a lipid substrate²⁶ (Fig. 4a). Most of our cryo-EM data have been obtained in presence of PI(4,5)P₂, however no clear density for this lipid was observed in the density maps. To identify the phosphoinositide binding site in ATP8B1 we therefore determined the cryo-EM structures of the two full-length E2P_{autoinhibited} states and the E2P_active state in the presence of PI(3,4,5)P_3, for which ATP8B1 has a higher apparent affinity²⁶. For the three structures, a lipid density could be observed in a positively charged cavity formed by TM5, TM7, TM8, and TM10 (Fig. 4b, Supplementary Figs. 21–22a), similar to the PI(4)P and PI(3,5)P₂ binding pockets of Drs2p^12,32 and ATP13A2³⁴, respectively (Supplementary Fig. 21). The observed density is weaker compared to the surrounding protein, potentially indicating partial occupancy of the phosphoinositide binding site and/or flexibility of the ligand.

Therefore, we performed molecular dynamics (MD) simulations of two systems based on the structure of the E2P_{autoinhibited} “closed” conformation. One with PI(3,4,5)P₃ and another with a non-phosphorylated PI molecule, matching the position of the modeled PI(3,4,5)P₃ molecule as a starting point. Both systems contained a native-like lipid bilayer composition mimicking the plasma membrane. We then ran a set of five independent 600-ns-long simulations for each system and analyzed the dynamics and interactions between the lipids and the surrounding protein residues. During the equilibration procedure, both lipids slightly shifted from the refined position in the cryo-EM structures (RMSD between carbon atoms of inositol and phosphate atoms: 1.90 Å, Supplementary Fig. 22b). However, both PI(3,4,5)P₃ and PI remained bound within the pocket formed by TM5, 7, 8 and 10 during the entire 600 ns in all simulations, suggesting stable binding of the two lipids within the cavity. To analyze the interaction between the protein and the lipids, we calculated hydrogen-bonded populations over the last 200 ns of each trajectory. Although the total number of interacting protein residues is similar for the two systems (12 and 15 for PI(3,4,5)P₃ and PI, respectively), we found that PI(3,4,5)P₃ forms a more stable contact with the protein compared to PI, as indicated by an overall increase of hydrogen-bond frequency (Fig. 5a). For PI(3,4,5)P₃, the most frequent contacts are established with residues R1032 (0.97), R1164 (0.96) and R952 (0.88). The same residues are also the most frequent interaction partners for PI but with substantially lower populations: R1032 (0.59), R1164 (0.19), and R952 (0.59). This supports a model for the lipid-protein interactions at the PI(3,4,5)P₃ binding site based primarily on the interactions between positively charged arginine and lysine residues and the negatively charged phosphates on the lipid headgroup.

**Fig. 5: Molecular dynamics studies of the PI(3,4,5)P₃ binding site.**

Next, we studied the dynamics of PI and PI(3,4,5)P₃ in the cavity by monitoring the positions of the center of mass (COM) of the inositol ring relative to the TM region surrounding the lipid-binding site. We found that the hydrogen-bonding patterns are closely related to the dynamics of the headgroups of the two different lipids (Fig. 5b, c): PI shows substantially higher mobility in all three directions, whereas PI(3,4,5)P₃ moves mainly along the z-axis (perpendicular to the membrane plane) inside the cavity. To visualize this, we clustered³⁵ the different binding poses based on the orientations of the inositol. For PI, we found a total of 43 clusters with 19%, 13%, and 13% of the structures populating the top three clusters (45% combined), reflecting the large mobility of PI (Fig. 5b–e, Supplementary Fig. 22c). In comparison, we found only a total of 9 clusters for PI(3,4,5)P₃ with 56%, 26%, and 10% of the structures in the top three clusters (92% combined); in line with a more stable binding of PI(3,4,5)P₃. When we examined the solvation properties of the cavity with respect to the bound lipid, we observed on average more water molecules inside the cavity with PI(3,4,5)P₃ bound (57 ± 9) than with PI bound (47 ± 7) (Supplementary Fig. 22d). Additionally, our simulations indicate that on average 2.7 ± 1.0 K⁺ ions populate the binding site when a PI(3,4,5)P₃ is present, whereas no K⁺ ions are observed in the presence of PI (Supplementary Fig. 22e). This suggests that the PI(3,4,5)P₃ head group binds additional counterions, despite numerous positively charged residues at the binding site. Altogether, our simulations suggest that PI(3,4,5)P₃ binds in a more stable fashion, where dynamics are restricted to mostly “up and down” movements (along a z-axis perpendicular to the membrane plane), whereas PI exhibits increased 3D mobility.

The transport site of ATP8B1 can accommodate a wide range of glycerophospholipids

The transport substrate lipid POPC observed in the E2P_active structure adopts a similar conformation as observed for other substrate-bound P4-ATPases^30,36. The PC headgroup is positioned at the end of a groove formed by TM1, 2, 3, and 4 (Figs. 3 and 6a). The acyl chains of the lipid are not interacting with the protein. Strikingly, the cavity surrounding the choline moiety formed by residues from TM1 and TM2 is wider than in other P4-ATPases characterized so far (Fig. 6a and Supplementary Fig. 23). We therefore hypothesized that with such a cavity, ATP8B1 should be able to bind a broad range of phospholipid headgroups. To test this hypothesis, we evaluated the ability of the main phospholipids found in eukaryotic membranes to stimulate the ATPase activity of ATP8B1 (Fig. 6b). C-terminally truncated ATP8B1 was incubated with increasing lipid/DDM ratios, and a saturating concentration of PI(3,4,5)P₃. All tested glycerophospholipids stimulated ATP8B1 ATPase activity, whereas SM did not (Fig. 6b). Lipids with smaller head groups (PE, PG, PC, or PA) seemed to be more activating than the negatively charged PS or the bulky PI.

**Fig. 6: ATP8B1 lipid substrate specificity, binding, and occlusion.**

To better understand how ATP8B1 can accommodate such different lipid substrates, we determined the structure of the state associated with lipid occlusion and dephosphorylation using different substrate lipids. For this purpose, we used vanadate as an inhibitor that mimics the transition state of dephosphorylation of the catalytic aspartate. In this E2┅Pi lipid-occluded conformation, the A-domain has rotated and the catalytic E234 residue, from the conserved DGET motif of the A-domain, now coordinates a water molecule for in-line attack of the aspartyl phosphate moiety in the P-domain. The vanadate most likely adopts a VO₃⁻ planar conformation (metavanadate), further ligated by the aspartate and the water molecule, mimicking the phosphate leaving group³⁷ (Fig. 1). In this conformation, TM1 and 2 now enclose the substrate lipid, blocking the entry groove to prevent the entry of a new lipid (Fig. 3). We collected cryo-EM data sets in the presence of three lipid substrates spanning different chemical properties: PC as a standard lipid, a negatively charged PS, and a bulky PI (Fig. 6c–e, Supplementary Fig. 24 and Supplementary Movie 1). The analysis of the lipid occlusion sites in each case revealed the presence of highly ordered water network interacting with the lipid headgroup and the protein (Fig. 6f). Interestingly, the structures showed that all three lipids are occluded in a similar manner, and side chains of the residues forming the occlusion site adopted virtually identical structures despite the presence of the charged carboxylic acid moiety of PS or the bulkier inositol moiety of PI (Fig. 6g). However, the water network differs among the different lipid occlusion sites (Fig. 6h), which could also explain the differences observed when comparing the ATPase activities. One may speculate that the organization of this water network is part of the triggering event of the allosteric transmission through TM2 that enables the A-domain rotation required for the dephosphorylation event.

Sn-2 ester moiety is critical for substrate recognition

Given that all major glycerophospholipids tested were able to stimulate ATPase activity, unlike sphingomyelin, and can be recognized and occluded by the ATP8B1-CDC50A complex, we then sought to identify which part of the lipid is essential for substrate recognition. In the outward-open E2P_active state, the phosphoglycerol backbone of PC was found close to the PISL motif of ATP8B1. Interestingly, in this conformation, a water molecule is observed interacting with the S403 side chain (Fig. 7a, Supplementary Fig. 24a). For the E2┅Pi lipid-occluded state, however, the lipid is bound deeper in the pocket and the water molecule now displaced (Fig. 7b, Supplementary Fig. 24b–d). In the E2┅Pi lipid-occluded conformation, the S403 side chain is less well defined but the different possible rotamers of the S403 can now interact with the phosphate or with the sn-2 ester carbonyl more tightly (Fig. 7b, Supplementary Fig. 24b–d). To evaluate the importance of the sn-2 ester carbonyl interaction, we tested the ability of lipid analogs to stimulate the ATPase activity of the ATP8B1-CDC50A complex (Fig. 7c). All lipids tested share the same choline headgroup and only differ in their chemical linkage between the phosphate group and the acyl chains. Plasmanylcholine, which contains an ether bond in sn-1 and an ester bond in sn-2, appeared to stimulate the ATPase activity almost as efficiently as PC (ester-PC). In contrast, ether PC, which lacks the ester bonds in sn-1 and sn-2 positions, failed to efficiently activate ATP8B1, achieving only 10% of the activity observed in the presence of PC at an equivalent lipid/DDM ratio. As previously described²⁶, lyso-PC, which lacks the fatty acid chain in sn-2 position, can stimulate the ATPase activity, but to a much lower extent than PC. Together, our data strongly suggest that the presence of an acyl chain in sn-2 position, and more specifically of an ester bond in the sn-2 position, is a critical determinant for lipid recognition by ATP8B1.

**Fig. 7: The sn-2 ester bond of the transport lipid is essential for substrate recognition.**

Discussion

A water molecule at the transport site in P4- and P5-ATPases

Biochemical and structural studies of P4-ATPases have shown substrate-independent phosphorylation from ATP unlike for cation pumps where phosphorylation is tightly coupled to cation binding and occlusion. Thorough characterization of bovine ATP8A2-CDC50A and yeast Drs2p-Cdc50p also showed that phosphorylation was not dependent on a range of cations and quite insensitive to pH^11,38. Lipid transport of ATP8A2 on solid-supported membranes revealed a net outward charge transfer upon PS transport, also supporting substrate-independent phosphorylation^39,40. Our studies agree with this model, but we additionally identify a neutral water molecule in the transport site of P4-ATPases, which interacts with polar side chains in the absence of a substrate lipid headgroup (Fig. 2c, Supplementary Fig. 17). Interestingly, the water molecule stays bound in the outward-open E2P_active form of ATP8B1, where an incoming substrate lipid makes first interactions with the lipid-binding site. The water molecule is then chased upon occlusion of the substrate lipid, and the lipid headgroup directly interacts with the S403 side chain of the PISL motif, i.e. replacing the water molecule. The role and dynamics of the water structure and how it is involved in the P4-ATPase cycle invite further studies by e.g. MD simulations. Interestingly, the structural and functional data obtained on P5A- and P5B-ATPases such as human ATP13A2 also show that substrate transport (of transmembrane alpha helices and polyamines, respectively) is coupled to dephosphorylation, whereas phosphorylation is independent of other substrates, i.e., similar to P4-ATPases^{31,34,41,42,43}. Congruent with this, the polyamine transport site of human ATP13A2 also binds water molecules in the E1 to E1P half-cycle³¹ (Supplementary Fig. 18). From an evolutionary point of view, the presence of a water binding site in P4- and P5-ATPases might result from the modification of a cation site of their last common ancestor, thus acquiring the ability to undergo phosphorylation from ATP in a de facto spontaneous mechanism.

Autoinhibition is compatible with lipid-binding, but not occlusion

Regulation through N- and C-terminal tails is a feature shared among various P-type ATPases, but structurally it was first described for the Drs2p-Cdc50p flippase complex, which revealed how the C-terminal tail occupies the nucleotide-binding pocket and locks the cytoplasmic domains³². A similar autoregulation mechanism applies to ATP8A1 and ATP8B1^7,26,27. C-terminal autoregulation locks the A-domain thereby restricting TM1-2 movements required for the opening of the outward-oriented lipid entry pathway (Supplementary Fig. 20). We show here that despite the presence of the N- and C-terminal tails at the interface of the cytoplasmic domains, the A-domain maintains some level of movement and TM1-2 can rotate to open the lipid entry pathway. However, lipid occlusion and dephosphorylation cannot occur. Thus, our data indicate that the E2P_{autoinhibited} state of ATP8B1 explores both open and closed conformations, but only the open state can initiate lipid binding. In our experimental conditions (4 °C in LMNG/CHS/PI(3,4,5)P₃ mixed micelles), the closed conformation seems to be favored, but the equilibrium could be poised toward the open conformation in a native membrane. Further structural investigations of ATP8B1 in a more native environment will be needed to evaluate sampling between lipid-free and lipid-binding conformations in the autoinhibited form and how lipids might trigger further activation.

ATP8B1 activation by phosphoinositides

ATP8B1 is directly activated by PI(3,4,5)P₃, although other phosphoinositides such as the more abundant PI(4,5)P₂ can also promote its ATPase activity²⁶. From the cryo-EM data we identified a PI(3,4,5)P₃ binding site in a positively charged cavity lined by TM8, 9, and 10. Additionally, the binding site geometry remains unchanged through all conformations of the catalytic cycle, suggesting that the regulatory lipid binds in a conformation-independent manner (Supplementary Fig. 25a). Our MD data show a clear interaction and stabilization of PI(3,4,5)P₃ compared to PI in the identified pocket. However, the cryo-EM density map corresponding to the PI(3,4,5)P₃ molecule was weaker compared to the rest of the protein which suggests flexibility or a partial occupancy of the site. This might also reflect the difference in experimental conditions between ATPase activity assays (37 °C in DDM) and cryo-EM sample preparation (4 °C in LMNG). Although not addressed here, the physiological role of this activation remains unclear. Given the localization of ATP8B1 at the apical membrane of epithelial cells, PI(4,5)P₂ represents the most abundant phosphoinositide compared to PI(3,4,5)P₃⁴⁴, which on the other hand displays a higher affinity²⁶_. Synthesis and degradation of phosphoinositides is a highly dynamic process related to cell signaling, and the cascades regulating ATP8B1 have not been identified yet. Moreover, questions remain open on the exact mechanism by which phosphoinositides activate ATP8B1, and which specific steps of the catalytic cycle are stimulated. For Drs2p-Cdc50p, PI(4)P stabilizes an amphipathic helix, destabilizes binding of the autoinhibitory C-terminal tail, and stimulates dephosphorylation of E2P coupled to lipid occlusion and transport^32,45. Yet, C-terminally truncated Drs2p-Cdc50p requires PI(4)P for full activity^32,46. Similarly, truncated versions of ATP8B1 also require phosphoinositides for activation indicating that if phosphoinositides play a role in relief of autoinhibition, they also have an additional function. Our studies do not provide immediate clues on the direct link between phosphoinositide binding and activation of ATP8B1, but it is likely linked to biased dynamics favoring transport substrate recognition and/or occlusion in the E2 half-cycle, as we could previously observe that the ATP-dependent E1 to E2P half-cycle is not modulated by phosphoinositides²⁶. In our study we could observe that PI(3,4,5)P₃ interacts tightly with the R952, which is located five residues after K957 on TM5, the latter being close to the lipid transport site and in interaction with the previously described water molecule (Supplementary Fig. 25b). However, comparison of our MD simulations did not reveal striking differences in R952 and TM5 orientation, whether PI or PI(3,4,5)P₃ is bound. Future work should aim to elucidate the mechanism by which phosphoinositides activate flippases, and to identify the signaling pathways involved in the activation of ATP8B1 in the cell.

ATP8B1 has a broad lipid specificity

So far, no consensus has been established regarding the transport substrate of ATP8B1^{20,21,22,23,24,25,26,27}. Our structures of the E2P_active and E2┅Pi states bound to lipids show that different lipid head groups may be accommodated at the lipid-binding site and more importantly, that they can be occluded. This is consistent with all of them being able to stimulate ATPase activity (Fig. 6b), and strongly suggests that they are also transport substrates. The different sizes and chemical properties between the lipid headgroups bound at the transport site may lead to differences in kinetics of binding and occlusion and thus explain why they stimulate ATPase activity in a different manner. In comparison, the lipid headgroup binding cavity of other P4-ATPases is more restrictive compared to ATP8B1 (Supplementary Fig. 23). A comparison of ATP11C and ATP8B1 in lipid-bound E2P states (i.e., where lipid recognition takes place prior to occlusion) shows that TM1 and TM2 are oriented differently. In ATP11C, TM1-2 are closer to TM3-4, thus diminishing the size of the groove for lipid-binding (Supplementary Fig. 26 and Supplementary Fig. 27). In line with this, a screen for glycerophospholipids showed that for ATP11C, only PS and PE stimulated ATPase activity and were transported^24,36,47. A sequence alignment of ATP8B1 to PS/PE-specific flippases such as ATP11C shows a high degree of conservation at the lipid-binding cavity (Supplementary Fig. 27). Only few residues differ significantly, in particular W390, but surprisingly, they are not located directly at the lipid headgroup binding site. Rather, the bulky W390 side chain keeps TM1-2 at a further distance from TM3-4 and would provide a wider and more permissive groove for lipid headgroup recognition and occlusion. Previous studies on ATP8A1 showed that ether-PS is not transported across the erythrocyte membrane⁴⁸. Here, we provide evidence that the carbonyl group of the sn-2 ester bond plays a critical role in substrate-induced ATPase activity in ATP8B1, most likely due to the interactions it forms with S403 of the conserved PISL motif (Fig. 7b). Noteworthy, a S403Y mutation has been found in patients suffering from PFIC1²⁹, and this mutation has been shown to dramatically diminish the ATPase activity of the purified ATP8B1-CDC50B complex²⁷. Sphingosine-based lipids, like SM, are not substrates of ATP8B1. In sphingolipids, the sn-2 ester bond is replaced by an amide bond, which engages in intra-molecular hydrogen bonds with the phosphate group⁴⁹. In our three different E2┅Pi lipid-occluded structures, the phosphate moiety of the lipid headgroup interacts simultaneously with S403 and a highly ordered water molecule (Fig. 7b and Supplementary Movie 1). Hence, the internal H-bond of SM would disfavor the interactions observed for glycerophospholipids. Thus, our study broadens the lipid specificity of ATP8B1 to sn-1-mono-ether phospholipids, in line with a recently identified link between ATP8B2 and plasmalogens homeostasis⁵⁰. It should be mentioned that we cannot exclude the possibility that sn-2-mono-ether phospholipids could also be recognized by ATP8B1, although probably not as effectively as sn-1-mono-ether phospholipids due to the proximity of the sn-2 ester bond with S403 observed in our structures. Transport studies of such non-natural sn-2-mono-ether phospholipids by P4-ATPases could potentially pave the way to the development of lipid-based drugs that would be better recognized compared to previously evaluated anti-cancer alkylphosphocholine analogs such as miltefosine and edelfosine²¹.

In summary, we present a series of nine cryo-EM structures of the ATP8B1-CDC50A complex with accompanying functional studies and MD simulations, deciphering most of its catalytic cycle. The structures highlight a so far unrecognized water molecule bound at the lipid transport site during the autophosphorylation part of the catalytic cycle. Additionally, we observe two different autoinhibited states, a closed and an outward-open conformation which are likely in a dynamic equilibrium. Finally, our work reveals important determinants of lipid specificity in P4-ATPases and explains earlier discrepancies about the ATP8B1 transport substrate. Future work should aim at deciphering the subtle but crucial effects of phosphoinositide binding and activation.

Methods

Co-expression of the hATP8B1-hCDC50A complex

hATP8B1 (Uniprot: O43520; A1152T natural variant) and hCDC50A (Uniprot: Q9NV96) were co-expressed in a protease deficient Saccharomyces cerevisiae W303.1b/Δpep4 (MATα, leu2-3, his3-11, ura3-1, ade2-1, Δpep4, can^r, cir⁺) yeast strain as previously described²⁶. To generate a C-terminal cleavable construct, an overlap extension PCR strategy was used to insert 8 aminoacids forming a HRV 3C protease site (LEVLFQGP, see Supplementary Table 2 for primer sequences) in between R1184 and K1186 of ATP8B1 sequence. Yeast strains and plasmids are available upon request.

Purification of the ATP8B1-CDC50A complex

Yeast cells were resuspended in buffer A (50 mM Tris-HCl pH 7.5, 1 mM EDTA, 0.6 M sorbitol) supplemented with protease inhibitors (2 mM PMSF, 2 µg.mL⁻¹ leupeptine, 2 µg.mL⁻¹ pepstatin, 2 µg.mL⁻¹ chymostatin). The cells were subsequently broken with 0.5 mm glass beads using bead beaters. The crude extract was then spun down at 3,000 g for 30 min at 6 °C, to remove cell debris and nuclei. The membrane fraction was pelleted at 100,000 g for 90 min at 4 °C. The resulting yeast membrane pellets were finally resuspended at about 50 mg.mL⁻¹ of total protein in buffer A supplemented with protease inhibitors (1 mM PMSF, 1 µg.mL⁻¹ leupeptine, 1 µg.mL⁻¹ pepstatin, 1 µg.mL⁻¹ chymostatin). Membranes were diluted to 5.5 mg.mL⁻¹ of total protein in ice-cold buffer B (50 mM MOPS-Tris at pH 7, 100 mM KCl, 1 mM DTT, 20% (w/v) glycerol and 5 mM MgCl_2,), supplemented with protease inhibitors (1 mM PMSF, 1 µg.mL⁻¹ leupeptine, 1 µg.mL⁻¹ pepstatin, 1 µg.mL⁻¹ chymostatin). The suspension was then supplemented with 15 mg.mL⁻¹ DDM and 5 mg.mL⁻¹ CHS and incubated 1 h at 4 °C. Insoluble material was pelleted by centrifugation at 125,000 g for 1 h at 4 °C. The supernatant, containing solubilized proteins, was applied onto a streptavidin-sepharose resin and incubated for 2 h at 6 °C. The resin was washed twice with six resin volumes of ice-cold buffer B supplemented with 0.2 mg.mL⁻¹ LMNG and 0.02 mg.mL⁻¹ CHS and protease inhibitors (1 mM PMSF, 1 µg.mL⁻¹ leupeptine, 1 µg.mL⁻¹ pepstatin, 1 µg.mL⁻¹ chymostatin). The resin was then washed thrice with six resin volumes of ice-cold buffer B supplemented with 0.1 mg.mL⁻¹ LMNG and 0.01 mg.mL⁻¹ CHS. To delipidate the purified ATP8B1-CDC50A complex, each washing step was performed in batch with a prolonged incubation time of at least 15 min, to allow the slowly diffusing lipids to be incorporated in detergent micelles⁵¹. Elution was performed by the addition of 60 µg of purified TEV protease per mL of resin and overnight incubation at 6 °C. The eluted fraction was concentrated using a Vivaspin unit (100 kDa MWCO) prior to injection on a size-exclusion Superose 6 10/300GL increase column equilibrated with buffer C (50 mM MOPS-Tris pH 7, 100 mM KCl, 1 mM DTT, 5 mM MgCl₂) supplemented with 0.03 mg.mL⁻¹ LMNG and 0.003 mg.mL⁻¹ CHS. For 3C protease site-containing version of ATP8B1, 100 µg of purified 3C protease (1:20 mol:mol ATP8B1:3C protease) was added to the concentrated sample and incubated 18 h at 6 °C and 1 h at 30 °C prior to size-exclusion chromatography.

Lipids

Lipids used for ATPase activity measurement and cryo-EM sample preparation were POPC (Avanti Polar Lipid Inc., 850457 P), POPS (Avanti Polar Lipid Inc., 840034 P), soy PI (Avanti Polar Lipid Inc., 840044 P), POPA (Avanti Polar Lipid Inc., 840857 P), POPE (Avanti Polar Lipid Inc., 850757 P), Egg PG (Avanti Polar Lipid Inc., 841138 P), Egg Sphingomyelin (Avanti Polar Lipid Inc., 860061 P), 18:0 Lyso-PC (Avanti Polar Lipid Inc., 855775 P), C16-18:1 Sn-1 Ether PC ( = Plasmanylcholine, Avanti Polar Lipid Inc., 878112 P), C16:0-18:1 Diether PC (Avanti Polar Lipid Inc., 999983 P), Brain PI(4,5)P₂ (Avanti Polar Lipid Inc., 840046 P) and diC16 PI(3,4,5)P₃ (Echelon, P-3916).

Cryo-EM sample preparation

Different conformations of the ATP8B1-CDC50A complex were stabilized in the following way: For the determination of the E1-ATP (AMPPCP), E1P-ADP (ADP+AlF₄⁻), E1P (AlF₄⁻), E2┅Pi (PC) (VO₃⁻ + POPC), E2┅Pi (PS) (VO₃⁻ + POPS) and E2┅Pi (PI) (VO₃⁻ + soy PI) states, the C-terminally truncated construct was incubated with PI(4,5)P₂ 0.05 mg.mL⁻¹ during 3C protease treatment. The sample was then injected on a size-exclusion Superose 6 10/300GL increase column equilibrated with buffer C supplemented with 0.03 mg.mL⁻¹ LMNG and 0.003 mg.mL⁻¹ CHS. The purified complex was then concentrated to 0.8-5 mg.mL⁻¹ and incubated with PI(4,5)P₂ 0.0015 mg.mL⁻¹ and respectively with 1 mM AMPPCP / 3 mM ADP + 2 mM AlCl₃ + 10 mM NaF / 2 mM AlCl₃ + 10 mM NaF / 1 mM Na₃VO₄ + 0.003 mg.mL⁻¹ POPC / 1 mM Na₃VO₄ + 0.003 mg.mL⁻¹ POPS / 1 mM Na₃VO₄ + 0.003 mg.mL⁻¹ soy PI for 1 h on ice. For the E2P_active state: the C-terminally truncated construct was incubated with 0.01 mg.mL⁻¹ POPC and 0.002 mg.mL⁻¹ PI(3,4,5)P₃ during elution from the streptavidin resin by TEV protease. The sample was then injected on a size-exclusion Superose 6 10/300GL increase column equilibrated with buffer C supplemented with 0.003 mg.mL⁻¹ POPC and 0.0015 mg.mL⁻¹ PI(3,4,5)P₃. The purified complex was then concentrated to 5 mg.mL⁻¹ and incubated with 1 mM BeSO₄ and 4 mM KF for 1 h on ice. For the E2P_{autoinhibited} states: the WT full-length construct was incubated with 0.002 mg.mL⁻¹ PI(3,4,5)P₃ during elution from the streptavidin resin by TEV. After concentration the sample was incubated with 2 mM ATP. The sample was then injected on a size-exclusion Superose 6 10/300GL increase column equilibrated with buffer C supplemented with 0.0015 mg.mL⁻¹ PI(3,4,5)P₃. For this sample, CHS was omitted from all steps of the purification as we initially hypothesized that CHS might compete with phosphoinositides. The purified complex was then concentrated to 3 mg.mL⁻¹. In all cases, 3 µL of the final sample was added to freshly glow-discharged (45 s at 15 mA) C-flat Holey Carbon grids, CF-1.2/1.3–4 C (Protochips), which were subsequently vitrified at 4 °C and 100% humidity on a Vitrobot IV (Thermo Fisher Scientific).

Cryo-EM data collection

All data were collected on a Titan Krios G3i (EMBION Danish National cryo-EM Facility – Aarhus node) with X-FEG operated at 300 kV and equipped with a Gatan K3 camera and a Bioquantum energy filter using a slit width of 20 eV. Movies were collected using aberration-free image shift data collection (AFIS) in EPU (Thermo Fisher Scientific) as 1.5 second exposures in super-resolution mode at a calibrated physical pixel size of 0.647 Å/pixel (magnification of 130,000x) with a total electron dose of 60 e^-/Å².

Cryo-EM data processing

For each data set a detailed procedure of the data processing workflow is available in Supplementary Figs. 3–11. Briefly, processing was performed in cryoSPARC⁵² (v3). Patch Motion Correction and Patch CTF were performed before low-quality micrographs (e.g. micrographs with crystalline ice, high motion) were discarded. Particles were initially picked using template picking on all movies. Particles were extracted in a 384 or 416-pixel box and Fourier cropped four times to a 96- or 104-pixel box (2.59 Å/pixel). Ab initio references were produced using a subset of all particles selected from 2D class. One protein-like reference and multiple junk references were used in multiple rounds of heterogeneous refinement. Selected particles were then re-extracted in an appropriate pixel box depending on the maximum resolution achieved with non-uniform refinement with a first re-extraction in a 208, 240 or 256-pixel box. The final particle stack was used to perform a final non-uniform refinement combined with per-particle CTF refinement⁵³. The resulting map was used to generate a mask to run 3D variability analysis to investigate sample flexibility and heterogeneity⁵⁴. For the two E2P autoinhibited states, conformational heterogeneity was detected, and the two final particles stack were subjected to a final non-uniform refinement combined with per-particle CTF refinement.

Model building

The ATP8B1-CDC50A models were built using the previously published structure of the complex (PDB: 7PY4) as templates. The cryo-EM maps were sharpened using the Autosharpen tool in PHENIX⁵⁵. The models were manually generated and relevant ligands were added with COOT⁵⁶ before real-space refinement in PHENIX⁵⁷. For real space refinement, hydrogen addition and ligand restraints were generated by the ReadySet tool in PHENIX. Model validation was performed using MolProbity⁵⁸ in PHENIX, and relevant metrics are listed in Supplementary Table 1.

Molecular dynamic simulation and analysis

The starting structures of the ATP8B1/CDC50A complex for the MD simulations were based on the cryo-EM model of the E2P_{autoinhibited} “closed” state (PDB:8OX7). We modeled missing loops (28–62, 72–105, 714–786, and 814–832) using MODELLER⁵⁹ (v10.1). The template for loop 2 (72–105) was based on the corresponding AlphaFold2 model⁶⁰ because of the presence of a local hairpin structure with reasonable confidence and confirmed by the density observed in the E2P-active conformation. We modeled the phosphorylated D454 with two negative charges and parameterized it as previously described⁶¹. We kept the Mg²⁺ ion and three disulfide bonds in CDC50A (C91-C104, C94-C102, and C157-C171) but removed glycans. E230 and E843 were protonated and K957 was deprotonated based on predictions using PROPKA3⁶² to match the experimental conditions at pH 7. The IP9 molecule in the cryo-EM structure was replaced by a PI or PI(3,4,5)P₃, modeled as a SAPI (18:0/20:4) and SAPI34 (18:0/20:4), respectively. We set up the system using the CHARMM-GUI Membrane Builder web tool⁶³ and parameterized the system with the Charmm36m force field⁶⁴ in conjunction with the TIP3 water model⁶⁵ using the CHARMM-GUI Input Generator⁶⁶. The protein complex was embedded in a bilayer, and we chose the lipid composition of the cytosolic (inner/lower) leaflet and luminal (outer/upper) leaflet to mimic a mammalian plasma membrane^67,68. The lower leaflet consists of 15% cholesterol (CHOL), 20% phosphatidylcholine (13% PLPC and 7% POPC), 33% phosphatidylethanolamine (YOPE), 23% phosphatidic acid (YOPA), 3% phosphatidylserine (POPS), and 6.0% phosphatidylinositol (3% SAPI, 1% SAPI14, 1% SAPI2D, and 1% SAPI34). The upper leaflet is composed of 39% cholesterol (CHOL), 46% phosphatidylcholine (31% PLPC and 15% POPC), and 15% sphingomyelin (SM). The membrane spans an area of 14 × 14 nm, totaling about 300 lipids in each leaflet. We solvated the systems with about 92,000 water molecules, neutralized the systems, and matched experimental ion concentrations of 100 mM KCl and 5 mM MgCl₂ with corresponding ions. The total atom number was around 370,000 and the equilibrated box dimensions in the x, y, and z dimension were 13.5 nm, 13.5 nm, and 19.9 nm, respectively.

We minimized the systems with 5000 steps of steepest descent and positional restraints on protein backbone side chain atoms and lipid atoms. We performed two steps of equilibration in the NVT ensemble (250 ps each) using the Berendsen thermostat⁶⁹ and a constant temperature of 310 K. This was followed by 4 more steps of equilibration in the NPT ensemble of 500 ps each by introducing semi-isotropic pressure coupling using the Berendsen barostat. Positional restraints were gradually released during equilibration steps. We performed the production run in the NPT ensemble at 310 K using the Noose-Hoover thermostat and a coupling constant of 1 ps. The pressure was kept constant at 1 bar with semi-isotropic pressure coupling using the Parrinello-Rahman barostat⁷⁰, a 5 ps time coupling constant, and an isothermal compressibility of 4.5.10⁻⁵bar ⁻¹. We applied a 1.2 nm cutoff for the van der Waals interactions using a switching function from 1.0 nm and a cutoff of 1.2 nm for short-range electrostatic interactions while long-range electrostatics were treated by Particle-mesh Ewald summation⁷¹. The Verlet algorithm was used to generate a pair list with buffer and the time step for integration was 2 fs. For each system (PI bound and PI(3,4,5)P₃-bound), we performed a set of five 600 ns long production runs. Finally, we analyzed the last 200 ns of each simulation, pooled the five replicates, and used a frame every 1 ns. All simulations were run using GROMACS⁷² (v2020/3). Analysis was performed using GROMACS v 2019/4 and python using the libraries MDAnalysis^73,74 and GetContacts (https://getcontacts.github.io/). To calculate relative positions of the inositol ring, we rotated and translated the trajectory by aligning Cɑ atoms of TM residues in proximity to the lipid-binding site and used the COM of the corresponding atoms as a reference. The binding poses were clustered using the gromos algorithm³⁵ based on the C atoms in the inositol ring using a cutoff of 0.1 nm.

ATPase activity

The rate of ATP hydrolysis was monitored continuously using an enzyme-coupled assay, coupling ATP hydrolysis to a drop in NADH absorption at 340 nm. ATPase activity was measured at 37 °C in buffer C supplemented with 1 mM ATP, 1 mM phosphoenolpyruvate, 0.04 mg.mL⁻¹ pyruvate kinase, 0.1 mg.mL⁻¹ lactate dehydrogenase, 370 µM NADH, 1 mg.mL⁻¹ DDM (2 mM), 0.025 mg.mL⁻¹ PI(3,4,5)P₃ and variable amounts of other phospholipids. In these experiments, the purified C-terminally truncated ATP8B1-CDC50A complex was added at a final concentration 2.3 µg.mL⁻¹. The NADH absorption decay was measured in a 96-well plate with a SpectraMax®i3 microplate reader in kinetic mode. Conversion from NADH oxidation rates expressed in mAU.s⁻¹ to ATPase activities expressed in µmol.min⁻¹.mg⁻¹ was based on the extinction coefficient of NADH at 340 nm.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Cryo-EM density maps have been deposited in the Electron Microscopy Data Bank under the accession codes EMD-17256 (E1-ATP), EMD-17257 (E1P-ADP), EMD-17258 (E1P), EMD-17259 (E2P_{autoinhibited} “closed”), EMD-17260 (E2P_{autoinhibited} “open”), EMD-17261 (E2P_active), EMD-17262 (E2┅Pi (PS)), EMD-17262 (E2┅Pi (PC)) and EMD-17263 (E2┅Pi (PI)). Atomic coordinates have been deposited in the Protein Data Bank under IDs 8OX4 [https://doi.org/10.2210/pdb8OX4/pdb] (E1-ATP), 8OX5 [https://doi.org/10.2210/pdb8OX5/pdb] (E1P-ADP), 8OX6 [https://doi.org/10.2210/pdb8OX6/pdb] (E1P), 8OX7 [https://doi.org/10.2210/pdb8OX7/pdb] (E2P_{autoinhibited} “closed”), 8OX8 [https://doi.org/10.2210/pdb8OX8/pdb] (E2P_{autoinhibited} “open”), 8OX9 [https://doi.org/10.2210/pdb8OX9/pdb] (E2P_active), 8OXA [https://doi.org/10.2210/pdb8OXA/pdb] (E2---Pi (PS)), 8OXA (E2┅Pi (PC)) and 8OXC [https://doi.org/10.2210/pdb8OXC/pdb] (E2┅Pi (PI). MD data and analysis have been deposited to the University of Copenhagen Electronic Research Data Archive (ERDA) and are available from: https://doi.org/10.17894/ucph.44e191c6-97ad-43ef-944f-be2ad95329fd. The data underlying Figs. 4a, 6b, and 7c are provided as a Source Data file. Source data are provided with this paper.

References

Clarke, R. J., Hossain, K. R. & Cao, K. Physiological roles of transverse lipid asymmetry of animal membranes. Biochim. Biophys. Acta BBA 1862, 183382 (2020).
Article CAS Google Scholar
Doktorova, M., Symons, J. L. & Levental, I. Structural and functional consequences of reversible lipid asymmetry in living membranes. Nat. Chem. Biol. 16, 1321–1330 (2020).
Article CAS PubMed PubMed Central Google Scholar
Andersen, J. P. et al. P4-ATPases as phospholipid flippases—structure, function, and enigmas. Front. Physiol. 7, 275 (2016).
Scortecci, J. F. et al. Cryo-EM structures of the ABCA4 importer reveal mechanisms underlying substrate binding and Stargardt disease. Nat. Commun. 12, 5902 (2021).
Palmgren, M. G. & Nissen, P. P-type ATPases. Annu. Rev. Biophys. 40, 243–266 (2011).
Article CAS PubMed Google Scholar
Vestergaard, A. L. et al. Critical roles of isoleucine-364 and adjacent residues in a hydrophobic gate control of phospholipid transport by the mammalian P4-ATPase ATP8A2. Proc. Natl Acad. Sci. 111, E1334–E1343 (2014).
Article CAS PubMed PubMed Central Google Scholar
Hiraizumi, M., Yamashita, K., Nishizawa, T. & Nureki, O. Cryo-EM structures capture the transport cycle of the P4-ATPase flippase. Science 365, 1149–1155 (2019).
Article ADS CAS PubMed Google Scholar
Albers, R. W. Biochemical aspects of active transport. Annu. Rev. Biochem. 36, 727–756 (1967).
Article CAS PubMed Google Scholar
Post, R. L., Hegyvary, C. & Kume, S. Activation by adenosine triphosphate in the phosphorylation kinetics of sodium and potassium ion transport adenosine triphosphatase. J. Biol. Chem. 247, 6530–6540 (1972).
Article CAS PubMed Google Scholar
Olesen, C., Sørensen, T. L.-M., Nielsen, R. C., Møller, J. V. & Nissen, P. Dephosphorylation of the calcium pump coupled to counterion occlusion. Science 306, 2251–2255 (2004).
Article ADS CAS PubMed Google Scholar
Coleman, J. A., Vestergaard, A. L., Molday, R. S., Vilsen, B. & Andersen, J. P. Critical role of a transmembrane lysine in aminophospholipid transport by mammalian photoreceptor P4-ATPase ATP8A2. Proc. Natl Acad. Sci. USA 109, 1449–1454 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Timcenko, M. et al. Structural basis of substrate-independent phosphorylation in a P4-ATPase lipid flippase. J. Mol. Biol. 433, 167062 (2021).
Article CAS PubMed Google Scholar
Lenoir, G., Williamson, P., Puts, C. F. & Holthuis, J. C. M. Cdc50p plays a vital role in the ATPase reaction cycle of the putative aminophospholipid transporter Drs2p. J. Biol. Chem. 284, 17956–17967 (2009).
Article CAS PubMed PubMed Central Google Scholar
Bryde, S. et al. CDC50 proteins are critical components of the human class-1 P4-ATPase transport machinery. J. Biol. Chem. 285, 40562–40572 (2010).
Article CAS PubMed PubMed Central Google Scholar
Segawa, K., Kurata, S. & Nagata, S. The CDC50A extracellular domain is required for forming a functional complex with and chaperoning phospholipid flippases to the plasma membrane. J. Biol. Chem. 293, 2172–2182 (2018).
Article CAS PubMed Google Scholar
Bull, L. N. et al. A gene encoding a P-type ATPase mutated in two forms of hereditary cholestasis. Nat. Genet. 18, 219–224 (1998).
Article CAS PubMed Google Scholar
Groen, A. et al. Complementary functions of the flippase ATP8B1 and the floppase ABCB4 in maintaining canalicular membrane integrity. Gastroenterology 141, 1927–1937.e1–4 (2011).
Article CAS PubMed Google Scholar
Dumitrescu, L. et al. Genetic variants and functional pathways associated with resilience to Alzheimer’s disease. Brain 143, 2561–2575 (2020).
Article PubMed PubMed Central Google Scholar
Holstege, H. et al. Exome sequencing identifies rare damaging variants in ATP8B4 and ABCA1 as risk factors for Alzheimer’s disease. Nat. Genet. 54, 1786–1794 (2022).
Article CAS PubMed PubMed Central Google Scholar
Paulusma, C. C. et al. ATP8B1 requires an accessory protein for endoplasmic reticulum exit and plasma membrane lipid flippase activity. Hepatol. Baltim. 47, 268–278 (2008).
Article CAS Google Scholar
Muñoz-Martínez, F., Torres, C., Castanys, S. & Gamarro, F. CDC50A plays a key role in the uptake of the anticancer drug perifosine in human carcinoma cells. Biochem. Pharmacol. 80, 793–800 (2010).
Article PubMed Google Scholar
Ray, N. B. et al. Dynamic regulation of cardiolipin by the lipid pump Atp8b1 determines the severity of lung injury in experimental pneumonia. Nat. Med. 16, 1120–1127 (2010).
Article CAS PubMed PubMed Central Google Scholar
Takatsu, H. et al. Phospholipid flippase activities and substrate specificities of human type IV P-type ATPases localized to the plasma membrane. J. Biol. Chem. 289, 33543–33556 (2014).
Article CAS PubMed Google Scholar
Segawa, K., Kurata, S. & Nagata, S. Human type IV P-type ATPases that work as plasma membrane phospholipid flippases and their regulation by caspase and calcium. J. Biol. Chem. 291, 762–772 (2016).
Article CAS PubMed Google Scholar
Mizutani, A. et al. Assessment of adenosine triphosphatase phospholipid transporting 8B1 (ATP8B1) function in patients with cholestasis with ATP8B1 deficiency by using peripheral blood monocyte-derived macrophages. Hepatol. Commun. 5, 52–62 (2021).
Article CAS PubMed Google Scholar
Dieudonné, T. et al. Autoinhibition and regulation by phosphoinositides of ATP8B1, a human lipid flippase associated with intrahepatic cholestatic disorders. eLife 11, e75272 (2022).
Article PubMed PubMed Central Google Scholar
Cheng, M.-T. et al. Structural insights into the activation of autoinhibited human lipid flippase ATP8B1 upon substrate binding. Proc. Natl Acad. Sci. USA 119, e2118656119 (2022).
Article CAS PubMed PubMed Central Google Scholar
Jidenko, M., Lenoir, G., Fuentes, J. M., le Maire, M. & Jaxel, C. Expression in yeast and purification of a membrane protein, SERCA1a, using a biotinylated acceptor domain. Protein Expr. Purif. 48, 32–42 (2006).
Article CAS PubMed Google Scholar
Klomp, L. W. J. et al. Characterization of mutations in ATP8B1 associated with hereditary cholestasis. Hepatology 40, 27–38 (2004).
Article ADS CAS PubMed Google Scholar
Nakanishi, H. et al. Transport cycle of plasma membrane flippase ATP11C by Cryo-EM. Cell Rep. 32, 108208 (2020).
Article CAS PubMed Google Scholar
Sim, S. I., von Bülow, S., Hummer, G. & Park, E. Structural basis of polyamine transport by human ATP13A2 (PARK9). Mol. Cell 81, 4635–4649.e8 (2021).
Article CAS PubMed PubMed Central Google Scholar
Timcenko, M. et al. Structure and autoregulation of a P4-ATPase lipid flippase. Nature 571, 366–370 (2019).
Article CAS PubMed Google Scholar
Xu, J., He, Y., Wu, X. & Li, L. Conformational changes of a phosphatidylcholine flippase in lipid membranes. Cell Rep. 38, 110518 (2022).
Article CAS PubMed Google Scholar
Tillinghast, J., Drury, S., Bowser, D., Benn, A. & Lee, K. P. K. Structural mechanisms for gating and ion selectivity of the human polyamine transporter ATP13A2. Mol. Cell 81, 4650–4662.e4 (2021).
Article CAS PubMed Google Scholar
Daura, X. et al. Peptide folding: when simulation meets experiment. Angew. Chem. Int. Ed. 38, 236–240 (1999).
Article CAS Google Scholar
Nakanishi, H. et al. Crystal structure of a human plasma membrane phospholipid flippase. J. Biol. Chem. 295, 10180–10194 (2020).
Article CAS PubMed PubMed Central Google Scholar
Clausen, J. D. et al. Crystal structure of the vanadate-inhibited Ca(2+)-ATPase. Structure 24, 617–623 (2016).
Article CAS PubMed Google Scholar
Jacquot, A. et al. Phosphatidylserine stimulation of Drs2p Cdc50p lipid translocase dephosphorylation is controlled by phosphatidylinositol-4-phosphate. J. Biol. Chem. 287, 13249–13261 (2012).
Article CAS PubMed PubMed Central Google Scholar
Tadini-Buoninsegni, F., Mikkelsen, S. A., Mogensen, L. S., Molday, R. S. & Andersen, J. P. Phosphatidylserine flipping by the P4-ATPase ATP8A2 is electrogenic. Proc. Natl Acad. Sci. USA 116, 16332–16337 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Tadini-Buoninsegni, F. et al. Electrogenic reaction step and phospholipid translocation pathway of the mammalian P4-ATPase ATP8A2. FEBS Lett. 597, 495–503 (2023).
Article CAS PubMed Google Scholar
van Veen, S. et al. ATP13A2 deficiency disrupts lysosomal polyamine export. Nature 578, 419–424 (2020).
Article ADS PubMed Google Scholar
McKenna, M. J. et al. The endoplasmic reticulum P5A-ATPase is a transmembrane helix dislocase. Science 369, eabc5809 (2020).
Article CAS PubMed PubMed Central Google Scholar
Tomita, A. et al. Cryo-EM reveals mechanistic insights into lipid-facilitated polyamine export by human ATP13A2. Mol. Cell 81, 4799–4809.e5 (2021).
Article CAS PubMed PubMed Central Google Scholar
Dickson, E. J. & Hille, B. Understanding phosphoinositides: rare, dynamic, and essential membrane phospholipids. Biochem. J. 476, 1–23 (2019).
Article CAS PubMed Google Scholar
Azouaoui, H. et al. Coordinated overexpression in yeast of a P4-ATPase and its associated Cdc50 subunit: the case of the Drs2p/Cdc50p lipid flippase complex. In: P-Type ATPases: Methods and Protocols (ed. Bublitz, M.) 37–55 (Springer, 2016) https://doi.org/10.1007/978-1-4939-3179-8_6.
Azouaoui, H. et al. High phosphatidylinositol 4-phosphate (PI4P)-dependent ATPase activity for the Drs2p-Cdc50p flippase after removal of its N- and C-terminal extensions. J. Biol. Chem. 292, 7954–7970 (2017).
Article CAS PubMed PubMed Central Google Scholar
Wang, J. et al. Proteomic analysis and functional characterization of P4-ATPase phospholipid flippases from murine tissues. Sci. Rep. 8, 10795 (2018).
Article ADS PubMed PubMed Central Google Scholar
Fellmann, P. et al. Transmembrane movement of diether phospholipids in human erythrocytes and human fibroblasts. Biochemistry 39, 4994–5003 (2000).
Article CAS PubMed Google Scholar
Slotte, J. P. The importance of hydrogen bonding in sphingomyelin’s membrane interactions with co-lipids. Biochim. Biophys. Acta 1858, 304–310 (2016).
Article CAS PubMed Google Scholar
Honsho, M., Mawatari, S. & Fujiki, Y. ATP8B2-mediated asymmetric distribution of plasmalogens regulates plasmalogen homeostasis and plays a role in intracellular signaling. Front. Mol. Biosci. 9, 915457 (2022).
Article CAS PubMed PubMed Central Google Scholar
Montigny, C. et al. Slow phospholipid exchange between a detergent-solubilized membrane protein and lipid-detergent mixed micelles: brominated phospholipids as tools to follow its kinetics. PLOS One 12, e0170481 (2017).
Article PubMed PubMed Central Google Scholar
Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat. Methods 14, 290–296 (2017).
Article CAS PubMed Google Scholar
Punjani, A., Zhang, H. & Fleet, D. J. Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction. Nat. Methods 17, 1214–1221 (2020).
Article CAS PubMed Google Scholar
Punjani, A. & Fleet, D. J. 3D variability analysis: resolving continuous flexibility and discrete heterogeneity from single particle cryo-EM. J. Struct. Biol. 213, 107702 (2021).
Article CAS PubMed Google Scholar
Terwilliger, T. C., Sobolev, O. V., Afonine, P. V. & Adams, P. D. Automated map sharpening by maximization of detail and connectivity. Acta Crystallogr. Sect. Struct. Biol. 74, 545–559 (2018).
Article ADS CAS Google Scholar
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D. Biol. Crystallogr. 66, 486–501 (2010).
Article ADS CAS PubMed PubMed Central Google Scholar
Afonine, P. V. et al. Real-space refinement in PHENIX for cryo-EM and crystallography. Acta Crystallogr. Sect. Struct. Biol. 74, 531–544 (2018).
Article ADS CAS Google Scholar
Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D. Biol. Crystallogr. 66, 12–21 (2010).
Article ADS CAS PubMed Google Scholar
Webb, B. & Sali, A. Comparative protein structure modeling using MODELLER. Curr. Protoc. Bioinformatics 54, 5.6.1−5.6.37 (2016).
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Das, A., Rui, H., Nakamoto, R. & Roux, B. Conformational transitions and alternating-access mechanism in the sarcoplasmic reticulum calcium pump. J. Mol. Biol. 429, 647–666 (2017).
Article CAS PubMed PubMed Central Google Scholar
Olsson, M. H. M., Søndergaard, C. R., Rostkowski, M. & Jensen, J. H. PROPKA3: consistent treatment of internal and surface residues in empirical pKa predictions. J. Chem. Theory Comput. 7, 525–537 (2011).
Article CAS PubMed Google Scholar
Wu, E. L. et al. CHARMM-GUI Membrane Builder toward realistic biological membrane simulations. J. Comput. Chem. 35, 1997–2004 (2014).
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
Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935 (1983).
Article ADS CAS Google Scholar
Lee, J. et al. CHARMM-GUI Input Generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM simulations using the CHARMM36 additive force field. J. Chem. Theory Comput. 12, 405–413 (2016).
Article CAS PubMed Google Scholar
Lorent, J. H. et al. Plasma membranes are asymmetric in lipid unsaturation, packing and protein shape. Nat. Chem. Biol. 16, 644–652 (2020).
Article CAS PubMed PubMed Central Google Scholar
van Meer, G., Voelker, D. R. & Feigenson, G. W. Membrane lipids: where they are and how they behave. Nat. Rev. Mol. Cell Biol. 9, 112–124 (2008).
Article PubMed PubMed Central Google Scholar
Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F., DiNola, A. & Haak, J. R. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81, 3684–3690 (1984).
Article ADS CAS Google Scholar
Parrinello, M. & Rahman, A. Polymorphic transitions in single crystals: a new molecular dynamics method. J. Appl. Phys. 52, 7182–7190 (1981).
Article ADS CAS Google Scholar
Essmann, U. et al. A smooth particle mesh Ewald method. J. Chem. Phys. 103, 8577–8593 (1995).
Article ADS CAS 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
Michaud-Agrawal, N., Denning, E. J., Woolf, T. B. & Beckstein, O. MDAnalysis: a toolkit for the analysis of molecular dynamics simulations. J. Comput. Chem. 32, 2319–2327 (2011).
Article CAS PubMed PubMed Central Google Scholar
Gowers, R. J. et al. MDAnalysis: a python package for the rapid analysis of molecular dynamics simulations. In: Proc. 15th Python Sci. Conf. 98–105, https://doi.org/10.25080/Majora-629e541a-00e (2016).

Download references

Acknowledgements

We thank Thomas Boesen, Andreas Bøggild, and Taner Drace for technical support during EM data collection at the EMBION Danish National cryo-EM facility of Aarhus University (5072-00025B, Danish Agency for Research and Higher Education) and Jesper Lykkegaard Karlsen for scientific computing support. We are grateful to Anna Marie Nielsen, Tanja Klymchuk, and Bente Andersen for their technical assistance. We also wish to thank, Tomás Heger, Line Marie Christiansen, Ronja Driller, Cédric Montigny, Christine Jaxel, and David Stokes for discussion and advice. T.D. also warmly thank Camille and Côme for their patience during the revision process. This work was supported by a Marie Skłodowska-Curie Actions Individual Fellowship — LivFlip (101024542) and by a Fondation Recherche Médicale (ARF202209015714) grant to T.D., by Engineering and Physical Sciences Research Council grants (EP/X035603 and EP/V030779) to S. K., by an ANR grant (ANR-14-CE09-0022) to G.L., by the French Infrastructure for Integrated Structural Biology (FRISBI; ANR-10-INSB-05) to G.L., by the Center National de la Recherche Scientifique (CNRS) to G.L., by a Lundbeckfonden Fellow grant (R335-2019-2053) to J.A.L, by the Lundbeck Foundation to the BRAINSTRUC structural biology initiative (155-2015-2666) to K.L.-L. and P.N., and by a Lundbeckfonden Professorship grant (R310-2018-3713) to P.N.

Author information

Thibaud Dieudonné
Present address: Université Paris-Saclay, CEA, CNRS, Institute for Integrative Biology of the Cell (I2BC), 91198, Gif-sur-Yvette, France

Authors and Affiliations

DANDRITE, Nordic EMBL Partnership for Molecular Medicine, Department of Molecular Biology and Genetics, Aarhus University, Aarhus, Denmark
Thibaud Dieudonné, Michelle Juknaviciute Laursen, Charlott Stock, Rasmus Kock Flygaard & Poul Nissen
Structural Biology and NMR Laboratory & Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark
Felix Kümmerer & Kresten Lindorff-Larsen
Department of Biochemistry, University of Oxford, Oxford, UK
Syma Khalid
Université Paris-Saclay, CEA, CNRS, Institute for Integrative Biology of the Cell (I2BC), 91198, Gif-sur-Yvette, France
Guillaume Lenoir
Department of Molecular Biology and Genetics, Aarhus University, Aarhus, Denmark
Joseph A. Lyons
Interdisciplinary Nanoscience Centre (iNANO) Aarhus University, Aarhus, Denmark
Joseph A. Lyons

Authors

Thibaud Dieudonné
View author publications
You can also search for this author in PubMed Google Scholar
Felix Kümmerer
View author publications
You can also search for this author in PubMed Google Scholar
Michelle Juknaviciute Laursen
View author publications
You can also search for this author in PubMed Google Scholar
Charlott Stock
View author publications
You can also search for this author in PubMed Google Scholar
Rasmus Kock Flygaard
View author publications
You can also search for this author in PubMed Google Scholar
Syma Khalid
View author publications
You can also search for this author in PubMed Google Scholar
Guillaume Lenoir
View author publications
You can also search for this author in PubMed Google Scholar
Joseph A. Lyons
View author publications
You can also search for this author in PubMed Google Scholar
Kresten Lindorff-Larsen
View author publications
You can also search for this author in PubMed Google Scholar
Poul Nissen
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

T.D. and P.N. conceived the project. T.D., F.K., K.L.-L., and P.N. designed the research. T.D. and F.K. performed the experiments. T.D., F.K., M.J.L, C.S., R.K.F., S.K., G.L., J.A.L., K.L.-L., and P.N. analyzed the data. T.D. drafted the manuscript. All authors contributed to the manuscript.

Corresponding authors

Correspondence to Thibaud Dieudonné or Poul Nissen.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Eunyong Park, Benoit Roux, 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

Reporting Summary

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 license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license 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 license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Dieudonné, T., Kümmerer, F., Laursen, M.J. et al. Activation and substrate specificity of the human P4-ATPase ATP8B1. Nat Commun 14, 7492 (2023). https://doi.org/10.1038/s41467-023-42828-9

Download citation

Received: 12 April 2023
Accepted: 23 October 2023
Published: 18 November 2023
DOI: https://doi.org/10.1038/s41467-023-42828-9

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.