a trimeric rab7 gef controls npc1-dependent lysosomal ... · 88 mc1-ypt7 complex suggests a gef...

1

A trimeric Rab7 GEF controls NPC1-dependent lysosomal 1

cholesterol export 2

3

Dick J.H. van den Boomen1*, Agata Sienkiewicz1, Ilana Berlin2, Marlieke L.M. Jongsma2, 4

Daphne M. van Elsland2, J. Paul Luzio3, Jacques J.C. Neefjes2 and Paul J. Lehner1* 5

6

1 Cambridge Institute for Therapeutic Immunology & Infectious Disease, University of 7

Cambridge, Cambridge, United Kingdom. 8

2 Leiden University Medical Centre, Leiden University, Leiden, the Netherlands. 9

3 Cambridge Institute for Medical Research, University of Cambridge, Cambridge, United 10

Kingdom 11

* Corresponding authors: (DJHB: [email protected], PJL: [email protected], correspondence 12

address: Cambridge Institute for Therapeutic Immunology & Infectious Disease, University of 13

Cambridge, Jeffrey Cheah Biomedical Centre, Cambridge Biomedical Campus, Puddicombe 14

Way, Cambridge CB2 0AW, United Kingdom) 15

16

Running title: A Rab7 GEF controls cholesterol export 17

18

Key words: C18orf8, Mon1, Ccz1, Rab7, GEF, LDL, cholesterol, NPC1, lysosome 19

20

.CC-BY 4.0 International licenseavailable under awas not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint (whichthis version posted November 8, 2019. ; https://doi.org/10.1101/835686doi: bioRxiv preprint

mailto:[email protected]

mailto:[email protected]

https://doi.org/10.1101/835686

http://creativecommons.org/licenses/by/4.0/

2

Summary 21

Cholesterol import in mammalian cells is mediated by the LDL receptor pathway. Here, using 22

an endogenous cholesterol reporter in a genome-wide CRISPR screen we identify >70 genes 23

involved in LDL-cholesterol import. We characterise C18orf8 as a core component of the 24

mammalian Mon1-Ccz1 guanidine exchange factor (GEF) for Rab7, required for complex 25

stability and function. C18orf8-deficient cells lack Rab7 activation and show severe defects in 26

late endosome morphology and endosomal LDL trafficking, resulting in cellular cholesterol 27

deficiency. Unexpectedly, free cholesterol accumulates within swollen lysosomes, suggesting 28

a critical additional defect in lysosomal cholesterol export. We find that active Rab7 interacts 29

with the NPC1 cholesterol transporter and licenses lysosomal cholesterol export. This process 30

is abolished in C18orf8-, Ccz1- and Mon1A/B-deficient cells and restored by a constitutively 31

active Rab7. The trimeric Mon1-Ccz1-C18orf8 (MCC) GEF therefore plays a central role in 32

cellular cholesterol homeostasis coordinating Rab7 activation, endosomal LDL trafficking and 33

NPC1-dependent lysosomal cholesterol export. 34

35



https://doi.org/10.1101/835686


3

Introduction 36

Cholesterol is an essential component of all mammalian lipid membranes and disturbances in 37

its homeostasis associate with a variety of diseases, including Niemann Pick type C, familial 38

hypercholesterolemia, atherosclerosis and obesity. Cellular cholesterol homeostasis is 39

transcriptionally regulated by the SREBP2 (sterol response element binding protein 2) 40

transcription factor. When cholesterol is abundant SREBP2 and its trafficking factor SCAP are 41

retained in the endoplasmic reticulum (ER), whereas upon sterol depletion they are released 42

to the Golgi, where SREBP2 is cleaved and traffics to the nucleus to activate the transcription 43

of genes involved in de novo cholesterol biosynthesis (e.g. HMGCS1, HMGCR, SQLE) and 44

LDL-cholesterol uptake (e.g. LDLR, NPC1/2) (1). SREBP2 thus couples cholesterol import 45

and biosynthesis to cellular cholesterol availability. 46

47

Cholesterol is taken up into mammalian cells in the form of low-density lipoprotein (LDL) 48

particles, composed predominantly of cholesterol esters (CEs), cholesterol and apolipoprotein 49

B-100 (2). LDL binds to the cell surface LDL receptor (LDLR) and the LDLR-LDL complex is 50

internalised via clathrin-mediated endocytosis (3). In the acidic pH of sorting endosomes, LDL 51

dissociates from LDLR, which recycles back to the plasma membrane. LDL is itself transported 52

to a late endosomal / lysosomal (LE/Ly) compartment, where cholesterol is released from CEs 53

by lysosomal acid lipase (LAL/LIPA). Free cholesterol then binds the Niemann Pick C2 (NPC2) 54

carrier protein and is transferred to the Niemann Pick C1 (NPC1) transporter for subsequent 55

export to other organelles (4). 56

57

NPC1 is a central mediator of lysosomal cholesterol export. Mutations in NPC1 and NPC2 58

cause Niemann Pick type C, a lethal lysosomal storage disease characterised by lysosomal 59

cholesterol accumulation in multiple organ systems. NPC1 encodes a complex polytopic 60

glycoprotein embedded in the lysosomal membrane (5). Cholesterol-loaded NPC2 binds the 61



https://doi.org/10.1101/835686


4

NPC1 middle luminal domain (6, 7) and delivers its cargo to the mobile N-terminal domain (8, 62

9), from where cholesterol is likely transferred to the sterol-sensing domain (SSD). Whether 63

cholesterol is transported through NPC1 in a channel-like manner or simply inserted into the 64

luminal membrane leaflet remains unclear (10). NPC1 is believed to be constitutively active 65

and no regulators of its activity have been identified. Once transported across the lysosomal 66

membrane, cholesterol transfer to other organelles is mediated by lipid transfer proteins 67

(LTPs) at inter-organelle membrane contact sites (MCS) (11). Although a variety of proteins 68

have been implicated in MCS formation with LE/Ly (12-15), the identity of the direct carrier 69

transporting cholesterol from LE/Ly remains controversial. 70

71

The LE/Ly compartment plays a central role in the cellular LDL-cholesterol uptake pathway. 72

LE homeostasis and substrate trafficking are regulated by the small GTPase Rab7 (16). Rab7 73

activity is controlled by its nucleotide status: its activation requires GDP-to-GTP exchange by 74

a guanidine exchange factor (GEF), whereas GTP hydrolysis induced by GTPase activating 75

proteins (GAPs) triggers Rab7 inactivation. Active GTP-bound Rab7 associates with LE 76

membranes and recruits Rab7 effector proteins which in mammalian cells include the 77

endosome motility factor RILP (17, 18), the cholesterol binding protein ORP1L (19, 20), the 78

VPS34 phosphatidylinositol (PtdIns) 3-kinase regulators Rubicon and WDR91 (21, 22) and 79

the retromer components VPS26 and VPS35 (23, 24). 80

81

Temporal and spatial control of Rab7 activation is of major importance. At least three GAPs – 82

TBC1D5, TBC1D15 and Armus (24-26) are implicated in Rab7 inactivation, whereas activation 83

of the yeast Rab7 homologue Ypt7 is mediated by the Mon1-Ccz1 (MC1) complex (27, 28). 84

MC1 is recruited to endosomal membranes by the phospholipid PtdIns(3)P (29) and its 85

activation of Rab7 drives Rab5-to-Rab7 conversion, endosome maturation and fusion with the 86

vacuolar/lysosomal compartment (30-33). A partial crystal structure of the C. thermophilum 87



https://doi.org/10.1101/835686


5

MC1-Ypt7 complex suggests a GEF model in which MC1 binding to Ypt7 induces magnesium 88

expulsion and GDP dissociation from the Ypt7 active site (34). Mammals have two Mon1 89

orthologues – Mon1A and Mon1B – and a single Ccz1 orthologue. Although mammalian MC1 90

also likely acts as a Rab7 GEF (35, 36), its in vivo function remains poorly characterised. 91

92

In this study, we performed a genome-wide CRISPR screen for essential genes in cholesterol 93

homeostasis using an endogenous SREBP2-dependent cholesterol reporter. We identified 94

>70 genes mainly involved in the LDL-cholesterol uptake pathway, including the poorly 95

characterised C18orf8. C18orf8 is a novel core component of the mammalian MC1 complex, 96

essential for complex stability and function. C18orf8-deficient cells exhibit severe defects in 97

Rab7 activation and LDL trafficking, concomitant with swelling of the LE/Ly compartment and 98

marked lysosomal cholesterol accumulation. We show that active Rab7 interacts with the 99

NPC1 cholesterol transporter to license lysosomal cholesterol export - a pathway deficient in 100

C18orf8-, Ccz1- and Mon1A/B-deficient cells and restored by a constitutively active Rab7 101

(Q67L). Our findings therefore identify a central role for the trimeric Mon1-Ccz1-C18orf8 102

(MCC) GEF in cellular LDL-cholesterol uptake, coordinating Rab7 activation with LDL 103

trafficking and NPC1-dependent lysosomal cholesterol export. 104

105



https://doi.org/10.1101/835686


6

Results 106

107

Generation of an endogenous, SREBP2-dependent, fluorescent cholesterol reporter 108

To screen for genes that maintain cellular cholesterol homeostasis, we engineered a cell line 109

expressing an endogenous fluorescent reporter responsive to intracellular cholesterol levels. 110

HMG-CoA synthase 1 (HMGCS1) catalyses the second step of cholesterol biosynthesis, the 111

conversion of acetoacetyl-CoA and acetyl-CoA to HMG-CoA. With two sterol-response 112

elements (SRE) in its promoter, HMGCS1 expression is highly SREBP2 responsive and thus 113

cholesterol sensitive (37), rendering it well suited to monitor cellular cholesterol levels. We 114

used CRISPR technology to knock-in the bright fluorescent protein Clover (38) into the 115

endogenous HMGCS1 locus yielding an endogenous HMGCS1-Clover fusion protein (Fig 1A, 116

S1A). Sterol depletion of the HMGCS1-Clover cell line, using mevastatin and lipoprotein-117

depleted serum (LPDS), increased basal HMGCS1-Clover expression 9-fold (Fig 1C, S1B, 118

S1C) and revealed the expected cytoplasmic localisation of the HMGCS1-Clover fusion 119

protein (Fig 1B). Expression returned to baseline following overnight sterol repletion (Fig 120

S1D). Sterol-depletion induced HMGCS1-Clover expression was abolished upon knockout of 121

SREBP2, but not its close relative SREBP1 (Fig 1C, S1E). SREBP2-deficiency also slightly 122

decreased steady-state HMGCS1-Clover levels (Fig S1F), suggesting low-level SREBP2 123

activation under standard growth conditions. Unlike HMG-CoA reductase (HMGCR), 124

HMGCS1 expression is not regulated by sterol-induced degradation (Fig S1G). HMGCS1-125

Clover is therefore a sensitive endogenous, SREBP2-dependent, transcriptional reporter for 126

intracellular cholesterol levels. 127

128

A genome-wide CRISPR screen identifies components essential for cellular cholesterol 129

homeostasis 130



https://doi.org/10.1101/835686


7

To identify genes critical to the maintenance of cellular cholesterol levels, we undertook a 131

genome-wide CRISPR screen using our HMGCS1-Clover reporter as an intracellular 132

cholesterol sensor. CRISPR knockout of these genes is predicted to induce spontaneous 133

cholesterol depletion, SREBP2 activation and HMGCS1-Clover upregulation, despite sterol 134

replete (high cholesterol) culture conditions. These cells thus develop a spontaneous 135

HMGCS1-Cloverhigh phenotype. HMGCS1-Clover reporter cells were targeted with a genome-136

wide CRISPR library containing 10 sgRNAs per gene (total sgRNA size 220,000) (39) and 137

were sorted in two successive rounds of flow cytometry for the rare HMGCS1-Cloverhigh 138

phenotype (Fig 1D). This yielded an enriched population with a 60% HMGCS1-Cloverhigh 139

phenotype under sterol replete conditions. Next-generation sequencing of the population 140

revealed sgRNA enrichment for 76 genes (P<10-5) (Fig 1E, Table S1). The results of our 141

HMGCS1-Clover screen emphasise the central role of LDL-cholesterol uptake and membrane 142

trafficking in cellular cholesterol homeostasis, with top hits including LDLR and NPC1/2. 143

144

Functionally, the hits can be grouped in categories (Fig 1E) that include: i) protein folding and 145

glycosylation (yellow, e.g. HSP90B1, GNTAP, SLC35A1); ii) trafficking through the early 146

secretory pathway (pink, e.g. ARF1, CHP1, COG3/4, EXOC5/7/9); iii) endocytosis (purple, 147

e.g. AP2M1/S1, EHD1, FCHO2, DNM2); iv) endosomal trafficking (green, e.g. RabGEF1, 148

Rab7A, V-ATPase (ATP6AP1/AP2V1A/V1B2/V1H), ESCRT (PTPN23, HRS), CORVET 149

(VPS8/16/18/TGFBRAP1), HOPS (VPS16/18/39/41), FERARI (ZFYVE20/RBSN, VPS45), 150

PI3K (BECN, UVRAG, PIK3R4), PIKfyve (FIG4, VAC14)); and v) cholesterol import (blue, e.g. 151

NPC1/2). These categories reflect successive stages in LDL-cholesterol import (Fig 1F), 152

respectively: i) LDLR folding and glycosylation; ii) trafficking of LDLR to the cell surface; iii) 153

endocytosis of the LDL-LDLR complex; iv) trafficking of internalised LDL to the LE/Ly 154

compartment; and v) lysosomal cholesterol release and NPC1-dependent lysosomal 155

cholesterol export. Besides the LDLR pathway, our screen identified components of the 156

SREBP transcriptional machinery (Fig 1E, orange, INSIG1, SREBP1) and several poorly 157



https://doi.org/10.1101/835686


8

characterised gene products (grey/red e.g. CLU/APOJ, EGR3, ADAM28, FAM160B2, 158

STMN4, C18orf8). To validate our screen results, we confirmed a subset of hits (LDLR, NPC1, 159

AP21, ANKFY1, VPS16, RBSN/ZFYVE20, INSIG1, SREBF1) using individual sgRNAs. All 160

sgRNA-treated cells showed elevated HMGCS1-Clover expression (Fig S2), suggesting 161

defective LDL-cholesterol import and/or spontaneous SREBP2 activation. 162

163

Our cholesterol reporter screen therefore provides a comprehensive overview and unique 164

insight into known and unknown factors essential for LDL-cholesterol import into mammalian 165

cells (Fig 1F), and emphasises the critical role of endo-/lysosomal trafficking within this 166

process. 167

168

C18orf8 is required for endo-lysosomal LDL-cholesterol uptake 169

A prominent uncharacterised gene in our screen is C18orf8 (Fig 1E, red), which encodes a 170

72kDa soluble protein, conserved from animals to plants with no identifiable yeast orthologue. 171

Structure prediction (Phyre2, HHPred) suggests high structural similarity to the N-terminus of 172

clathrin heavy chain, with an N-terminal -propeller composed of WD40 repeats and a C-173

terminal -helical domain (Fig S4E). Organelle proteomics have previously identified C18orf8 174

as a lysosome (40) and endosome (41) associated protein. 175

176

To further characterise C18orf8 function, we generated C18orf8 knockout clones using two 177

independent sgRNAs (Fig 2A). Consistent with our screening results, C18orf8-deficient cell 178

clones showed elevated HMGCS1-Clover expression under sterol-replete culture conditions 179

(Fig 2B, red line) – implying spontaneous cholesterol depletion – and this phenotype was 180

reversed upon re-expression of HA-tagged C18orf8 (C18orf8-3xHA, green line). Cholesterol 181

deficiency could result from defective exogenous LDL-cholesterol uptake or defective 182



https://doi.org/10.1101/835686


9

endogenous biosynthesis. To differentiate these pathways, we inhibited exogenous uptake by 183

culturing cells in LPDS or blocked endogenous biosynthesis with mevastatin (Fig 2C). Wild-184

type reporter HeLa cells showed no response to mevastatin alone (Fig 2C, left panel, red 185

line), upregulated HMGCS1-Clover in LPDS (green line) and showed maximum reporter 186

induction in LPDS and mevastatin (blue line). In contrast, C18orf8-deficient cells showed 187

elevated steady-state HMGCS1-Clover expression (Fig 2C, middle/right panels, black 188

lines), were unresponsive to LPDS (green lines), but addition of mevastatin alone induced 189

maximum HMGCS1-Clover induction (red lines) with no further increase in LPDS + 190

mevastatin (blue line). These results show that while wild-type HeLa cells rely predominantly 191

on external cholesterol and switch to endogenous biosynthesis when LDL is unavailable, 192

C18orf8-deficient cells are completely reliant on endogenous cholesterol biosynthesis under 193

all conditions. This effect is likely explained by a defect in LDL-cholesterol uptake. 194

195

C18orf8-deficient cells showed normal LDLR cell surface expression (Fig S3A) and uptake of 196

fluorescent Dil-LDL (Fig 2D). However, whereas in wild-type cells, Dil-LDL fluorescence 197

disappeared within 3 hours of pulse-labelling, in C18orf8-deficient cells Dil-LDL accumulated 198

during the 3-hour chase (Fig 2E), a phenotype rescued upon complementation with C18orf8-199

3xHA. To determine whether this defect is LDL-specific, or derives from a general endo-200

lysosomal trafficking defect, we stimulated cells with fluorescent-labelled EGF. Similar to Dil-201

LDL, EGF was degraded within 3 hours in wild-type cells, but accumulated during the 3h chase 202

in C18orf8-deficient cells (Fig S3B). C18orf8-deficient cells thus show a general defect in 203

endo-lysosomal degradation that affects both LDL and EGF and results in their reliance on de 204

novo cholesterol biosynthesis for cholesterol supply. 205

206

C18orf8-deficient cells are defective in late endosome morphology and early-to-late 207

substrate trafficking 208



https://doi.org/10.1101/835686


10

Consistent with a defect in LDL/EGF degradation, C18orf8-deficient cells showed a severe 209

disruption of endosome morphology, in particular of the LE/Ly compartment (Fig 3A, 3B, 210

S3C). Perinuclear Rab7+ LEs were markedly swollen and surrounded by equally swollen 211

LAMP1+ LE/Lys, with EEA1+/Rab5+ EEs clustered between Rab7+ LEs in the perinuclear 212

region. By electron microscopy (EM), the swollen LE/Lys appeared as enlarged multivesicular 213

bodies (MVBs) with a high intraluminal vesicle (ILV) content and a 3-fold increase in average 214

diameter (Fig S3D). Complementation of C18orf8-deficient cells with C18orf8xHA was slow, 215

but after 14 days restored endosome morphology (Fig S3E) and function (Fig 2E) back to 216

wild-type. This prolonged recovery likely reflects the severity of the cellular phenotype. 217

218

To assess how this abnormal morphology affects substrate trafficking, we pulse-labelled wild-219

type and C18orf8-deficient cells with fluorescent-labelled EGF. In both cell lines EGF moved 220

to EEA1+ EE within ~20min (Fig 3C). However, whereas in wild-type cells EGF disappeared 221

from EE at 1-3 hours and was degraded (Fig 3C, top panel), in C18orf8-deficient cells no 222

degradation occurred and EGF remained in EEA1+ EE for the duration of the 3 hours chase 223

(bottom panel), implying a defect in early-to-late substrate trafficking. Indeed, inhibiting EGF 224

degradation with Leupeptin, E-64d and Pepstatin showed EGF entry into LAMP1+ LE/Ly at 1-225

3 hours in wild-type (Fig 3D, top panel), but not C18orf8-deficient cells, where co-localisation 226

remained minimal throughout the chase (bottom panel). As seen with EGF, C18orf8-deficient 227

cells also accumulated Dil-LDL in EEA1+ EE for the duration of a 3 hours chase (Fig 3E) and 228

trafficking of the fluid phase dye Sulforhodamine 101 (SR101) (42) into lysotracker+ LE/Ly 229

was delayed (Fig S3F). C18orf8 therefore plays an essential role in early-to-late and/or late 230

endosomal trafficking that precedes substrate degradation in lysosomes. 231

232

C18orf8 is an integral component of the Mon1-Ccz1 complex 233



https://doi.org/10.1101/835686


11

To further characterise C18orf8 function, we identified C18orf8-interacting proteins by mass 234

spectrometry. Pull-down of N- or C-terminal HA-tagged C18orf8 revealed strong interactions 235

with three proteins – Ccz1, Mon1A and Mon1B – the three components of the mammalian 236

MC1 complex (Fig 4A). The interaction between overexpressed C18orf8 and endogenous 237

Ccz1 and Mon1B was readily confirmed by immunoblot (Fig S4A). To visualise the interaction 238

between endogenous proteins, we used CRISPR technology to knock-in a 3xMyc-tag into the 239

C18orf8 locus, yielding an endogenous C18orf8-3xMyc fusion (Fig S4B, C). Pull-down of 240

C18orf8-3xMyc precipitated Ccz1 and Mon1B (Fig 4B), and conversely immune precipitation 241

of endogenous Mon1B (Fig 4B) or Ccz1 (Fig S4D) revealed the endogenous 3xMyc-tagged 242

C18orf8. C18orf8 is composed of an N-terminal WD40 and C-terminal -helical domain. To 243

identify the MC1-interaction site in C18orf8, we expressed mScarlet-tagged single domains 244

(Fig S4E). Immune precipitation revealed an interaction of Ccz1 and Mon1b with the C-245

terminal (AA 354-657), but not N-terminal (AA 1-362) domain of C18orf8 (Fig S4F). 246

247

The robust interaction of C18orf8 with all three MC1 components suggests C18orf8 is either 248

a regulator or integral component of the mammalian MC1 complex. Interestingly, while 249

C18orf8-deficient cells expressed normal levels of Ccz1, they showed a marked reduction in 250

Mon1B expression (Fig 4C), which was restored by re-expression of full-length C18orf8 (Fig 251

4C), its C-terminal MC1-interaction domain (Fig S4F) or proteasome inhibition (Fig S4G). 252

Mon1B expression was also decreased in Ccz1-deficient cells, while conversely endogenous 253

C18orf8 expression was reduced in Ccz1 and Mon1A/B double-deficient cells (Fig 4D). 254

Mon1B stability is therefore dependent on C18orf8 and Ccz1 and in the absence of either, 255

Mon1B is proteasomally degraded. Similarly, C18orf8 stability depends on Mon1A/B and 256

Ccz1, whereas Ccz1 remains relatively stable in the absence of both other components (Fig 257

4D). We conclude that mammalian Mon1, Ccz1 and C18orf8 form a stable trimeric complex, 258

the MCC complex, of which C18orf8 is a core component. 259

260



https://doi.org/10.1101/835686


12

To determine whether the function of C18orf8 overlaps with mammalian MC1, we knocked-261

out Ccz1, Mon1A, Mon1B and Rab7 in our HMGCS1-Clover reporter line. Similar to C18orf8 262

depletion, loss of Ccz1 or a combination of Mon1A and Mon1B, but neither alone, increased 263

HMGCS1-Clover expression (Fig 4E), as did knockout of Rab7, a prominent hit from our 264

screen (Fig 1E, 4E). This phenotype was restored by complementation of knockouts with their 265

respective wild-type proteins (Fig S4I), with Mon1A/B-double-deficient cells rescued by 266

Mon1A or Mon1B alone. Like C18orf8-deficient cells, cells deficient for Ccz1- or Mon1A/B-267

deficient contained abnormal, enlarged Rab7+ and LAMP1+ LE/Ly compartments (Fig 4F). 268

This phenotypic similarity implies C18orf8 is required for both MCC stability and functionality. 269

Importantly, although the C18orf8 -helical domain is necessary and sufficient for MCC 270

binding and Mon1B stabilization (Fig S4F), it is insufficient for full complementation of the 271

C18orf8-deficiency phenotype which required full-length C18orf8 (Fig S4H). C18orf8 is 272

therefore not simply a stabilizing factor but an active integral component of the MCC complex. 273

274

The Mon1-Ccz1-C18orf8 complex is responsible for mammalian Rab7 activation 275

Yeast MC1 acts as an activating GEF for the yeast Rab7 homologue Ypt7 (28). To assess 276

whether the mammalian MCC complex has a similar function, we expressed HA-tagged wild-277

type, dominant-negative (T22N) and constitutively active (Q67L) Rab7 in C18orf8-3xMyc 278

knock-in cells and probed its interaction with endogenous MCC components. Consistent with 279

a role in Rab7 activation, C18orf8-3xMyc, Ccz1 and Mon1B preferentially bound the inactive 280

Rab7-T22N, but not a constitutively active Rab7-Q67L mutant (Fig 5A). Rab7 activation 281

promotes binding and LE recruitment of its effectors RILP and ORP1L (17-20) and effector 282

binding is commonly used to determine the activation status of Rab GTPases. In wild-type 283

cells, an interaction between FLAG-tagged RILP and endogenous Rab7 was readily detected 284

- indicating the presence of an active GTP-bound Rab7, whereas this interaction was lost in 285

C18orf8-, Ccz1- and Mon1A/B-deficient cell lines (Fig 5B). MCC-deficient cells are thus 286

unable to activate Rab7, suggesting MCC acts as a mammalian Rab7 GEF. Indeed, HA-287



https://doi.org/10.1101/835686


13

tagged RILP was efficiently recruited to LAMP1+ LE/Ly in wild-type, but not in C18orf8-288

deficient cells, where RILP redistributed to the cytoplasm (Fig 5C). LE recruitment of the Rab7 289

effector ORP1L was also strongly reduced (Fig S5A). Consistent with RILP’s function in 290

endosome mobilisation, LE motility was sharply decreased in C18orf8-deficient cells (Fig 291

S5B). C18orf8, Ccz1 and Mon1A/B are therefore required for Rab7 activation and functional 292

recruitment of the Rab7 effectors RILP and ORP1L in mammalian cells. 293

294

C18orf8 function can be by-passed by a constitutively active Rab7 or depletion of 295

Rab7GAPs 296

The activation of Rab GTPases by RabGEFs is counteracted by RabGAPs that enhance 297

intrinsic GTPase activity and revert Rab GTPases back to an inactive state. Our data suggest 298

that LE defects in C18orf8-deficient cells are secondary to defective Rab7 activation. We 299

therefore asked whether LE function could be rescued by concomitantly blocking one or more 300

Rab7GAPs; or by the use of a constitutively active Rab7-Q67L mutant that lacks intrinsic 301

GTPase activity. Knockdown of the Rab7GAPs TBC1D5 and TBC1D15, but not either alone, 302

rescued HMGCS1-Clover levels in C18orf8-deficient cells back to wild-type (Fig 5D) and a 303

similar rescue was obtained by expression of the constitutively active Rab7-Q67L, but not the 304

inactive Rab7-T22N (Fig 5E). Rab7-Q67L also restored the swollen LE/Ly phenotype (Fig 5F, 305

green) and Dil-LDL degradation (red) in C18orf8-deficient cells. In contrast, the T22N mutant 306

augmented both phenotypes. The trimeric Mon1-Ccz1-C18orf8 (MCC) complex therefore acts 307

as an activating GEF for mammalian Rab7 and its essential role in late endosomal trafficking 308

and LDL-cholesterol uptake can be by-passed with a constitutively active Rab7 (Q67L) or by 309

Rab7GAP depletion. 310

311

MCC-deficient cells accumulate free cholesterol in their lysosomal compartment. 312



https://doi.org/10.1101/835686


14

How does a defect in Rab7 activation result in cellular cholesterol deficiency? C18orf8-313

deficient cells showed a severe defect in LDL trafficking (Fig 2E, 3E). We assumed this defect 314

would cause impaired cholesterol ester (CE) hydrolysis, decreased cholesterol release and, 315

ultimately, cellular cholesterol deficiency. To ascertain the fate of LDL-derived cholesterol, we 316

stained our cells with Filipin III, a bacterial compound that specifically binds free cholesterol 317

and differentiates it from CEs in LDL (43). Remarkably, C18orf8-deficient cells did not show a 318

decrease in Filipin staining, but rather accumulated free cholesterol in their swollen Rab7+ 319

and LAMP1+ LE/Ly compartment (Fig 6A, Fig 6C). Cholesterol accumulation resolved upon 320

complementation with C18orf8-3xHA (Fig 6A) and was also observed in Ccz1- and Mon1A/B-321

deficient cells (Fig 6B). At the ultra-structural level cholesterol accumulation was confirmed 322

by Theonellamides (TNM) staining (44, 45) which enriched within the enlarged MVBs of 323

C18orf8-deficient cells (Fig 6D). The accumulation of free cholesterol in LE/Ly of MCC-324

deficient cells suggests that defective lysosomal cholesterol export, as opposed to LDL 325

trafficking, is primarily responsible for the cellular cholesterol deficiency in MCC-deficient cells. 326

Indeed the MCC-deficiency phenotype is strikingly similar to that observed in cells defective 327

for the NPC1 cholesterol transporter (Fig 6E). 328

329

Active Rab7 interacts with the NPC1 cholesterol transporter 330

The NPC1 transporter is critical for lysosomal export of LDL-derived cholesterol and cellular 331

cholesterol homeostasis (Fig 1E, 1F) and NPC1 mutations are the primary cause of Niemann 332

Pick type C lysosomal storage disease. C18orf8-deficient cells show normal to elevated levels 333

of NPC1 expression (Fig S6A), with increased NPC1 expression in Ccz1- and Mon1A/B-334

deficient cells (Fig S6B). In wild-type cells, NPC1 resides predominantly in the LAMP1+ Ly 335

compartment (Fig S6C), where it co-localises with NPC2 (Fig S6D) and this localisation was 336

unaltered in C18orf8-, Ccz1- and Mon1A/B-deficient cells (Fig S6C, D). Since NPC1 337

expression and localisation appeared normal, we used mass spectrometry to identify NPC1 338

interactions partners. Remarkably, among the most abundant interaction partners identified in 339



https://doi.org/10.1101/835686


15

NPC1 immune precipitations is Rab7 itself (Fig 6F). This interaction was not only confirmed 340

in Rab7:NPC1 immune precipitations but NPC1 preferentially bound to the constitutively active 341

Rab7-Q67L but not the inactive Rab7-T22N mutant (Fig 6G), as could be expected for a 342

functionally important Rab7 interaction. Furthermore, consistent with an activation-dependent 343

event, a robust interaction was observed between NPC1 and Rab7 in wild-type cells and this 344

interaction was lost in C18orf8-, Ccz1- or Mon1A/B-cells that are defective in Rab7 activation 345

(Fig 6H). The Rab7-NPC1 interaction is maintained by an inactive NPC1-P692S mutant and 346

is therefore independent of NPC1 cholesterol export function (Fig 6I). Indeed Rab7 binding to 347

NPC1 remains largely unaffected by treatment of cells with LPDS treatment, which decreases, 348

and the NPC1 inhibitor U18666A which increases lysosomal cholesterol levels (Fig 6J). 349

350

Rab7 activation by the MCC GEF drives NPC1-dependent cholesterol export 351

Loss of the Rab7-NPC1 interaction in MCC-deficient cells correlates with lysosomal 352

cholesterol accumulation, suggesting Rab7 activation is essential for NPC1-dependent 353

lysosomal cholesterol export. To test this hypothesis directly and avoid potential caveats 354

created by LDL trafficking defects, we set up a lysosomal cholesterol export assay analogous 355

to a pulse-chase. Cholesterol export was initially blocked using the reversible NPC1 inhibitor 356

U18666A (46) to induce free cholesterol accumulation in LE/Ly (pulse). The inhibitor was then 357

washed out in LDL-free medium (LPDS) which allowed lysosomal cholesterol efflux in the 358

absence of further LDL-cholesterol uptake (chase). Cholesterol release from LE/Ly was 359

monitored by Filipin staining (Fig 7A). During a 24h chase period, cholesterol was completely 360

exported from CD63+ LE/Ly in wild-type cells, whereas export was abolished in cells deficient 361

for the NPC1 transporter (Fig 7A, B). Similar to NPC1 deficiency, Ly cholesterol export was 362

blocked in C18orf8-, Ccz1- and Mon1A/B -deficient cells in which Filipin staining remained co-363

localised with CD63 during the 24h chase (Fig 7A, B). To confirm that the cholesterol export 364

defect in MCC-deficient cells depends on Rab7 activation, C18orf8-deficient cells were 365

complemented with different Rab7 mutants. While the cholesterol accumulation resolved upon 366



https://doi.org/10.1101/835686


16

expression of the constitutively active Rab7-Q67L, no resolution was seen with either the wild-367

type or inactive Rab7-T22N (Fig 7C), confirming Rab7 dependency of the cholesterol export 368

defect. 369

370

NPC1 function is compromised in type C Niemann Pick (NPC) disease in which lysosomal 371

cholesterol accumulates in multiple organs. The common NPC1I1061T mutation destabilizes the 372

NPC1 transporter, resulting in its ER retention, and subsequent degradation (47). However, 373

any transporter reaching its correct lysosomal location remains functional (48). We asked 374

whether increased Rab7 expression improves NPC1 function, promoting clearance of 375

lysosomal cholesterol in mutant NPC1 fibroblasts. Remarkably, lentiviral overexpression of a 376

GFP-tagged wild-type Rab7 strongly reduced cholesterol accumulation in NPC1I1061T/I1061T 377

patient fibroblasts (Fig 7D), suggesting Rab7 activity is limiting for lysosomal cholesterol 378

export in NPC disease. 379

380

Taken together our results show that Rab7 and its trimeric Mon1-Ccz1-C18orf8 (MCC) GEF 381

control cellular cholesterol homeostasis through coordinated regulation of LDL trafficking and 382

NPC1-dependent lysosomal cholesterol export, making it a potential therapeutic target in 383

Niemann Pick disease. 384

385



https://doi.org/10.1101/835686


17

Discussion 386

The endocytic uptake of LDL-cholesterol is a major source of cellular cholesterol. Using an 387

endogenous cholesterol sensor, HMGCS1-Clover, we screened for and identified >70 genes 388

involved in the cellular LDL-cholesterol uptake pathway. Of these, we characterised C18orf8 389

as a novel core component of the Mon1-Ccz1 (MC1) GEF for mammalian Rab7. C18orf8-390

deficient cells are defective Rab7 activation, resulting in impaired LDL trafficking and cellular 391

cholesterol depletion. The unexpected accumulation of free cholesterol in the swollen LE/Ly 392

compartment of C18orf8-deficient cells, suggested an additional critical role for Rab7 in 393

cholesterol export out of lysosomes. We show that active Rab7 binds the NPC1 cholesterol 394

transporter and licenses lysosomal cholesterol export. This pathway is defective in C18orf8-, 395

Ccz1- and Mon1A/B-deficient cells and restored by a constitutively active Rab7 (Q67L). The 396

trimeric Mon1-Ccz1-C18orf8 (MCC) complex thus plays a central role in cellular LDL-397

cholesterol uptake, coordinating Rab7 activation, LDL trafficking and NPC1-dependent 398

lysosomal cholesterol export. 399

400

A CRISPR screen provides a comprehensive overview of trafficking factors required for 401

cellular LDL-cholesterol import 402

Genome-wide CRISPR screens provide a powerful tool to interrogate intracellular pathways. 403

To avoid artefacts associated with exogenous reporters, we generated an endogenous 404

SREBP2 knock-in reporter, HMGCS1-Clover, and screened for genes regulating cellular 405

cholesterol homeostasis. The predominant pathway identified by our screen is LDL-406

cholesterol import, as HeLa cells, like many tissue-culture cells, rely predominantly on 407

cholesterol import rather than de novo biosynthesis. The screen identified multiple stages in 408

the LDLR-LDL trafficking pathway (Fig 1E, F), including i) LDLR folding and trafficking to the 409

cell surface; ii) endocytosis of the LDLR-LDL complex; iii) LDL trafficking to the LE/Ly 410

compartment; and iv) NPC1 dependent cholesterol export from LE/Ly. Functional complexes 411



https://doi.org/10.1101/835686


18

identified include COG, Exocyst, AP2, V-ATPase, PI3K, PIKfyve, CORVET, HOPS, FERARI 412

and GARP; poorly characterised genes include the chaperone Clusterin (also known as 413

Apolipoprotein J) and the metalloprotease ADAM28. Our functional genetic approach 414

therefore reveals a comprehensive genome-wide analysis of essential cellular components 415

required for LDL-cholesterol import, and identifies novel candidate genes in this pathway. 416

417

C18orf8 is a core component of the Mon1-Ccz1-C18orf8 (MCC) GEF for mammalian 418

Rab7 419

Spatial and temporal control of Rab7 activation is critical to coordinate substrate trafficking 420

and degradation in LE/Ly (16). In yeast the dimeric Mon1-Ccz1 (MC1) GEF activates the Rab7 421

homologue Ypt7 (27, 28) and a similar function has been suggested for mammalian MC1 (35, 422

36). We find that a stable trimeric complex of Mon1A/B, Ccz1 and C18orf8 (MCC) is required 423

for Rab7 activation in mammalian cells, as defined by effector recruitment. C18orf8, Mon1 and 424

Ccz1 are conserved from animals to plants, but a C18orf8 orthologue is missing from 425

laboratory yeast and most other fungi, either due to divergence of the endocytic machinery or 426

because C18orf8 function is incorporated into other components. 427

428

Rab7 is a key regulator of LE homeostasis and C18orf8-, Ccz1- and Mon1A/B-deficient cells 429

show a broad range of LE defects including LE morphology, endosomal substrate trafficking, 430

lysosomal degradation and cholesterol export. Consistent with a previous Rab7 depletion 431

study (49), C18orf8-deficient cells accumulate enlarged MVBs; and consistent with current 432

Rab5-to-Rab7 conversion models (32) we observe a delay in early-to-late endosomal 433

trafficking. Whereas LDL trafficking was previously suggested to involve only Mon1B (50), we 434

find that only the combined depletion of Mon1A/B causes cholesterol deficiency, suggesting 435

redundancy of Mon1 homologues. 436

437



https://doi.org/10.1101/835686


19

Besides LE trafficking, Rab7 is an important factor in autophagy (51). During preparation of 438

this manuscript, a role for C18orf8 was reported as a regulator of autophagic flux (52). Like 439

us, this group reports a robust interaction between C18orf8 and mammalian MC1. Without 440

direct evidence of Rab7 activation, C18orf8 is however postulated to be a MC1 regulator. We 441

find C18orf8 is not a regulator, but a core component of the trimeric mammalian MCC complex, 442

which acts as a Rab7 GEF. Stability of the three core components is interlinked, with Mon1B 443

stability dependent on C18orf8 and Ccz1, and C18orf8 stability dependent on Ccz1 and 444

Mon1B. Ccz1 appears relatively stable in the absence of the other components and may 445

therefore form a scaffold for Mon1A/B and C18orf8 assembly. The intricate relationship 446

between C18orf8 and Mon1A/B is intriguing as Phyre2 prediction suggests C18orf8 adopts a 447

clathrin-like fold, while Mon1A/B shows structural similarity to the AP2 subunit (53). Whether 448

this structural similarity has functional consequences, remains to be established. 449

450

What is the role of C18orf8 within the MCC GEF complex? 451

The structural interdependence of the MCC subunits complicates in vivo studies to assess the 452

direct role of C18orf8 in mammalian MCC. Yeast Ccz1p and Mon1p are necessary and 453

sufficient for Ypt7 binding and GDP-to-GTP exchange, with most of the MC1-Ypt7 binding 454

interface mediated by Mon1p (28, 34). Residues critical for Ypt7 binding are conserved in 455

mammalian MC1 and a dimeric complex of mammalian Mon1A and Ccz1 was shown to have 456

limited in vitro Rab7 GEF activity (36). While this complex might have retained residual 457

C18orf8 due to its isolation from mammalian cells, future studies will address whether C18orf8 458

contributes towards in vitro MC1 GEF activity. 459

460

In vivo, C18orf8 function goes beyond being a scaffold for MCC stabilisation. The C18orf8 C-461

terminal -helical domain is sufficient to bind MC1 and stabilize Mon1B, but insufficient to fully 462

restore MCC function (Fig S4F, I). The C18orf8 N-terminus thus has an independent function 463



https://doi.org/10.1101/835686


20

and may bind other trafficking components and/or membrane lipids. The timing of Rab7 464

activation is strictly regulated to coordinate Rab5-to-Rab7 conversion and likely requires 465

control of MCC by upstream components (32). Membrane association allosterically activates 466

yeast MC1 and increases its GEF activity ~1600-fold (29). Like yeast Mon1p and Ccz1p, 467

mammalian Mon1A binds PtdIns3P and PtdSer (27), while positively charged residues in the 468

N-terminal -propeller of C18orf8 are well-positioned to bind negatively charged 469

phospholipids. In analogy to AP2 (54), PtdIns binding might open the MCC complex towards 470

Rab7 binding, thus allosterically activating MCC upon membrane association. 471

472

Rab7 controls NPC1-dependent cholesterol export from late endosomes 473

Cholesterol is hydrolysed from CEs in a LE/Ly compartment, and NPC1 is essential for its 474

subsequent export to the ER, mitochondria and plasma membrane. Increased expression of 475

our SREBP2 reporter indicates C18orf8-, Ccz1- and Mon1A/B-deficient cells are starved of 476

ER cholesterol, yet paradoxically they accumulate free cholesterol in a swollen LE/Ly 477

compartment. This suggested a novel role for Rab7 in lysosomal cholesterol export. We find 478

an active Rab7 binds the lysosomal cholesterol transporter NPC1 and licenses its export of 479

LDL-derived cholesterol from LE/Ly. The inability of MCC-deficient cells to activate Rab7, 480

results in a Niemann-Pick-like phenotype, even though NPC1/NPC2 expression and 481

localisation are intact. Rab7 therefore functions as a novel regulator of NPC1 and its export of 482

lysosomal cholesterol. 483

484

The mechanism by which Rab7 controls NPC1-dependent lysosomal cholesterol export 485

remains unclear. Rab7 may (i) directly activate the NPC1 transporter, thus licensing lysosomal 486

cholesterol export (Fig 7E, panel a); (ii) assemble a lipid transfer hub down-stream of NPC1 487

to allow inter-organelle cholesterol transfer at membrane contact sites (MCS) (panel b); or 488

(iii) act by a combination of both. 489



https://doi.org/10.1101/835686


21

490

Activation of the NPC1 transporter by Rab7 491

The activity of many transporters is tightly regulated, for instance the lysosomal calcium 492

channel TRPML1 is specifically activated by PtdIns(3,5)P2 at LE/Ly (55). Even so, the NPC1 493

transporter is believed to be constitutively active and to date no NPC1 regulators have been 494

identified. So why would Rab7 regulate NPC1 activity? Inappropriate cholesterol export by a 495

nascent NPC1 could harm the composition and functionality of non-lysosomal compartments. 496

To prevent cholesterol export outside LE/Ly, the interaction of cholesterol-loaded NPC2 with 497

NPC1 depends on the acidic lysosomal pH (7). Activation of NPC1-dependent cholesterol 498

export by Rab7 might further restrict cholesterol transfer to LE/Ly. The Rab7-NPC1 interaction 499

is independent of lysosomal cholesterol content. Even so, variations in Rab7 activity might 500

tune lysosomal cholesterol export to cellular or organelle-specific conditions, such as MCS 501

formation or organelle division. Future studies will aim to identify the cytoplasmic Rab7 binding 502

site on NPC1 and its putative function in regulating NPC1 activity. 503

504

A role for Rab7 in cholesterol transfer at membrane contact sites 505

Besides NPC1 activation, Rab7 might control lysosomal cholesterol export through the 506

formation of inter-organelle membrane contact sites (MCS) and/or the recruitment of lipid 507

transfer proteins (LTPs) down-stream of NPC1. At least three Rab7 effectors - ORP1L, 508

Protrudin (ZFYVE27) and PDZD8 - are involved in ER-to-LE MCS formation (19, 20, 56, 57) 509

and Rab7 itself has been implicated in ER-to-mitochondria MCS formation (58). MCS are of 510

key importance for rapid inter-organelle lipid exchange, yet whether Rab7-mediated MCS are 511

involved in cholesterol transfer remains controversial. The Rab7 effector ORP1L was reported 512

to act as a LTP linking the ER and LE/Ly and transporting cholesterol towards the ER in a 513

PtdIns(4,5)P2 / PtdIns(3,4)P2-dependent manner (12, 59). Its depletion however only causes 514

a minor steady-state change in lysosomal cholesterol, as does depletion of other LTPs and 515

tethering proteins implicated, including STARD3, ORP5 and SYT7 (13, 14, 60). This suggests 516

redundancy between LTP(s) or an uncharacterised LTP required for lysosomal cholesterol 517



https://doi.org/10.1101/835686


22

export. In contrast, our C18orf8-, Ccz1- and Mon1A/B-deficient cells are entirely defective in 518

lysosomal cholesterol export and fully phenocopy an NPC1-deficiency phenotype. The LTP(s) 519

responsible for lysosomal cholesterol export may therefore be Rab7 effectors. Future studies 520

will aim to identify the Rab7-dependent LTP(s) responsible for lysosomal cholesterol 521

accumulation in MCC-deficient cells. 522

523

The MCC Rab7 GEF coordinates LDL trafficking and lysosomal cholesterol export 524

LDL-cholesterol uptake is a highly efficient process, with each LDLR molecule estimated to 525

endocytose one LDL particle containing ~1600 CE/cholesterol molecules every 10min (3). 526

Down-stream LDL processing must be tightly coordinated to prevent a backlog in the endocytic 527

pathway. CE hydrolysis therefore likely commences en route to NPC1+ Ly to allow the large 528

LDL particle to unpack; whereas the arrival and export of cholesterol in Ly must be linked to 529

the formation and transport capacity of MCS. As a central regulator or LE homeostasis, Rab7 530

is well suited to coordinate these processes. Indeed, MCC-deficient cells show defects in LDL 531

trafficking and lysosomal cholesterol export, indicating both are linked to the Rab7 nucleotide 532

cycle. We propose that Rab7 and its MCC GEF are central regulators of cellular LDL-533

cholesterol uptake, coordinating LDL trafficking, NPC1-dependent lysosomal cholesterol 534

export and lipid transfer at MCS. Besides cholesterol, the Rab7 lipid transfer hub might 535

mediate lysosomal egress of other lipids, with VPS13A/C prominent Rab7 effectors (61, 62). 536

537

Rab7 functionality in Niemann Pick disease 538

Mutations in NPC1 and NPC2 cause the lethal Niemann Pick type C (NPC) lysosomal storage 539

disease. Disease-associated mutations commonly destabilize NPC1, leading to its ER 540

retention and proteasome- or lysosome-mediated degradation (47, 63). The most prevalent 541

NPC1 mutation, I1061T, destabilizes NPC1, yet the mutant protein retains its cholesterol 542

export function and the disease phenotype can be restored by overexpression of the mutant 543

NPC1 (48). Confirming earlier observations (64), we show lentiviral overexpression of a GFP-544

tagged wild-type Rab7 also reduces cholesterol accumulation in NPC1I10161T/I1061T patient-545



https://doi.org/10.1101/835686


23

derived primary fibroblasts (Fig 7D). Whereas this rescue was previously ascribed to 546

enhanced vesicular trafficking, we suggest that Rab7 overexpression directly enhances 547

NPC1-dependent lysosomal cholesterol export. This effect depends on an active Rab7 and is 548

not observed with the inactive Rab7-T22N, whereas the hyperactive Rab7-Q67L shows an 549

intermediate phenotype (Fig S7). A modest increase in Rab7 activity therefore increases 550

lysosomal cholesterol clearance in NPC1-mutant cells while optimal clearance requires both 551

Rab7 activation and inactivation in a time- and location-dependent manner. 552

553

In summary, we have identified a novel function for the Rab7 GTPase and its trimeric Mon1-554

Ccz1-C18orf8 (MCC) GEF in the coordination of late endosomal LDL trafficking and NPC1-555

dependent lysosomal cholesterol export. Future studies will aim to further characterise the 556

mechanism behind Rab7 function in lysosomal cholesterol export and explore its therapeutic 557

potential in Niemann Pick disease. 558

559



https://doi.org/10.1101/835686


24

Acknowledgments 560

We are grateful to the following people for their help in this study: Dr. Michael Bassik (Stanford 561

University) for kind donation of the genome-wide CRISPR/Cas9 sgRNA library and Prof. Ron 562

Kopito and Dr. Matthew Porteus (Stanford University) for the pDonor CRISPR knock-in 563

plasmid. FACS experiments were enabled by Dr. Reinard Schulte and the CIMR FACS core 564

facility team and Confocal microscopy by Matthew Gratian and Mark Bowen. Stuart Bloor 565

kindly assisted with the Illumina sequencing, Dr. James Williamson analysed samples by mass 566

spectrometry and Dr. Nick Matheson critically reviewed the manuscript. Special thanks to 567

Zhengzheng Sophia Liang for fruitful discussions and moral support. This work was financially 568

supported by a Wellcome Trust Principal Research Fellowship to P.J.L (084957/Z/08/Z). JPL 569

was supported by MRC research grant MR/R0009015/1. The Cambridge Institute for Medical 570

Research (CIMR) was in receipt of a Wellcome Trust strategic award (100140). The authors 571

declare no competing financial interests. 572

573

Author contributions 574

DJHB and PJL conceived the project. Experiments were carried out by DJHB, AS, IB, MLMJ 575

and DE. DJHB analysed the data and prepared the figures. DJHB and PJL wrote the 576

manuscript. JJCN and JPL advised on the project and critically reviewed the manuscript. 577

578

Conflict of interest 579

The authors declare that they have no conflict of interest. 580



https://doi.org/10.1101/835686


25

Figure legends 581

582

Figure 1: A genome-wide CRISPR screen identifies essential factors in cellular 583

cholesterol homeostasis. (A - C) Characterisation of a HMGCS1-Clover CRISPR knock-in 584

in HeLa cells. (A) Immunoblotting of a reporter clone shows the endogenous HMGCS1-Clover 585

fusion protein is detected by both HMGCS1- and GFP-specific antibody staining. (B) Sterol 586

depletion induces HMGCS1-Clover expression. HeLa HMGCS1-Clover cells were sterol 587

depleted overnight and HMGCS1-Clover expression was analysed by confocal microscopy. 588

(C) HMGCS1 upregulation is SREBP2-dependent. HeLa HMGCS1-Clover CAS9 cells were 589

transfected with sgRNA against SREBF2, and after 7 days, their response to sterol depletion 590

was determined by flow cytometry. (D - F) A genome-wide CRISPR screen identifies essential 591

factors in cellular cholesterol homeostasis. (D) HeLa HMGCS1-Clover/Cas9 cells were 592

transduced with a genome-wide sgRNA library (22,000 sgRNAs). Rare HMGCS1- Cloverhigh 593

cells were isolated using two rounds of cell sorting, resulting in a 60% enriched HMGCS1-594

Cloverhigh population. (E) Illumina sequencing of sgRNAs in the isolated HMGCS1-Cloverhigh 595

population shows sgRNA enrichment for genes involved in cholesterol uptake (LDLR, NPC1, 596

NPC2; blue), protein folding and glycosylation (yellow), membrane trafficking of LDLR/LDL 597

(pink, purple, green) and SREBP2 function (orange). Genes with sgRNA enrichment at P<10-598

5 are indicated. (F) A schematic representation of hits from the genome-wide CRISPR screen 599

for cholesterol regulators highlights the central role for LDL-cholesterol import for cellular 600

cholesterol homeostasis. Hits involved in pathways other than those indicated can be found 601

in Table S1. 602

603

Figure 2: C18orf8 is required for endosomal LDL-cholesterol uptake. (A, B) C18orf8-604

deficient cells show spontaneous cholesterol deficiency. (A) Wild-type and C18orf8-deficient 605

clones were lysed, endogenous C18orf8 immunoprecipitated, and detected by a C18orf8-606



https://doi.org/10.1101/835686


26

specific antibody. (B) C18orf8-deficient HMGCS1-Clover clones were transduced with HA-607

tagged C18orf8 (C18orf8-3xHA; green lines) or empty vector (red lines) and HMGCS1-Clover 608

expression was determined by flow cytometry at day 18 using wild-type HMGCS1-Clover cells 609

as a control (black lines). (C) C18orf8-deficient cells are dependent on endogenous 610

cholesterol biosynthesis. Wild-type and C18orf8-deficient HMGCS1-Clover cells were either 611

LPDS depleted to block exogenous LDL-cholesterol uptake (green lines), treated with 612

mevastatin to block endogenous cholesterol biosynthesis (red lines), or a combination of both 613

treatments (blue lines), after which cells were analysed by flow cytometry for HMGCS1-Clover 614

expression. (E, F) C18orf8-deficient cells show defective endo-lysosomal LDL degradation. 615

Wild-type, C18orf8-deficient or C18orf8-deficient cells complemented with C18orf8-3xHA 616

were starved for 1 hour and pulse-labelled with fluorescent Dil-LDL, grown for the 5 min (E) or 617

180 min (F), fixed and visualised by confocal microscopy. Exposure times were kept constant 618

between individual conditions at given time points. Scale bars = 10µm. 619

620

Figure 3: C18orf8-deficient cells show severe defects in late endosome morphology 621

and early-to-late endosomal trafficking. (A, B) C18orf8-deficient cells show clustering of 622

EE and swelling of LE/Ly. Confocal microscopy comparing EEA1, Rab7 and LAMP1in wild-623

type versus C18orf8-deficient cells. (C - E) C18orf8-deficient cells have severe defects in 624

early-to-late endosomal trafficking. (C, D) Wild-type and C18orf8-deficient cells were starved 625

for 1 hour, pulse-labelled with AlexaFluor555-conjugated EGF, incubated for the indicated 626

times, fixed and stained for EEA1 (blue) and LAMP1 (green). In (D) a cocktail of protease 627

inhibitors (Leupeptin/E-64d/Leupeptin) was added during starve and chase to block EGF 628

degradation. EEA1+ and LAMP1+ vesicles were identified using Volocity software and the 629

percentage of EGF co-localising with these structures was determined from 5-6 images per 630

condition with at least eight cells per image (** P<0.01, *** P<0.001). (E) C18orf8-deficient 631

cells were stimulated with Dil-LDL, chased for 3 hours, fixed and stained for EEA1 (blue) and 632

LAMP1 (green). Scale bars = 10µm. 633



https://doi.org/10.1101/835686


27

634

Figure 4: C18orf8 forms an integral component of the Mon1-Ccz1 (MC1) complex, 635

essential for complex stability and function. (A, B) C18orf8 interacts with the Mon1-Ccz1 636

complex. (A) Immune precipitations of exogenous 3xHA-C18orf8 (N-term) or C18orf8-3xHA 637

(C-term) were analysed by mass spectrometry. Proteins detected by >3 peptides in both 638

C18orf8 samples and absent from control, are indicated. (B) Immune precipitation of an 639

endogenous C18orf8-3xMyc fusion or Mon1B reveals a reciprocal interaction between 640

C18orf8, Ccz1 and Mon1B (see also Fig S4C, D). (C - D) C18orf8, Mon1B and Ccz1 show 641

reciprocal stabilisation. Immunoblot analysis of lysates from (C) wild-type, C18orf8-deficient 642

and C18orf8-deficient cells complemented with C18orf8-3xHA; or (D) wild-type, C18orf8-, 643

Ccz1- and Mon1A/B-deficient cells. For endogenous C18orf8 detection, C18orf8 was immune 644

precipitated prior to immunoblotting (D). (E - F) Ccz1- and Mon1A/B-deficient cells show 645

cholesterol deficiency and disruption of LE/Ly morphology. (E) HMGCS1-Clover CAS9 cells 646

were transfected with sgRNAs against Mon1A, Mon1B, Mon1A and Mon1B, Ccz1 or Rab7, 647

grown for 14 days, treated overnight with mevastatin and analysed for HMGCS1-Clover 648

expression. (F) Wild-type, Ccz1- and Mon1A/B-deficient cells were stained intracellularly for 649

EEA1, Rab7 and LAMP1. Scale bars = 10µm. 650

651

Figure 5: The trimeric Ccz1-Mon1-C18orf8 (MCC) complex activates mammalian Rab7. 652

(A) The MCC complex binds an inactive Rab7 (T22N). Immune precipitations of 2xHA-tagged 653

wild-type, T22N or Q67L Rab7 from C18orf8-3xMyc knock-in cells, were analysed by 654

immunoblot using Myc, Ccz1 and Mon1B specific antibodies. (B, C) C18orf8-, Ccz1- and 655

Mon1A/B-deficient cells lack activation-dependent recruitment of Rab7 effectors. (B) Immune 656

precipitations of 3xFLAG-RILP from wild-type, C18orf8-, Ccz1- or Mon1A/B-deficient cells 657

were analysed by immunoblotting for endogenous Rab7. (C) Wild-type and C18orf8-deficient 658

cells were transfected with HA-RILP or ORP1L (Fig S5A) and stained intracellularly for HA 659

(green) and LAMP1 (magenta). Manders correlation was determined for 6-10 cells from two 660



https://doi.org/10.1101/835686


28

independent experiments (*** P<0.001). (D – F) Cholesterol and trafficking defects in C18orf8 661

deficient cells can be rescued by knockdown of Rab7GAPs or expression of a constitutively 662

active Rab7 (Q67L). (D) HMGCS1-Clover expression was determined at day 8 after 663

transduction with shRNAs against TBC1D5, TBC1D15 or both. (E, F) C18orf8-deficient cells 664

were transduced with 2xHA-tagged Rab7-T22N, -Q67L or an empty vector and either (E) 665

analysed at day 10 by flow cytometry for HMGCS1-Clover expression, or (F) pulse-labelled 666

with AlexaFluor 555 labelled EGF, incubated for 3 hours and stained intracellularly for LAMP1 667

and EEA1. Scale bars = 10µm. 668

669

Figure 6: MCC-deficient cells lack an activation-dependent interaction between Rab7 670

and NPC1 and accumulate lysosomal cholesterol. (A – C) MCC-deficient cells accumulate 671

lysosomal cholesterol and phenocopy NPC1-deficiency. (A) Filipin staining of wild-type, 672

C18orf8-deficient and complemented C18orf8-deficient cells; or (C) wild-type, Ccz1- and 673

Mon1A/B-deficient cells. (B) Filipin co-staining with the LE/Ly markers Rab7 and LAMP1 in 674

C18orf8-deficient cells. (D) Theonellamides (TNM) immuno-gold labelling of C18orf8-deficient 675

cells, visualised by EM. (E) Filipin staining of wild-type, C18orf8- and NPC1-deficient cells. 676

Scale bars = 10µm. (F, G) Rab7 interacts with the NPC1 cholesterol transporter in activation-677

dependent manner. (F) Immune-precipitation of HA-tagged NPC1 and detection of NPC1-678

interacting proteins using mass spectrometry. Interaction partners detected with >2 peptides 679

are indicated by abundance. (G) Immune precipitations of HA-tagged wild-type, dominant-680

negative (T22N) or constitutively active Rab7 (Q67L) reveal an activation-dependent 681

interaction between Rab7 and endogenous NPC1. (H) The Rab7-NPC1 interaction is lost in 682

MCC-deficient cells that lack Rab7 activation. Wild-type, C18orf8-, Ccz1 and Mon1A/B-683

deficient cells were stably transduced with the inactive NPC1-P692S-HA. HA-tagged NPC1 684

was immune precipitated and immune blotted for endogenous Rab7. The inactive NPC1-685

P692S was used to prevent altering lysosomal cholesterol content. (I, J) The Rab7-NPC1 686

interaction is independent of NPC1 activity or lysosomal cholesterol levels. (I) NPC1-deficient 687



https://doi.org/10.1101/835686


29

cells were complemented with HA-tagged wild-type or inactive P692S-mutant NPC1 and 688

NPC1-HA immune-precipitations were analysed by immune blotting for endogenous Rab7. (J) 689

Wild-type NPC1-HA complemented cells were treated with LPDS to decrease, or U18666A to 690

increase lysosomal cholesterol levels and the NPC1-Rab7 interaction was probed using 691

immune precipitation. 692

693

Figure 7: Rab7 activation by the MCC GEF controls NPC1-dependent lysosomal 694

cholesterol export. (A, B) Lysosomal cholesterol export is abolished in NPC1-, C18orf8-, 695

Ccz1- and Mon1A/B-deficient cells. (A) Wild-type, NPC1-, C18orf8-, Ccz1 and Mon1A/B-696

deficient cells were treated for 24 hours with the NPC1 inhibitor U18666A to increase 697

lysosomal cholesterol (pulse, top panels), followed by a 24 hours chase in the presence of 698

LPDS and mevastatin (lower panels). Lysosomal cholesterol accumulation was visualised 699

using Filipin co-staining with the LE/Ly marker CD63. (B) Colocalisation of Filipin with CD63 700

was plotted as Pearson correlation, calculated from 3 independent experiments with 6 701

representative fields per experiment and >8 cells per field. ** p<0.01, *** p<0.001. (C) 702

Cholesterol accumulation in C18orf8-deficient cells is abolished by overexpression of a 703

hyperactive Rab7. C18orf8-deficient cells were transduced with a wild-type, dominant-704

negative (T22N) or hyperactive Rab7 (Q67L) or empty vector and co-stained with Filipin and 705

anti-CD63. (D) Rab7 overexpression rescues lysosomal cholesterol accumulation in NPC 706

patient fibroblasts. NPC1I1061T/I1061T primary patient fibroblasts were transduced with a GFP-707

tagged wild-type Rab7 and analysed at day 7 for cholesterol accumulation using Filipin 708

staining. Fibroblasts from a healthy individual were used as a control. Representative images 709

are shown from 2 independent experiments. (E) Model for MCC and Rab7 function in 710

lysosomal cholesterol export. The trimeric Mon1-Ccz1-C18orf8 (MCC) GEF activates 711

mammalian Rab7, which then binds to the NPC1 cholesterol transporter and either a) directly 712

activates NPC1 cholesterol export function; or b) assembles a down-stream membrane 713

contact site (MCS) at which a yet-uncharacterised lipid transfer protein (LTP) mediates 714



https://doi.org/10.1101/835686


30

cholesterol transfer to the ER and/or plasma membrane. A combined Rab7 function in NPC1 715

activation and MCS formation would assure lysosomal cholesterol is only exported once a 716

down-stream lipid transfer module is assembled. 717

718



https://doi.org/10.1101/835686


31

Materials and Methods 719

720

Plasmids 721

pHRSIN.pSFFV and pHRSIN.pCMV lentiviral expression constructs harbouring pGK 722

Puromycin, pGK Hygromycin or pSV40 Blasticidin resistance have been reported previously 723

(65). The C18orf8 open reading frame was cloned from HeLa cell cDNA, modified with an N- 724

or C-terminal 3xHA-tag and inserted into pHRSIN.pSFFV pGK Hygro and pHRSIN.pSFFV 725

pGK Puro respectively. Ccz1, Mon1A and Mon1B were cloned from IMAGE clones 726

(Dharmacon), modified with a C-terminal 3xMyc tag and inserted into pHRSIN.pSFFV pGK 727

Hygro. Mon1A was corrected by PCR to add a missing N-terminus. 2xHA-Rab7 constructs 728

were cloned from GFP-HA-Rab7 wild-type, T22N and Q67L constructs (Addgene #28047, 729

#28048, #28049, created by Prof. Qing Zhong, University of California Berkeley, USA (21)) 730

and inserted into pHRSIN.pCMV pGK Puro. HA-RILP and HA-ORP1L constructs have been 731

described previously (66, 67). RILP was recloned with an N-terminal 3xFLAG tag and inserted 732

into pHRSIN.pSFFV pGK Puro. FAM160B2 was amplified from human stem cell cDNA and 733

inserted in pHRSIN.pSFFV pGK Puro with a C-terminal 3xHA-tag. 734

735

The pDonor vector used for C-terminal knock-in was a kind gift from Prof. Ron Kopito and Dr. 736

Matthew Porteus (Stanford University, USA) and modified with Clover (Addgene #49089, 737

created by Prof. Kurt Beam, University of Colarado (38)) or a 3xMyc C-terminal tag replacing 738

the original TAP-tag. 750-800bp flanking arms for HMGCS1 or C18orf8 were cloned from 739

HeLa cell genomic DNA and inserted into pDonor using Gibson Assembly (NEB), yielding 740

pDonor HMGCS1-Clover Puro and pDonor C18orf8-3xMyc Puro respectively. To allow 741

selection for multiple-allele insertion, the puromycin resistance cassette of pDonor was 742

replaced with hygromycin or blasticidin resistance cassettes, yielding pDonor HMGCS1-743

Clover Hygro and pDonor HMGCS1-Clover Blast respectively. 744

745



https://doi.org/10.1101/835686


32

The pHRSIN.pSFFV 3xFLAG-NLS-CAS9-NLS pSV40 Blast construct used for stable CAS9 746

expression was described previously (68). The genome-wide Bassik sgRNA library (10 747

sgRNAs per gene, total 220,000 sgRNAs) was a kind gift from Dr. Mike Basssik (Stanford 748

University, USA, (39)). sgRNAs for individual genes were cloned into a modified pKLV.U6 pGK 749

Puro-2A-BFP vector originally created by Dr. Kosuke Yusa (Wellcome Sanger Institute, 750

Hinxton, UK). 751

752

sgRNA sequences used were: SREBF1 sg1 GGTCACAGTGGTCGTTACAG, sg2 753

CCTGTAGAGAAGCCTCCCGG; SREBF2 sg1 AGTGCAACGGTCATTCACCC, sg2 754

CTCACCGTCGATGTCTCCCA, sg3 AGCCGGGCGATGGACGACAG; LDLR sg1 755

CGGCGAATGCATCACCC, NPC1 sg1 CGTCAGCGTCCTTCCCACAC; AP2M1 sg1 756

CCTGTCTCGGAATTCTGT; ANKFY1 sg1 TATTCGCTTCTACCAGA, sg2 757

TAGCGGGTACTCTGTCT, sg3 ACGCAGCCAAACGCCTGTAC, sg4 758

GTACAGGCGTTTGGCTGCGT; VPS16 sg1 GCAGTGGAAGAGTGGACCCG; ZFYVE20 sg1 759

CTCAGCTCTGAATAATCGGG; INSIG1 sg1 CGTAGCTAGAAAAGCTA, sg2 760

CTGCTGTCCCGCAGCAGGG, sg3 CAGCCCCTACCCCAACACC, sg4 761

TCAACCTGCTGCAGATCCAG; C18orf8 sg1 GCGGCCGGTGCAGTTCGAGA, sg2 762

TTCCGAAAGAGACATCGCAA; Mon1A sg1 GTTGGGGCCCGTGTAGTAAA; Mon1B sg1 763

GCAGGTATAGAGCTCGAATT; Ccz1 sg1 AATGAGAAGATTAGAAATGT; Rab7A sg1 764

AGGCGTTCCAGACGATTGCA; 2m sg1 GGCCGAGATGTCTCGCTCCG. 765

766

shRNAs were cloned into the lentiviral pHRSiren vector harbouring puromycin or hygromycin 767

resistance cassettes. shRNA sequences used were: TBC1D5 sh1 768

GAAGCCATATCGCAGAGCTA, TBC1D15 sh1 AATGGGACATGGTTAATACAGTT, non-769

targeting sh1 GGGTATCGACGATTACAAA. 770

771

Antibodies 772



https://doi.org/10.1101/835686


33

Primary antibodies used include: mouse anti-HA-tag (16B12, Biolegend, #901501), rabbit anti-773

HA-tag (Poly9023, Biolegend, #923501), rat anti-HA-tag (3F10, Sigma-Aldrich, #ROAHAHA), 774

rabbit anti-Myc-tag (71D10, Cell Signalling, #2278), mouse anti-Myc-tag (9B11, Cell 775

Signalling, #2276), mouse anti-FLAG-tag (M2, Sigma-Aldrich, #F1804), rabbit anti-GFP-tag 776

(Thermo-Fisher, #A11122), rabbit anti-Rab5 (C8B1, Cell Signalling, #3547), rabbit anti-Rab7 777

(D95F2, Cell Signalling, #9367), rabbit anti-Rab7 (EPR7589) AlexaFluor647 conjugate 778

(Abcam, #ab198337), mouse anti-C18orf8 (OTI4E4, Novus, #NBP2-01949), rabbit anti-779

C18orf8 (Proteintech, #20111-1-AP), rabbit anti-Mon1B (Novus, #NBP1-92131), rabbit anti-780

Mon1B (Proteintech, #17638-1-AP), mouse anti-Ccz1 (B-7, Santa Cruz, #sc-514290), mouse 781

anti-EEA1 (14/EEA1, BD Biosciences, #610456), mouse anti-LAMP1 (H4A3, Biolegend, 782

#328601), mouse anti-LAMP1 (H4A3) BV421 conjugate (BD Biosciences, #562623), mouse 783

anti-LAMP1 (H4A3) AlexaFluor647 conjugate (Biolegend, #121610), mouse anti-CD63 (MEM-784

259, Thermo-Fisher, #MA1-19281), mouse anti-HMGCS1 (A6, Santa Cruz, #sc-166763), 785

mouse anti-HMGCR (C1, Santa Cruz, #sc-271595), mouse anti-LDLR (C7) PE-conjugate (BD 786

Biosciences, #565653), Mouse anti-alpha-tubulin (DM1A, Thermo-Fisher, #62204), mouse 787

anti-beta-actin (AC-74, Sigma-Aldrich, #A5316), rabbit anti-BODIPY-FL (Thermo-Fisher, 788

#A5770). 789

790

Secondary antibodies used include: goat-anti-rabbit IgG (H+L) AlexaFluor546 and 791

AlexaFluorPlus647 (Thermo-Fisher, #A11035, #A32733), goat-anti-mouse IgG (H+L) 792

AlexaFluor568 and AlexaFluor647 (Thermo-Fisher, #A11031, #A21236), goat-anti-mouse 793

IgG1 AlexaFluor546 (Thermo-Fisher, #A21123), donkey anti-rat IgG (H+L) CF568 (Biotium, # 794

20092), donkey anti-mouse IgG (H+L) AlexaFluor647 (Thermo-Fisher #A31571), goat-anti-795

mouse IgG2a BV421 (Jackson Immunoresearch, #115-675-206), goat-anti-mouse (H+L) 796

HRP-conjugate (Jackson Immunoresearch, #115-035-146), goat-anti-rabbit (H+L) HRP-797

conjugate (Jackson Immunoresearch, #111-035-144), Mouse TrueBlot ULTRA HRP-798

conjugate (Rockland, #18-8817-33). 799

800



https://doi.org/10.1101/835686


34

For immune precipitation: EZview Red anti-HA, anti-c-Myc and anti-FLAG M2 Affinity Gel 801

(Sigma-Aldrich, #E6779, #E6654, #F2426). Protein A and Protein G-Sepharose (Sigma-802

Aldrich, #P3391, #P3296), IgG sepharose 6 Fast Flow (#GE17-0969-01). 803

804

Cell lines 805

HeLa and 293T cells were maintained respectively in RPMI-1640 and IMDM (Sigma-Aldrich) 806

supplemented with 10% fetal calf serum (FCS). For sterol depletion cells were washed in PBS 807

and incubated overnight in RPMI supplement with 5% lipoprotein-depleted serum (LPDS, 808

Biosera), 10M mevastatin (Sigma-Aldrich) and 50M mevalonate (Sigma-Aldrich). 809

NPC1I1061t/I1061T patient-derived primary fibroblasts (GM18453) and healthy controls 810

(GM08399) were obtained from Coriell Institute for Medical Research (New Jersey, USA) and 811

grown in MEM with Earl’s salt (Sigma-Aldrich) supplemented with 15% FCS and non-essential 812

amino acids (Gibco). 813

814

CRISPR knock-out lines were created by transient transfection of HeLa cells using TransIT 815

HeLa Monster (Mirus). Transfected cells were selected on puromycin 24-72h post-transfection 816

and single-cell cloned >7 days post-transfection. Knock-out clones were characterised by flow 817

cytometry and Western blotting. 818

819

Stable protein overexpression was achieved using lentiviral transduction. Briefly, 293T cells 820

were co-transfected in a 1:1 ratio with a lentiviral expression vector (pHRSIN/pHRSiren/pKLV) 821

and the packaging vectors pMD.G and pCMVR8.91 using TransIT-293 (Mirus). Supernatant 822

was harvested at 48h post-transfection and transferred onto target cells. Cells were spun 823

45min at 1800rpm to enhance infectivity and incubated with virus overnight. Transduced cells 824

were selected for stable transgene expression with appropriate antibiotics from 48h post-825

transduction. 826

827

CRISPR knock-in of C-terminal tags 828



https://doi.org/10.1101/835686


35

HeLa cells were transiently transfected with sgRNA, CAS9 and pDonor vector using TransIT 829

HeLa Monster (Mirus). To create a C18orf8-3xMyc knock-in a single donor vector (pDonor 830

C18orf8-3xMyc Hygro) was used, whereas for HMGCS1-Clover knock-in a combination of 831

three donor plasmids (pDonor HMGCS1-Clover Puro, Hygro, Blast) was used to select for 832

multiple-allele integration. Transfected cells were transiently selected for sgRNA expression 833

using puromycin at 24-72h post-transfection, followed by selection of the knock-in cassette 834

using hygromycin or a combination of puromycin, hygromycin and blasticidin at 7 days post-835

transfection. Selected cells were transiently transfected with Cre recombinase to remove 836

resistance cassettes and single-cell cloned. Single cell clones were characterised using flow 837

cytometry and Western blotting. 838

839

Genome-wide CRISPR screening 840

HeLa HMGCS1-Clover cells were stably transduced with Cas9. CRISPR sgRNA library 841

lentivirus was produced as indicated above and titrated on HeLa HMGCS1-Clover/Cas9 cells. 842

For screening, 1x108 cells were transduced at 30% infectivity (>100-fold coverage), puromycin 843

selected, grown for 9 days and sorted by FACS for a HMGCS1-Cloverhigh phenotype. Sorted 844

cells were grown for another 9 days and sorted a second time for the same phenotype. Cells 845

were finally grown for another 5 days, harvested and genomic DNA was extracted using a 846

Genra Puregene Core kit A (Qiagen). Integrated sgRNAs were amplified via two rounds of 847

PCR, using the following primer sets: outer 5’-848

AGGCTTGGATTTCTATAACTTCGTATAGCATACATTATAC-3’ and 5’-849

ACATGCATGGCGGTAATACGGTTATC-3’; inner 5’-850

AATGATACGGCGACCACCGAGATCTACACTCTCTTGTGGAAAGGACGAAACACCG-3’ 851

and 5’-852

CAAGCAGAAGACGGCATACGAGATnnnnnnnGTGACTGGAGTTCAGACGTGTGCTCTTC853

CGATCCGACTCGGTGCCACTTTTTC-3’. The second (inner) PCR introduced adaptors for 854

Illumina sequencing, with nnnnnnn indicating variable index sequences. PCR products were 855

purified using AMPure XP beads (Agencourt), quantified on a DNA-1000 chip (Agilent) and 856



https://doi.org/10.1101/835686


36

sequenced on a Miniseq sequencer (Illumina). sgRNA abundance in the sorted sample was 857

compared against a library sample of similar age (68) and sgRNA enrichment was calculated 858

using the RSA algorithm under default settings (69). Hits with p<10-5 were manually annotated 859

into function pathways (cholesterol import, endocytosis, endosomal trafficking, ER/Golgi 860

trafficking, protein folding/glycosylation, SREBP pathway) (Fig 2C, Fig 3). The full dataset is 861

available in Table S1. 862

863

Metabolic labelling and pulse-chase 864

Cells were sterol-depleted overnight, starved for 30 min at 37°C in methionine-free, cysteine-865

free RPMI containing 5% dialysed LPDS, labelled with [35S]methionine/[35S]cysteine) 866

(Amersham) for 10 min and then chased in RPMI with 10% FCS supplemented with 2 µg/ml 867

25-hydroxy-cholesterol and 20 µg/ml cholesterol. Samples taken at the indicated time-points 868

were lysed in 1% Digitonin/TBS as above. Immunoprecipitations were performed as above, 869

washed with 1% Tx-100/TBS and samples separated by SDS-PAGE and processed for 870

autoradiography with a Cyclone scanner (Perkin-Elmer). 871

872

Immunoblotting and Immunoprecipitation 873

For immunoblotting, cells were lysed in 1% IGEPAL (Sigma-Aldrich) in TBS pH 7.4 with Roche 874

complete protease inhibitor or in 2% SDS in 50mM Tris pH7.4 in the presence of Benzonase 875

(Sigma-Aldrich). Post-nuclear supernatants were heated at 70°C in SDS sample buffer, 876

separated by SDS-PAGE and transferred to PVDF membranes (Millipore). Membranes were 877

probed with the indicated antibodies and reactive bands visualised with ECL, Supersignal 878

West Pico or West Dura (Thermo Scientific). 879

880

For immunoprecipations, cells were lysed in 1% digitonin, 1% IGEPAL or 0.5% IGEPAL in 881

TBS pH 7.4 with Roche protease inhibitor. Samples were precleared with protein A/IgG-882

Sepharose and incubated with primary antibody and protein A/protein G-Sepharose or 883

antibody conjugated agarose for at least 2 hrs. After 5 washes in 0.2% detergent, proteins 884



https://doi.org/10.1101/835686


37

were eluted in 2% SDS, 50mM Tris pH7.4, separated by SDS-PAGE and immunoblotted as 885

described. For RILP/Rab7 immune precipitations, lysis buffers were supplemented with 10mM 886

MgCl2 and 1mM EDTA. 887

888

Flow cytometry 889

Cells were washed in PBS, detached by trypsinising for 5 min, pelleted and where necessary 890

antibody stained for 30min on ice. After washing in ice-cold PBS, cells were analysed on a 891

FACS Calibur or Fortessa (BD Biosciences) and analysed in FlowJo. 892

893

Immune fluorescence 894

Cells were grown on 10mm coverslips, fixed for 15 min in 4% formaldehyde (Polysciences), 895

washed 3 times with PBS and incubated 10 min with 15mM Glycine. Cells were permeabilised 896

for 1 hour with 0.05% saponin, 5% goat serum in PBS and stained for 1-2 hours with primary 897

antibody as indicated in 3% BSA, 0.05% saponin, PBS. Coverslips were washed 3 times 5 898

min in 0.1% BSA, 0.05% saponin, PBS and incubated 1h with fluorochrome-conjugated 899

secondary, washed 3 more times and embedded in Prolong Antifade Gold (Thermo Scientific). 900

Stainings with Rab5 and Rab7 antibody (Cell Signalling) antibody were performed overnight 901

in 3% BSA, 0.3% Tx-100, PBS at 4°C, followed by staining with additional primary antibody 902

and secondary antibodies as above. 903

904

For Dil-LDL co-staining, cells were permeabilised for 40 sec in 0.01% digitonin (Merck) and 905

stained in 3% BSA without detergent. For Filipin staining, cells were fixed and incubated 1 906

hour with 0.05mg/ml Filipin III (Cayman) in 0.5% BSA. For Filipin co-staining, primary and 907

secondary antibody stainings were performed in the presence of Filipin and without detergent. 908

Fluorescent staining was recorded on a Zeiss LSM880 Confocal microscope using a 63x oil 909

objective and analysed using Zen software (Zeiss). 910

911



https://doi.org/10.1101/835686


38

For ORP1L/RILP-LAMP1 co-localisation, cells were transiently transfected with HA-tagged 912

RILP or ORP1L constructs using Effectine (Qiagen), harvested at 24 hours post-transfection 913

and fixed in 3.7% formaldehyde in PBS for 15-20 min. After 10min permeabilized with 0.1% 914

TritonX-100 in PBS, coverslips were stained with HA and LAMP1 primary antibodies and 915

donkey anti-rat CF568 and donkey anti-mouse Alexa647 secondary antibodies and mounted. 916

Samples were imaged on a Leica SP8 microscope adapted with a HCX PL 63x 1.32 oil 917

objective, solid-state lasers, and HyD detectors. Colocalization was reported as Mander’s 918

coefficient calculated using JACoP plug-in for ImageJ on the basis of 2 independent 919

experiments. 920

921

LDL and EGF trafficking assays 922

For LDL trafficking assays, cells were starved 30min in RPMI with 5% LPDS at 37°C, labelled 923

15 min at 4°C with 20 g/ml Dil-LDL in RPMI with 5% LPDS and 10mM HEPES, washed and 924

chased for indicated times in full medium. For EGF trafficking, cells were starved for 1 hour in 925

HBSS (Gibco), incubated with 100ng/ml EGF-AlexaFluor 555 for 5 min at 4°C, washed and 926

chased in full medium. At indicated times, cells fixed in 4% formaldehyde and either mounted 927

directly or stained for indicated markers as described above. Fluorescent staining was 928

recorded using a Zeiss LSM880 Confocal microscope with a 63x oil objective. EEA1+ and 929

LAMP1+ vesicles were annotated in an automated fashion using Volocity software (Perkin-930

Elmer) and the percentage EGF within these vesicles was determined as a percentage of total 931

EGF fluorescence for 5-6 fields per sample with >8 cells per field. To include luminal content 932

of enlarged LAMP1+ vesicles area filling function was enabled and enlarged vesicles were 933

trimmed back by one pickle to prevent artificial area merging in areas of high vesicle density. 934

Statistical significance was determined using two-tailed Student’s T-Test with **p<0.01 and 935

***p<0.001 and error bars reflecting standard error of mean (SEM). 936

937

Lysosomal cholesterol export assay 938



https://doi.org/10.1101/835686


39

Cells were seeded on coverslips and treated for 24 hours with 2 M U18666A (Santa Cruz) in 939

RPMI with 10% FCS, washed with PBS and either fixed directly with 4% paraformaldehyde 940

(pulse); or incubated for 24 hours chase in RPMI with 5% LPDS, 10M mevastatin and 50M 941

mevalonate (chase). Coverslips were stained using Filipin and anti-CD63 antibody and 942

analysed using a Zeiss LSM880 Confocal microscope with a 63x oil objective. Pearson’s 943

correlation between Filipin and CD63 staining was determined using Volocity software (Perkin-944

Elmer) for 6 fields per sample with >8 cells per field. Statistical significance was determined 945

using two-tailed Student’s T-Test with **p<0.01 and ***p<0.001 and error bars reflecting 946

standard error of mean (SEM). 947

948

949

Fluid phase endocytosis 950

Bulk endocytosis and delivery of extracellular materials to the late endosomal compartment 951

was measured using the dye Sulforhodamine 101 (SR101, Sigma), following published 952

protocols (42, 70). Briefly, Cells were seeded in live cell imaging dishes (MatTek), 953

preincubated with Lysotracker Green (Molecular Probes) for 30 min, and SR101 dye was 954

added directly to the cell medium at 25µg/ml. Samples were monitored by time-lapse on a 955

Leica SP8 microscope adapted with a climate control chamber using a HCX PL 63x 1.32 oil 956

objective and images taken at multiple positions at 2 min intervals. SR101 colocalisation with 957

the Lysotracker-positive compartment is reported as Mander’s coefficient calculated using 958

JACoP plug-in for ImageJ on the basis of 2 independent experiments. Total SR101 entering 959

both cell lines was quantified at indicated time points by flow cytometry. Error bars indicate 960

SD of the mean. Statistical evaluations report on Student’s T Test (analysis of two groups), 961

with *p<0.05, **p<0.01, ***p<0.001. 962

963

Electron microscopy 964



https://doi.org/10.1101/835686


40

Samples were prepared for cryo sectioning as described elsewhere (71, 72). Briefly, C18orf8-965

deficient cells and wild-type cells were fixed for either 2h in freshly prepared 2% 966

paraformaldehyde and 0.2% gluteraldehyde in 0.1M phosphate buffer, or for 24h in freshly 967

prepared 2% paraformaldehyde in 0.1 M phosphate buffer. Fixed cells were scraped, 968

embedded in 12% gelatin (type A, bloom 300, Sigma) and cut with a razor blade into 0.5 mm3 969

cubes. The sample blocks were infiltrated in phosphate buffer containing 2.3 M sucrose. 970

Sucrose-infiltrated sample blocks were mounted on aluminium pins and plunged in liquid 971

nitrogen. The vitrified samples were stored under liquid nitrogen. 972

973

Ultrathin cell sections of 75 nm were obtained essentially as described elsewhere (72). Briefly, 974

the frozen sample was mounted in a cryo-ultramicrotome (Leica). The sample was trimmed to 975

yield a squared block with a front face of about 300 x 250 μm (Diatome trimming tool). Using 976

a diamond knife (Diatome) and antistatic devise (Leica) a ribbon of 75 nm thick sections was 977

produced that was retrieved from the cryo-chamber with the lift-up hinge method (73). A 978

droplet of 1.15 M sucrose was used for section retrieval. Obtained sections were transferred 979

to a specimen grid previously coated with formvar and carbon. 980

981

Grids containing thawed cryo sections of cells fixed with 2% paraformaldehyde and 0.2% 982

glutaraldehyde were incubated on the surface of 2% gelatin at 37 °C. Subsequently grids were 983

rinsed to remove the gelatin and sucrose and were embedded in 1.8% methylcellulose and 984

0.6% uranyl acetate. In case of additional gold-labelling, sections of cells fixed with 2% 985

paraformaldehyde were incubated on drops with 1 μM TNM-BODIPY for 30 minutes on ice, 986

washed, blocked with 1% BSA in PBS, and then incubated with rabbit polyclonal anti-BODPIY 987

antibody (Molecular Probes, Invitrogen, UK) (45). Sections were subsequently labelled with 988

10 nm protein A‐coated gold particles (CMC, Utrecht University). EM imaging was performed 989

with an Tecnai 20 transmission electron microscope (FEI) operated at 120 kV acceleration 990

voltage. 991

992



https://doi.org/10.1101/835686


41

Identification of C18orf8 and NPC1 interaction partners by mass spectrometry (MS). 993

HeLa cells, or HeLa cells expressing 3xHA-C18orf8, C18orf8-3xHA or NPC1-3xHA were lysed 994

in 1% digitonin, HA-tagged proteins were immune precipitated for 2.5 hours at 4°C using 995

Ezview anti-HA agarose (Sigma-Aldrich) as described above, eluted using 0.5 g/ml HA 996

peptide (Sigma-Aldrich) and denatured using 2% SDS, 50mM Tris pH7.4. Eluted samples 997

were reduced with 10mM TCEP for 10mins RT and alkylated with 40mM Iodoacetamide for 998

20mins RT in the dark. Reduced and alkylated samples were then submitted to digestion using 999

the SP3 method (PMID: 25358341). Briefly, carboxylate coated paramagnetic beads are 1000

added to the sample and protein is bound to the beads by acidification with formic acid and 1001

addition of acetonitrile (ACN, final 50%). The beads are then washed sequentially with 100% 1002

ACN, 70% Ethanol (twice) and 100% ACN. 10uL of a buffer of TEAB (Triethylammonium 1003

bicarbonate) pH8 and 0.1% Sodium deoxycholate (SDC) is then added to the washed beads 1004

along with 100ng trypsin. Samples were then incubated overnight at 37 degrees with periodic 1005

shaking at 2000rpm. After digestion, peptides are immobilised on beads by addition of 200uL 1006

ACN and washed twice with 100uL ACN before eluting in 19uL 2% DMSO and removing the 1007

eluted peptide from the beads. 1008

1009

MS data acquisition 1010

Samples were acidified by addition of 1uL 10% TFA and the whole, 20uL sample injected. 1011

Data were acquired on an Orbitrap Fusion mass spectrometer (Thermo Scientific) coupled to 1012

an Ultimate 3000 RSLC nano UHPLC (Thermo Scientific). Samples were loaded at 10 μl/min 1013

for 5 min on to an Acclaim PepMap C18 cartridge trap column (300 um × 5 mm, 5 um particle 1014

size) in 0.1% TFA. After loading a linear gradient of 3–32% solvent B over 60 or 90min was 1015

used for sample separation with a column of the same stationary phase (75 µm × 75 cm, 2 1016

µm particle size) before washing at 90% B and re-equilibration. Solvents were A: 0.1% FA and 1017

B:ACN/0.1% FA. MS settings were as follows. MS1: Quadrupole isolation, 120’000 resolution, 1018

5e5 AGC target, 50 ms maximum injection time, ions injected for all parallelisable time. MS2: 1019

Quadrupole isolation at an isolation width of m/z 0.7, HCD fragmentation (NCE 34) with the 1020



https://doi.org/10.1101/835686


42

ion trap scanning out in rapid mode from, 8e3 AGC target, 250 ms maximum injection time, 1021

ions accumulated for all parallelisable time. Target cycle time was 2s. 1022

1023

MS data analysis 1024

Spectra were searched by Mascot within Proteome Discoverer 2.1 in two rounds of searching. 1025

First search was against the UniProt Human reference proteome (26/09/17) and compendium 1026

of common contaminants (GPM). The second search took all unmatched spectra from the first 1027

search and searched against the human trEMBL database (Uniprot, 26/0917). The following 1028

search parameters were used. MS1 Tol: 10 ppm, MS2 Tol: 0.6 Da, Fixed mods: 1029

Carbamindomethyl (C) Var mods: Oxidation (M), Enzyme: Trypsin (/P). PSM FDR was 1030

calculated using Mascot percolator and was controlled at 0.01% for ‘high’ confidence PSMs 1031

and 0.05% for ‘medium’ confidence PSMs. Proteins were quantified using the Minora feature 1032

detector within Proteome Discoverer and values normalised against the median protein 1033

abundance of the whole sample. 1034

1035



https://doi.org/10.1101/835686


43

Supplementary figure legends 1036

1037

Figure S1: Characterisation of the HMGCS1-Clover cholesterol reporter (related to Fig 1038

1). Creation of a HMGCS1-Clover CRISPR knock-in in HeLa cells. (A) Schematic for 1039

generation of an endogenous HMGCS1-Clover fusion protein. (B, C) Sterol depletion induces 1040

HMGCS1-Clover expression. HeLa HMGCS1-Clover cells were sterol depleted overnight and 1041

HMGCS1-Clover expression analysed by flow cytometry (B) or immune blotting using GFP 1042

(Clover) and HMGCS1-specfic antibodies. (D) Overnight supplementation with cholesterol and 1043

25-hydroxy-cholesterol reverses HMGCS1-Clover upregulation in sterol-depleted HMGCS1-1044

Clover cells (E, F) Sterol-depletion induced HMGCS1 up-regulation depends on SREBP2. 1045

HMGCS1-Clover CAS9 cells were transfected with two independent sgRNAs against 1046

SREBF1, SREBF2 or both, sterol depleted overnight where indicated and analysed by flow 1047

cytometry for HMGCS1-Clover expression at day 8. (G) HMGCS1 protein stability is not 1048

cholesterol-sensitive. HeLa cells were sterol depleted overnight, pulse-labelled with 35S-1049

Met/Cys and chased in the presence of excess cholesterol/25-hydroxy-cholesterol for the 1050

indicated times. After cell lysis, HMGCS1 or HMGCR was immune precipitated, separated by 1051

SDS-PAGE and analysed by autoradiography. 1052

1053

Figure S2: Validation of select genetic screening hits using the HMGCS1-Clover 1054

reporter (related to Fig 1). CRISPR knockout of select screening hits induces spontaneous 1055

HMGCS1-Clover up-regulation. HMGCS1-Clover CAS9 cells were transfected with sgRNAs 1056

against indicated targets and HMGCS1-Clover expression was analysed by flow cytometry at 1057

day 5 (AP21), day 7 (LDLR, NPC1, INSIG1), day 8 (ANKFY1, SREBF2) or day 9 (ZFYVE20, 1058

VPS16). 1059

1060



https://doi.org/10.1101/835686


44

Figure S3: C18orf8-deficient cells show severe defects in late endosome morphology 1061

and trafficking (related to Fig 2, 3). (A) C18orf8-deficient cells have normal cell surface LDL 1062

receptor (LDLR) expression. Wild-type or C18orf8-deficient cells were labelled for cell surface 1063

LDLR and analysed by flow cytometry. (B) C18orf8-deficient cells are defective in EGF 1064

degradation. Wild-type or C18orf8-deficient cells were pulse-labelled with AlexaFluor-555 1065

conjugated EGF, incubated for 3 hours and fixed. Remaining EGF-555 fluorescence was 1066

visualised by confocal microscopy. (C) C18orf8-deficient cells show clustering of early 1067

endosomes (EE). Rab5 staining of wild-type and C18orf8-deficient cells. (D) C18orf8-deficient 1068

cells accumulated enlarged multivesicular bodies (MVBs). (MVBs). Electron microscopy of 1069

wild-type and C18orf8-deficient cells. MVB size was quantified in 30 cells (**** P<0.0001). (E) 1070

Complementation of endosomal phenotypes in C18orf8-deficient cells. Wild-type cells, 1071

C18orf8-deficient cells and C18orf8-deficient cells complemented with C18orf8-3xHA were 1072

labelled for EEA1, Rab7 and LAMP1 and visualised by confocal microscopy. (F) C18orf8-1073

deficient cells show delayed SR101 trafficking into the late endosomal compartment. Wild-1074

type or C18orf8-deficient cells were labelled with lysotracker (magenta), incubated with 1075

Sulforhodamine 101 dye (SR101, green) and imaged live for 3.5h at 1.5 min intervals. Select 1076

frame overlays from one of 5 positions imaged per experiment per condition are shown. Scale 1077

bars = 10µm. Quantification of SR101 trafficking into the lysotracker+ LE compartment over 1078

time is reported as Manders correlation (top right graph) determined from two independent 1079

experiment using 3-4 fields (containing 10-30 cells per field of view) per experiment (*** 1080

P<0.001). SR101 uptake was determined at indicated times by flow cytometry (bottom right 1081

graph). 1082

1083

Figure S4: C18orf8 interacts with the mammalian Mon1-Ccz1 (MC1) complex and is 1084

essential for MC1 complex stability (related to Fig 4). (A – C) C18orf8 interacts with the 1085

mammalian MC1 complex. (A) Immune precipitation of overexpressed HA-tagged C18orf8 1086

shows an interaction with Ccz1 and Mon1B. (B) Schematic for generation of an endogenous 1087



https://doi.org/10.1101/835686


45

C18orf8-3xMyc fusion protein in HeLa cells. (C) Myc immune precipitation and immunoblotting 1088

of a knock-in clone shows the endogenous C18orf8-3xMyc fusion protein, detected by both 1089

C18orf8 and Myc antibodies. (D) Immune precipitation using a Ccz1-specific antibody but not 1090

an isotype control shows an interaction between Ccz1, endogenous C18orf8-3xMyc and 1091

Mon1B in C18orf8-3xMyc ki cells. (E, F) The C18orf8 C-terminal -helical domain interacts 1092

with and stabilizes mammalian MC1. (E) Depiction of C18orf8 domain structure and mutants 1093

created. (F) C18orf8-deficient cells were complemented with mScarlet-labelled C18orf8 full-1094

length (wt), C-terminal (AA 1-362) and N-terminal truncations (AA 354-657). mScarlet proteins 1095

were immune precipitated and immune precipitations and input controls were analysed by 1096

immune blotting using Mon1B, Ccz1 and mScarlet-specific antibodies. Note wild-type and N-1097

terminal truncations (AA 354-657) stabilize Mon1B expression in input controls. (G) Mon1B 1098

expression in C18orf8-deficient cells is rescued by overnight proteasome inhibition 1099

(bortezomib). (H) The C18orf8 C-terminus alone is insufficient to complement C18orf8 1100

functions. C18orf8-deficient cells were complemented with mScarlet-labelled C18orf8 full-1101

length (wt), C-terminal (AA 1-362) and N-terminal truncations (AA 354-657) and analysed by 1102

flow cytometry for HMGCS1-Clover expression. (I) Complementation of Ccz1-, Mon1A/B- and 1103

Rab7-deficient cells. Ccz1-, Mon1A/B- and Rab7-deficient cells with transduced with 1104

respectively Ccz1-3xMyc; Mon1A-3xMyc, Mon1B-3xMyc or both; and 2xHA-Rab7-wt, -T22N 1105

or –Q67L and HMGCS1-Clover expression was analysed using wild-type HMGCS1-Clover 1106

cells as a control. 1107

1108

Figure S5: C18orf8-deficient cells show impaired ORP1L recruitment and reduced late 1109

endosome motility (related to Fig 5). (A) Wild-type and C18orf8-deficient cells were 1110

transfected with HA-ORP1L and stained intracellularly for HA and LAMP1. Manders 1111

correlation was determined for 6-10 cells from two independent experiments (** P<0.01). (B) 1112

Wild-type or C18orf8-deficient cells were labelled with lysotracker and lysotracker+ 1113

endosomes (white) were traced for 380 sec using live cell microscopy at 3 sec intervals. 1114



https://doi.org/10.1101/835686


46

Displacement of individual endosomes was calculated using TrackMate for Fiji and is depicted 1115

with maximum velocity of travel indicated by a heat map (blue: no motility, red: maximum 1116

motility attainable under selected tracking parameters). 1117

1118

Figure S6: NPC1 expression and localisation is unperturbed in C18orf8-, Ccz1- and 1119

Mon1A/B-deficient cells (related to Fig 6). (A, B) MCC-deficient cells show normal to 1120

increased NPC1 expression. Immune blotting for endogenous NPC1 in (A) wild-type, C18orf8-1121

deficient and complemented C18orf8-deficient cell clones; or (B) wild-type, Ccz1- and 1122

Mon1A/B-double deficient cells. (C, D) NPC1 localisation is largely unperturbed in MCC-1123

deficient cells. (C) NPC1 localises to LAMP1+ LE/Ly in MCC-deficient cells. Wild-type, 1124

C18orf8-, Ccz1- and Mon1A/B-deficient cells were transduced with the inactive NPC1-P692S-1125

3xHA and stained for HA and LAMP1. (D) NPC1 co-localises with the NPC2 cholesterol carrier 1126

in MCC-deficient cells. Wild-type, C18orf8-, Ccz1- and Mon1A/B-deficient cells were stably 1127

transduced with NPC1-P692S-CFP and NPC2-mCherry and monitored using confocal 1128

microscopy. (E) Complementation of NPC1-deficient cells. NPC1-deficient cells were 1129

transduced with wild-type or P692S mutant NPC1-3xHA and analysed by flow cytometry for 1130

HMGCS1-Clover expression. 1131

1132

Figure S7: Rab7 overexpression reduces cholesterol accumulation in Niemann Pick 1133

primary patient fibroblasts (related to Fig 7). NPC1I1061T/I1061T primary patient fibroblasts 1134

were stably transduced with GFP-Rab7 wild-type (wt), dominant negative (T22N) or 1135

constitutively active (Q67L) or empty vector and labelled for free cholesterol at day 7 using 1136

Filipin staining. Fibroblast from a healthy individual were used as a control. Overexpression of 1137

wild-type Rab7 reduces cholesterol accumulation in NPC patient fibroblasts, whereas Rab7-1138

T22N does not and Rab7-Q67L gives an intermediate phenotype. Representative images from 1139

2 independent experiments are shown. 1140



https://doi.org/10.1101/835686


47

1141

Table S1: Full RSA analysis of the HMGCS1-Clover screen for cholesterol regulators 1142

(related to Fig 1). sgRNA enrichment for individual genes in the HMGCS1-Cloverhigh 1143

population compared to a library population, is indicated by RSA score and P-value. Genes 1144

enriched with p<10-5 are annotated into function pathways (See also Fig 1E, F). 1145

1146

1147



https://doi.org/10.1101/835686


48

References: 1148

1. Brown MS & Goldstein JL (1997) The SREBP pathway: regulation of cholesterol metabolism 1149 by proteolysis of a membrane-bound transcription factor. Cell 89(3):331-340. 1150

2. Lund-Katz S & Phillips MC (1986) Packing of cholesterol molecules in human low-density 1151 lipoprotein. Biochemistry 25(7):1562-1568. 1152

3. Goldstein JL & Brown MS (2009) The LDL receptor. Arterioscler Thromb Vasc Biol 29(4):431-1153 438. 1154

4. Pfeffer SR (2019) NPC intracellular cholesterol transporter 1 (NPC1)-mediated cholesterol 1155 export from lysosomes. J Biol Chem 294(5):1706-1709. 1156

5. Watari H, et al. (1999) Mutations in the leucine zipper motif and sterol-sensing domain 1157 inactivate the Niemann-Pick C1 glycoprotein. J Biol Chem 274(31):21861-21866. 1158

6. Li X, Saha P, Li J, Blobel G, & Pfeffer SR (2016) Clues to the mechanism of cholesterol transfer 1159 from the structure of NPC1 middle lumenal domain bound to NPC2. Proc Natl Acad Sci U S A 1160 113(36):10079-10084. 1161

7. Deffieu MS & Pfeffer SR (2011) Niemann-Pick type C 1 function requires lumenal domain 1162 residues that mediate cholesterol-dependent NPC2 binding. Proc Natl Acad Sci U S A 1163 108(47):18932-18936. 1164

8. Infante RE, et al. (2008) NPC2 facilitates bidirectional transfer of cholesterol between NPC1 1165 and lipid bilayers, a step in cholesterol egress from lysosomes. Proc Natl Acad Sci U S A 1166 105(40):15287-15292. 1167

9. Kwon HJ, et al. (2009) Structure of N-terminal domain of NPC1 reveals distinct subdomains 1168 for binding and transfer of cholesterol. Cell 137(7):1213-1224. 1169

10. Pfeffer SR (2016) Clues to NPC1-mediated cholesterol export from lysosomes. Proc Natl Acad 1170 Sci U S A 113(29):7941-7943. 1171

11. Luo J, Jiang L, Yang H, & Song BL (2017) Routes and mechanisms of post-endosomal 1172 cholesterol trafficking: A story that never ends. Traffic 18(4):209-217. 1173

12. Zhao K & Ridgway ND (2017) Oxysterol-Binding Protein-Related Protein 1L Regulates 1174 Cholesterol Egress from the Endo-Lysosomal System. Cell Rep 19(9):1807-1818. 1175

13. Du X, et al. (2011) A role for oxysterol-binding protein-related protein 5 in endosomal 1176 cholesterol trafficking. J Cell Biol 192(1):121-135. 1177

14. Chu BB, et al. (2015) Cholesterol transport through lysosome-peroxisome membrane 1178 contacts. Cell 161(2):291-306. 1179

15. Zhang M, et al. (2002) MLN64 mediates mobilization of lysosomal cholesterol to 1180 steroidogenic mitochondria. J Biol Chem 277(36):33300-33310. 1181

16. Guerra F & Bucci C (2016) Multiple Roles of the Small GTPase Rab7. Cells 5(3). 1182 17. Cantalupo G, Alifano P, Roberti V, Bruni CB, & Bucci C (2001) Rab-interacting lysosomal 1183

protein (RILP): the Rab7 effector required for transport to lysosomes. EMBO J 20(4):683-693. 1184 18. Jordens I, et al. (2001) The Rab7 effector protein RILP controls lysosomal transport by 1185

inducing the recruitment of dynein-dynactin motors. Curr Biol 11(21):1680-1685. 1186 19. Johansson M, Lehto M, Tanhuanpaa K, Cover TL, & Olkkonen VM (2005) The oxysterol-1187

binding protein homologue ORP1L interacts with Rab7 and alters functional properties of 1188 late endocytic compartments. Mol Biol Cell 16(12):5480-5492. 1189

20. Rocha N, et al. (2009) Cholesterol sensor ORP1L contacts the ER protein VAP to control 1190 Rab7-RILP-p150 Glued and late endosome positioning. J Cell Biol 185(7):1209-1225. 1191

21. Sun Q, Westphal W, Wong KN, Tan I, & Zhong Q (2010) Rubicon controls endosome 1192 maturation as a Rab7 effector. Proc Natl Acad Sci U S A 107(45):19338-19343. 1193

22. Liu K, et al. (2017) WDR91 is a Rab7 effector required for neuronal development. J Cell Biol 1194 216(10):3307-3321. 1195



https://doi.org/10.1101/835686


49

23. Rojas R, et al. (2008) Regulation of retromer recruitment to endosomes by sequential action 1196 of Rab5 and Rab7. J Cell Biol 183(3):513-526. 1197

24. Seaman MN, Harbour ME, Tattersall D, Read E, & Bright N (2009) Membrane recruitment of 1198 the cargo-selective retromer subcomplex is catalysed by the small GTPase Rab7 and 1199 inhibited by the Rab-GAP TBC1D5. J Cell Sci 122(Pt 14):2371-2382. 1200

25. Zhang XM, Walsh B, Mitchell CA, & Rowe T (2005) TBC domain family, member 15 is a novel 1201 mammalian Rab GTPase-activating protein with substrate preference for Rab7. Biochem 1202 Biophys Res Commun 335(1):154-161. 1203

26. Frasa MA, et al. (2010) Armus is a Rac1 effector that inactivates Rab7 and regulates E-1204 cadherin degradation. Curr Biol 20(3):198-208. 1205

27. Cabrera M, et al. (2014) The Mon1-Ccz1 GEF activates the Rab7 GTPase Ypt7 via a longin-1206 fold-Rab interface and association with PI3P-positive membranes. J Cell Sci 127(Pt 5):1043-1207 1051. 1208

28. Nordmann M, et al. (2010) The Mon1-Ccz1 complex is the GEF of the late endosomal Rab7 1209 homolog Ypt7. Curr Biol 20(18):1654-1659. 1210

29. Lawrence G, et al. (2014) Dynamic association of the PI3P-interacting Mon1-Ccz1 GEF with 1211 vacuoles is controlled through its phosphorylation by the type 1 casein kinase Yck3. Mol Biol 1212 Cell 25(10):1608-1619. 1213

30. Wang CW, Stromhaug PE, Shima J, & Klionsky DJ (2002) The Ccz1-Mon1 protein complex is 1214 required for the late step of multiple vacuole delivery pathways. J Biol Chem 277(49):47917-1215 47927. 1216

31. Kinchen JM & Ravichandran KS (2010) Identification of two evolutionarily conserved genes 1217 regulating processing of engulfed apoptotic cells. Nature 464(7289):778-782. 1218

32. Poteryaev D, Datta S, Ackema K, Zerial M, & Spang A (2010) Identification of the switch in 1219 early-to-late endosome transition. Cell 141(3):497-508. 1220

33. Langemeyer L, Perz A, Kummel D, & Ungermann C (2018) A guanine nucleotide exchange 1221 factor (GEF) limits Rab GTPase-driven membrane fusion. J Biol Chem 293(2):731-739. 1222

34. Kiontke S, et al. (2017) Architecture and mechanism of the late endosomal Rab7-like Ypt7 1223 guanine nucleotide exchange factor complex Mon1-Ccz1. Nat Commun 8:14034. 1224

35. Yasuda S, et al. (2016) Mon1-Ccz1 activates Rab7 only on late endosomes and dissociates 1225 from the lysosome in mammalian cells. J Cell Sci 129(2):329-340. 1226

36. Gerondopoulos A, Langemeyer L, Liang JR, Linford A, & Barr FA (2012) BLOC-3 mutated in 1227 Hermansky-Pudlak syndrome is a Rab32/38 guanine nucleotide exchange factor. Curr Biol 1228 22(22):2135-2139. 1229

37. Inoue J, Sato R, & Maeda M (1998) Multiple DNA elements for sterol regulatory element-1230 binding protein and NF-Y are responsible for sterol-regulated transcription of the genes for 1231 human 3-hydroxy-3-methylglutaryl coenzyme A synthase and squalene synthase. J Biochem 1232 123(6):1191-1198. 1233

38. Lam AJ, et al. (2012) Improving FRET dynamic range with bright green and red fluorescent 1234 proteins. Nat Methods 9(10):1005-1012. 1235

39. Morgens DW, et al. (2017) Genome-scale measurement of off-target activity using Cas9 1236 toxicity in high-throughput screens. Nat Commun 8:15178. 1237

40. Schroder B, et al. (2007) Integral and associated lysosomal membrane proteins. Traffic 1238 8(12):1676-1686. 1239

41. Itzhak DN, Tyanova S, Cox J, & Borner GH (2016) Global, quantitative and dynamic mapping 1240 of protein subcellular localization. Elife 5. 1241

42. Wubbolts R, et al. (1996) Direct vesicular transport of MHC class II molecules from lysosomal 1242 structures to the cell surface. J Cell Biol 135(3):611-622. 1243

43. Maxfield FR & Wustner D (2012) Analysis of cholesterol trafficking with fluorescent probes. 1244 Methods Cell Biol 108:367-393. 1245



https://doi.org/10.1101/835686


50

44. Nishimura S, et al. (2013) Visualization of sterol-rich membrane domains with fluorescently-1246 labeled theonellamides. PLoS One 8(12):e83716. 1247

45. Edgar JR, Manna PT, Nishimura S, Banting G, & Robinson MS (2016) Tetherin is an exosomal 1248 tether. Elife 5. 1249

46. Lu F, et al. (2015) Identification of NPC1 as the target of U18666A, an inhibitor of lysosomal 1250 cholesterol export and Ebola infection. Elife 4. 1251

47. Schultz ML, et al. (2018) Coordinate regulation of mutant NPC1 degradation by selective ER 1252 autophagy and MARCH6-dependent ERAD. Nat Commun 9(1):3671. 1253

48. Gelsthorpe ME, et al. (2008) Niemann-Pick type C1 I1061T mutant encodes a functional 1254 protein that is selected for endoplasmic reticulum-associated degradation due to protein 1255 misfolding. J Biol Chem 283(13):8229-8236. 1256

49. Vanlandingham PA & Ceresa BP (2009) Rab7 regulates late endocytic trafficking downstream 1257 of multivesicular body biogenesis and cargo sequestration. J Biol Chem 284(18):12110-1258 12124. 1259

50. Shao X, et al. (2016) Numb regulates vesicular docking for homotypic fusion of early 1260 endosomes via membrane recruitment of Mon1b. Cell Res 26(5):593-612. 1261

51. Ao X, Zou L, & Wu Y (2014) Regulation of autophagy by the Rab GTPase network. Cell Death 1262 Differ 21(3):348-358. 1263

52. Pontano Vaites L, Paulo JA, Huttlin EL, & Harper JW (2017) Systematic analysis of human 1264 cells lacking ATG8 proteins uncovers roles for GABARAPs and the CCZ1/MON1 regulator 1265 C18orf8/RMC1 in macro and selective autophagic flux. Mol Cell Biol. 1266

53. Hirst J, et al. (2014) Characterization of TSET, an ancient and widespread membrane 1267 trafficking complex. Elife 3:e02866. 1268

54. Jackson LP, et al. (2010) A large-scale conformational change couples membrane 1269 recruitment to cargo binding in the AP2 clathrin adaptor complex. Cell 141(7):1220-1229. 1270

55. Dong XP, et al. (2010) PI(3,5)P(2) controls membrane trafficking by direct activation of 1271 mucolipin Ca(2+) release channels in the endolysosome. Nat Commun 1:38. 1272

56. Raiborg C, et al. (2015) Repeated ER-endosome contacts promote endosome translocation 1273 and neurite outgrowth. Nature 520(7546):234-238. 1274

57. Guillen-Samander A, Bian X, & De Camilli P (2019) PDZD8 mediates a Rab7-dependent 1275 interaction of the ER with late endosomes and lysosomes. Proc Natl Acad Sci U S A. 1276

58. Wong YC, Ysselstein D, & Krainc D (2018) Mitochondria-lysosome contacts regulate 1277 mitochondrial fission via RAB7 GTP hydrolysis. Nature 554(7692):382-386. 1278

59. Dong J, et al. (2019) Allosteric enhancement of ORP1-mediated cholesterol transport by 1279 PI(4,5)P2/PI(3,4)P2. Nat Commun 10(1):829. 1280

60. Wilhelm LP, et al. (2017) STARD3 mediates endoplasmic reticulum-to-endosome cholesterol 1281 transport at membrane contact sites. EMBO J 36(10):1412-1433. 1282

61. Kumar N, et al. (2018) VPS13A and VPS13C are lipid transport proteins differentially localized 1283 at ER contact sites. J Cell Biol 217(10):3625-3639. 1284

62. Munoz-Braceras S, Tornero-Ecija AR, Vincent O, & Escalante R (2019) VPS13A is closely 1285 associated with mitochondria and is required for efficient lysosomal degradation. Dis Model 1286 Mech 12(2). 1287

63. Shammas H, Kuech EM, Rizk S, Das AM, & Naim HY (2019) Different Niemann-Pick C1 1288 Genotypes Generate Protein Phenotypes that Vary in their Intracellular Processing, 1289 Trafficking and Localization. Sci Rep 9(1):5292. 1290

64. Choudhury A, et al. (2002) Rab proteins mediate Golgi transport of caveola-internalized 1291 glycosphingolipids and correct lipid trafficking in Niemann-Pick C cells. J Clin Invest 1292 109(12):1541-1550. 1293

65. van den Boomen DJ, et al. (2014) TMEM129 is a Derlin-1 associated ERAD E3 ligase essential 1294 for virus-induced degradation of MHC-I. Proc Natl Acad Sci U S A 111(31):11425-11430. 1295



https://doi.org/10.1101/835686


51

66. Wijdeven RH, et al. (2016) Cholesterol and ORP1L-mediated ER contact sites control 1296 autophagosome transport and fusion with the endocytic pathway. Nat Commun 7:11808. 1297

67. van der Kant R, et al. (2013) Late endosomal transport and tethering are coupled processes 1298 controlled by RILP and the cholesterol sensor ORP1L. J Cell Sci 126(Pt 15):3462-3474. 1299

68. Timms RT, et al. (2016) Genetic dissection of mammalian ERAD through comparative haploid 1300 and CRISPR forward genetic screens. Nat Commun 7:11786. 1301

69. Konig R, et al. (2007) A probability-based approach for the analysis of large-scale RNAi 1302 screens. Nat Methods 4(10):847-849. 1303

70. Jongsma ML, et al. (2016) An ER-Associated Pathway Defines Endosomal Architecture for 1304 Controlled Cargo Transport. Cell 166(1):152-166. 1305

71. Peters PJ, Bos E, & Griekspoor A (2006) Cryo-immunogold electron microscopy. Current 1306 protocols in cell biology / editorial board, Juan S. Bonifacino ... [et al.] Chapter 4:Unit 4 7. 1307

72. van Elsland DM, et al. (2016) Detection of bioorthogonal groups by correlative light and 1308 electron microscopy allows imaging of degraded bacteria in phagocytes. Chemical Science 1309 7(1):752-758. 1310

73. Bos E, et al. (2011) A new approach to improve the quality of ultrathin cryo-sections; its use 1311 for immunogold EM and correlative electron cryo-tomography. J Struct Biol 175(1):62-72. 1312

1313



https://doi.org/10.1101/835686


ADAM28

ANKFY1

AP2M1

AP2S1

ARF1

ATP2A2

ATP6AP1

ATP6AP2

ATP6V1AATP6V1B2

ATP6V1H

BECN1

C18orf8

C19orf25

C1GALT1C1CABLES1

CHP1CLU

COG3

COG4

COPB1

DNM2

EGR3

EHD1

EXOC5EXOC7EXOC8

FAM160B2FCHO2

FIG4GALNT11

GNPTAB

GNPTG

HGS

HSP90B1

INSIG1

LDLR

NAPG

NBAS

NPC1

NPC2

NPRL3

NUGGC

PBK

PIK3R4

PNOCPSAP

PTPN23

RAB7A

RABEP1

RABGEF1

RER1RP11-343C2.9

SFPQSLC33A1

SLC35A1

SREBF1

STMN4TGFBRAP1

TNFRSF10AUBR2

USP8UVRAG

VAC14

VPS11

VPS16VPS18

VPS39

VPS41

VPS45

VPS52

VPS53

VPS54VPS8WDR7

ZFYVE20

0

5

10

15

20

25

30

35

0 5000 10000 15000 20000

Figure 1: A genome‐wide CRISPR screen identifies essential factors in cellularcholesterol homeostasis.

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

B

HMGCS1-Clover

Ctrl Sterol depl.

ECholesterol importEndocytosisEndosomal traffickingER/Golgi traffickingProtein folding/glycosylationSREBP pathway

A

wt

HM

GC

S1-

Clo

ver

ki

Clover

HMGCS1

kDa

-actin50

37

75

50

75

C

HeLaHMGCS1‐Clover kiSterol depletedSterol depl. + sgSREBP2

Cholesterol importNPC1/NPC2

Endosomal sortingV‐ATPase (ATP6AP1/AP2/ATP6V1A/V1B2/V1H), ESCRT (HGS, PTPN23)

PI3K (BECN1, UVRAG, PIK3R4), PIKFYVE (FIG4, VAC14)CORVET (VPS8/16/18/TGFBRAP1), HOPS (VPS16/18/39/41),

FERARI (ZFYVE20, VPS45)RABGEF1, RABEP1, C18ORF8, RAB7A

ANKFY1, USP8GARP (VPS52/53/54)

ReceptorLDLR

LDL

golgi

Protein folding & glycosylationHSP90B1, GALNT (GALNT11/T1C1)GNPT (GNPTAB/G), SLC33A1/35A1

ER

ER‐Golgi trafficking COPB1, RER1, NAPG, NBAS

ARF1, CHP1

early endosome

maturinglate endosome

(MVB)

lysosome

free cholesterol

NPC1

NPC2

EndocytosisEHD1, FCHO2

DNM2, AP2 (AP2M1/S1)

Golgi & exocytosisCOG (COG3/4)

EXOCYST (EXOC5/7/8)ARF1, WDR7

SREBP2INSIG1

LDLR

HMGCS1‐Clover(reporter)

F

D

Genes (alphabetical)

Sig

nif

ican

ce

(-L

og

(P))

Library Post‐sort 1

sort 1 sort 2

Post‐sort 2

librarylibrary sterol depl.sorted population

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover



https://doi.org/10.1101/835686


wt

C18orf8

KO #1

C18orf8

KO #2

75

kDA

IP: C18orf8

Blot: C18orf8

A C18orf8 KO #1 C18orf8 KO #2B

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

wild‐typeC18orf8 KO+ empty vector

C18orf8 KO+ C18orf8‐3xHA

wild‐type C18orf8 KO #1 C18orf8 KO #2C

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

no stimulusLPDS alonestatin aloneLPDS + statin

Figure 2: C18orf8 is required for endosomal LDL‐cholesterol uptake

wild-type C18orf8 KOC18orf8 KO+ C18orf8-3HAD wild-type C18orf8 KO

+ C18orf8-3HA

+ C18orf8-3HA

180 min Dil-LDL

Dil-LDL

E

Dil-LDL, 5min chase Dil-LDL, 180min chase

Dil-LDL



https://doi.org/10.1101/835686


Figure 3: C18orf8‐deficient cells show severe defects in late endosome morphology and early‐to‐late endosomal trafficking.

0 min 20 min 60 min 180 min

LAMP1 - EEA1 - EGF-555

wild-type

C18orf8 KO

C

0

5

10

15

20

25

30

35

0 20 60 180

Co

loca

lisat

ion

EG

Fw

ith

EE

A1+

EE

Time (min)

*** ***

wild-typeC18orf8 KO

C18orf8 KO

LAMP1Dil-LDLEEA1 Merge

E

LAMP1Rab7

wild-type

C18orf8 KO

EEA1AEEA1 Rab7 LAMP1 Merge

wild-type

C18orf8KO

B

0

10

20

30

40

50

60 120 180

Co

loca

lisa

tio

nE

GF

Wit

h L

AM

P1+

LE

/Ly

Time (min)

** ******

wild-typeC18orf8 KO

wild-type

C18orf8 KO

60 min 120 min 180 min

LAMP1 - EEA1 - EGF-555

+ Leupeptin

, E‐64d, P

epstatin

D



https://doi.org/10.1101/835686


Figure 4: C18orf8 forms an integral component of the Mon1‐Ccz1 (MC1) complex, essential for complex stability and function.

ProteinNo of pept

C18-HANo of pept

HA-C18

CCZ1 26 33MON1B 25 25MON1A 19 22HUWE1 10 4TRIM28 4 3ARID1A 4 3

A B

Mon1B

Ccz1

Rab7

-actin

- - + - + : C18orf8-3HA

wt C18orf8KO #1

C18orf8 KO #2

HA(C18orf8)

C

75

50

75

20

25

50

F wild-type Ccz1 KO Mon1A/B KO

EEA1Rab7

LAMP1

D

C18orf8

IP: C18orf8

KO:

75

Mon1B75

Ccz150

Tubulin

input

50

E

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

Ccz1

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

Mon1A

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

Mon1B

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

Mon1A/B

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

Rab7

Ctrltarget sgRNA

IP: myc(endogenous

C18orf8)

input

Mon1B

myc(C18orf8 ki)

Ccz1

IP: Mon1B

50

kDa

75

50

75

50

75

50

75



https://doi.org/10.1101/835686


Figure 5: The trimeric Ccz1‐Mon1‐C18orf8 (MCC) complex activates mammalian Rab7.

25

20

Input IP: HA (Rab7)

Rab7endog.2xHA

75 Myc(C18orf8 ki)

50 Ccz1

50

75

Mon1B

2xHA-Rab7: – wt T22N Q67L – wt T22N Q67L

A

C

C

HA-RILPLAMP1

wild-type C18orf8 KO

B

Rab7

FLAG (RILP)

KO:

input

IP: FLAG (RILP)

Rab7

FLAG (RILP)

20

25

50

75

20

25

50

75


+ empty vector + Rab7-Q67L + Rab7-T22N

Dil-LDLEEA1 LAMP1

F

0.0

0.2

0.4

0.6

0.8

RILP

Man

der

s co

rrel

atio

n

wild-type

C18orf8 KO

***

wild-typeC18 KO + shScr.C18 KO + sh1D5C18 KO + sh1D15C18 KO + sh1D5/1D15

D

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

E

wild-typeC18 KO + empty vectorC18 KO + Rab7-T22NC18 KO + Rab7-Q67L



https://doi.org/10.1101/835686


B+ empty vector + C18orf8-3HA


Filipin

A

C18orf8 KO detail

200nm500nm

TNM‐BF

D

C

Filipin

wild-type Ccz1 KO Mon1A/B KO

E

ProteinNo of pept

Abund.

NPC1 12 8x1011

HACD3 10 1x1010

RAB7A 14 8x109

UBB 3 7x109

ARL8B 6 6x109

PI4K2A 24 5x109

F

NPC1

HA(Rab7)

IP: HA (Rab7)

Myc(C18orf8 ki)

input

G

IP: HA (NPC1)

HA (NPC1)

KO:

Rab7

input

25

20

250

150

H

Figure 6: MCC‐deficient cells lack an activation‐dependent interaction between Rab7 and NPC1 and accumulate lysosomal cholesterol.

IP: HA (NPC1)

HA (NPC1)

Rab7

input

25

20

250

150

J

‐ + + + ‐ + + + :NPC1‐3HA

IP: HA (NPC1)

HA (NPC1)

Rab7

input

25

20

250

150

I

Filipin Rab7

LAMP1 Overlay

C18orf8 KO

wild-type NPC1 KOC18orf8 KO

Filipin



https://doi.org/10.1101/835686


Figure 7: Rab7 activation by the MCC GEF controls NPC1‐dependent lysosomal cholesterol export.

A

C + empty vector

C18orf8 KOwild-type

+ 2HA-Rab7-wt + 2HA-Rab7-T22N + 2HA-Rab7-Q67L

FilipinCD63

0

0.2

0.4

0.6

0.8

1

wt NPC1 C18orf8 Ccz1 Mon1A/B

Pea

rso

n c

orr

elat

ion

Fili

pin

-C

D63

** *** **** **

*****

**

KO:

Late endosome / Lysosome

free cholesterol

NPC1

NPC2

GTP

Rab

7

LTP(?)

SREBP2INSIG1

HMGCS1‐Clover(reporter)

ER

b)

‘active’NPC1

Cholesterol export

‘inactive’NPC1

NPC2

GTPGDP

C18orf8

Rab

7

Rab

7

NPC2

GTP

Rab

7

a) Late endosome / Lysosome

MCC GEF

‘active’Rab7

+ empty vector

NPC1I1061T/I1061T

patient fibroblastswild-type

fibroblasts

+ GFP-Rab7

Filipin

FilipinCD63

wild-type C18orf8 KO Ccz1 KO Mon1A/B KO

+ U18666A(NPC1 inh.)

24h LPDS chase

NPC1 KO

B

D

E



https://doi.org/10.1101/835686


puropUbClover

pA

HMGCS1-1 ORF 3’ UTR

pA

LoxP LoxP

HMGCS1 ORF 3’ UTRStop

myc

Cre Cre

CRISPR

A B

CtrlHMGCS1-CloverHMGCS1-Clover

+ sterol depletion

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

75

Ctr

l

Ste

rol d

epl.

C

Clover(short exp.)

HMGCS1

kDa

-actin50

37

75

75

Clover(long exp.)

Figure S1: Characterisation of the HMGCS1‐Clover cholesterol reporter (related to Fig 1).

Steady-stateSterol depletedSterol depleted

+ chol/HO-chol104

0

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

D

Ctrl steady-statesgRNA SREBP steady-stateCtrl sterol depleted

F

Ctrl steady-stateCtrl sterol depletedsgRNA SREBP sterol depleted

SREBP1 sg2SREBP1 sg1

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

SREBP2 sg1

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

SREBP2 sg2

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

100 103101 102

HMGCS1‐Clover

E SREBP2 sg1

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

SREBP2 sg2

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

G

HMGCR

Chase (min): 0 30 60 120

HMGCS155

46

97

kDa

35S‐pulse chase analysis(10 min pulse)



https://doi.org/10.1101/835686


Figure S2: Validation of select genetic screening hits using the HMGCS1‐Clovercholesterol reporter (related to Fig 1).

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

INSIG1

LDLR NPC1 AP21 ANKFY1

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

VPS16 ZFYVE20 (RBSN)

Ctrltarget sgRNA

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

SREBF1



https://doi.org/10.1101/835686


Figure S3: C18orf8‐deficient cells show severe defects in late endosome morphology andtrafficking (related to Fig 2, 3)

0.0

0.2

0.4

0.6

0.8

0 1 2 3 4

Man

der

sco

rrel

atio

n

lyso

trac

ker

-S

R10

1

Time (hours)

0

1

2

3

4

0 1 2

SR

101

up

take

Time (hours)wild-typeC18orf8 KO

wild-type C18orf8 KO C18orf8 KO detail

0

200

400

600

800

1000

wild-type C18orf8KO

MV

B d

iam

eter

(n

m)

****

D

A

1040

60

100

80

40

20

100 103101 102

Surface LDLR

wild‐typeC18orf8 KO #1C18orf8 KO #2

B

EGF-555


EGF-555, 3h chase

E + empty vector + C18orf8-3HA


EEA1Rab7LAMP1

F


Rab5

C



https://doi.org/10.1101/835686


mSc

mSc

mSc

WD40 ‐helicalC18orf81‐657

1‐362

354‐657Mon1B

‐ ‐ + ‐ + : Bortezomib

wtC18orf8 KO #1

C18orf8KO #2

Ccz1

-actin

G

50

50

75

Mon1B

mScarlet(C18orf8)

IP: mScarlet(C18orf8)

Ccz1

input

F

H

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

C18(1‐657)‐mSc C18(1‐362)‐mSc mSc‐C18(354‐657)

wild‐type C18orf8 KO C18orf8 KO + C18orf8 mutant

Figure S4: C18orf8 interacts with the mammalian Mon1‐Ccz1 (MC1) complex and is essential for MC1 complex stability (related to Fig 4).

A

IP: HA (C18orf8)input

w/o

3HA

-C18

orf

8

C18

orf

8-3H

A

w/o

3HA

-C18

orf

8

C18

orf

8-3H

A

50

75

50

75

50

Mon1B

HA(C18orf8)

Ccz1

50

kDA

D IP:

input

IgG

Ccz

1

myc(end. C18orf8)

Ccz1

Mon1B

kDA

50

75

50

75

IP

50

75

50

75

puropUb

pA

C18orf8 ORF 3’ UTR

pA

LoxP LoxP

C18orf8 ORF 3’ UTRStop

3xmyc

Cre Cre

CRISPR

pA

C

B

IP: myc(endogenous

C18orf8)

input

myc(endogenous

C18orf8)

7575

75 C18orf8

– + – + :C18orf8-3xmyc ki

E

wild‐typeCcz1 KOKO + Ccz1

Ccz1

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

Mon1A+B

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

wild‐typeMon1A/B KOKO + Mon1AKO + Mon1BKO + Mon1A+B

Rab7

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

I

wild‐typeRab7 KOKO + Rab7‐wtKO + Rab7‐T22NKO + Rab7‐Q67L



https://doi.org/10.1101/835686


HA-ORP1LLAMP1


0.0

0.1

0.2

0.3

0.4

ORP1L

Man

der

s co

rrel

atio

n

wild-type

C18orf8 KO

**

Figure S5: C18orf8‐deficient cells show impaired ORP1L recruitment and reduced late endosome motility (related to Fig 5).

A

B



https://doi.org/10.1101/835686


wt

- - + - + -

C18 KO #1

C18 KO #2

NPC1KO

-Actin

NPC1250

150

50

37

C18orf8‐3HA

A B

150

VCP

NPC1

KO:

Figure S6: NPC1 expression and localisation is unperturbed in C18orf8‐, Ccz1‐ and Mon1A/B‐deficient cells (related to Fig 6).

NPC1‐CFP

NPC2‐mCherry

Merge

wild‐type C18orf8 KO Ccz1 KO Mon1A/B KOD

NPC1‐HA

wild‐type C18orf8 KO Ccz1 KO Mon1A/B KOC

LAMP1

Merge

NPC1

1040

60

100

80

40

20

100 103101 102

HMGCS1‐Clover

wild‐typeNPC1 KO NPC1 KO + wt‐NPC1‐3HANPC1 KO + NPC1‐P692S‐3HA

E



https://doi.org/10.1101/835686


Figure S7: Rab7 overexpression reduces cholesterol accumulation in Niemann Pick patient fibroblasts (related to Fig 7).

+ empty vector

NPC1I1061T/I1061T

patient fibroblastswild-type

fibroblasts

+ GFP-Rab7-wt

Filipin

+ GFP-Rab7-T22N + GFP-Rab7-Q67L



https://doi.org/10.1101/835686


a trimeric rab7 gef controls npc1-dependent lysosomal ... · 88 mc1-ypt7 complex suggests a gef...

Documents