Molecular basis of the inositol deacylase PGAP1 involved in quality control of GPI-AP biogenesis

The secretion and quality control of glycosylphosphatidylinositol-anchored proteins (GPI-APs) necessitates post-attachment remodeling initiated by the evolutionarily conserved PGAP1, which deacylates the inositol in nascent GPI-APs. Impairment of PGAP1 activity leads to developmental diseases in humans and fatality and infertility in animals. Here, we present three PGAP1 structures (2.66−2.84 Å), revealing its 10-transmembrane architecture and product-enzyme interaction details. PGAP1 holds GPI-AP acyl chains in an optimally organized, guitar-shaped cavity with apparent energetic penalties from hydrophobic-hydrophilic mismatches. However, abundant glycan-mediated interactions in the lumen counterbalance these repulsions, likely conferring substrate fidelity and preventing off-target hydrolysis of bulk membrane lipids. Structural and biochemical analyses uncover a serine hydrolase-type catalysis with atypical features and imply mechanisms for substrate entrance and product release involving a drawing compass movement of GPI-APs. Our findings advance the mechanistic understanding of GPI-AP remodeling.


Fig. S1 | Schematic of GPI-AP biosynthesis and PGAP1 function. a
The synthesis phase of GPI-AP biogenesis in humans and yeast.Substrates and byproducts of each reaction are indicated with grey text.Enzymes catalyzing the reactions are indicated with black text with the yeast homologs in brackets.The two steps that may be skipped in humans are indicated with a grey arrow.A question marker denotes unknown participants.b, c The remodeling phase of GPI-AP biogenesis in humans (b) and yeast (c).PGAP1/Bst1, the object of this study, is highlighted in red.a-c are redrawn based on Ref. 1 with the recent information from Refs. 2,3.A question marker denotes unknown information.d A model depicting the cross-talk between PGAP1 and the calnexin cycle during the quality control process of GPI-APs.Various components are explained in the figure.The ER membrane is shown as a thick grey line.The simplified schematic is redrawn based on the model proposed in Ref. 4 .e, f Schematic of the PGAP1-mediated ERAD pathway (e) and RESET or RESET-like pathways (f) for the degradation of misfolded GPI-APs.
7][8] .CANN, calnexin; CMP, cytidine monophosphate; CSP, C-terminal signal peptide; CW, cell wall; DAG, diacylglycerol; DolP, dolichol phosphate; ER, endoplasmic reticulum; ERAD, ER-associated degradation; EtNP, ethanolamine phosphate; FA, fatty acid; GalNAc, acetylgalactosamine; GDP, guanidine phosphate; GlcN, glucosamine; GlcNAc, acetylglucosamine; Glu, glucose; GPI-AP, glycosylphosphatidylinositol-anchored protein; Ino, inositol; MAG, monoacylglycerol; Man, mannose; PE, phosphatidylethonalamine; PM, plasma membrane; RESET, rapid ER (endoplasmic reticulum) stress-induced export; UDP, uridine diphosphate.H443N in complex with TGP2.Absorbance at 280 nm (for all proteins, black), 493 nm (for TGP2/3, green), and 585 nm (for mCherry-tagged enzyme, magenta) were simultaneously monitored.The gel image was captured using a smartphone with a portable transilluminator.g FSEC of the reaction mix from the prolonged assay (see Methods) in the absence (i) and presence (ii) of cPGAP1 S327A.FSEC profile for the sample treated with (red) and without (black) PI-PLC were recorded.h SEC and SDS-PAGE (inset) of cPGAP1 S327A in complex with TGP2.Absorbance at 280 nm (for all proteins, black), 493 nm (for TGP2/3, green), and 585 nm (for mCherry-tagged enzyme, magenta) were simultaneously monitored.The gel image was captured using a smartphone with a portable transilluminator.The brightness of the in-gel fluorescence images in f and h was adjusted as a whole to enhance visibility.Molecular weight of markers are indicated on the side of each gel.i FSEC of the supernatant fraction of the crude membranes from hPGAP1-KO cells co-expressing TGP3 with S327A (magenta) or S327A/H443N (blue) treated with (solid) or without PI-PLC (dash).The 3.1-mL peak corresponds to TGP0 release from TGP2 (the product of PGAP1) by PI-PLC.

Fig. S2 |
Fig. S2 | Sequence alignment of PGAP1 orthologs.a Sequence alignment of PGAP1 from Chaetomium thermophilum, human, and Saccharomyces cerevisiae.Asterisk, colon, and dot indicate identical, conserved, and semi-conserved substitutions, respectively.Secondary structure segments are numbered in accord with Fig. 2a, 2c, 5a, and Fig. S7a.Catalytic triad and oxyanion residues are indicated with a red triangle and a blue square, respectively.The first 130 residues of cPGAP1 and the first 44 residues of yPGAP1, which are predicted to be disordered, are not shown due to their poor sequence homology.b Sequence alignment for segments around the conserved cysteine pair (red).Protein sequences were selected from evolutionarily representative species.Uniprot IDs of the sequences and their sequence identity/similarity (%) to cPGAP1 are shown along with the binomial nomenclature and common names.Asterisk, colon, and dot indicate identical, conserved, and semi-conserved substitutions, respectively.Secondary elements were labeled for easier location of the corresponding elements in Fig.S7a.AH, amphipathic helix; TMH, transmembrane helix.

Fig. S3 |
Fig. S3 | Summary of assays.a Principal of the assays.PGAP1 deacylates GPI-AP3 to form GPI-AP2, which can be converted to the water-soluble GPI-AP0.This conversion leads to loss of surface staining in FACS, and delayed elution in FSEC.In yeast, the PGAP1 reaction leads to lethality in sec13-1 cells.b The FACS assay.The surface staining of GPI-AP3 in PGAP1-KO cells is unaffected by PI-PLC treatment unless cells are transfected with functional PGAP1 constructs.c Cellgrowth assay for PGAP1 function in yeast cells.In wildtype (WT) cells (i), PGAP1 inhibits bulk flow vesicles, and GPI-APs are exported by COP vesicles involving Sec13.The sec13-1 cells (ii) fail to assemble COP vesicles at restrictive temperatures and are, therefore, not viable.This lethality is bypassed by the loss of PGAP1 function, as bulk vesicles are no longer suppressed (iii).d Flowchart of

Fig. S5 |
Fig. S5 | Cryo-EM data processing and high-quality density/model fitting exemplary views of the wildtype cPGAP1 in lipid nanodiscs.a Representative cryo-EM micrograph and selected 2D class averages.b The workflow of classification and refinement.c Top, the nominal resolution of PGAP1 determined by the gold-standard Fourier shell correlation (FSC) curve using the FSC=0.143criterion; middle, angular distribution heatmap calculated in Cryosparc; bottom, local resolution evaluation.d Cryo-EM map density and model of representative protein parts and lipids/detergents.CHS, cholesteryl hemisuccinate; POPS, 1palmitoyl-2-oleioyl-phosphatidylserine; TMH, transmembrane helix.

Fig. S6 |
Fig. S6 | Cryo-EM data processing and high-quality density/model fitting exemplary views of liganded cPGAP1 mutants.a Representative cryo-EM micrograph of cPGAP1 H443N in detergents and selected 2D class averages.b The workflow of classification and refinement.c Top left, the nominal resolution of PGAP1 H443N determined by the gold-standard Fourier shell correlation (FSC) curve using the FSC=0.143criterion; bottom left, angular distribution heatmap calculated in Cryosparc; right, local resolution evaluation.d Cryo-EM map density and model of representative protein parts and lipids/detergents.CHL, cholesterol; PLM, palmitic acid; TMH, transmembrane helix.e-h Data processing and density views of cPGAP1 S327A in the same order as a-d.

Fig. S7 |
Fig. S7 | Topology and the jelly-roll domains of cPGAP1.a Topology of cPGAP1.Residues at the beginning of each domain are labeled.A star (yellow) near the catalytic triad (red dots) marks the active site where the products are bound.A brown dumbbell indicates disulfide bonds.Dashed lines mark unresolved regions, with the starting and ending residues indicated.AH, amphipathic helix; ER, endoplasmic reticulum; TM, transmembrane.b Comparison of the jelly-roll domains between cPGAP1 (top) and the AlphaFold2-predicted 9 hPGAP1 (bottom left) (Uniprot ID Q75T13) and yPGAP1 (bottom right) (Uniprot ID P4357).

Fig. S8 |
Fig. S8 | Integrity of cPGAP1 mutants on SDS-PAGE.cPGAP1 wildtype (WT) and the indicated mutants were analyzed by SDS-PAGE.Gels were imaged on a Typhoon machine (FLA-9000 controlled by the software Image Reader FLA-9000 Ver.1.0,GE Healthcare) for in-gel fluorescence with the setting for green fluorescence (for the GFP-tagged markers) and for cPGAP1 constructs (mCherrytagged).The resulting images are merged but the two parts are shown separately to indicate the different fluorescence settings.Molecular weight of the home-made standards 10 are indicated on the left.Results are from a single experiment.

Fig. S9 |
Fig. S9 | Fitting TGP3 into the TGP2/fatty acid density suggests substrate recognition mechanisms.a Chemical structure of TGP3.EtNP2 and TGP are not drawn as they were invisible in the product-bound structures.Various parts are color-coded to match that in Fig. 4a.b The fitting of the substrate TGP3 (blue) into the density of products in cPGAP1 H443N .TGP2 (product 1) is colored alternatingly with magenta and cyan, while palmitic acid (PLM, product 2) is colored green.c Expanded view of the catalytic site.The catalytic triad (orange), TGP3 (blue), TGP2 (magenta), and PLM (green) are shown as stick representations.EtNP, ethanolamine phosphate; G3P, glycerol 3-phosphate; GlcN, glucosamine; Ino, inositol; Man, mannose; PLM, palmitic acid; TGP, thermostable green fluorescence protein.

Fig. S10 |
Fig. S10 | Simplified LigPlot 11 view of the product-cPGAP1 interactions.Hydrophobic interactions are indicated by eye slashes and H-bonds are indicated by dashed lines.The palmitic acid (PLM) is colored green.The GPI moiety is colored alternatively and protein subunits are shaded by the color scheme used in Fig. 2a, 2b, and Fig. 4. EtNP, ethanolamine phosphate; GlcN, glucosamine; GPI, glycosylphosphatidylinositol; Ino, inositol; Man, mannose; PLM, palmitic acid.

Fig. S11 |
Fig. S11 | Simplified schematics of constructs used in this study.a The construct that expressses cPGAP1 for the structural determination of cPGAP1 apo (i), and for the assays of all cPGAP1 variants (Strep-tag) and the structural study of cPGAP1 H443N and S327A (His-tag) (ii).b-d The construct (i) and the fragment (ii) used for the generation of the sec13-1 mutant (b), the replacement of yPGAP1 with cPGAP1 (c) or yPGAP1 itself (d).e The construct used for the generation of yPGAP1-KO cells.Positive and negative numbers indicate nucleotide positions downstream (3'direction) and upstream (5'-direction) of open reading frame (ORF) of indicated genes, respectively.HIS and TRP are genes encoding the nutrient markers for selection of positive integration.In b-d, fragments (ii) were obtained by Gibson assembly of several fragments amplified by polymerase chain reaction (PCR) using the constructs in (i) as the template (See Methods).KO, knockout; TGP, thermostable fluorescence protein.

Fig. S12 |
Fig. S12 | Gating strategies for fluorescence-activated cell sorting.a The cells were first gated to select living cells and single cells.The expression of TGP-fused hPGAP1/cPGAP1/mutants was gated by the fluorescence of TGP (488/525 nm).This population was further analyzed for PE-positivity (561/585 nm) as an indication of the cell surface staining of CD59 via its antibody (Fig.1b, 2e, 3f).b Cells were first gated to select living cells and single cells.Cells expressing Flagtagged TGP3/TGP2/TGP0 were gated by TGP fluorescence to eliminate nonexpressing cells (total TGP expression).The TGP-positive cells were further analyzed by APC fluorescence (638/660 nm from anti-Flag antibodies), which indicates cell surface expression of TGP3/TGP2 (Fig.S4b).c Cells were first gated

Fig. S13 |
Fig. S13 | Uncropped images for Fig. S4. a Fig. S4c.A single and a double asterisk denotes probable monomer and dimer of TGP3.b Fig. S4e.c Fig. S4f.d Fig. S4h.Both the Coomassie Blue images (i) and in-gel fluorescence images (ii) of each gel are shown.Lanes are labelled only for those shown in Fig. S4.Molecular weights of the home-made GFP markers (GM) and standard markers (M) are indicated on the side (left, GFP markers; right, standard markers).In-gel fluorescence images were taken with a portable transilluminator.The Coomassiestained gel images were captured using a smartphone.