Extended Data Fig. 9: TTYH2 in complex with APOE-lipoprotein in cell-derived vesicles and functional investigations of lipid transport.

a, Representative motion and CTF-corrected micrograph (total micrographs=27,306) of TTYH2 in cell-derived vesicles with APOE-lipoprotein discs added at 45 µM. b, Data processing workflow. An initial set of manually picked particles were used for the training of a Topaz model. Topaz-picked particles were first extracted in a small box size of 280 pixels to aid the alignment of TTYH2 in vesicle membranes. 2D classification revealed classes of TTYH2 with additional density corresponding to ApoE. Classes with APOE were selected for an ab-initio reconstruction, which was 3D refined and yielded a map of TTYH2 residing in the vesicle membrane with extra density at the top of TTYH2. The particles were further sorted using 3D classification in cryoSPARC and classes with extended APOE density were picked for particle re-extraction in a bigger box of 440 pixels. The particles were refined to yield a final reconstruction at 10.7 Å. c, Angular distribution of particle orientations. d, Fourier shell correlation (FSC) plot of the final refined cryo-EM density map of the TTYH2 in complex with APOE(lipidated) in cell-derived vesicles at 10.7 Å with pronounced density of APOE-lipoprotein discs. e, Scheme of a lipid transfer assay monitoring the decay of NBD-fluorescence from labeled donor lipids reconstituted into APOE-containing lipoprotein particles quenched by Rhodamine-labeled acceptor lipids in liposomes containing 85% DPPC and DGS-NTA(Ni) lipids. f, Traces of the NBD fluorescence from lipoproteins assembled with a His-tagged construct of APOE leading to the tethering of the lipoprotein particle to the surface of the liposome is shown (as average and s.e.m. of three technical replicates) in comparison to a single trace of lipoprotein particles assembled from untagged APOE. The mean fluorescence decay recorded from proteoliposomes containing TTYH2 and its protein-free controls displayed in Fig. 5b are shown as dashed lines for reference. ‘*’ indicates addition of NBD-labeled lipoprotein discs, ‘#’ addition of Triton X-100 leading to the dissolution of liposomes and lipoproteins. g, Lipid transfer into TTYH3-containing proteoliposomes is only moderately enhanced compared to mock liposomes. Data show mean of three experiments from a single reconstitution. Mean values of TTYH2 and mock (displayed in Fig. 5b) are shown for comparison. ‘*’ indicates addition of NBD-labeled lipoprotein discs, ‘#’ addition of Triton X-100 leading to the dissolution of liposomes and lipoproteins. h, Rate constants from a fit of the fluorescence decay of mock liposomes (n = 4 samples from 2 independent reconstitutions) and TTYH3-containing proteoliposomes (n = 3 samples from 1 reconstitution) to a single exponential function defining the rate of non-specific transfer. The two rate constants obtained for the fit of the TTYH2-mediated fluorescence decay correspond to nonspecific (slow rate, n = 5 samples from 2 independent reconstitutions) and TTYH2-enhanced transfer (fast rate, n = 5 samples from 2 independent reconstitutions). g, h, Errors are s.e.m. Differences between the TTYH2-enhanced transfer rate constant (fast) and nonspecific lipid transfer rate constants were analyzed in a two-sample two-sided t test and found to be significant (TTYH2 (slow) p = 0.00004, mock p = 0.00003, TTYH3 p = 0.00005). i-l, Proteoliposome-based and cellular assays testing the ability of TTYH2 to scramble lipids. i, Scheme of a proteoliposome-based lipid scrambling assay monitoring the irreversible bleaching of NBD-labeled lipids located on the outside by the membrane-impermeable reducing agent dithionite. j, Results from proteoliposome-based lipid scrambling assays. Shown are the averaged differences in the fluorescence compared to mock reconstituted liposomes from the same lipid batch 200 s after the addition of dithionite. TTYH2 liposomes were reconstituted by two different methods into either preformed and destabilized liposomes (n = 4 from 2 independent reconstitutions) or liposomes assembled from solubilized lipids (n = 3 from 1 reconstitution). The data for the Ca²⁺-activated scramblase TMEM16F in absence and presence of Ca²⁺, as reported previously⁵⁰, are shown for comparison (line shows the average of n = 3 repeats from 1 reconstitution). Values of TTYH2-containing liposomes are similar to inactive TMEM16F in its Ca²⁺ free state⁶³ and thus does not show any indication of scrambling. Errors are s.e.m. Differences in the liposome-based scrambling assay data of solubilized and destabilized liposome preparations containing TTYH2 compared to previously obtained data from TMEM16F in absence and presence of Ca²⁺ were analyzed in a two-sample two-sided t test. Differences were found to be non-significant compared to inactive TMEM16F in absence of Ca²⁺ (p values are 0.32 and 0.85 for destabilized and solubilized preparations, respectively), and significant compared to active TMEM16F in presence of Ca²⁺ (p values are 0.0003 and 0.007 for destabilized and solubilized preparations, respectively). k, Scheme of a cellular scrambling assay monitoring the exposure of phosphatidyl serine (PS) to the cell surface by fluorescently labeled Annexin V. l, Results from cellular scrambling assays where protein constructs are overexpressed in a HEK293 TMEM16F knockout cell line by transient transfection. Surface expression of TTYH2 under such conditions was confirmed previously¹. Scrambling is assayed under resting Ca²⁺ concentrations where the activating TMEM16F mutant F518H but not WT TMEM16F mediates basal scrambling⁵⁰. No activity was detected for TTYH2 under such conditions. For each datapoint, Annexin V fluorescence data was averaged from multiple cells and examined over 3 independent experiments, errors are s.e.m. Differences in the cellular scrambling assay data to the constitutively active TMEM16F mutant F518H were analyzed in a two-sample two-sided t test and found to be significant for mock, TTYH2 and TMEM16F (TTYH2 p = 0.03, mock p = 0.02, TMEM16F p = 0.03). Differences between data from TTYH2 and TMEM16F samples were found to be non-significant with the p value of 0.17. m, Confocal fluorescent microscopy images of HEK293 cells showing the colocalization of TTYH2, (green, stained with a polyclonal antibody targeting the protein) and fluorescently labeled PE delivered via APOE-lipoprotein particles applied to the cell media. Shown are images of the individual channels and their combination (center, right). Locations with overlap of both fluorescent markers are indicated by circles. The illustration on the right shows co-localization (white signals) after correcting for the mean intensity of extracellular regions and applying an adjusted background subtraction to eliminate false or random co-localizations. The analysis was performed using the open-source software ImageJ_Coloc2 (https://imagej.net/plugins/coloc-2). The diagrams in e, i and k were created using BioRender (https://www.biorender.com).

Source data