The architecture of membrane structures involved in hepatitis C virus genome replication revealed in close-to-native conditions by cryo-electron tomography

22 Hepatitis C virus (HCV) infection induces extensive rearrangements of host 23 cytoplasmic membranes, leading to the formation of multiple membranous structures 24 that facilitate RNA replication. Current knowledge of these membranous structures 25 has largely relied on correlative light and electron microscopy (CLEM) techniques 26 using chemical fixation and resin embedding. To overcome these limitations, cryo -27 preserved cell s were prepared using cryo -focused ion beam (cryo -FIB) milling and 28 cryo-ultramicrotomy. For the first time, the contents within the membranous structures 29 have been observed in-situ using cryo-electron tomography (cryo -ET) performed on 30 lamellae (prepared via cryo -FIB) and on ultrathin sections (prepared via cryo-31 ultramicrotomy) from HCV subgenomic replicon harbouring cells. Observations from 32 112 cryo-electron tomograms of cryo -FIB-derived samples revealed the presence of 33 densities within the inner vesicles of a subset of single -, double-membrane vesicles 34 (SMVs, and DMVs respectively), as well as within multi -vesicular bodies (MVBs) , 35 which might represent the viral replication machinery . Notably, this study represents 36 the first direct visualisation of the arrangement of non-structural proteins within a multi-37 membrane vesicle ( MMV) observed from c ryo-electron microscopy of vitreous 38 sections (CEMOVIS) . The cryo -ET methodologies established here lay the 39 groundwork for future investigations into the architecture of the HCV replication 40 complex, leveraging advanced computational tools for deeper structural and functional 41 analysis. 42 43 44 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 3

45 Hepatitis C virus (HCV) is a positive sense RNA virus that primarily infects hepatocytes 46 in the liver, leading to both acute and chronic hepatitis. The latter results in progressive 47 liver damage, including cirrhosis, liver failure, and hepatocellular carcinoma (Martinello 48 et al., 2023) . In common with other p ositive-sense RNA viruses, HCV induces 49 cytoplasmic membrane rearrangements in the host cell to form specialised structures 50 that support genome replication (Wolff et al., 2020). Current ultrastructural insights into 51 host membrane rearrangements during HCV infection have primarily come from 52 fluorescence microscopy (FM), electron microscopy (EM), and soft X -ray microscopy 53 (Ferraris et al., 2010; Romero -Brey et al., 2012; Paul et al., 2014; Lee et al., 2019) . 54 For example, these studies have shown that the membrane rearrangements arise from 55 the ER and autophagy pathways (Mohl et al., 2016; Pérez-Berná et al., 2016). These 56 membranes form vesicular networks, observed during both viral infection and after 57 transfection with subgenomic replicons (SGR), which are often referred to as viral 58 replication factories or the membranous web (MW) (Gosert et al., 2003; Moradpour et 59 al., 2003; Ferraris et al., 2010) . The MW comprises various virus-induced vesicles, 60 and includ es single, double, and multi-membrane vesicles (SMV, DMV, MMV, 61 respectively), multivesicular bodies (MVBs) and lipid droplets (LDs) (Romero-Brey et 62 al., 2012). These structures have been proposed as potential replication organelles 63 (ROs), serving as platforms for the replicase machinery. 64 HCV RNA replication is orchestrated by the five non-structural proteins NS3, NS4A, 65 NS4B, NS5A and NS 5B, which are necessary and sufficient to form MW in the 66 cytoplasm of infected or transfected cells (Lohmann et al., 1999; Gosert et al., 2003). 67 NS3/4A functions as a protease and helicase, facilitating HCV polyprotein cleavage 68 and the unwinding of RNA secondary structures (Bartenschlager et al., 1995). NS4B 69 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 4 contributes to MW formation alongside NS5A (Moradpour et al., 2003) . NS5A is a 70 multifunctional viral protein involved in viral replication and assembly (Hughes et al., 71 2009); notably it binds NS5B and modulates its RNA-dependent RNA polymerase 72 activity (Shirota et al., 2002) . Specifically, within the MW, DMVs have been primarily 73 hypothesised as the main HCV replication site through immunogold labelling of NS3 74 and NS5A. Thus far antibodies against NS5B have not been used to designate a 75 specific vesicle by immunogold labelling. Additionally, double-stranded RNA (dsRNA) 76 (a marker for RNA replication ) showed colocali sation with NS5A by 77 immunofluorescence and correlated with the production of DMVs over the course of 78 infection, further supporting DMVs as active sites of HCV RNA synthesis (Romero-79 Brey et al., 2012). Furthermore, another study employed an SGR containing an NS5A-80 mCherry fusion protein and suggested that the majority of NS5A -positive regions 81 (~35%) were in DMVs and MMVs (Grünvogel et al., 2018). However, the same studies 82 also showed that immunogold labelling of NS3 and NS5A also localised to SMVs, LDs 83 and ER (Romero-Brey et al., 2012), and that a relevant proportion of NS5A -mCherry 84 regions corresponded to MVBs (~20%), ER (~7%), LDs (~7%) and mitochondria 85 (~5%) (Grünvogel et al., 2018) . Additionally, chronically infected liver tissue samples 86 from patients does not present any DMVs, and the MW appeared primarily as clusters 87 of SMVs closely associated with ER and LDs (Blanchard and Roingeard, 2018). 88 Overall, a lthough DMVs are suggested as the primary ROs for HCV , the type of 89 membranous structure utilised as the major HCV replication site is still inconclusive. 90 To date, the EM methods employed to study cellular architecture following infection 91 (or in cells harbouring an SGR) have yielded insights into membrane rearrangements 92 and the roles of individual NS proteins in these processes (Ferraris et al., 2010; 93 Romero-Brey et al., 2012). However, the sample preparation methods for visualisation 94 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 5 by EM utilised chemical fixation, high pressure freezing with freeze substitution, resin 95 embedding and sectioning techniques . T herefore, we explored the recent 96 advancements in cryogenic sample preparation techniques and cryo -electron 97 tomography (cryo-ET), which facilitate imaging in close-to-native conditions, to gain 98 insight into the ultrastructure of these membranous structures in cells harbouring SGR. 99 Overall, our results confirm the presence of all the above -mentioned vesicles within 100 the HCV MW, and supports a model in which internal densities, potentially 101 corresponding to viral and/or cellular components, accumulate inside inner vesicles 102 (InVs), present inside both DMVs and SMVs, and in which NS5A-eGFP accumulates 103 around MMVs within cells stably harbouring an HCV SGR. 104 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 6

105 Genome sequences 106 The HCV genome sequence for JFH-1 (genotype 2a) used in the analysis was sourced 107 from GenBank with accession number AB047639.1. The JFH-1 SGR (Wakita et al., 108 2005) was modified to replace the firefly luciferase (FF-luc) reporter with a fusion of 109 FF-luc and neomycin phosphotransferase (termed Feo) to create the pSGR JFH-1 110 Feo NS3-5B construct (Wyles et al., 2009; Ross-Thriepland et al., 2015) and a green 111 fluorescent protein (GFP) was inserted in NS5A-domain III (using insertion site P418) 112 (Moradpour et al., 2004). The final construct used for this study was SGR JFH -1 Feo 113 NS3-5B (NS5A-eGFP), hereafter referred to as SGR-NS5A-eGFP. 114 Linearisation of plasmid DNA for generating in-vitro RNA transcripts 115 Plasmid DNA (~5 µg) was digested with XbaI (NEB) in CutSmart buffer (NEB) at 37 116 °C for 1 hour and purified by phenol-chloroform extraction. The purified linearised DNA 117 (~1 µg) was used for RNA synthesis using Ribomax Express T7 kit (Promega) and 118 purified using an RNA clean and concentrator kit (Zymo Rese arch). The RNA 119 transcript integrity was analysed on a 1% agarose gel, and concentration was 120 measured using a Nanodrop spectrophotometer. 121 Cell culture 122 Huh7 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Sigma) 123 supplemented with 10% foetal bovine serum (FBS, Gibco), 50 Units/ml penicillin, 50 124 μg/ml streptomycin, 1% nonessential amino acids (NEAA, Lonza) and 2.8% HEPES 125 (Lonza). Cells were maintained in a humified incubator at 37 °C with 5% CO2. 126 127 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 7 To generate stable SGR-NS5A-eGFP harbouring Huh7 cells, 2 µg of replicon RNA 128 was electroporated into 2 x 10 6 cells using a square -wave protocol (260 V, 25 ms). 129 Electroporated cells were mixed with non-electroporated cells to a final density of 1.8 130 x 10⁵ cells and seeded in 6 -well plates. After 48 h of incubation, cells were washed 131 with PBS and subjected to selection with 700 µg/ml G418. Following multiple selection 132 rounds, the G418 concentration was reduced to 400 µg/ml for maintenance of the 133 polyclonal cell population. 134 Fluorescence microscopy 135 To prepare coverslips for confocal fluorescence microscopy, 22 mm coverslips were 136 cleaned with dH ₂O, treated with 1M HCl, washed, incubated in ethanol, dried, and 137 autoclaved. Mock or SGR -expressing cells (5 x 10⁴) were seeded on coverslips or 138 sorted by FACS for fluorescence-based selection. Sorted cells were incubated for ~21 139 h, fixed with 4% formaldehyde, permeabilized with 0.2% Triton X -100, and blocked 140 with 3% BSA-PBS. Coverslips were then stained with BODIPY dye (1:1000), washed, 141 and mounted with ProLong Gold Antifade reagent with DAPI. Mounted slides were 142 incubated for 2.5 h at room temperature and stored at 4 °C. 143 Z-stack confocal microscopy images were captured using a Zeiss LSM880 upright 144 confocal microscope. After acquisition, the images were processed with Fiji software 145 (Schindelin et al., 2012). Wide-field fluorescent images were acquired with the EVOS 146 microscope (Thermo Fisher Scientific). 147 Resin embedding and sectioning for cellular EM 148 Huh7 cells (control and stably expressing SGR -NS5A-eGFP) were trypsinised and 149 pelleted. The resulting cell pellet was fixed in with 2.5% glutaraldehyde (EM Grade, 150 Agar Scientific) in 0.1M sodium phosphate buffer for at least 2.5 h at room 151 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 8 temperature. The cell pellet was washed twice for 30 min in 0.1M sodium phosphate 152 buffer, post-fixed with 1% osmium tetroxide in 0.1M phosphate buffer for 1 h and 153 washed again twice for 30 min with the same buffer. Dehydration was achieved by 154 incubating the pellet for 20 min in increasing concentrations of ethanol: 20%, 40%, 155 60%, 80%, and twice in 100%. The pellet was then treated with propylene oxide twice 156 for 20 min each. A freshly prepared resin mixture of 50% propylene oxide and 50% 157 Araldite epoxy resin was then added to the pellet and allowed to infiltrate for several h 158 to overnight (Luft, 1961). This was followed by infiltration with a 25% propylene oxide 159 and 75% Araldite epoxy resin mixture for 3 h, and with 100% Araldite epoxy resin for 160 3-8 h. Polymerization was then carried overnight at 60 °C. Ultrathin sections (80 -100 161 nm) were collected on slotted copper EM grids with formvar support and stained with 162 saturated uranyl acetate for 2 h and Reynolds lead citrate for 15 min. 163 Image acquisition using the FEI Tecnai T12 TEM 164 Cellular EM images from ultrathin sections prepared through resin embedding and 165 sectioning were captured using the FEI Tecnai T12 TEM operated at 120 kV. 166 Cell culture workflow on EM grids 167 For cryo-ET, Quantifoil Finder 200 mesh gold grids R1.2/1.3 (Gilder G200F1) and C -168 flat 200 mesh gold 1.2/1.3 (Electron Microscopy Sciences) were utilised. The grids 169 were glow discharged in a Quorum GloQube (Quorum Technologies) to enhance 170 hydrophilicity and impart a negative charge. A negative polarity cycle of 20 mA was 171 applied for 30-60 seconds. 172 Next, the grids were positioned in the centre of a glass-bottom cell culture dish (Nunc 173 TM, 150680). Grids were functionalised with 7 μl of fibronectin (Sigma -Aldrich) at a 174 concentration of 25 μg/ml and incubated at 37 °C for 1 h. Concurrently, fluorescence-175 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 9 activated cell sorting (FACS) was performed to collect 5 x 10 3 and 10 x 103 cells and 176 were pelleted and resuspended in 50 μl of complete DMEM before being seeded onto 177 the EM grids at various cell densities. Following a 20 min incubation at room 178 temperature, 2 ml of complete DMEM was added to each dish, and the cells were 179 incubated at 37 °C with 5% CO 2 for 20-24 h to promote optimal cell attachment and 180 spread. 181 Plunge freezing of cells on EM grids 182 After culture, the cells on the EM grids were plunge -frozen using a Leica EM GP 183 Automatic plunge freezer (Leica Microsystems) in liquid ethane at -180 °C and 184 transferred in liquid nitrogen. To facilitate blotting, 3 μl of 1X PBS was applied to the 185 side of the grid containing the cells, and 0.5 μl was applied to the back of the grid 186 within the humidifier chamber. The chamber was maintained at 90% humidity and 8 187 °C, with blotting times adjusted between 5 and 8 seconds for each sample. The grids 188 were blotted using No.1 Whatman paper, then stored in liquid nitrogen. Subsequently, 189 clipped into C -rings and placed into autogrids for screening and data collection 190 compatible with the FEI Titan Krios TEM (Thermo Fisher Scientific). 191 Sample thinning by cryo-focussed ion milling-scanning EM (cryo-FIB-SEM) 192 The lamellae generation was carried out at the Electron BioImaging Centre (eBIC) at 193 Diamond Light Source, utili sing the Scios cryo-FIB-SEM (Thermo Fisher Scientific) 194 and Aquilos cryo-FIB-SEM (Thermo Fisher Scientific). Lamellae were prepared using 195 a focused gallium ion beam on a dual -beam focused ion beam -scanning electron 196 microscope (FIB-SEM) at a stage temperature of -180 °C. Cells located in the centr e 197 of the grid squares were selected for milling. To achieve uniform milling of the cells 198 during the process, an organo-metallic platinum layer was applied prior to milling. This 199 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 10 was followed by a thin layer, or splutter coat, of Pt to enhance SEM imaging. The rough 200 milling was conducted in three gradual steps at a stage angle of 12° using a gallium 201 beam, followed by a final fine milling step to achieve the desired nominal thickness of 202 the lamellae. The final beam current used when milling was performed on SCIOS was 203 37 pA, while on Aquilos was 50 pA. Progress was monitored using SEM on the SCIOS 204 system (operating at 2.2-10 kV with ETD and T2 detectors) and on the Aquilos system 205 (operating at 2-5 kV with ETD detectors and Auto TEM). 206 Sample vitrification by high-pressure freezing 207 SGR-NS5A-eGFP cells were vitrified using a high -pressure freezing method. 208 Approximately 1 x 106 cells per ml were resuspended in 100 µl of cryoprotectant (20% 209 w/v Dextran 40,000 in DMEM). The sample was placed into the 100 µm deep wells of 210 an Au type-A specimen carrier (Leica, catalogue no.16770152) , ensuring the wells 211 were slightly overfilled to prevent air bubbles, and then covered with the flat side of a 212 lipid-coated (L-⍺-Phosphatidylinecholine, Sigma, 61771 ) Au type -B carrier (Leica, 213 catalogue no.16770153 ). Excess liquid was absorbed with a filter paper, and the 214 sample was then high-pressure frozen at approximately 2100 bar and -190 °C for 300 215 ms using a Leica EM ICE High-Pressure Freezer (Leica Microsystems). 216 Cryogenic fluorescence microscopy 217 High-pressure frozen, cryo -preserved carriers and cryo -EM grids containing SGR -218 NS5A-eGFP cells were screened and imaged using a Leica THUNDER Imager EM 219 cryo CLEM (Leica Microsystems) equipped with a HC PL APO 50x/0.9 NA cryo 220 objective, an Orca Flash 4.0 V2 sCMOS camera (Hamamatsu Photonics), and a Solar 221 Light Engine (Lumencor). Z-stack images were acquired with a frame size of 2048 x 222 2048 pixels, at 30% intensity and an exposure time of 0.2 seconds, utilizing the LASX 223 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 11 software (Leica Microsystems). The resulting images were processed using Fiji 224 software (Schindelin et al., 2012). 225 Sample thinning by cryo-ultramicrotomy 226 High pressure frozen samples were sectioned at cryogenic temperatures ( -160 °C) 227 using a Leica UC7 equipped with an FC7 chamber , trimming (Diatome, Trim20), and 228 cryo-sectioning (Diatome, cryo immune, 3 mm ) Diamond knifes, and micro -229 manipulators (Studer et al., 2014) . Images were captured with a stereomicroscope 230 inside the FC7 chamber and correlated with fluorescent images obtained from the 231 Leica Thunder Imager (Leica Microsystems) to precisely identify the regions for cryo-232 ET. A trapezoidal block of tissue measuring 150 x 100 x 40 µm was shaped around 233 fluorescent targets, from which ribbons of either 70 or 40 nm thickness were produced. 234 These ribbons were then adhered to Quantifoil R2/2, Cu 300 mesh grids (EMS) using 235 an electrostatic gun, after the grids had been glow discharged for 60 seconds at 30 236 mA by glow discharge (Easy Glow, Cressington). 237 Image correlation of cryo-EM and cryo-FM maps 238 Maps software (Thermo Fisher Scientific) was utilised to correlate fluorescent images 239 on-the-fly at the TEM. Low magnification electron micrographs (atlas map: 125X, pixel 240 size 819.2 Å, spot size: 7, illuminated area: 1100 µm) were aligned based on broken 241 grid squares and distinctive patterns embedded in the centre of Quantifoil grids. At 242 medium magnification (overview: 580X, pixel size 224.4 Å, spot size: 6, illuminated 243 area: 313 µm), specific grid squares were matched. At high magnifications (search 244 map: 8700X, pixel size 28.8 Å, spot size: 10, illuminated area: 12 µm), holes in the 245 carbon support film served as fiducial markers to facilitate correlation. The correlation 246 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 12 between the fluorescent images and electron micrographs after tomogram collection 247 was achieved using the Matlab script Correlate.tex (Schorb and Briggs, 2014). 248 Cryo-ET 249 Cryo-ET was conducted using the FEI Titan Krios TEM (Thermo Fisher Scientific), 250 operated at 300 kV and equipped with a Falcon 4i direct electron detector and a 251 Selectris imaging filter (Thermo Fisher Scientific) in energy -filtered TEM (EFTEM) 252 nanoprobe mode. 253 Two different acquisition schemes were employed for imaging lamellae. The first one, 254 involved a magnification of 42,000 X, corresponding to a pixel size of 3.0 Å at the 255 specimen level, using Tomography 5.11 or 5.13 software (Thermo Fisher Scientific). 256 A dose-symmetric acquisition scheme (Hagen et al., 2017) was implemented from -257 60º to 60º in 2º increments, maintaining a constant electron dose of ~1.5 e−/Ų per 258 projection, corresponding to a total dose of ~95 e−/Ų. The target defocus was 259 between -3 and -5 μm and an energy filter slit set to 10 eV was used. Individual 260 projections were captured in counting mode through dose fractionation. Four frames 261 per projection were aligned and summed 'on-the-fly' using custom Python scripts that 262 utilized commands from the MotionCor2 (Zheng et al., 2017) and IMOD (Kremer et al., 263 1996) software packages. The second acquisition scheme was similar to the first one, 264 with the following differences: the magnification employed was 64,000 X, with a 265 corresponding pixel size of 1.9 Å. The electron dose was ~2.9 e−/Ų per projection, 266 corresponding to a total dose of ~120 e−/Ų. The tilt range covered was from 71.2° to 267 -46.8° in 3° increments, starting at an angle of 12.14°. Eight frames per projection were 268 collected. 269 270 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 13 Similarly, tilt series for the cryo-ultrathin sections were collected at a magnification of 271 53,000 X, with a corresponding pixel size of 2.40 Å. A dose -symmetric acquisition 272

was employed from -60° to 60° in 3° increments, maintaining a constant 273 electron dose of ~2.2 e−/Ų per projection, resulting in a total electron dose of ~91 274 e−/Ų. The target defocus ranged from -3 to -5 μm, and an energy filter slit of 10 eV 275 was utilized. Six frames per projection were collected. 276 Tomogram reconstruction 277 Raw images were motion corrected using MotionCor2 (Zheng et al., 2017) . The tilt 278 series images were aligned and reconstructed via the eTOMO interface of the IMOD 279 software package (Kremer et al., 1996) . For lamellae dataset 1, tomograms were 280 generated using fiducial -less alignment. All tomograms were reconstructed with 281 weighted back-projection. For lamellae dataset 2 and the tomograms of cryo-ultrathin 282 sections, reconstruction was carried out using Aretomo software (Zheng et al., 2022). 283 All tomograms were binned 4 times. Tomograms from datasets 1 and 2 are presented 284 after filtering with 3D Gaussian blur function in Fiji. For the CEMOVIS dataset, images 285 are presented without filtering. 286 Statistical analysis on diameter of the vesicles 287 To evaluate the size of each membranous structure, identifiable vesicles and cellular 288 features were summarized for each tomogram based on the existing literature. The 289 data on the lamellae were collected without specific correlative guidance for the NS5A-290 eGFP foci, leading to the analysis of all measurable membrane structures (SMV, DMV, 291 MMV, MVB, LD) across 112 tomograms (combined dataset) . The predominant 292 vesicles visible in the tomograms and measurable included SMVs, followed by DMVs, 293 LDs, MVBs, MMVs, and mitochondria. For each structure, both the minor and major 294 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 14 axes were measured using IMOD’s ‘measure’ function with the diameter calculated by 295 averaging these two measurements. The data w as exported to GraphPad Prism f or 296 presentation. 297 298 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 15

299 Establishing a system to study HCV MW 300 Most ultrastructural studies examining HCV MW have utilised the JFH-1 virus or JFH-301 1-SGR-NS5A-eGFP, supporting its use for structural studies (Ferraris et al., 2010; 302 Romero-Brey et al., 2012; Ferraris et al., 2013; Lee et al., 2019). NS5A plays a critical 303 role in both replication and assembly of the virus and is distributed as puncta 304 throughout the cytoplasm (Eyre et al., 2014), which makes it a key protein to track and 305 study HCV replication. 306 To validate the use of SGR-NS5A-eGFP (Figure 1A) harbouring cells for subsequent 307 cryo-EM based analysis , the localisation of NS5A -eGFP in proximity to the LDs, as 308 previously reported (Lee et al., 2019), was confirmed by confocal microscopy (Figure 309 1B). To confirm architectural differences between control Huh7 cells and those stably 310 harbouring the SGR-NS5A-eGFP, resin -embedded cell samples were sectioned, 311 stained, and imaged by transmission electron microscope (TEM) prior to implementing 312 a cryo-electron tomography (cryo-ET) workflow. Overall, noticeable differences were 313 observed in the cellular architecture between control cells and cells harbouring SGR-314 NS5A-eGFP. Whereas control cells lacked any evidence of intracellular membrane 315 reorganisation, stably harbouring cells contained apparent MWs (Figures 1C and 1D). 316 Therefore, cells stably harbouring SGR-NS5A-eGFP were employed to establish in-317 situ cryo-ET workflows and to gain knowledge into the organisation of the different 318 types of membranous vesicles present within the MW of HCV. 319 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 16 320 Figure 1: Huh7 stably harbouring SGR-NS5A-eGFP show membrane rearrangements. A) Schematic of the HCV SGR-NS5A-eGFP employed in this study. B) Confocal image of .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 17 Huh7 cells harbouring SGR -NS5A-eGFP. Cells were fixed with 4% formaldehyde . LDs stained with 1:1000 dilution of BODIPY (558/568) for 1 hour at room temperature and nuclei stained with ProLong gold antifade mountant with DAPI. NS5A-eGFP (green), LDs (red) and nuclei of cells stained with DAPI (blue) are shown . Scale bar: 20 µm . C and D) Representative TEM of ultrathin sections of Huh7 cells (C) and Huh7 cells harbouring HCV SGR-NS5A-eGFP (D). MW: Membranous web, LD: Lipid droplets, Nu: Nucleus, ER: Endoplasmic reticulum. Scale bars: C1,5 µm; D1,4 µm; C2 and D2, 2 µm; C3 and D3, 1µm. 321 Sample preparation workflow to generate lamellae by cryo-FIB for cryo-ET 322 Next, to gain insight into the architecture of the HCV MW in the SGR-NS5A-eGFP 323 harbouring cells in close-to-native conditions, two different approaches were explored: 324 cryo-focused ion beam m illing (cryo-FIB) and cryo-electron microscopy of vitreous 325 sections (CEMOVIS). 326 To establish a cryo-FIB workflow, a homogenous population of cells harbouring SGR-327 NS5A-eGFP were fluorescently activated cell sort ed (FACS) and selected for the 328 brightest eGFP intensities (Figure 2A). These cells were seeded on negatively glow-329 discharged and fibronectin-functionalised gold grids (Figure 2B). Since cell adherence 330 was variable on each grid , even after functionalisation , optimal cell distribution was 331 analysed by widefield fluorescence microscopy before selecting grids for plunge 332 freezing (Figure 2B and C). Subsequently, grids were screened by cryo -EM for ice 333 thickness and target cells suitable for cryo -FIB milling were selected (Figure 2D ). 334 Generally, the success rate of grids with optimum cell targets and ice thickness was ~ 335 50% per session. Cryo-FIB was then performed and the presence of LDs in the 336 lamella, as observed through SEM imaging, confirmed that the lamella’s thickness was 337 suitable for cryo-ET imaging (Figures 2E and F). 338 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 18 Figure 2: Schematic sample preparation workflow to generate lamellae by cryo -FIB for cryo-ET of Huh7 cells stably harbouring SGR-NS5A-eGFP. Schematics (A, C and E) and images of workflow (B, D and F) to prepare samples for cryo-FIB and cryo-ET imaging of HCV MW. A) Schematic of sample preparation for grid preparation. Cell cultures of Huh7 cells stably harbouring SGR-NS5A-eGFP (left) were FACS sorted (centre) and cultured on glow-discharged and fibronectin-functionalised gold EM grids (right). B) The cell adherence and distribution on the grid was analysed by fluorescence microscopy. Scale bars: 1 mm (centre) and 0.5 mm (right). C) Schematic of plunge freezing process . D) Ice thickness of plunge-frozen grids and identification of cell targets suitable for cryo-FIB were assessed by .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 19 cryo-EM. Scale bars: 100 µm. E) Schematic of cryo -FIB milling to generate lamella e for cryo-ET. F) SEM images showing the process of cryo -FIB: top, SEM ( F1) and ion ( F2) images of a target cell prior to cryo -FIB; bottom, SEM image of a lamella in which LDs are apparent (F3) and low magnification SEM image of a lamella (F4). Scale bars: C1, 30 µm; C2 and C3, 10 µm; and C4, 200 µm. 339 Quantitative and qualitative analysis of HCV MW using the cryo-FIB/cryo-ET 340 workflow 341 To investigate the native architecture of HCV the different membranous vesicles within 342 the MW , cryo-ET was applied to the lamellae generated from SGR-NS5A-eGFP 343 harbouring cells. To locate regions for tilt-series acquisition, search maps were utilised 344 (Figure 3A), looking for areas containing MWs. A total of 112 reconstructed tomograms 345 from two datasets were analysed. Among these, 50% contained MVBs, 39% included 346 SMVs, 29% featured DMVs , 15% contained LDs, 12% included ER, 11% featured 347 MMVs, 9% exhibited extensive MW, and 5% contained mitochondria. 348 Based on the hypothesis that DMVs are the main sites of replication (Romero-Brey et 349 al., 2012; Romero-Brey and Bartenschlager, 2015), the initial investigation centred on 350 studying the ultrastructure of DMVs and their surrounding environment. DMVs were 351 often surrounded by double-membrane tubules (DMTs ), SMVs, and MVBs . Most of 352 the DMVs examined had a closed configuration (i.e. no pores connecting the inside of 353 the DMVs to the cytoplasm were observed; Figure 3B-1), with only one instance (1 out 354 of 112 tomograms, consisted of two open DMVs) exhibiting an aperture facing towards 355 the cytosol (Figure 3B-2). 356 As an initial characterisation of the different vesicles, the diameter of all membranous 357 structures that were clearly visible (n = 199) was measured (Figure 3C). The majority 358 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 20 of SMVs, DMVs and inner vesicles ( InVs, found within DMVs and SMVs) had a 359 diameter between 50 and 300 nm, whilst LDs displayed a wide r range of diameter , 360 ranging from 100 to 500 nm. MVBs, the most abundant type of vesicles imaged in the 361 datasets, were much larger and mostly above 400 nm in diameter . However, their 362 larger size rendered them only partially visible within the tomograms, making it 363 challenging to accurately measure their diameter. For this reason, the diameter of only 364 20 MVBs was measured. Finally, MMVs were scarce, observed only in 11 tomograms. 365 However, only 3 could be measured and had a diameter similar to that of MVBs. 366 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 21 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 22 Figure 3: HCV membranous structures imaged by cryo-FIB/cryo-ET of Huh7 cells harbouring SGR-NS5A-eGFP. A) Representative search map ( magnification 15,000X) of a lamella containing membranous structures around LDs. An area highlighting specific structures is shown in the inset. Scale bars: search map, 2 µm; inset, 500 nm. B) Representative sections through reconstructed tomograms showing the architecture of HCV MW. Scale bars: 200 nm. The white arrow indicates the opening of a DMV towards the cytosol. C) Graph depicting the diameters of the 199 observed membranous structures . SMV: Single membrane vesicle . DMV: Double membrane vesicle. MMV: Multi membrane vesicle. InV: Inner vesicle. LD: Lipid droplets. MVB: Multivesicular bodies. Mito: Mitochondria. Next, differences in the general distribution of densities within DMVs and SMVs were 367 examined. Most DMVs (92.8%) either contained faint densities that could be a result 368 of the inherent noise within cellular cryo -electron tomograms or appeared empty 369 (Figure 4A-1,2). Only 7.14% DMVs contained patches of densities on the inside of the 370 membrane that could potentially represent an assembly of the replicase component s 371 (Figure 4A-3, 4B-1). Finally, 17.8% DMV contained an InV (either a single or double 372 membrane, Figure 4B-1,2,3). 373 On the other hand, most SMVs (51.4%, Figure 4C-1, 4D-1) contained faint densities 374 or appeared empty making it difficult to co nfirm their contents , 28.15% contained 375 homogeneous densities throughout the vesicle (Figure 4 C-2). Only 2.9% SMV 376 contained patches of densities (Figure 4 C-3), which could indicate the presence of 377 either cellular or viral proteins . Similar to DMVs, 19.4% of SMVs contained InVs; 378 Figures 4D-1,2,3). Out of 31 InVs measured for diameter (6 inside DMVs and 20 inside 379 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 23 SMVs), 38% of them contained densities, 32% of them could not be confirmed but 380 possibly have densities, 25% of them appeared empty. 381 Previous reports by i mmunogold labelling suggested that both NS3 and NS5A 382 primarily labelled rER and SMVs with 50-70 nm diameter and to a lesser extent on 383 DMVs (Romero-Brey et al., 2012) and th us at least some of these densities could 384 potentially be NS3 and NS5A proteins. Overall, this analysis suggests that DMVs and 385 SMVs are similar in terms of content, but a larger percentage of SMVs contained 386 internal densities, which might correspond to viral and/or cellular proteins. Strikingly, 387 the only membranous structure that contained a significant percentage of inner 388 densities were InVs. 389 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 24 Figure 4: Differences in densities present within DMVs and SMVs . A) Representative tomographic slice s showing the differences in the content within DMV s. B) Representative tomographic slices showing InVs within DMVs and their contents. C) Representative tomographic .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 25 slices showing the differences in the content within SMVs. D) Representative tomographic slices showing InVs within the SMV and their content. Scale bars: A1,2; B2,3; C1,2 and D1,2,3: 100 nm, A3, B1and C-3: 160 nm. White arrows indicate densities present within the vesicles. SMVs: Single membrane vesicles. DMVs: Double membrane vesicles. InV- Inner vesicle 390 NS5A-eGFP preferentially locates around MMVs 391 So far, a clear organisation of the replicase machinery within specific structures has 392 not been established using the cryo-FIB milled cryo-ET datasets, due to limitations in 393 utilising NS5A -eGFP fluorescence for guided tilt -series collection on the lamellae. 394 Therefore, an alternative workflow was developed , incorporating CEMOVIS . This 395 approach aimed to gather evidence regarding the types of structures close to NS5A, 396 potentially housing the replication complex in cells harbouring SGR-NS5A-eGFP 397 (Figure 5) . An advantage of this approach is its ability to employ a heterogeneous 398 population of SGR-NS5A-eGFP cells, thereby eliminating the need for FACS sorting. 399 NS5A-eGFP puncta are detectable in cryo-ultrathin sections following cryo-400 ultramicrotomy and can thus serve as markers for guided tilt-series acquisition. 401 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 26 Figure 5: Schematic sample preparation workflow to generate cryo-ultrathin sections by CEMOVIS of Huh7 cells stably harbouring SGR-NS5A-eGFP. A) Cells were initially mixed with a cryoprotectant. B) Subsequently, cells were added to gold carriers and high-pressure frozen and imaged by cryo-fluorescence microscopy (C). Scale bar : 500 µm. Finally, cells were thinned by CEMOVIS (D) to generate ultrathin sections (E). Scale bar: 2 µm. 402 Using this approach, ultrathin sections of varying thicknesses (100 nm, 70 nm, and 40 403 nm) were evaluated, with the 40 nm sections displaying clear cellular features, 404 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 27 whereas the thicker sections exhibited low contrast . Therefore, 12 tilt-series were 405 collected from a single 40 nm section , guided by NS5A-eGFP foci and subsequently 406 correlated (Figure 6A). 407 Reconstructed tomograms from 12 NS5A -eGFP tilt-series contained 83.3% MMVs, 408 33.3% ER, 25% DMVs, 25% mitochondria, 16.6% LD and 8.3% MVB. 12 tomograms 409 primarily revealed 39 MMVs with notable densities inside (4 of which are illustrated in 410 Figure 6B-E) and only 3 DMV identified. Similar densities were observed within and 411 around DMVs as noted in the cryo-FIB-milled dataset. Interestingly, the proportion of 412 DMVs and MMVs was reversed between tomograms obtained using the CEMOVIS 413 and cryo-FIB workflows. In the cryo -FIB datasets, which were not guided by NS5A -414 eGFP presence, DMVs were more prevalent, comprising 13% of the vesicles (2 7 415 DMVs out of 199 structures measured). In contrast, MMVs dominated in the CEMOVIS 416 datasets, comprising 80% of the vesicles (39 MMVs out of 49 structures identified) , 417 which were exclusively collected in regions exhibiting NS5A-eGFP signals. Strikingly, 418 one of the MMVs displayed a heterogeneous arrangement of proteins, potentially 419 indicative of a replication complex assembly (Figure 6E). Notably, this represents the 420 most organised assembly of cryo-densities observed across all datasets, underscoring 421 the value of fluorescence-guided tilt-series collection on thin cryo-specimens. 422 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 28 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 29 Figure 6: NS5A-eGFP-guided acquisition of HCV MW observed on Huh7 Cells stably expressing HCV SGR-NS5A-eGFP by CEMOVIS and cryo-ET. A) Left, low magnification search map (125 X) of a 40 nm cryo-ultrathin ribbon with regions of NS5A-eGFP signal shown in white boxes. Centre and right, the area of NS5A-eGFP foci used for tilt-series collection is shown with green circles and labelled with designated letters that correspond to the search maps and tomograms below. Scale bar: 2 µm. B-E) Left, cryo- EM search map (580 X) with highlighted position by a white circle where the tilt series was acquired. Centre and right, tomographic sections at low (centre) and high magnification (right) of membranous structures observed in the tomograms acquired at the specified NS5A-eGFP signals. Scale bars left, 100 nm; centre, 200 nm; right, 100 nm. MMVs: Multi membrane vesicles. DMV: Double membrane vesicle. 423

424 The use of direct-acting antivirals (DAAs) has dramatically improved the life of HCV-425 infected patients, resulting in a reduction in the global burden of HCV from 170 million 426 individuals 10 years ago to the current estimate of 50 million (WHO, 2024). The targets 427 for DAAs (NS3 protease, NS5A and NS5B RNA -dependent RNA polymerase) are all 428 directly involved in virus genome replication . It is thus important to understand th e 429 molecular details of this process as this will shed light on the mode of action of DAAs 430 and may contribute to an understanding of DAA resistance which is becoming 431 increasingly common. In this regard one important avenue of research has focused on 432 understanding how HCV modifies the host membranes to establish its replication 433 complex. From 2002 to 2019 a number of research groups explored the membranous 434 structures within the MW in relation to HCV infection or the presence of SGR, using a 435 mix of light and electron microscopy techniques (Ferraris et al., 2010; Romero-Brey et 436 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 30 al., 2012; Paul et al., 2014; Berger et al., 2014; Mohl et al., 2016; Pérez -Berná et al., 437 2016; Lee et al., 2019). However, these studies relied on chemical fixatives and resin-438 embedded sections, limiting direct visualisation of key replication components such as 439 dsRNA and protein arrangements within membranous structures. Furthermore, even 440 though DMVs have been identified as the most probable structures to harbour the 441 replication complex (Romero-Brey et al., 2012), the location of the replication complex 442 is still under debate based on direct visualisation of assembly of viral and cellular 443 proteins. Here, we aimed to develop a cryo-ET workflow to visualise HCV MW in close-444 to-native conditions in the absence of chemical fixation, and to explore the localisation 445 of the HCV replication complex within it. A detailed workflow was developed, spanning 446 from sample preparation to data analysis, incorporating the latest cryo-ET techniques, 447 and t wo cryogenic sample preparation techniques were established for an in-situ 448 investigation of the membranous structures in cells containing an SGR (HCV genotype 449 2a): cryo -FIB milling (Lam and Villa, 2021) and CEMOVIS (Chlanda and Sachse, 450 2014). 451 HCV-induced host membrane rearrangements are driven by viral proteins. In SGR 452 harbouring cells host membrane modifications similar to those in HCV -infected cells 453 were observed (Romero-Brey et al., 2012). To track viral proteins, we utilised an SGR-454 NS5A-eGFP replicon. This construct is the only HCV system with a fluorescently 455 tagged non-structural protein. However, one limitation of this approach is that NS5A 456 is involved in both replication and assembly (Eyre et al., 2014), suggesting that some 457 observed NS5A-eGFP signals may not be exclusively associated with replication sites. 458 Datasets acquired using cryo -FIB milling followed by cryo -ET (Wagner et al., 2020) 459 without fluorescence guidance enabled the exploration of the MW induced by the 460 SGR-NS5A-eGFP in Huh7 cells. Analysis of these datasets revealed numerous DMVs 461 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 31 and SMVs. While most DMVs appeared empty or contained densities at background 462 levels, SMVs frequently exhibited either patchy densities or were entirely filled. 463 Additionally, both DMVs and SMVs occasionally contained in ner vesicles (InVs) with 464 densities present within . These InVs have been previously described as self -465 invaginations of SMVs (Romero-Brey et al., 2012) or vesicles in cluster (ViCs) (Ferraris 466 et al., 2013). However, they may also represent exosomes containing dsRNA or non-467 structural proteins that fuse with larger vesicles, potentially serving as replication sites 468 (Ramakrishnaiah et al., 2013; Bukong et al., 2014; Yin et al., 2022) . Given that 469 positive-sense RNA virus replication sites are expected to contain viral replicase and 470 RNA (Wolff et al., 2020) , our findings suggest that SMVs and InVs may serve as 471 primary sites for HCV SGR replication. This is further supported by EM studies of 472 chronically infected HCV patient cells, which identified SMV s and not DMV around 473 LDs and the ER (Blanchard and Roingeard, 2018). 474 In any case, the m ost frequently observed vesicles in the cryo-FIB/cryo-ET dataset 475 were the MVBs, consistent with EM analysis of replicon cells transfected with NS5A-476 mCherry (Grünvogel et al., 2018) , suggesting their involvement in the process of 477 replication. The aggregation of MVBs, found primarily near SMVs, DMVs, and LDs, 478 with diameter sizes ranging from 400–800 nm (though most exceeded 1000 nm) were 479 previously described as single -membrane compartments containing multiple circular 480 units with dense cores (Ferraris et al., 2010). The presence of densities within SMVs 481 and DMVs within MVB also aligns with the previous observations of NS5A -mCherry 482 localised inside MVBs (Grünvogel et al., 2018) and with the dense lumen within 483 concentric units (Ferraris et al., 2010). The smaller SMVs (30 –150 nm) observed 484 within MVBs resemble exosomes in size. Previous studies have demonstrated that 485 exosomes derived from HCV-infected cells contain viral RNA, proteins, and complete 486 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 32 virions (Yin et al., 2022). Notably, these exosomes facilitate the transfer of HCV RNA 487 to uninfected cells, even in the presence of neutralising antibodies (Ramakrishnaiah 488 et al., 2013). This suggests that MVBs play a role in HCV RNA dissemination. On rare 489 instances, tomograms captured SMVs invaginating into MVBs (not shown), supporting 490 the hypothesis that MVBs may transport vesicles containing replicase machinery or 491 dsRNA to neighbouring cells (Grünvogel et al., 2018) . Perhaps, the accumulation of 492 MVBs in SGR harbouring cells is due to the lack of viral assembly in this system, thus 493 depriving nascent genomic RNA of a ‘final destination’? Nonetheless, this observation 494 suggests a potential role for MVBs in HCV replication 495 Previous immunofluorescence studies indicated that HCV replication vesicle 496 membranes originate from the ER but also colocalise with markers of early and late 497 endosomes, coat protein complex (COP) vesicles, mitochondria, and LDs (Romero-498 Brey et al., 2012) , as well as lysosomes (Matsui et al., 2021) . The measured SMV 499 diameter (50–300 nm) aligns with that of endosomes (100 –500 nm) and lysosomes 500 (200–300 nm). If these SMVs are derived from early endosomes, their internal 501 densities may represent NS3 and NS5A, while surface -exposed densities may 502 correspond to GFP-Rab21 (Romero-Brey et al., 2012). Alternatively, if they originate 503 from lysosomes, internal densities may include NS5A and LAMP -2A (Matsui et al., 504 2021). 505 Notably, in CEMOVIS datasets guided by NS5A -eGFP fluorescence, M MV 506 accumulation was observed. One tomogram revealed densities arranged in a regular 507 pattern, potentially representing replicase machinery assembly on the inner leaflet of 508 an MMV. Overall, this study contributes to the development of cryo in-situ workflows 509 for investigating HCV-induced membranous structures. Our findings show that DMVs 510 appear mostly empty, while SMVs and InVs contain significant densities. Additionally, 511 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 33 MVBs emerge as the most abundant vesicles in MW imaging at random, reinforcing 512 their potential role in HCV replication and RNA transfer. The most abundant vesicles 513 among NS5A-eGFP fluorescence-guided cryo-ET were MMVs, suggesting that are 514 involved in replication in cells harbouring SGR-NS5A-eGFP. 515 In conclusion, these findings provide new insights into the spatial organi sation of NS 516 or cellular proteins within membranous structures in replicon -transfected cells, 517 revealing structural details previously inaccessible through conventional electron 518 microscopy. As cryo-ET continues to evolve with advancing computational techniques, 519 it will further refine our understanding of NS protein interactions with cellular proteins, 520 and HCV RNA genome organisation within these membranous structures. 521 522

523 UMS was supported by a Wellcome PhD studentship (222370/Z/21/Z). HCV studies 524 in the MH laboratory w ere supported by a Wellcome Investigator Award 525 (096670/Z/11/Z) and an MRC project grant (MR/S001026/1) . JF was supported by 526 grant PID2023-149259NB-I00, funded by MICIU/AEI/10.13039/501100011033 and by 527 “ERDF A way of making Europe” . The funders had no role in study design, data 528 collection and analysis, decision to publish, or preparation of the manuscript. 529 We acknowledge the technical contributions of the University of Leeds Bioimaging 530 Facility, especially Dr Ruth Hughes and Dr Sally Boxall, and the Astbury Biostructure 531 Laboratory Electron Microscopy facility, especially Mr Martin Fuller , Dr Rebecca 532 Thompson; and the Electron BioImaging Centre (eBIC) at Diamond Light Source, 533 especially Dr James Gilchrist. 534 535 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 34 Author Contributions 536 UMS, JF and MH planned the study. UMS, TJO’S, and YH performed the experiments. 537 UMS and JF analysed the data. UMS, JF and MH wrote the main manuscript text. All 538 authors reviewed the manuscript. 539 Competing interests 540 The authors declare no competing interests. 541 Data availability 542 The datasets generated during and/or analysed during the current study are available 543 from the corresponding authors on reasonable request. 544 545 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 35

546 547 Bartenschlager, R., Lohmann, V., Wilkinson, T. and Koch, J.O. 1995. Complex 548 formation between the NS3 serine -type proteinase of the hepatitis C virus 549 and NS4A and its importance for polyprotein maturation. Journal of Virology. 550 69(12), pp.7519–7528. 551 Berger, C., Romero -Brey, I., Radujkovic, D., Terreux, R., Zayas, M., Paul, D., 552 Harak, C., Hoppe, S., Gao, M., Penin, F., Lohmann, V. and Bartenschlager, 553 R. 2014. Daclatasvir -like inhibitors of NS5A block early biogenesis of 554 hepatitis C virus-induced membranous replication factories, independent of 555 RNA replication. Gastroenterology. 147(5), pp.1094–105.e25. 556 Blanchard, E. and Roingeard, P. 2018. The Hepatitis C Virus -Induced 557 Membranous Web in Liver Tissue. Cells. 7(11). 558 Bukong, T.N., Momen -Heravi, F., Kodys, K., Bala, S. and Szabo, G. 2014. 559 Exosomes from hepatitis C infected patients transmit HCV infection and 560 contain replication competent viral RNA in complex with Ago2 -miR122-561 HSP90. PLoS Pathogens. 10(10), p.e1004424. 562 Chlanda, P. and Sachse, M. 2014. Cryo-electron microscopy of vitreous sections. 563

in Molecular Biology. 1117, pp.193–214. 564 Eyre, N.S., Fiches, G.N., Aloia, A.L., Helbig, K.J., McCartney, E.M., McErlean, 565 C.S.P., Li, K., Aggarwal, A., Turville, S.G. and Beard, M.R. 2014. Dynamic 566 Imaging of the Hepatitis C Virus NS5A Protein during a Productive Infection. 567 Journal of Virology. 88(7), pp.3636–3652. 568 Ferraris, P., Beaumont, E., Uzbekov, R., Brand, D. biogenesis of host cell 569 membrane rearrangements induced by hepatitis C. virus infection, Gaillard, 570 J., Blanchard, E. and Roingeard, P. 2013. Sequential biogenesis of host cell 571 membrane rearrangements induced by hepatitis C virus infection. Cellular 572 and Molecular Life Sciences. 70(7), pp.1297–1306. 573 Ferraris, P., Blanchard, E. and Roingeard, P. 2010. Ultrastructural and 574 biochemical analyses of hepatitis C virus-associated host cell membranes. J. 575 Gen. Virol. 91(9), pp.2230–2237. 576 Gosert, R., Egger, D., Lohmann, V., Bartenschlager, R., Blum, H.E., Bienz, K. and 577 Moradpour, D. 2003. Identification of the hepatitis C virus RNA replication 578 complex in Huh-7 cells harboring subgenomic replicons. Journal of Virology. 579 77(9), pp.5487–5492. 580 Grünvogel, O., Colasanti, O., Lee, J. -Y., Klöss, V., Belouzard, S., Reustle, A., 581 Esser-Nobis, K., Hesebeck-Brinckmann, J., Mutz, P., Hoffmann, K., Mehrabi, 582 A., Koschny, R., Vondran, F.W.R., Gotthardt, D., Schnitzler, P., Neumann -583 Haefelin, C., Thimme, R., Binder, M., Bartenschlager, R., Dubuisson, J., 584 Dalpke, A.H. and Lohmann, V. 2018. Secretion of Hepatitis C Virus 585 Replication Intermediates Reduces Activation of Toll-Like Receptor 3 in 586 Hepatocytes. Gastroenterology. 154(8), pp.2237-2251.e16. 587 Hagen, W.J.H., Wan, W. and Briggs, J.A.G. 2017. Implementation of a cryo -588 electron tomography tilt-scheme optimized for high resolution subtomogram 589 averaging. Journal of Structural Biology. 197(2), pp.191–198. 590 Hughes, M., Griffin, S. and Harris, M. 2009. Domain III of NS5A contributes to 591 both RNA replication and assembly of hepatitis C virus particles. J Gen. Virol. 592 90(6), pp.1329–1334. 593 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 36 Kremer, J.R., Mastronarde, D.N. and McIntosh, J.R. 1996. Computer visualization 594 of three-dimensional image data using IMOD. Journal of Structural Biology. 595 116(1), pp.71–76. 596 Lam, V. and Villa, E. 2021. Practical Approaches for Cryo -FIB Milling and 597 Applications for Cellular Cryo -Electron Tomography BT - cryoEM: Methods 598 and Protocols In: T. Gonen and B. L. Nannenga, eds. New York, NY: Springer 599 US, pp.49–82. 600 Lee, J.Y., Cortese, M., Haselmann, U., Tabata, K., Romero -Brey, I., Funaya, C., 601 Schieber, N.L., Qiang, Y., Bartenschlager, M., Kallis, S., Ritter, C., Rohr, K., 602 Schwab, Y., Ruggieri, A. and Bartenschlager, R. 2019a. Spatiotemporal 603 Coupling of the Hepatitis C Virus Replication Cycle by Creating a Lipid 604 Droplet- Proximal Membranous Replication Compartment. Cell Reports . 605 27(12), pp.3602-3617.e5. 606 Lohmann, V., Körner, F., Koch, J. -O., Herian, U., Theilmann, L. and 607 Bartenschlager, R. 1999. Replication of Subgenomic Hepatitis C Virus RNAs 608 in a Hepatoma Cell Line. Science. 285(5424), pp.110–113. 609 Luft, J.H. 1961. Improvements in epoxy resin embedding methods. Journal of 610 Biophysical and Biochemical Cytology. 9(2), pp.409–414. 611 Martinello, M., Solomon, S.S., Terrault, N.A. and Dore, G.J. 2023. Hepatitis C. 612 The Lancet. 402(10407), pp.1085–1096. 613 Matsui, C., Yuliandari, P., Deng, L., Abe, T. and Shoji, I. 2021. The Role of 614 Chaperone-Mediated Autophagy in Hepatitis C Virus-Induced Pathogenesis. 615 Frontiers in Cellular and Infection Microbiology. 11, p.796664. 616 Mohl, B.-P., Bartlett, C., Mankouri, J. and Harris, M. 2016. Early events in the 617 generation of autophagosomes are required for the formation of membrane 618 structures involved in hepatitis C virus genome replication. J Gen. Virol. 619 97(3), pp.680–693. 620 Moradpour, D., Evans, M.J., Gosert, R., Yuan, Z., Blum, H.E., Goff, S.P., 621 Lindenbach, B.D. and Rice, C.M. 2004. Insertion of Green Fluorescent 622 Protein into Nonstructural Protein 5A Allows Direct Visualization of Functional 623 Hepatitis C Virus Replication Complexes. Journal of Virology . 78(14), 624 pp.7400–7409. 625 Moradpour, D., Gosert, R., Egger, D., Penin, F., Blum, H.E. and Bienz, K. 2003. 626 Membrane association of hepatitis C virus nonstructural proteins and 627 identification of the membrane alteration that harbors the viral replication 628 complex. Antiviral Research. 60(2), pp.103–109. 629 Paul, D., Madan, V. and Bartenschlager, R. 2014 . Hepatitis C virus RNA 630 replication and assembly: Living on the fat of the land. Cell Host and Microbe. 631 16(5), pp.569–579. 632 Pérez-Berná, A.J., Rodríguez, M.J., Chichón, F.J., Friesland, M.F., Sorrentino, A., 633 Carrascosa, J.L., Pereiro, E. and Gastaminza, P. 2016. Structural Changes 634 In Cells Imaged by Soft X -ray Cryo -Tomography During Hepatitis C Virus 635 Infection. ACS nano. 10(7), pp.6597–6611. 636 Ramakrishnaiah, V., Thumann, C., Fofana, I., Habersetzer, F., Pan, Q., de Ruiter, 637 P.E., Willemsen, R., Demmers, J.A.A., Stalin Raj, V., Jenster, G., 638 Kwekkeboom, J., Tilanus, H.W., Haagmans, B.L., Baumert, T.F. and van der 639 Laan, L.J.W. 2013. Exosome -mediated transmission of hepatitis C virus 640 between human hepatoma Huh7.5 cells. Proceedings of the National 641 Academy of Sciences of the United States of America . 110(32), pp.13109–642 13113. 643 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 37 Romero-Brey, I. and Bartenschlager , R. 2015. Viral infection at high 644 magnification: 3D electron microscopy methods to analyze the architecture 645 of infected cells. Viruses. 7(12), pp.6316–6345. 646 Romero-Brey, I., Merz, A., Chiramel, A., Lee, J.Y., Chlanda, P., Haselman, U., 647 Santarella-Mellwig, R., Habermann, A., Hoppe, S., Kallis, S., Walther, P., 648 Antony, C., Krijnse -Locker, J. and Bartenschlager, R. 2012. Three -649 Dimensional Architecture and Biogenesis of Membrane Structures 650 Associated with Hepatitis C Virus Replication. PLoS Pathogens. 8(12). 651 Ross-Thriepland, D., Mankouri, J. and Harris, M. 2015. Serine Phosphorylation of 652 the Hepatitis C Virus NS5A Protein Controls the Establishment of Replication 653 Complexes. Journal of Virology. 89(6), pp.3123–3135. 654 Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, 655 T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., Tinevez, J.-Y., White, 656 D.J., Hartenstein, V., Eliceiri, K., Tomancak, P. and Cardona, A. 2012. Fiji: 657 an open-source platform for biological-image analysis. Nature Methods. 9(7), 658 pp.676–682. 659 Schorb, M. and Briggs, J.A.G. 2014. Correlated cryo -fluorescence and cryo -660 electron microscopy with high spatial precision and improved sensitivity. 661 Ultramicroscopy. 143, pp.24–32. 662 Shirota, Y., Luo, H., Qin, W., Kaneko, S., Yamashita, T., Kobayashi, K. and 663 Murakami, S. 2002. Hepatitis C virus (HCV) NS5A binds RNA -dependent 664 RNA polymerase (RdRP) NS5B and modulates RNA -dependent RNA 665 polymerase activity. J Biol. Chem. 277(13), pp.11149–11155. 666 Studer, D., Klein, A., Iacovache, I., Gnaegi, H. and Zuber, B. 2014. A new tool 667 based on two micromanipulators facilitates the handling of ultrathin 668 cryosection ribbons. Journal of Structural Biology. 185(1), pp.125–128. 669 Wagner, F.R., Watanabe, R., Schampers, R., Singh, D., Persoon, H., Schaffer, 670 M., Fruhstorfer, P., Plitzko, J. and Villa, E. 2020. Preparing samples from 671 whole cells using focused -ion-beam milling for cryo -electron tomography. 672 Nature Protocols. 15(6), pp.2041–2070. 673 Wakita, T., Pietschmann, T., Kato, T., Date, T., Miyamoto, M., Zhao, Z., Murthy, 674 K., Habermann, A., Kräusslich, H.-G., Mizokami, M., Bartenschlager, R. and 675 Liang, T.J. 2005. Production of infectious hepatitis C virus in tissue culture 676 from a cloned viral genome. Nature Medicine. 11(7), pp.791–796. 677 WHO 2024. Hepatitis C. Available from: https://www.who.int/news -room/fact-678 sheets/detail/hepatitis-c. 679 Wolff, G., Melia, C.E., Snijder, E.J. and Bárcena, M. 2020. Double -Membrane 680 Vesicles as Platforms for Viral Replication. Trends in Microbiology . 28(12), 681 pp.1022–1033. 682 Wyles, D.L., Kaihara, K.A., Korba, B.E., Schooley, R.T., Beadle, J.R. and 683 Hostetler, K.Y. 2009. The octadecyloxyethyl ester of (S) -9-[3-hydroxy-2-684 (phosphonomethoxy) propyl] adenine is a potent and selective inhibitor of 685 hepatitis C virus replication in genotype 1A, 1B, and 2A replicons. 686 Antimicrobial Agents and Chemotherapy. 53(6), pp.2660–2662. 687 Yin, Y., Zhao, Y., Chen, Q., Chen, Y. and Mao, L. 2022. Dual roles and potential 688 applications of exosomes in HCV infections. Front. Microbiol. 13, p.1044832. 689 Zheng, S., Wolff, G., Greenan, G., Chen, Z., Faas, F.G.A., Bárcena, M., Koster, 690 A.J., Cheng, Y. and Agard, D.A. 2022. AreTomo: An integrated software 691 package for automated marker -free, motion -corrected cryo -electron 692 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint HCV replication imaged in close-to-native conditions 38 tomographic alignment and reconstruction. Journal of Structural Biology: X . 693 6, p.100068. 694 Zheng, S.Q., Palovcak, E., Armache, J. -P., Verba, K.A., Cheng, Y. and Agard, 695 D.A. 2017. MotionCor2: anisotropic correction of beam -induced motion for 696 improved cryo-electron microscopy. Nature Methods. 14(4), pp.331–332. 697 698 699 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 24, 2025. ; https://doi.org/10.1101/2025.04.24.650446doi: bioRxiv preprint