Molecular insights into human phosphatidylserine synthase 1 reveal its inhibition promotes LDL uptake

Abstract

In mammalian cells, two phosphatidylserine (PS) synthases drive PS synthesis. Gain-of-function mutations in the Ptdss1 gene lead to heightened PS production, causing Lenz-Majewski syndrome (LMS). Recently, pharmacological inhibition of PSS1 has been shown to suppress tumorigenesis. Here, we report the cryo-EM structures of wild-type human PSS1 (PSS1^WT), the LMS-causing Pro269Ser mutant (PSS1^P269S), and PSS1^WT in complex with its inhibitor DS55980254. PSS1 contains 10 transmembrane helices (TMs), with TMs 4-8 forming a catalytic core in the luminal leaflet. These structures revealed a working mechanism of PSS1 akin to the postulated mechanisms of the membrane-bound O-acyltransferase family. Additionally, we showed that both PS and DS55980254 can allosterically inhibit PSS1 and that inhibition by DS55980254 activates the SREBP pathways, thus enhancing the expression of LDL receptors and increasing cellular LDL uptake. This work uncovers a mechanism of mammalian PS synthesis and suggests that selective PSS1 inhibitors have the potential to lower blood cholesterol levels.

Keywords: DS55980254; LDL; LDL receptors; MBOAT; PSS1; cholesterol trafficking; phosphatidylserine.

Figure 1. Functional characterization and overall structure of human PSS1

(A) Schematic of PS synthesis. The structures of serine, choline and ethanolamine are shown. (B) The activity of PSS1^WT and PSS1^P269S. Data are mean ± S.D. (n=3). P values (two-sided) were calculated by Student’s t tests using GraphPad Prism v.9; **, P<0.01; ****, P<0.0001. (C) DS55980254 (termed as “DS”) inhibits the activity of PSS1^WT and PSS1^P269S in vitro. Data are mean ± S.D. (n=3). The 2D chemical structure of DS is shown. The IC₅₀ of DS to PSS1^WT is 1.65 μM (the deviation range is 1.22–2.26 μM). The IC₅₀ of DS to PSS1^P269S is 1.27 μM (the deviation range is 0.87–1.87 μM). (D) Overall structure showing PSS1^WT dimer viewed from the side of the membrane. The cryo-EM map of the putative PE is shown. (E) View of the dimer from the cytosol. The cryo-EM map of the putative PI is shown. (F) View of the dimer from the lumen. TMs1-3 and TM9 are located at the interface and are involved in dimer assembly. TM5 and TM6 contribute to engage the calcium (green sphere). (G) Structural details of the calcium binding site. (H) Structural details of PI-mediated dimeric interface. The putative PC (gray sticks), PS (yellow sticks), PI (red sticks) and PE (cyan sticks) are shown. The interactions between calcium and residues are indicated by dashed lines. R349 from another protomer is underlined. See also Figures S1 and S2.

Figure 2. The cytosolic PS-binding sites

(A) Luminal view of two PS-binding sites. One PSS1 monomer is colored in blue, and the other is colored in cyan. The PS molecules are shown by yellow sticks. The residues that contribute to the rigidity of the sites are indicated. Pro269 is highlighted in red. PSs are only labeled in one subunit, but present in both subunits with C2 symmetry. (B) Electrostatic surface representation of the cytosolic PS-binding sites. (C) Interaction details of the PS1-binding site. The cryo-EM map of PS1 is shown. (D) Interaction details of the PS2-binding site. Residues are represented as sticks; dashed lines represent hydrophilic interactions with the distance. The cryo-EM map of PS2 is shown. See also Figure S2.

Figure 3. Overall structure of PSS1^P269S

(A) Overall structure showing PSS1^P269S dimer viewed from the side of the membrane. (B) The comparison of TMs in PSS1^P269S (yellow, dark green ball) and PSS1^WT (blue, light green ball). (C) R.M.S.F. of linker between PH1 and TM1 (upper) and PH4 (lower) in PSS1^WT with (blue) and without PS binding (green). Data of each residue is an average of two monomers in three parallel groups: $\sqrt{\frac{{\sum }_{\text{traj}=1}^3{\sum }_{\text{monomer}=1}^2{(RMS{F}_{\text{monomer},\text{traj}})}^2}{6}}$ (D) Luminal view of PSS1^P269S compared to PSS1^WT. The conformational changes are indicated by arrows. (E) Structural comparison of the entrances of the phospholipid substrate. (F) Structural comparison of the linker between TM5 and TM6. (G) Changes in the distance between representative amino acid residues (C_α) and the calcium ion with and without PS binding (from 200 ns MD). ΔDistance=Distance in PSS^WT without PS binding – Distance in PSS^WT with PS binding. Data is an average of three parallel groups. Each group contains two PSS1 molecules in one homodimer. See also Figures S1, S2 and S4.

Figure 4. The catalytic mechanism of PSS1

(A) and (B) Electrostatic surface representation of catalytic cavity of PSS1^WT (A in blue) and PSS1^P269S (B in yellow). (C) The docking model of serine (gray sticks) in the calcium binding site of PSS1^P269S. (D) A working model of PSS1-meidated PS synthesis. (E) Functional validation of the catalytic residue and lipid entrance. Data are mean ± S.D. (n=3). (F) Salt bridge between Lys179 and Glu301 induces a shift of Phe168. Residues are represented as sticks; dashed lines represent hydrophilic interactions. See also Figures S1, S2 and S7.

Figure 5. The inhibition of PSS1 by its specific inhibitor DS55980254

(A) Overall structure of DS55980254 (labeled as “DS”) bound PSS1^WT dimer viewed from the side of the membrane. The cryo-EM map of DS (sticks) is shown. The residues 245–275 were colored in magenta. (B) Interaction details between DS and PSS1. Dashed lines represent hydrophilic interactions. (C) Functional validation of the DS55980254-binding residues. Data are mean ± S.D. (n=3 independent experiments). P value (two-sided) was calculated by Student’s t tests using GraphPad Prism v.9; ns, P > 0.05. (D) The comparison of TM7 and TM8 between the inhibitor-bound PSS1^WT (light blue) and apo PSS1^WT (blue). The conformational changes are indicated by red arrows. (E) R.M.S.F. of residues 165–300 in PSS1^WT with DS (yellow), PS (blue) and without DS and PS (green) during 200 ns MD simulations. Data of each residue is an average of two monomers in three parallel groups: $\sqrt{\frac{{\sum }_{\text{traj}=1}^3{\sum }_{\text{monomer}=1}^2{(RMS{F}_{\text{monomer},\text{traj}})}^2}{6}}$ See also Figures S1, S2 and S5.

Figure 6. PSS1 deficiency triggers the SREBP-2 cleavage and reduces the ER cholesterol concentration

(A) The SREBP-2 processing with and without DS55980254 (labeled as “DS”) in Ptdss1^−/− CHO-K1 cells. On day 0, cells were set up in 6-well plates at a density of 25 × 10⁴ cells/well. On day 1, cells were switched to 1 ml medium A containing 10% FCS and 100 μl baculovirus expressed hPSS1^WT or hPSS1^H172A were supplemented into the medium. The inhibitor (DS) was supplemented into the medium at a final concentration of 1 μM at the time of infection and was kept at a constant concentration in each buffer throughout the experiment. On day 2, cells were switched to cholesterol-depletion medium A containing 1% hydroxypropyl-β-cyclodextrin (HPCD), which can deplete the free cholesterol in cells. After incubation for 1 hour, cells received cholesterol-depletion medium A in the absence or presence of 100 μg protein/ml LDL. After 6 hours, cells were harvested for immunoblotting of SREBP-2, Flag and Histone H3. P, precursor. N, nuclear. SREBP cleavage was quantified using ImageJ. For each lane, the ratio of nuclear to total SREBP-2 (nuclear + precursor) was calculated. (B) Diagram of ER membrane fractionation scheme. (C) and (D) denote major fractions recovered and analyzed by immunoblot analysis. CHO-K1 cells (C) and Ptdss1^−/− CHO-K1 cells (D) were treated according to the fractionation scheme as described in METHOD DETAILS. Aliquots representing equal volumes of each fraction (A–F) were subjected to immunoblot analysis for the indicated organelle markers. (E) Cholesterol content of the purified ER membranes in CHO-K1 cells and Ptdss1^−/− CHO-K1 cells. Lipids were extracted from the purified ER, and the amounts of cholesterol and phospholipids were quantified as described in METHOD DETAILS. Mole % of cholesterol in the ER from different cells was calculated as the ratio of cholesterol to phospholipids plus cholesterol. Data are mean ± S.D. (n=4). P value (two-sided) was calculated by Student’s t tests using GraphPad Prism v.9; **, P<0.01. (F) Working model of PSS1 inhibitors in simulating LDL uptake.

Figure 7. DS55980254 stimulates LDLR expression and LDL uptake without causing cholesterol accumulation in lysosomes or overexpression of HMGCR in the ER

(A) The SREBP-2 processing and (B) LDLR mRNA level with and without DS55980254 (labeled as “DS”) in CHO-K1 cells. On day 0, cells were set up in 6-well plates at a density of 15 × 10⁴ cells/well. On day 1, DS was supplemented into the medium at different concentration as indicated and was kept at a constant concentration in each buffer throughout the experiment. On day 2, cells were switched to cholesterol-depletion medium A containing 1% hydroxypropyl-β-cyclodextrin (HPCD). After incubation for 1 hour, the cells received cholesterol-depletion medium A in presence of 100 μg protein/ml LDL. After 18 hours, cells were harvested for immunoblotting of SREBP-2, Flag and Histone H3 and measuring LDLR mRNA level. P, precursor. N, nuclear. SREBP cleavage was quantified using ImageJ. For each lane, the ratio of nuclear to total SREBP-2 (nuclear + precursor) was calculated. Data of LDL mRNA level is presented as mean ± S.D. (n=3 biological replicates, with each point representing the average of three technical replicates). (C) The SREBP-2 processing, (D) surface LDLR expression level and (E) BODIPY FL-LDL uptake with and without DS in SV589j cells. On day 0, cells were set up in 6-well plates at a density of 25 × 10⁴ cells/well. After plating for 6 hours, cells were supplemented with DMSO (solvent control) or 1μM DS, which was kept at a constant concentration in each buffer throughout the experiment. On day 1, cells were switched to cholesterol-depletion medium B. After incubation for 16 hours, the cells received cholesterol-depletion medium B in presence of 50 μg protein/ml LDL. After 24 hours, cells were harvested for immunoblotting of SREBP-2, Flag and Histone H3 or by incubation with EDTA, washed, incubated with PE-anti-LDLR, and subjected to flow cytometry. The rest of cells were supplemented with 5 μg protein/ml BODIPY FL-LDL. After 2 hours, cells were harvested for flow cytometry. Data are mean ± S.D. (n=3). P value (two-sided) was calculated by Student’s t tests using GraphPad Prism v.9; **, P<0.01. (F) Cholesterol distribution in CHO-K1 cells with different compounds treatment. On day 0, CHO-K1 cells were plated on 12-mm glass coverslips in medium A with 5% FCS. On day 1, cells were incubated with DMSO (solvent control), 1μM DS or 1μM U18666A. Both compounds were kept at a constant concentration in each buffer throughout the experiment. On day 2, cells were switched to cholesterol-depletion medium A. After a 12-hour incubation, cells were switched to medium B containing 5% LPDS, 10 μM compactin, and 200 μM mevalonate. After 6 h, cells were imaged using filipin (Scale bar: 20 μm.). (G) Measurement of sterol synthesis. On day 0, CHO-K1 cells were set up in 6 cm plates at a density of 4 × 10⁵ cells/plate. On day 1, cells were incubated with DMSO (solvent control), 1 μM DS or 1 μM compactin (CPN). Compounds were kept at a constant concentration in each buffer throughout the experiment. On day 2, cells were switched to medium A supplemented with 5% FCS and 3 μCi/ml [¹⁴C] acetate as well as cold acetate at a final concentration of 0.1 mM. After 4 hours, cells were harvested for the metabolic labeling study and immunoblotting of HMG-CoA reductase, and Histone H3. Data are mean ± S.D. (n=3). P values (two-sided) were calculated by Student’s t tests using GraphPad Prism v.9; *, P<0.05. See also Figure S6.