Tetherin antagonism by SARS-CoV-2 enhances virus release: multiple mechanisms including ORF3a-mediated defective retrograde traffic

The antiviral restriction factor, tetherin, blocks the release of several different families of enveloped viruses, including the Coronaviridae. Tetherin is an interferon-induced protein that forms parallel homodimers between the host cell and viral particles, linking viruses to the surface of infected cells and inhibiting their release. We demonstrated that SARS-CoV-2 infection causes tetherin downregulation, and that tetherin depletion from cells enhances SARS-CoV-2 viral titres. We investigated the potential viral proteins involved in abrogating tetherin function and found that SARS-CoV-2 ORF3a reduces tetherin localisation within biosynthetic organelles via reduced retrograde recycling and increases tetherin localisation to late endocytic organelles. By removing tetherin from the Coronavirus budding compartments, ORF3a enhances virus release. We also found expression of Spike protein caused a reduction in cellular tetherin levels. Our results confirm that tetherin acts as a host restriction factor for SARS-CoV-2 and highlight the multiple distinct mechanisms by which SARS-CoV-2 subverts tetherin function.

severe acute respiratory syndrome. 48 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 Introduction 71 The causative agent of coronavirus disease 2019  is the enveloped 72 coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [1,2]. angiotensin-converting enzyme-2 (ACE2) [3]. Unlike that of 2002-2003 (hereafter named SARS-CoV-1), the SARS-CoV-2 spike protein contains a polybasic 88 furin cleavage site which facilitates the cleavage of the spike into two proteins, S1 89 and S2 that remain non-covalently associated [4,5]. The S2 fragment is further 90 primed by the serine protease TMPRSS2 [3], whilst the S1 fragment binds 91 Neuropilin-1 [6,7], facilitating virus entry and infection. Coronaviruses enter 92 TMPRSS2-positive cells by direct fusion at the plasma membrane, and are 93 endocytosed by TMPRSS2-negative cells [8], following which their envelope fuses 94 within late endosomes/lysosomes [9], liberating the viral nucleocapsid to the cytosol 95 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 6 downregulates tetherin differ, they all reduce tetherin abundance from the organelle 121 in which respective viruses bud, and/or reduce tetherin levels upon the plasma 122 membrane. 123 124 Of the previously described coronaviruses, HCoV-229E and SARS-CoV-1 have been 125 shown to undergo viral restriction by tetherin [17,18]. Two SARS-CoV-1 proteins 126 have been shown to antagonise tetherin resulting in a concomitant increase in virion 127 spreadthe ORF7a protein and spike glycoprotein [18,19]. However, several 128 questions remain about the mechanisms surrounding tetherin antagonism by 129 coronaviruses. It is also unclear exactly how and where tetherin forms such tethers, 130 as in coronaviruses, unlike other viruses that undergo tetherin-dependent restriction 131 (such as HIV-1), membrane envelopment occurs in the biosynthetic pathway and not 132 at the plasma membrane. The mechanism by which sarbecoviruses dysregulate 133 tetherin remain unclear. 134

135
Here, we show that tetherin is directly responsible for tethering of nascent enveloped 136 SARS-CoV-2 virions to infected cell surfaces. We demonstrate using primary cells 137 and immortalised cell lines that SARS-CoV-2 infection causes a dramatic 138 downregulation of tetherin, and that loss of tetherin aids SARS-CoV-2 viral spread 139 and infection. We examine the effect of expression of individual SARS-CoV-2 140 proteins on tetherin antagonism and find that ORF3a redirects tetherin away from the 141 biosynthetic pathway and to the endolysosomal pathway by inhibiting retrograde 142 retrieval. The reduction of tetherin within the biosynthetic pathway limits its 143 incorportation to forming virions, with subsequent enhancement in virus release. We 144 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 also demonstrate that Spike expression causes tetherin downregulation, as has 145 previously been described for SARS-CoV-1.  was confirmed by Western blotting (Figure 1B). Using a clinical isolate of SARS- To examine whether SARS-CoV-2 virions were tethered to the cell surface, we 188 performed transmission electron microscopy. In infected HeLa+ACE2 cells, SARS-189 CoV-2 virions could be found clustered on the plasma membrane of cells, although 190 tethered virions were frequently polarised to discrete areas, rather than distributed 191 evenly along the plasma membrane ( Figure 1E, Supplemental Figure 1C).  containing tubulovesicular organelles were often polarised towards sites of 193 significant surface-associated virus. 194 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in Electron microscopy also verified the presence of double membrane vesicles 196 (DMVs) in infected cells, and typical Golgi cisternae were not present in infected 197 cells ( Figure 1E). To check whether the formation of DMVs and aberration to the 198 biosynthetic machinery was causing a global downregulation of surface proteins, we 199 stained infected cells for the surface protein beta2microglobulin (Supplemental 200 Figure 1D), but no obvious loss in plasma membrane staining was observed in 201 infected cells. Surface labelling immunogold electron microscopy (see Methods) 202 revealed tetherin molecules to be found between SARS-CoV-2 virions ( Figure 1F) 203 and virions were verified as being SARS-CoV-2 using an anti-SARS-CoV-2 spike 204 antibody ( Figure 1G). 205

206
The human alveolar epithelial cell line, A549, expresses low levels of ACE2 207 endogenously, although ectopic overexpression of ACE2 has been used to facilitate 208 betacoronavirus entry [23,24]. A549 cells stably expressing ACE2, designated 209 A549+ACE2, were generated by lentiviral transduction (Figure 2A) and these cells 210 were amenable to SARS-CoV-2 infection ( Figure 2B, uninfected cells shown with 211 asterisk). A549 cells do not express tetherin at steady state, although its expression 212 can be induced through stimulation with IFN alpha (IFNα) [13,25] (Supplemental 213 Figure 2A), following which tetherin is localised to the plasma membrane and to 214 intracellular compartments. Immunofluorescence analysis of A549+ACE2 cells 215 infected with SARS-CoV-2 revealed a dramatic loss of tetherin as revealed by the 216 loss of tetherin in SARS-CoV-2 infected cells ( Figure 2C, uninfected cells shown 217 with asterisk). To examine what effect this near-total loss of tetherin had on virus 218 tethering, we again performed electron microscopy. SARS-CoV-2 infected, IFN 219 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 treated A549+ACE2 cells were characterised by significant intracellular remodelling, 220 but very few surface-associated virions were present, likely due to the significant 221 tetherin downregulation ( Figure 2D). Virion-containing DMVs were frequently 222 observed in the perinuclear region of infected cells, and these were associated with 223 dramatic membrane remodelling, including a loss of typical Golgi cisternae from cells 224 (Figures 2E, 2F).  Figure 2B). The differences in the amount of tetherin 234 downregulation between these cell lines may reflect either differences in kinetics of 235 infection, differences in resting levels of tetherin, or differences in host machinery 236 involved in the process of downregulation. These data are consistent with previous 237 observations suggesting that other coronaviruses downregulate tetherin [17][18][19]. 238 239

Tetherin loss aids SARS-CoV-2 viral spread 240
To determine whether tetherin plays a functional role in viral tethering, we performed 241 both one-step (MOI of 5) and multi-step (MOI of 1) growth curves, to measure both 242 released and intracellular virus production from infected WT HeLa+ACE2 and 243 Bst2KO HeLa+ACE2 ( Figure 3A). HeLa cells were used due to their high, 244 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in were able to be infected with SARS-CoV-2 ( Figure 3B). WT HeLa+ACE2 and 247 Bst2KO HeLa+ACE2 cells were infected at the respective MOI and released and 248 intracellular virus was harvested at the indicated time points (Figure 3C, 3D). steps, as minimal reinfection should occur. However, this approach results in the 255 vast majority of cells being infected with >1 virus particle. In addition to being less 256 physiologically relevant, this higher viral load may be able to overcome host 257 restriction factors and mask a phenotype. In this case, the released virus titre was 258 clearly higher in the Bst2KO HeLa+ACE2 cells (24, 48 and 72 hpi), evidencing the 259 tetherin-mediated restriction in the WT HeLa+ACE2 cells. However, intracellular 260 virions appeared to accumulate quickly in the Bst2KO HeLa+ACE2 cells at the early 261 time point of 24 hours; whether this is due to a lack of tetherin indirectly allowing 262 enhanced RNA replication or is a by-product of this model system remains to be 263 seen. It is obvious, however, that when the released virions are viewed as a 264 proportion of total infectious particles, the Bst2KO HeLa+ACE2 cells release 265 significantly more than WT HeLa+ACE2 cells at 24 and 48 hpi, due to their inability 266 to tether nascent virions. 267 268 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 By contrast, the multi-step growth curve utilising an MOI of 1 ( Figure 3D) has the 269 advantage of being more physiologically relevant and providing a more realistic 270 stoichiometric ratio of viral to host proteins, making it less likely to mask naturally 271 occurring interactions and their resulting phenotypes. However as approximately 272 37% of the cells will not be infected by the initial inoculum, infection of naïve cells will 273 continue to occur throughout the time course and the viral replication events cannot 274 be presumed to be aligned. In this case, the disparity between Bst2KO HeLa+ACE2 Together, these data demonstrate that tetherin acts to limit SARS-CoV-2 infection 278 and that SARS-CoV-2 acts to downregulate tetherin. These data support the notion 279 that tetherin exerts a broad restriction against numerous enveloped viruses, 280 regardless of whether budding occurs at the plasma membrane or within intracellular 281 compartments. As previously studied coronaviruses, including HCoV-229E and 282 CoV-2 [28], although the mechanism of antagonism is not well understood. SARS-290

SARS-
CoV-2 ORF7a has acquired a number of mutations which exist in and around the 291 transmembrane domain which plays a role in SARS-CoV-1 ORF7a localisation. We 292 found that SARS-CoV-1 ORF7a colocalises with the trans-Golgi marker TGN46, but 293 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in To determine if ORF7a expression affected tetherin localisation, abundance, 301 glycosylation or ability to form dimers, HeLa cells stably expressing SARS-CoV-1 302 ORF7a-FLAG or SARS-CoV2-ORF7a-FLAG were analysed by immunofluorescence, 303 Western blotting and flow cytometry. By immunofluorescence ORF7a-FLAG and 304 tetherin were not found to significantly overlap, but we did observe that the two 305 proteins were both localised to some degree in adjacent perinuclear organelles 306 ( Figure 4A). Antagonism of tetherin by ORF7a could occur by interfering with 307 tetherin's ability to form homodimers. SARS-CoV-1 ORF7a was previously shown to 308 alter glycosylation of ectopic tetherin without altering protein abundance, and the 309 glycosylation-defective tetherin showed impaired ability to restrict SARS-CoV-1 310 egress [18]. To determine whether SARS-CoV-2 ORF7a impaired endogenous 311 glycosylation or affected tetherin dimer formation, we performed Western blotting of 312 WT HeLa, Bst2KO HeLa, KO + C3A-tetherin, and WT HeLa cells stably expressing 313 either SARS-CoV-1 ORF7a-FLAG or SARS-CoV-2-ORF7a-FLAG ( Figure 4B). C3A-314 tetherin stable cells (expressed in a tetherin KO HeLa background), are unable to 315 form tetherin dimers and this can be observed by Western blotting. The stable 316 expression of either SARS-CoV-1 ORF7a or SARS-CoV-2 ORF7a did not impact 317 upon tetherin abundance, glycosylation, or the ability of tetherin to form dimers 318 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ; https://doi.org/10.1101/2021.01.06.425396 doi: bioRxiv preprint ( Figure 4B). Finally, flow cytometry determined that expression of SARS-CoV-1 or 319 SARS-CoV-2 ORF7a did not impact upon cell surface levels of tetherin ( Figure 4C). 320 Overall, these data show that ORF7a does not directly influence tetherin localisation, 321 abundance, glycosylation or dimer formation. 322 323

Tetherin is downregulated by SARS-CoV-2 Spike 324
SARS-CoV-1 Spike causes tetherin downregulation via lysosomal degradation [19]. 325 To interrogate the impact of SARS-CoV-2 Spike on tetherin, we generated an 326 epitope tagged Spike construct to aid our imaging studies. We placed a HA epitope cells, yet only minor downregulation by Spike, we performed a miniscreen of SARS-364 CoV-2 ORFs to identify additional SARS-CoV-2 proteins involved in tetherin 365 downregulation or antagonism. We transiently transfected HeLa cells with: ORF3a-366 CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in ORF9c, ORF10-Strep and confirmed their expression by intracellular flow cytometry 368 using an anti-Strep antibody (Supplemental Figure 5A). Surface tetherin levels 369 were analysed by flow cytometry and no significant tetherin downregulation was 370 observed upon expression of any ORF ( Figure 5A). The intracellular tetherin 371 localisation was analysed by confocal microscopy (Supplemental Figure 5B) and 372 we noticed a redistribution of tetherin towards punctate organelles upon expression 373 of ORF3a ( Figure 5B). 374 375 SARS-CoV-2 ORF3a is a viroporin that localises to and damages endosomes and 376 lysosomes [29]. As such, the observed increased presence of tetherin puncta 377 following ORF3a expression may be due to decreased lysosomal degradation. 378 Western blotting showed that expression of OFR3a had no impact on total levels of 379 endogenous tetherin protein ( Figure 5C). Flow cytometry confirmed minimal 380 differences in tetherin levels upon transient expression of ORF3a (Supplemental 381 Figure 6A). We noticed that tetherin localisation appeared altered upon ORF3a 382 expression and observed a decrease in perinuclear tetherin that appeared in and 383 around the Golgi (Figure 5D), and that tetherin localised to ORF3a-Strep positive 384 punctate organelles. We quantified the levels of colocalisation between tetherin and 385 the trans-Golgi marker, TGN46, and found a significant loss of tetherin from this 386 region upon expression of ORF3a ( Figure 5E). The loss of tetherin from the peri-387 Golgi area in ORF3a transfected cells was associated with an increase in tetherin 388 within LAMP1 positive punctate organelles (Figure 5F, 5G). Expression of ORF3a 389 also disrupted the distribution of numerous endosome-related markers including 390 CIMPR, VPS35, CD63, (Supplemental Figure 6B), and the mixing of early and late 391 endosomal markers EEA1 and cathepsin D (Supplemental Figure 6C). By TEM, 392 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in ORF3a expressing cells contained enlarged endolysosomes, consistent with defects 393 in endolysosomal homeostasis (Supplemental Figure 6D). 394

395
The ORF3a-mediated increase in tetherin abundance within lysosomes could be due 396 to defective lysosomal degradation but could also be due to altered intracellular 397 trafficking. In addition to the plasma membrane, tetherin is found in several  CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in and E and yields were significantly reduced upon addition of tetherin ( Figure 6A). 426 The amount of VLPs were greatly increased by the expression of ORF3a alone, 427 indicating that ORF3a may enhance VLP release independently of its antagonism of 428 tetherin. The co-expression of structural proteins, tetherin and ORF3a resulted in a 429 lower yield of VLP when compared with structural proteins and ORF3a only. 430

431
We reproducibly found that the expression of tetherin reduced cellular levels of 432 structural proteins (Figures 4G, 6A), and we hypothesised that tetherin expression 433 could promote endocytosis and subsequent degradation of VLPs. Tetherin restriction 434 of enveloped viruses prevents their egress, but also promotes internalisation of 435 tethered virions which are redirected to lysosomal degradation. To test this, we 436 expressed either wildtype tetherin or C3A-tetherin which is unable to form 437 homodimers and therefore unable to promote viral retention and reinternalisation. 438 Both WT and C3A-tetherin gave similar reductions in cellular levels of structural 439 proteins (Figure 6B), indicating that reduced structural protein levels were 440 independent of tethering, reinternalization and subsequent degradation. As 441 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in for SARS-CoV-2 restriction ( Figure 6C). The opposing relocalisation of tetherin by SARS-CoV-2away from the biosynthetic 458 organellessimilarly reflects the need of SARS-CoV-2 to remove tetherin from its 459 budding compartment. 460

461
Our data support that tetherin molecules become incorporated to SARS-CoV-2 462 virions and act to restrict virus release upon delivery of virions to the plasma 463 membrane. SARS-CoV-2 counteracts this restriction in a variety of ways, ensuring its 464 continued transmission. We have also shown that ORF3a antagonises tetherin by 465 altering its steady state distribution, and that VLP yields are enhanced upon ORF3a 466 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ; https://doi.org/10.1101/2021.01.06.425396 doi: bioRxiv preprint expression ( Figure 6A). We also demonstrate that Spike downregulates tetherin, 467 consistent with reports for SARS-CoV-1 [19], but we found this downregulation to be 468 insufficient to promote VLP release ( Figure 4G). 469

470
Tetherin is an IFN-stimulated gene (ISG) [13], and many cell types express low CoV-ORF3a impaired retrograde traffic, reducing the amount of tetherin retrieved to 485 biosynthetic organelles, and permitting VLP release (Figures 5D-I, 6A) CoV-2 lysosomal egress [35]. Secretion of lysosomal hydrolases has been observed 490 upon ORF3a expression [31] and whilst this may in-part be due to enhanced 491 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ; https://doi.org/10.1101/2021.01.06.425396 doi: bioRxiv preprint lysosome-plasma membrane fusion, our finding that ORF3a impairs CIMPR retrieval 492 may also be a contributing factor (Supplemental Figures 6B, 6F, 6G). 493 494 Precisely how ORF3a impedes retrograde traffic is unclear. Recruitment of VPS35, a 495 core retromer component, to membranes does not appear impeded by ORF3a 496 expression (Supplemental Figure 6B), indicating that tubule formation or scission 497 may be impaired. Endosomal acidification has been proposed to regulate tubule 498 formation, as a homeostatic mechanism to ensure tubulation doesn't occur too early 499 in the endocytic pathway or from incorrect organelles [37]. 500

501
The ORF3a-mediated defective retrograde trafficking is not specific to tetherin, and 502 likely affects numerous other protein cargos, some of which may aid the infectivity 503 and transmissibility of SARS-CoV-2. Our findings highlight defective retrograde traffic 504 as a novel mechanism of tetherin antagonism. Whether other enveloped viruses 505 which bud within the biosynthetic pathway employ similar strategies remains to be 506 investigated. 507 508 In addition to our identification of the role of ORF3a in tetherin antagonism, we also 509 investigated the roles of two previously identified tetherin antagonists -ORF7a and 510 Spike. We did not observe any difference in total tetherin levels, tetherin 511 glycosylation, ability to form dimers, or surface tetherin upon expression of SARS-512 CoV-2 ORF7a (Figures 4A, 4B, 4C). The intracellular localisation of tetherin also 513 appeared unaffected by ORF7a expression. However, others have demonstrated a 514 role for ORF7a in sarbecovirus infection and both SARS-CoV-1 and SARS-CoV-2 515 virus lacking ORF7a show impairments in virus replication in the presence of tetherin 516 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in  [18,38]. A direct interaction between SARS-CoV-1 ORF7a and SARS-CoV-2 ORF7a 517 and tetherin have been described [18,38], although the precise mechanism(s) by 518 which ORF7a antagonises tetherin remains enigmatic, as does its role in 519 pathogenesis. A number of SARS-CoV-2 variants have been described which Goat anti-Mouse IgG Alexa488/555 and Goat anti-rabbit IgG Alexa 488/555 566 (ThermoFisher) secondary antibodies were used for confocal microscopy. 567 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 Goat IRDye 680 anti-mouse, anti-rabbit, anti-goat, anti-rat and Goat IRDye 800 anti-568 mouse, anti-rabbit antibodies (Li-Cor) were used for Western blotting. 569 570 Cloning 571 pcDNA6B SARS-CoV-1 and SARS-CoV-2 ORF7a-FLAG constructs were a gift from 572 Professor Peihui Wang (Shandong University, China). To generate stable cell lines, 573 ORF7a-FLAG cDNA fragments were subcloned into pQCXIH retroviral vectors. 574 ss-HA-Spike was generated by cloning an HA epitope plus a Serine-Glycine linker 575 between residues S13 and Q14 of SARS-CoV-2 spike. Following translocation to the 576 ER lumen, cleavage of the signal sequence will render the HA tag at the N terminus 577 with PneumaCult-ALI medium in the basal chamber and the apical surface exposed, 614 giving an air liquid interface to stimulate cilia biogenesis. 615 616 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

SARS-CoV-2 infections of HAE cells 617
SARS-CoV-2 isolate hCoV-19/England/204501206/2020 (EPI_ISL_660791) was 618 isolated from swabs as described in [48]. The isolate was passaged twice in Vero 619 cells before being used to infect HAE cells. To remove the mucus layer from the 620 apical surface of HAE cells prior to infection, 200 μL of DMEM was added to the HAE 621 cells at 37 °C, 5% CO2 for 10 minutes. The cells were infected at a MOI of 0.01 with 622 inocula added to the apical chamber and incubated for 1 hour at 37 °C, 5% CO2 623 before removal of the inoculum and incubating for a further 48 hours. CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ; https://doi.org/10.1101/2021.01.06.425396 doi: bioRxiv preprint ACE2 stable HeLa and A549 cell lines were generated using the lentiviral pLVX-641 ACE2-Blasticidin construct from Dr Yohei Yamauchi (University of Bristol). Following 642 transduction, cells were selected with 10 g/mL blasticidin for 18 days. 643 Expression-inducible stable cell lines were generated using the pLVX-TetOne 644 system. pLVX-TetOne-Puro-ORF7a-2xStrep and pLVX-TetOne-Puro-2xStrep-645 ORF9c were a gift from Dr David Gordon (UCSF, USA). pLVX-TetOne-Puro-ss-HA-646 spike was generated as described above. Following transduction, cells underwent 647 antibiotic selected with 1 g/mL puromycin for 5 days. 648 HEK293T cells were transfected with lentiviral vectors (pLVX / pLVX-TetOne) plus 649 packaging plasmids pCMVR8.91 and pMD.VSVG using . were selected using 400 μg/mL Hygromycin B for 10 days. 663

SARS-CoV-2 infections of immortalised cell lines 665
. CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in isolate BetaCoV/ Australia/VIC01/2020 [22], which had been passaged once on Vero 667 cells following receipt from Public Health England. All cells were washed with PBS 668 before being infected with a single virus stock, diluted to the desired MOI with sera-669 free DMEM (supplemented with 25mM HEPES, penicillin (100 U/mL), streptomycin 670 (100 g/mL), 2 mM L-glutamine, 1 % non-essential amino acids). After one hour, the 671 inoculum was removed, and cells washed again with PBS. Infected cells were 672 maintained in DMEM, supplemented with the above-described additions plus 2% 673

FCS (virus growth media). 674
For immunofluorescence, cells were plated to glass-bottomed 24-well plates 675 (E0030741021, Eppendorf) and infected at an MOI of 0.5 and incubated for 24 676 hours, after which plates were submerged in 4% PFA/PBS for 20 min. 677 For conventional electron microscopy, cells were plated to plastic Thermanox (Nunc) 678 coverslips in 24-well plates and infected at an MOI of 0.5 and incubated for 24 hours, 679 after which plates were submerged in 2% PFA / 2.5% glutaraldehyde / 0.1M 680 cacodylate buffer for 20 minutes. 681 For surface labelling immunoEM, cells were plated to Thermanox coverslips, infected 682 at an MOI of 0.5 and incubated for 24 hours, and fixed with 4% PFA / 0.1M 683 cacodylate buffer for 20 minutes. 684 685

Conventional electron microscopy 686
Cells were fixed (described above) before being washed with 0.1M cacodylate 687 buffer. Cells were stained using 1% osmium tetroxide + 1.5% potassium ferrocyanide 688 for 1 hour before staining was enhanced with 1% tannic acid / 0.1M cacodylate 689 buffer for 45 minutes. Cells were washed, dehydrated and infiltrated with Epoxy 690 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 Immunofluorescence of cryostat sections 716 HAE cells were fixed in 4% PFA / PBS for 1 hour before embedding and freezing in 717 OCT (optimal cutting temperature) compound. ~15 μm thick sections were cut using 718 a cryostat and these permeabilised using 0.2% saponin in PBS for 20 minutes at 719 room temperature before incubating in blocking solution containing 0.02%, 1% BSA 720 in PBS for 30 minutes at room temperature. The cryostat sections were incubated 721 with primary antibodies in blocking solution for 2 hours at room temperature. 722 Subsequently, Alexa Fluor bound secondary antibodies and phalloidin 647 (Abcam, 723 ab176759) were applied to the sections for 1 hour at room temperature. Images 724 were acquired using a Leica SP8 confocal. Cells were seeded to glass coverslips in 24-well plates a day before uptake 736 experiments were performed. Antibody was diluted in pre-chilled 1% BSA / PBS (to 5 737 μg/mL) and 50μL drops pipetted to parafilm-covered blocks on ice. Coverslips were 738 removed from 24-well plates are placed cell-side down to diluted antibody drops on 739 ice. Cells were incubated on ice for 30 mins to allow antibodies to bind to proteins at 740 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021 the cell surface. Coverslips were then removed and placed back in 24-well plates, 741 and transferred back to a 37 o C, 5% CO2 incubator to allow antibody-antigen 742 complexes to be endocytosed and trafficked for 2 hours. Coverslips were then fixed 743 and processed for immunofluorescence microscopy using secondary antibodies 744 specific for internalised antibodies to detect antibody localisation. 745 746

Western blotting 747
Tetherin blots were performed using Laemmli sample buffer and run in non-reducing 748 conditions as previously described [25]. For all other blots, lysates were mixed with 749 4x NuPage LDS sample buffer (ThermoFisher). Gels were loaded to NuPage 4-12% 750 Bis-Tris precast gels (ThermoFisher) and transferred to PVDF membranes before 751 being blocked using 5% milk / PBS / 0.1% Tween. Primary antibodies and secondary 752 antibodies were diluted in PBS-tween. Blots were imaged using an Odyseey CLx . three freeze-thaw cycles to release intracellular virions. Following clarification, the 765 . CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in  CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

940
. CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

965
. CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

990
. CC-BY 4.0 International license perpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for this this version posted December 22, 2022. ;https://doi.org/10.1101https://doi.org/10. /2021