Interaction of NPC2 protein with Lysobisphosphatidic Acid is required for normal endolysosomal cholesterol trafficking

Unesterified cholesterol accumulation in the late endosomal/lysosomal (LE/LY) compartment is the cellular hallmark of Niemann-Pick C (NPC) disease, caused by defects in the genes encoding NPC1 or NPC2. We previously reported the dramatic stimulation of NPC2 cholesterol transport rates by the LE/LY phospholipid lysobisphosphatidic acid (LBPA) and in these studies sought to determine their functional relationship in normal LE/LY cholesterol egress. Here we demonstrate that NPC2 interacts directly with LBPA and identify the NPC2 hydrophobic knob domain as the site of interaction. Using its precursor phosphatidylglycerol (PG), we show that PG-induced LBPA enrichment results in clearance of accumulated cholesterol from NPC1-deficient cells but is ineffective in cells lacking functional NPC2. Together these studies reveal a heretofore unknown aspect of intracellular cholesterol trafficking, in which NPC2 and LBPA function together in an obligate step of sterol egress from the LE/LY compartment, which appears to be independent of NPC1.

Cholesterol is a small, hydrophobic molecule that is a vital building block of cell membranes and a 16 precursor for steroid hormones, bile salts, vitamin D, and oxysterol ligands for transcription factors. 17 Intracellular transport of cholesterol is a highly regulated but, as yet, incompletely understood process. 18 Perturbations can lead to detrimental outcomes such as in the lysosomal storage disorder Niemann Pick 19 Type C (NPC) disease, where LDL-derived cholesterol becomes trapped within the late 20 endosomal/lysosomal (LE/LY) system. The sterol particularly enriches LE/LY inner membranes, which 21 develop during endosome maturation as a means of compartmentalizing its contents (Gruenberg, 2001;22 Gruenberg, 2003;Matsuo et al., 2004). The accumulation of cholesterol in NPC disease is associated 23 with amassing of other lipids in the LE/LY, disruption of post-lysosomal cholesterol metabolism, and 24 ultimately clinical manifestations including organomegaly and neurological deterioration. In 95% of NPC 25 cases, mutations in the large LE/LY transmembrane protein, NPC1, prevent proper export of cholesterol 26 from the LE/LY to other cellular compartments. The remaining 5% of cases are caused by mutations in 27 the small, 132-amino acid, soluble LE/LY protein, NPC2 (Peake & Vance, 2010;Sokol et al., 2010). 28 29 Similarities in the cellular and clinical phenotypes resulting from either NPC1 or NPC2 deficiency have 30 led to the suggestion that these two proteins function cooperatively in normal LE/LY cholesterol 31 trafficking (Kwon et al., 2009;Sleat et al., 2004); a proposed model shows cholesterol directly 32 transferred from NPC2, in the LE/LY lumen, to NPC1, located in the limiting membrane of the 33 compartment (Estiu et al., 2013;Wang et al., 2010). Recent tertiary structural analyses identifying a 34 potential NPC2 interacting domain on the NPC1 protein, support this mode of cholesterol egress from 35 the LE/LY compartment (Zhao et al., 2016). It is further proposed that cholesterol transfer from NPC2 to 36 the luminally localized N-terminal domain of NPC1 allows sterol passage through the glycocalyx found at 37 the luminal surface of the LE/LY limiting membrane, via concerted effort from membrane glycoproteins 38 (Li et al., 2016). 39 40 In addition to functioning with NPC1, accumulating evidence suggests that NPC2 may also have NPC1-41 independent actions. Goldman and Krise, for instance, showed that NPC2 deficient fibroblasts exhibit 42 significant reductions in exocytosed dextran relative to NPC1 deficient cells. They additionally 43 demonstrated that treatment with the commonly used inducer of the NPC disease phenotype, 44 U18666A, further reduced exocytosis in NPC1 but not NPC2 cells, indicating a potential divergence in the 45 egress pathway of membrane-impermeable species utilized by the NPC1 and NPC2 proteins (Goldman & 46 Krise, 2010). It has also been shown that overexpression of ABCA1 reversed lysosomal cholesterol accumulation in NPC1 deficient but not NPC2 deficient cells (Boadu et al., 2012). Further, Karten and 48 colleagues demonstrated that endosome to mitochondrial cholesterol transport occurs efficiently in the 49 absence of NPC1 protein but is dependent upon the presence of functional NPC2 (Kennedy et al., 2012). 50 51 NPC2 binds cholesterol with a 1:1 stoichiometry . In 2003, Ko et al. showed that point 52 mutations preventing cholesterol binding also prevented NPC2-mediated LE/LY cholesterol efflux from 53 NPC2 deficient patient fibroblasts; cholesterol binding was therefore deemed essential to the function of 54 NPC2 (Ko et al., 2003). Interestingly, however, some NPC2 point mutants with normal cholesterol 55 binding affinity were nevertheless deficient in clearing intracellular cholesterol from NPC2 patient 56 fibroblasts, suggesting that the ability to bind cholesterol was necessary but not sufficient for the 57 cholesterol efflux function of NPC2 (Ko et al., 2003). Indeed, we later demonstrated that NPC2-58 mediated LE/LY cholesterol egress is likely due to its cholesterol transport activity, which involves direct 59 contact between the surface of NPC2 and membranes (Cheruku et al., 2006;Xu et al., 2008), and that 60 the residues identified by Ko et al (2003) as being unable to 'rescue' sterol accumulation in NPC2 patient 61 cells, despite normal binding, were in fact markedly defective in cholesterol transport (McCauliff et al., 62 2015). Of particular note was that residues which, when mutated, abrogated sterol transfer between 63 NPC2 and membranes, mapped not to a single domain but rather to several surface domains (McCauliff 64 et al., 2015). This suggested, in turn, that NPC2 may be able to interact with more than one membrane 65 simultaneously, and we and others have in fact shown that NPC2 promotes membrane-membrane 66 interactions (Abdul-Hammed et al., 2010;McCauliff et al., 2011;McCauliff et al., 2015). Importantly, 67 mutants with reduced sterol transfer rates were unable to clear accumulated cholesterol from NPC2 68 deficient fibroblasts and were also deficient in promoting membrane-membrane interactions (McCauliff 69 et al., 2015). Notably, all of these impairments occur despite normal cholesterol binding, indicating that 70 certain regions of the NPC2 surface may interact with membranes to effectively transfer cholesterol 71 (McCauliff et al., 2015). 72 73 Earlier, using model membranes, we demonstrated a remarkable, order of magnitude stimulation in 74 NPC2 cholesterol transfer rates by the incorporation of lysobisphosphatidic acid (LBPA), also known as 75 bis-monoacylglycerol phosphate (Cheruku et al., 2006;Xu et al., 2008). LBPA is a structural isomer of 76 phosphatidylglyerol with an atypical phospholipid stereoconfiguration. It is localized primarily to inner 77 LE/LY membranes and is thought to be involved not only in the formation of these internal membranes 78 and their architecture, but also in the sorting and efflux of LE/LY components, including cholesterol 79 (Gruenberg, 2003;Hullin-Matsuda et al., 2007;Kobayashi et al., 1998;Kobayashi et al., 1999). 80 Incubation of BHK cells and macrophages with an anti-LBPA monoclonal antibody resulted in cholesterol 81 accumulation in the LE/LY compartment, resembling the NPC phenotype (Delton-Vandenbroucke et al., 82 2007;Kobayashi et al., 1999). Interestingly, Chevallier et al. showed that by enriching NPC1-deficient 83 cells with exogenously added LBPA, the cholesterol accumulation was reversed (Chevallier et al., 2008). 84 Based on the dramatic effects of LBPA on NPC2-mediated cholesterol transfer, however, we 85 hypothesized a specific functional interaction between LBPA and NPC2, such that LBPA enrichment of 86 Our previous kinetics analyses strongly suggested that the mechanism of cholesterol transfer between 99 NPC2 and membranes was via protein-membrane interaction (Cheruku et al., 2006;McCauliff et al., 100 2015;Xu et al., 2008). NPC2 does not contain any apparent transmembrane domains, nor are there 101 experimentally documented membrane interactive domains to date. For de novo predictions we 102 therefore employed the Orientation of Proteins in Membranes (OPM) Database, a curated online 103 resource that predicts the spatial positions of known protein structures relative to the hydrophobic core 104 of a lipid bilayer (Lomize et al., 2012). Using the crystal structure of bovine NPC2 (PDB ID: 1NEP), a loop 105 domain consisting of hydrophobic residues as highlighted in Figure 1, was predicted to be highly 106 membrane interactive, with a ΔG of -4.6 kcal/mol. This domain corresponds to 56-HGIVMGIPV-64 and 107 consists primarily of the hydrophobic residues I58, V59, M60, I62, P63, and V64, plus the non-polar 108 residues G61 and G57. Structurally, this domain forms a "hydrophobic knob" which presents 109 prominently on the surface of the NPC2 protein. OPM predicts that this knob domain inserts into the 110 hydrophobic space of the membrane model, positioning the sterol binding pocket of NPC2 near the 111 membrane surface ( Figure 1A) and, thus, in proximity to membrane sterols. These computational 112 observations suggest that this knob domain may play a key role in the mechanism by which NPC2 is able 113 to transport cholesterol between inner membranes of the LE/LY compartment. Of note, and as shown in 114 Figure 1B, the primary sequence of the hydrophobic knob domain is conserved in mammalian NPC2 115 proteins but not in the yeast NPC2 homologue. Hydrophobicity scales for NPC2 as well as a  Doolittle plot of the hydrophobicity scores along the primary protein sequence indicate that the 117 hydrophobicity of the knob domain is conserved amongst mammalian NPC2 proteins, in contrast to the 118 low hydrophobicity of the yeast NPC2 protein ( Figure 1C). 119 120 LBPA markedly stimulates sterol transfer rates between NPC2 and membranes 121 We previously found that incorporation of 25 mol% LBPA in EPC membranes resulted in cholesterol 122 transfer rates from vesicles to NPC2 that were markedly accelerated relative to 100% EPC membranes 123 (Cheruku et al., 2006;Xu et al., 2008). LBPA accounts for approximately 15 mol% of total LE/LY 124 phospholipids, with the likelihood of higher lateral concentrations in the highly heterogeneous inner 125 LE/LY membranes (Kobayashi et al., 2002). Thus, we examined the rates of cholesterol transfer from 126 NPC2 to membranes as a function of increasing levels of LBPA. The results in Figure 2 indicate an 127 exponential relationship between the LBPA content of the vesicles and the NPC2 cholesterol transfer 128 rate; increasing the mol% of LBPA in SUV from 0 to 30% effectively increases the NPC2 cholesterol 129 transfer rates by approximately 100 fold. 130 131 LBPA restores normal NPC2 cholesterol transfer rates for proteins with mutations outside the 132 hydrophobic knob 133 We recently reported that point mutations in multiple regions on the NPC2 surface led to markedly 134 diminished rates of cholesterol transfer between NPC2 and zwitterionic PC membranes; the 135 hydrophobic knob domain was one of the regions found to be dramatically impacted (McCauliff et al., 136 2015). Here we examined the rates of cholesterol transfer from these and additional hydrophobic knob 137 domain point mutations to membranes containing 25 mol% LBPA. Point mutations were confirmed by 138 DNA sequencing and all mutant proteins were found to bind cholesterol similar to WT NPC2 (Friedland 139 et al., 2003;Ko et al., 2003), with submicromolar affinity (McCauliff et al., 2015). The results in Figure  140 3A show that when acceptor membranes included 25 mol% of LBPA, cholesterol transfer rates for NPC2 141 proteins with mutations in regions other than the hydrophobic knob were similar to those of WT. 142 Indeed, though mutations at H31, Q29, D113, and E108 exhibited sterol transfer rates to EPC 143 membranes that were ≤ 15% of WT NPC2, the inclusion of LBPA in acceptor membranes resulted in rates 144 of cholesterol transfer that were ≥ 85% of WT rates. By contrast, the I62 and V64 mutations in the 145 hydrophobic knob, which also resulted in markedly defective cholesterol transfer to EPC vesicles, were 146 unaffected by the inclusion of LBPA in the acceptor membranes, with cholesterol transfer by these 147 mutants remaining barely detectable. The G61A mutation, also in the hydrophobic knob, resulted in 148 cholesterol transfer deficiencies similar to the I62 and V64 mutants, though changes are less extreme; 149 sterol transfer to EPC vesicles was reduced by 70%, and it remained highly defective in the presence of 150 LBPA, with rates of cholesterol transfer of only 16% relative to WT NPC2. Mutations in hydrophobic 151 knob residues H56, G57, and I58 had little effect on cholesterol transfer rates to EPC vesicles, however, 152 unlike WT NPC2, these mutants were insensitive to the presence of LBPA in acceptor membranes. The 153 NPC2 residues where mutations cause large decreases in cholesterol transfer rates to EPC, are shown in 154 red in Figure 3B; multiple surface regions are highlighted. By contrast, Figure 3C shows the mutations 155 that were sensitive to LBPA inclusion, in green, vs. mutations which remained insensitive to membrane 156 LBPA, in red; the hydrophobic knob is clearly denoted as LBPA-sensitive. Overall, the mutagenesis results 157 strongly indicate that the hydrophobic knob domain of NPC2 is the LBPA-sensitive region on the protein 158 surface. 159

Effects of NPC2 mutations on vesicle-vesicle interaction also highlight the hydrophobic knob domain 161
We and others have shown that WT NPC2 promotes membrane-membrane interactions (Abdul-162 Hammed et al., 2010;Berzina et al., 2018;McCauliff et al., 2011;McCauliff et al., 2015). We further 163 showed that NPC2 point mutants with deficient cholesterol transfer abilities are also unable to cause 164 EPC membrane aggregation (McCauliff et al., 2015). In the present studies, we investigated whether the 165 presence of LBPA in the vesicles affected vesicle aggregation by WT and mutant NPC2 proteins. The 166 results in Table 1 show that inclusion of 25 mol% LBPA in LUVs resulted in a 16-fold increase in the rate 167 of membrane-membrane interaction by WT NPC2, relative to 100% EPC LUVs. Incorporation of LBPA 168 into membranes normalizes the membrane aggregation rates for NPC2 proteins with mutations outside 169 the hydrophobic knob, e.g. H31, D113, and Q29. By contrast, the hydrophobic knob domain mutations 170 were relatively insensitive to membrane LBPA. The results for this membrane-membrane interaction 171 assay map virtually identically onto the NPC2 structure as did those for the cholesterol transport rates, 172 as seen in Figure 3, again indicating that the hydrophobic knob of NPC2 is the LBPA-sensitive region of 173 the protein. 174

NPC2 interaction with LBPA and other phospholipids, and identification of the LBPA-interactive domain 176
To determine whether the relationship between LBPA and NPC2 in cholesterol trafficking involves direct 177 interactions, protein-lipid binding assays were conducted. For studies of WT NPC2 protein, custom LBPA 178 Snoopers (Avanti Polar Lipids) containing various LBPA isomers were incubated with WT NPC2 protein 179 and relative binding was assessed via densitometric analysis of an antibody-probed strip, as described in 180 Methods. The results shown in Figure 4A demonstrate that WT NPC2 binds to LBPA, showing greater 181 interaction with isomers containing oleoyl (C18:1) as opposed to myristoyl (C14:0) fatty acyl chains. 182 Interestingly, WT NPC2 exhibited the greatest degree of binding to the S,S 18:1 LBPA, which possess not 183 only the likely stereoconfiguration in mammals (Chevallier et al., 2000;Hayakawa et al., 2006;Kobayashi 184 et al., 2002), but also the predominant fatty acyl chains in most cell types studied thus far (Kobayashi et 185 al., 2002;Mason et al., 1972). NPC2 binding of the S,S 18:1 LBPA was also greater than binding to egg PC. 186

187
Binding of WT NPC2 to other typical membrane phospholipid species, in comparison to di-oleoyl LBPA, 188 was also examined using membranes spotted with di-oleoyl phospholipids. Figure 4B shows that NPC2 189 interacts more strongly with LBPA than with PC, PA, PG, and PS species. As two other anionic 190 phospholipids assayed, PG and PA, show even less interaction with NPC2 than does zwitterionic PC, the 191 mechanism by which NPC2 binds to LBPA is likely not solely dependent on electrostatic interactions. 192 193 Since the phospholipids are not necessarily present in a physiological orientation in protein-lipid overlay 194 assays, we further examined NPC2-lipid interaction using Homogenous Time Resolved Fluorescence 195 (HTRF), in which the phospholipids are present as lamellar structures. HTRF technology has been 196 recently demonstrated to be an effective and sensitive assay for lipid-protein interaction (Fleury et al., 197 2015). The NPC2-interaction with LBPA was substantially greater than with other negatively charged 198 lipids or with zwitterionic PC (Figure 4D For example, studies analyzing binding to various isomers of LBPA show that the hydrophobic knob 206 mutants I62N and V64A bound the S,S and S,R di-oleoyl isomers at only ~ 20% of WT levels, and the 207 C18:1 R,R isomer at approximately 40% of WT. In contrast, the Q29A and D113A mutants, in regions 208 outside the hydrophobic knob, bound nearly all LBPA isomers similar to WT; only Q29A was observed to 209 bind the C18:1 R,R isomer at approximately 50% relative to WT binding ( Figure 5A). The I62D and V64A 210 mutants exhibited greater interaction with the 18:1 Semi LBPA species, which has three oleoyl acyl 211 chains ( Figure 5A), binding approximately 70% relative WT NPC2. Mutations in other hydrophobic knob 212 residues also reduced the interaction of NPC2 with the di-oleyol S,S LBPA on blots using various 213 phospholipids. Indeed, similar to what was observed with the LBPA isomer blots, the H56, G57, I58, and 214 G61 mutants showed only 30 to 40% degree of interaction relative WT NPC2. In marked contrast, 215 surface mutations outside the hydrophobic knob had virtually no impact on NPC2 interaction with LBPA; 216 Q2A, H31A, D113, and E108 mutants exhibiting WT levels of interaction with LBPA ( Figure  In agreement with several previous reports, incubation of NPC2 deficient fibroblasts with WT NPC2 227 protein resulted in a dramatic decrease in filipin staining, reaching levels similar to healthy fibroblasts 228 ( Figure 6) (Ko et al., 2003;Liou et al., 2006;McCauliff et al., 2011;McCauliff et al., 2015). The H56A 229 mutant, with rates of sterol transfer and membrane aggregation similar to WT protein, also reduced 230 filipin stain area, similar to WT NPC2. In contrast, the G57D and I58A hydrophobic knob mutants, with 231 markedly attenuated cholesterol transfer and membrane aggregation rates, were unable to reverse 232 cholesterol accumulation in NPC2 cells; filipin staining remained at a level comparable to that of the 233 unsupplemented cells. G61A, also in the knob domain, was able to lessen cholesterol accumulation in 234 NPC2 cells to a moderate extent, and its defect in cholesterol transfer was also more modest than that 235 of other hydrophobic knob mutants (Figure 6). These results are in keeping with our previously reported 236 results for two other hydrophobic knob mutants, I62D and V64A (McCauliff et al., 2015) which are also 237 deficient in cholesterol transfer and membrane aggregation ability. The consistency between results of 238 the sterol transfer assays and membrane-membrane interaction assays, with the cholesterol clearance 239 in patient cells, strongly supports the physiological relevance of the structure-function studies, and 240 points to a particularly important role for the hydrophobic knob of NPC2 in effecting normal sterol To increase cell LBPA levels, NPC patient fibroblasts were incubated with PG, known to be its precursor 253 (Bouvier et al., 2009;Poorthuis & Hostetler, 1978;Thornburg et al., 1991). There is currently some 254 uncertainty as to the stereoconfiguration and acyl chain attachment site of LBPA in cells (Chevallier et 255 al., 2008;Goursot et al., 2010;Hayakawa et al., 2006), thus we reasoned that the endogenous synthesis 256 of LBPA from its precursor PG would allow for generation of the appropriate LBPA isomer, providing a 257 physiologically relevant model. The results in Figure 7 show that incubation of the cells with 100% PG 258 SUVs led to substantial increases in cellular content of LBPA in all fibroblast types; a nearly 6-fold 259 increase was observed in WT cells. In agreement with previous reports LBPA levels in NPC patient cells 260 were found to be increased relative to WT cells prior to enrichment (Chevallier et al., 2008;Davidson et 261 al., 2009;Sleat et al., 2004;Vanier, 1983); PG incubation resulted in 2-3 fold increases in NPC1-and 262 NPC2-deficient cells, respectively, relative to unsupplemented cells ( Figure 7A). Interestingly, the levels 263 of other cellular phospholipids remained unchanged, including PG itself ( Figure 7B). Increases in LBPA 264 were also visualized using the 6C4 anti-LBPA antibody, as described in Methods, and results showed 2-265 fold or greater increases in LBPA levels relative to unsupplemented cells ( Figure 7C). Direct addition of 266 LBPA SUVs also led to approximately 2 to 3-fold increases in cellular LBPA levels (data not shown), in 267 agreement with Chevallier et al (Chevallier et al., 2008). Supplementation with PC SUVs as a control had 268 no effect on the phospholipid composition of any of the cell types (data not shown). 269 270 We evaluated the effect of LBPA enrichment via PG supplementation on intracellular cholesterol 271 content in WT and NPC disease fibroblasts using filipin staining. Following PG supplementation, 272 cholesterol content remained apparently unchanged in the WT fibroblasts. NPC1 deficient fibroblasts 273 exhibited a dramatic reduction in cholesterol accumulation following PG supplementation/LBPA 274 enrichment, approaching levels observed for WT cells, similar to the direct addition of LBPA (Chevallier 275 et al., 2008). In marked contrast, and as we hypothesized based on NPC2-LBPA interactions and the 276 effects of LBPA on NPC2 sterol transfer rates, the cholesterol accumulation in NPC2 deficient fibroblast 277 persisted following PG supplementation despite increased LBPA content (Figure 8) In the present studies we discovered an unanticipated 284 impact of LBPA incorporation into membranes, in which some NPC2 mutants that were highly defective 285 in sterol transfer to phosphatidylcholine membranes, were essentially normalized when LBPA was 286 present. These "LBPA-sensitive" NPC2 mutations were, almost exclusively, present in surface domains 287 outside of the hydrophobic knob. NPC2 proteins with mutations in the hydrophobic knob, that were 288 highly defective in cholesterol transfer to phosphatidylcholine membranes, remained insensitive to LBPA 289 in the membranes (Figure 3). Based on this structure-based difference in LBPA sensitivity, we 290 hypothesized that increasing the LBPA levels in NPC2 deficient cells would enhance the activity of 291 mutants outside the hydrobphobic knob that responded to LBPA in the kinetic assays, whereas cellular 292 LBPA enrichment would not enhance the action of the hydrophobic knob mutants that were insensitive 293 to LBPA. To test these predictions, NPC2 deficient fibroblasts were incubated with purified wild type or 294 mutant NPC2 proteins, with or without PG supplementation, and cholesterol accumulation was assessed 295 by filipin staining as described. The results in Figure 9 demonstrate that NPC2 proteins with mutations 296 outside the hydrophobic knob, such as Q29A, D113A, and D72A, which were unable to reduce 297 cholesterol accumulation in unsupplemented NPC2 deficient cells, were indeed able to 'rescue' cells that 298 were enriched with LBPA via PG supplementation. By contrast, the hydrophobic knob mutants I62D and 299 G61A, which were insensitive to LBPA in cholesterol transfer assays, were unable to clear cholesterol 300 from LBPA-enriched NPC2 deficient cells. Thus, the results show that a combination of increased cellular 301 levels of LBPA and LBPA-sensitive NPC2 protein was able to rescue the NPC2 deficient cells, beyond the 302 ability of the mutant protein alone (Figure 9). As before, WT cells supplemented with PG alone showed 303 no change in cholesterol accumulation, and cells supplemented with purified WT NPC2, with or without 304 PG, showed significantly reduced filipin staining. The results in patient cells again mirror the results of 305 sterol kinetics experiments, suggesting that return to normal sterol trafficking can be achieved for NPC2 306 mutations outside the hydrophobic knob that are otherwise dysfunctional, when the LBPA content of 307 the cells is increased. 308 309 Discussion 310 LBPA was proposed to be involved in intracellular cholesterol trafficking based on the sterol 311 accumulation which accompanies incubation of cells with an anti-LBPA antibody (Kobayashi et al., 1999), 312 however the mechanism of LBPA action has remained unknown. In this study we demonstrate for the 313 first time that functional and likely direct interaction of NPC2 protein with LBPA is a required step in 314 normal cholesterol trafficking through the endosomal/lysosomal compartment. We show that NPC2 315 interacts with membrane LBPA, and that the hydrophobic knob domain is the site of NPC2-LBPA 316 interaction. This surface domain is located near the NPC2 cholesterol binding pocket, thus its insertion 317 into the bilayer would position the protein to efficiently exchange cholesterol with the membrane. 318 Recent molecular dynamic simulations show that LBPA, but not other phospholipids, may position NPC2 319 in an orientation that could promote protein-membrane sterol exchange (Enkavi et al., 2017). 320 A decade ago, the Gruenberg laboratory reported that viral-mediated supplementation of NPC1-321 deficient cells with exogenous LBPA reversed cholesterol accumulation in the diseased cells (Chevallier 322 et al., 2008 LBPA is reported to comprise only 1% of total cellular phospholipids, but about 15 mol% of total 337 phospholipids in the LE/LY (Chevallier et al., 2000;Kobayashi et al., 1998;Kobayashi et al., 2002). There 338 is no agreement regarding which isomeric form of LBPA is predominant in cells, as the fatty acids can be 339 linked at the sn-2 or sn-3 positions of each glycerol in the S or R conformation, though the sn-2, sn-2' 340 positions are currently favored (Chevallier et al., 2000;Kobayashi et al., 1998;Kobayashi et al., 2002;341 Mason et al., 1972;Matsuo et al., 2004). The 2,2'-LBPA was shown to be quite effective at mobilizing 342 cholesterol in NPC1 disease cells while the 3,3'-and semi-LBPA isoforms were unable to promote sterol 343 efflux. No difference in efficacy between S,S, S,R, and R,R LBPA isomers were noted, however, 344 suggesting that the conformation of LBPA has no bearing on its ability to reverse cholesterol 345 accumulation in NPC1 deficient cells (Chevallier et al., 2008). In agreement with this observation, we 346 previously showed that the S,S, S,R, and R,R configurations of LBPA had little to no effect on in vitro 347 cholesterol transfer rates by NPC2 protein (Xu et al., 2008). Similarly, in the present studies we 348 observed little variation in NPC2 binding to these different LBPA stereoisomers. The cellular fatty acyl 349 chain components of LBPA have also been found to vary (Bouvier et al., 2009), with oleic acid and 350 docosahexaenoic acid (DHA) reported to be selectively incorporated (Besson et al., 2006;Luquain et al., 351 2001). Here we found acyl chain-dependent differences in NPC2-LBPA interactions, with reduced 352 binding to the 14-carbon saturated dimyristoyl-LBPA species relative to the 18-carbon monounsaturated 353 dioleoyl-LBPA species. In prior work we showed that cholesterol transfer from NPC2 to membranes 354 containing dioleoyl-LBPA was > 2-fold faster than transfer to vesicles with dimyristoyl-LBPA (Xu et al., 355 2008). Taken together, the results suggest that the acyl composition but not the steroconfiguration of 356 If direct interactions between LBPA and NPC2 are integral to this mechanism of efflux, a rate 361 determining step may be their frequency of interaction. While the concentration of LE/LY LBPA has 362 been shown to increase in parallel with cholesterol and other lipids in NPC disease (Davidson et al., 363 2009;Sleat et al., 2004;Vanier, 1983), as also found here, it is possible that LBPA levels nevertheless 364 remain too low to provide support for NPC2-mediated transport of the elevated cholesterol load. 365 Gruenberg and colleagues noted the possibility of this limitation, showing that although cholesterol 366 laden NPC1 cells had elevated levels of LBPA, virus-mediated supplementation of exogenous LBPA was 367 effective at reversing the sterol accumulation phenotype (Chevallier et al., 2008). Our results using PG 368 supplementation to augment LBPA levels further indicate that LBPA may become limiting in NPC1 369 disease. Importantly, we also show here that LBPA enrichment cannot reverse cholesterol accumulation 370 caused by NPC2 deficiency. 371 In addition to the apparently obligate interaction of NPC2 with LBPA, LBPA has been found to display a 372 variety of other unique functions within the endo/lysosomal system that could potentially promote this 373 cooperative mechanism of cholesterol efflux. It has been shown to modulate phospholipid membrane 374 curvature, for instance, and is required for formation of the multivesicular structures within the LE/LY 375 (Matsuo et al., 2004), where cholesterol localizes in raft domains (Fivaz et al., 2002). LBPA has also been 376 shown to be necessary for proper dynamics and organization of contents within this inner-membrane 377 system (Kobayashi et al., 1998;Kobayashi et al., 1999). It has recently been proposed that LBPA may 378 influence cholesterol homeostasis beyond the confines of late endosomes and lysosomes, having been 379 shown to be necessary for lipid droplet formation via Wnt signaling within the endoplasmic reticulum, 380 where cholesteryl esters are synthesized . 381 To increase cellular LBPA levels in these studies, we used its presumed precursor and structural isomer, 382 phosphatidylglycerol. PG is thought to convert to LBPA along the endo/lysosomal system (Hullin-383 Matsuda et al., 2007;Poorthuis & Hostetler, 1978), and studies have demonstrated that exogenously 384 administered PG can be converted to LBPA in vivo (Somerharju & Renkonen, 1980), although the 385 anabolic pathway of this conversion remains unknown. The present demonstration that PG 386 supplementation specifically increases the LBPA content of all cells tested supports the hypothesis that 387 PG is a precursor to LBPA. Utilizing PG as a precursor also obviates any potential concern about the 388 LBPA isoform, as the cells presumably generate the physiologically accurate form. Cellular conversion of 389 PG to LBPA has also been demonstrated in mammalian alveolar macrophages (Waite et al., 1987), 390 lymphoblasts (Hullin-Matsuda et al., 2007) and RAW macrophages (Bouvier et al., 2009). 391 Based on the present results, we propose that the NPC2 hydrophobic 'knob' domain inserts into LBPA 393 enriched inner LE/LY membranes, interacting directly with the phospholipid. This, in turn, facilitates 394 rapid transfer of cholesterol from the membranes to NPC2. We speculate, given the ability of NPC2 to 395 promote membrane-membrane interaction, that LBPA and NPC2 are involved in the formation of 396 membrane contact sites which could potentially exist between closely apposed inner LE/LY membranes. 397 Membrane contact sites have been shown to be important in intermembrane lipid transfer (Helle et al., 398 2013;Holthuis & Levine, 2005;Prinz, 2014), although none have yet been specifically described within 399 the multivesicular LE/LY. The present results show that NPC2 promotes membrane-membrane 400 interaction, and further indicate the ability of NPC2 to bind to LBPA; this interaction may represent one 401 membrane contact point on inner LE/LY membranes. Our previous kinetic studies suggested that NPC2 402 interacts with other membrane phospholipids as well, and preliminary molecular dynamics simulations 403 indicate that unlike LBPA, where NPC2 interacts at the hydrophobic knob, these interactions occur at 404 NPC2 surface sites other than the knob domain (data not shown); such interactions could represent a 405 second contact point within the inner LE/LY membranes. Interestingly, it has now been demonstrated 406 that NPC2 also binds directly to NPC1 which resides in the limiting LE/LY membrane, and possibly also 407 with LAMP proteins in these same membranes (Li et al., 2016). These could also be potential tether 408 points for the NPC2 and would effectively bring inner LE/LY cholesterol-laden membranes in closer 409 proximity to the limiting LE/LY membrane, which cholesterol must ultimately cross to exit the 410 compartment. 411 The key finding of the present studies is that LBPA functions in cholesterol efflux, at least in part, by an 412 NPC2-dependent mechanism, as LBPA enrichment in NPC2-deficient fibroblasts was completely 413 ineffective at reversing cholesterol accumulation. This would suggest that an NPC2-dependent, NPC1-414 indepenedent pathway of cholesterol egress exists, in addition to the 'hydrophobic handoff' of 415 cholesterol from NPC2 to NPC1 (Deffieu & Pfeffer, 2011;Infante et al., 2008;Wang et al., 2010). As 416 noted earlier several lines of evidence support NPC2-dependent NPC1-independent cholesterol egress 417 from the LE/LY (Boadu et al., 2012;Goldman & Krise, 2010;Kennedy et al., 2012). Our present 418 demonstration of a critical functional interaction between NPC2 and LBPA also implies that in the 419 presence of intact NPC2, bypassing dysfunctional NPC1 may be achieved via LBPA enrichment. LBPA 420 enrichment may also be effective in NPC2 cases where the mutation is in a residue outside of the 421 hydrophobic knob. For example, a human mutation in D72 has been reported to be disease causing 422 (Biesecker et al., 2009), and we found here that PG supplementation/LBPA enrichment allowed NPC2 423 deficient cells to be effectively cleared by the D72A protein; supplementation with D72A-NPC2 was 424 entirely ineffective prior to LBPA enrichment. 425 Currently there is no cure for NPC disease. Pharmacological options are limited, and palliative care 426 remains the standard for treatment of the disease, focusing on increasing the length and quality of life 427 for affected patients. The present study demonstrates that, in vitro, NPC disease cells are able to 428 convert administered PG to LBPA. The increase in LBPA occurs concurrently with a decrease in 429 cholesterol accumulation in cells with NPC1 deficiency, which comprises 95% of NPC disease cases. 430 Thus, we propose that cellular LBPA enrichment is worth exploring as a possible therapy. Aerosolized 431 phospholipids such as dipalmitoyl phosphatidylcholine are widely used to enhance pulmonary drug 432 delivery (Duret et al., 2014), and can be adequately nebulized without losing compositional integrity 433 (Schreier et al., 1994). PG itself, administered intranasally, has been used as a therapy by the Voelker 434 group to effectively inhibit respiratory syncytial virus infection (Numata et al., 2010;Numata et al., 435 2013) and influenza A virus (Numata et al., 2012). Moreover, lipid based colloidal carriers are able to 436 cross the blood brain barrier (BBB) when administered intranasally (Ganesan et al., 2018;Mittal et al., 437 2014;Patel & Patel, 2017;Tapeinos et al., 2017) and have recently been shown to be efficient drug 438 delivery vehicles in the treatment of intrinsic brain tumors (van Woensel et al., 2013) and 439 neurodegenerative diseases including Alzheimer's (Agrawal et al., 2018;Tapeinos et al., 2017) and 440 Parkinson's (Tapeinos et al., 2017;Yang et al., 2016 (Sievers et al., 2011). Protein conservation was scored using a PAM250 scoring matrix, which is 457 extrapolated from comparisons of closely related proteins, similar to the current application (Pearson, 458 2013). Domain specific conservation of the hydrophobic knob between each NPC2 sequence was 459 analyzed by taking the sum of the conservation scores of each residue from 56 to 64, relative to the 460 human NPC2 sequence. Protein hydrophobicity was scored using the Kyte and Doolittle Amino acid 461 Hydropathicity scale (Kyte & Doolittle, 1982). Domain specific hydrophobicity of the hydrophobic knob 462 of each NPC2 sequence was analyzed by taking the sum of the Kyte and Doolittle Amino acid 463 Hydropathicity score of each residue from 56 to 64. Whole protein hydrophobicity of aligned sequences 464 were analyzed using the ProtScale Tool on the ExPASy server (Gasteiger et al., 2005), based on the Kyte 465 and Doolittle Amino acid Hydropathicity scale (Kyte & Doolittle, 1982). 466

512
Effect of LBPA on cholesterol transfer by NPC2 513 As detailed previously (McCauliff et al., 2011;McCauliff et al., 2015;Xu et al., 2008), transfer of 514 cholesterol from WT or mutant NPC2 protein to membranes was monitored by the dequenching of 515 tryptophan fluorescence over time using a stopped-flow mixing chamber interfaced with a 516 Spectrofluormeter SX20 (Applied Photophysics, Leatherhead, UK). Cholesterol transfer rates from 1 µM 517 WT NPC2 to 125 µM EPC membranes containing increasing mol percentages of LBPA (0%, 10%, 20% and 518 30%) were determined at 25°C. Additionally, to determine the effects of LBPA on the sterol transport 519 properties of mutant NPC2 proteins, transfer of cholesterol from 1 µM WT or mutant NPC2 to 125 µM 520 SUVs composed of either 100% EPC or 25 mol% LBPA/EPC was monitored at 25°C. Instrument settings to 521 ensure the absence of photobleaching were established before each experiment. Data were analyzed 522 with the Pro-Data SX software provided with the Applied Photophysics stopped-flow 523 spectrofluorometer, and the cholesterol transfer rates were obtained by single exponential fitting of the 524 curves, as previously described (McCauliff et al., 2015;Xu et al., 2008). 525 526

Membrane-membrane interaction 527
Effects of NPC2 on vesicle-vesicle interactions were assessed in two ways, both using light scattering 528 approaches. 200 µM LUVs were mixed with 1 µM WT or mutant NPC2 proteins in a 96-well plate 529 reader, and absorbance at 350 nm monitored every 10 seconds over a period of 30 minutes 530 (Petrusevska et al., 2013). Increases in A350nm (light scattering) are indicative of vesicle-vesicle 531 interaction, the rate of which was determined by a three-parameter hyperbolic fit of the data using 532 Sigma Plot software. Additionally, 750 µM LUVs were mixed with 1 µM WT or mutant NPC2 proteins in a 533 spectrophotometer (Hitachi U-2900, Pleasanton, CA) and A350nm was measured over a period of 60 534 seconds (Schulz et al., 2009); rates of vesicle-vesicle interaction were obtained by single exponential 535 fitting of the curves. 536 537

Lipid blot analysis of NPC2 interactions 538
To assess WT NPC2 binding to various LBPA isomers, LBPA Snoopers (Avanti), containing 1 μg spots of 539 pure LBPA isomers, were blocked with tris-buffered saline (TBS) (0.8% NaCl, 20mM Tris-HCl pH 7.4) + 3% 540 BSA (fatty-acid free), followed by a one hour incubation at room temperature with 5 μg of WT NPC2 in 541 TBS + 3% BSA, at a final concentration of 0.5 μg/ml protein. The protein solution was removed and the 542 Snoopers were washed with TBS. NPC2 bound to LBPA isomers was detected by incubating the 543 Snoopers with rabbit polyclonal anti-c-myc-tag antibody (RRID: AB_914457, GenScript, Piscataway, NJ) 544 at a concentration of 0.5μg/ml in TBS + 3% BSA for one hour at room temperature. Following removal 545 of the primary antibody, the strips were washed with TBS and incubated with anti-rabbit IgG HRP-546 conjugated antibodies (RRID: AB_772206, GE Healthcare,Pittsburgh, PA) at a 1:20,000 dilution in TBS + 547 3% BSA. After a one-hour incubation with the secondary antibody, the Snoopers were washed with TBS 548 + 0.05% Tween and developed with ECL reagents (GE Healthcare). 549 For further analysis of NPC2-lipid interaction, Hybond-C membranes (GE-Healthcare) were spotted with 550 either 500 pmol of 18:1 LBPA/BMP (S,R) (Avanti), 18:1 PA (Avanti), 18:1 PG (Avanti), 18:1 PA (Avanti), 551 and Egg PC, or with increasing concentrations of LBPA (125, 250, 375 and 500 pmol) to analyze binding 552 of WT NPC2 to various phospholipid species, or binding of NPC2 mutants to LBPA, respectively. 553 Following the protocol of Dowler et al (Dowler et al., 2002), each phospholipid was spotted in duplicate 554 and allowed to dry for one hour. Membranes were blocked for 1 hour in blocking buffer containing TBS 555 (50 mM Tris/HCl, pH 7.5, 150 mM NaCl) and 5% (w/v) non-fat dry milk. Membranes were then 556 incubated overnight at 4°C with either WT or mutant NPC2 protein diluted to a final concentration of 557 1ug/ml in TBS and 3% (w/v) non-fat dry milk. The membranes were then washed at room temperature 558 in TBST (0.1% Tween 20) six times for 5 minutes each, followed by incubation with mouse monoclonal 559 anti-myc antibody (RRID: AB_309938, Millipore) at a 1:2,000 dilution in TBS and 3% (w/v) milk. After 560 washing with TBST, the membrane was then incubated with anti-mouse IgG IRDye-800CW conjugated 561 antibody (LI-COR, Lincoln, NE) at a 1:10,000 dilution in TBS, 0.1% SDS, and 3% (w/v) milk. The 562 membranes were finally washed in TBST 12 times for 5 minutes each at room temperature before 563 acquiring images on the LI-COR Odyssey. 564 565

Homogenous Time Resolved Fluorescence (HTRF) 566
Assays were performed as described in Fleury, et al. (Fleury et al., 2015), with minor modifications. Elmer; λex = 320 nm, λem = 615 and 665 nm; 100 μs delay time). The HTRF ratio value was represented 579 as Ch1/Ch2*10,000 where Ch1 is the energy transfer signal at 665nm, and Ch2 is the europium cryptate 580 antibody signal at 615nm. The negative control wells contained donor and acceptor fluorochromes 581 without NPC2 or biotinylated lipid. The negative control (background) readout was subtracted from all 582 the sample readings. 583 584 Clearance of cellular cholesterol by NPC2 proteins 585 As detailed previously, a single dose of purified WT or mutant NPC2 protein was added to the media of 586 NPC2 mutant fibroblasts cultured on 8-well tissue culture slides (Nalgene), and allowed to incubate for 3 587 days. The final concentration of added protein was 0.4 nM. Cells were then fixed and stained with 0.05 588 mg/mL filipin III (Fisher) and subsequently imaged on a Nikon Eclipse E800 epifluorescence microscope 589 using a DAPI filter set. Filipin stain area was quantified with the accompanying NIS-Elements software 590 (Nikon Inc.) and results were calculated as the ratio of filipin area to total cell area as described 591 previously (McCauliff et al., 2011;McCauliff et al., 2015). 592 593 Cellular LBPA enrichment via PG supplementation 594 WT, NPC1-, and NPC2 mutant fibroblasts were cultured to confluence in 100mm petri dishes and 595 passaged by trypsinization at a 1:3 ratio. After 24-48 hours, media was removed and replaced by media 596 supplemented with either 30, 100, or 250 µM PG SUVs (Bouvier et al., 2009;Luquain-Costaz et al., 597 2013), or with 100 µM LBPA or PC SUVs. After 24 hours, cells were collected and 4-6 dishes of the same 598 treatment were pooled. Protein levels were analyzed using the Bradford method (Bradford, 1976). 599 Total cell lipids were extracted from 2 mL of 1 mg/mL protein via the method of Bligh and Dyer (Bligh & 600 Dyer, 1959), resuspended in 200 µL 2:1 chloroform:methanol, and were run on HPTLC plates (EMD 601 Chemicals, Inc) in a solvent of 65:35:5 chloroform:methanol(v/v):30% ammonium hydroxide(v/v) (Akgoc 602 et al., 2015). Lipid spots were quantified by densiometric analysis (ImageJ) from standard curves of 603 authentic standards. TLC results were confirmed with an anti-LBPA antibody (6C4), graciously provided 604 by Jean Gruenberg. Briefly, WT, NPC1-and NPC2 mutant fibroblasts were plated onto 8-well tissue 605 culture slide (BD Falcon) at a density of approximately 6,000 cells/well. Cells were exposed to media 606 containing 100 µM PG SUV, as above, and washed with PBS following the twenty-four hour incubation 607 period. Cells were fixed with 4% paraformaldehyde, rinsed with PBS, and exposed to the 6C4 primary 608 antibody at a 1:1000 dilution in PBS for at least one hour at room temperature or overnight at 4°C. 609 Following removal of the primary antibody, cells were rinsed with PBS, permeabilized with 50 µg/ml 610 saponin in a 10% FBS blocking solution for 5 minutes, rinsed with PBS again and finally exposed to anti-611 mouse IgG Alexa 488-conjugated secondary antibody (Abcam, Cambridge, MA) at a dilution of 1:200 in 612 PBS for one hour at room temperature. All cells were imaged on a Nikon Eclipse E800 epifluorescence 613 microscope using a FITC filter set to detect the Alexa-488 secondary antibody. Alexa-488 stain area was 614 quantified with the accompanying NIS-Elements software (Nikon Inc) and LBPA accumulation was 615 calculated as the ratio of antibody stain area to cell area. 616 617 Clearance of cholesterol by PG supplementation 618 WT, NPC1-, and NPC2 mutant fibroblasts were plated onto 8-well tissue culture slides (BD falcon) at a 619 density of approximately 20,000 cells per well and incubated at 37 o C, 5% CO2 for 24 hours. Culture 620 media was then removed and the cells were incubated with media supplemented with 100 µM PG SUVs 621 for 24 hours. Cells were subsequently fixed and stained with 0.05 mg/mL filipin III and anti-LBPA 622 primary antibodies (6C4), followed by an Alexa 488-conjugated secondary antibody, as described above. 623 In some experiments, NPC2 deficient fibroblasts were secondarily incubated for 24 hours with WT or 624 mutant NPC2 proteins at a final concentration of 0.4 nM, 24 hours after the 24 hour supplementation 625 with PG. Cells were imaged on a Nikon Eclipse E800 epifluorescence microscope using a DAPI filter set 626 to detect filipin and a FITC filter set to detect the Alexa-488 secondary antibody to LBPA. Filipin and 627 antibody stain areas were quantified with the accompanying NIS-Elements software (Nikon Inc.); 628 cholesterol and LBPA accumulation was calculated as the ratio of filipin or antibody area, respectively, to 629 cell area McCauliff et al., 2011;McCauliff et al., 2015).  Figure 1. OPM predicts NPC2 stably inserts into membranes via a conserved hydrophobic knob region. (A) Holo bovine NPC2 (PDB ID: 2HKA) is predicted by OPM to insert its prominent hydrophobic knob region (red ribbon) into the hydrophobic space of a model membrane, positioning the cholesterol in the sterol binding pocket in close proximity to the membrane surface (red line). (B) Multiple sequence alignment of human NPC2 (NCBI Accession: NP_006423.1), rat NPC2 (NP_775141.2) mouse NPC2 (NCBI Accession: NP_075898.1), bovine NPC2 (NCBI Accession: NP_776343.1), cat NPC2 (XP_003987882.1), chimpanzee NPC2 (NP_001009075.1) and the yeast NPC2 (NCBI Accession: KZV12184.1) were aligned with CLUSTAL Omega, and alignment for hydrophobic knob residues H56 to V64 are shown. Consensus sequences are in black and conserved residues are boxed. (C) Conservation scores and hydrophobicity scores for the hydrophobic knob, residues H56 to V64, were calculated based on the PAM250 scoring matrix and Kyte & Doolitle Hydrophobicity scale (Kyte & Doolittle, 1982). (D) The aligned NPC2 protein sequences were analyzed with the ProtScale Tool on ExPASy server based on the Kyte and Doolittle Amino acid Hydropathicity scale with a frame window of 15 residues (Gasteiger et al., 2005).  µM WT or mutant NPC2 to 125 µM 100% EPC or 25% LBPA/EPC vesicles was measured on an SX20 Stopped Flow Spectrofluorometer by monitoring the dequenching of NPC2 endogenous tryptophan fluorescence. All curves were well fit using a single exponential function using the Applied Photophysics Pro-Data Viewer software. Mutants with rates of cholesterol transfer less than 50% of WT NPC2 were considered to have defective transfer kinetics properties and their relative rates are italicized. Data are representative of 3 experiments, each consisting of 2-3 individual runs. Absolute and relative rates of transfer to each model membrane system, ±SE, are shown. (B) NPC2 point mutations resulting in defective cholesterol transport to 100% EPC vesicles are shown in red while mutations having little or no effect on NPC2 cholesterol transport properties relative to WT protein, are shown in green. (C) Point mutations with attenuated rates of cholesterol transport to 25% LBPA/EPC vesicles are shown in red while mutations having little or no effect, relative to WT protein, are shown in green. (A) WT NPC2 protein was incubated with strips (Snoopers) containing LBPA isomers. LBPA-bound protein was detected with anti-c-myc antibody as described under Methods, and degree of binding ± SE (n=3) is represented by the integrated density of the blots. (B) 500 pmol of various membrane phospholipids were spotted onto nitrocellulose strips and probed with WT NPC2-myc-his protein as described under Methods. Relative binding of WT NPC2 ± SE (n=5) is shown, represented by signal intensity detected with the LI-COR system. (C) Structures of the LBPA isomers. (D) 75nM of WT NPC2 protein was incubated with 1μM of the indicated biotin-C12-ether phospholipid, streptavidin-d2 conjugate and europium cryptate-labeled monoclonal anti-histidine antibody in detection buffer, as described in Methods. FRET signal between europium cryptate and streptavidin was detected with a HTRF capable Envision plate reader (λex = 320 nm, λem = 615 and 665 nm; 100 μs delay time; n=3). Figure 5. NPC2 binds to LBPA via the "hydrophobic-knob" domain. Binding of NPC2 WT and mutant proteins to (A) LBPA isomers and (B),18:1 (S,R) LBPA was detected using LBPA Snoopers and membranes spotted with 500 pmol phospholipid, respectively, as described under Methods. Relative binding is represented as (A) the integrated density of the blots, relative to WT NPC2, ±SE (n=3) and (B) the signal intensity detected with the LI-COR system ±SE (n=3) (C) HTRF analysis: WT or mutant NPC2 protein was incubated with biotin-C12-ether LBPA, streptavidin-d2 conjugate and europium cryptate-labeled anti-His antibody in detection buffer, as described in Methods. FRET signal between europium cryptate and streptavidin was detected with a HTRF capable Envision late reader as described in Methods. (D) FRET signal was analyzed with the One site-specific binding function (Graphpad) and the Bmax extrapolations were used to infer binding capacity between recombinant NPC2 protein and biotin-C12-ether LBPA. Results are representative of three experiments, with deviations < 20%.