SARS-CoV-2 ORF8 limits expression levels of Spike antigen

Survival from COVID-19 depends on the ability of the host to effectively neutralize virions and infected cells, a process largely driven by antibody-mediated immunity. However, with the newly emerging variants that evade Spike-targeting antibodies, re-infections and breakthrough infections are increasingly common. A full characterization of SARS-CoV-2 mechanisms counteracting antibody-mediated immunity is needed. Here, we report that ORF8 is a SARS-CoV-2 factor that controls cellular Spike antigen levels. ORF8 limits the availability of mature Spike by inhibiting host protein synthesis and retaining Spike at the endoplasmic reticulum, reducing cell-surface Spike levels and recognition by anti-SARS-CoV-2 antibodies. With limited Spike availability, ORF8 restricts Spike incorporation during viral assembly, reducing Spike levels in virions. Cell entry of these virions leaves fewer Spike molecules at the cell surface, limiting antibody recognition of infected cells. Our studies propose an ORF8-dependent SARS-CoV-2 strategy that allows immune evasion of infected cells for extended viral production.


97
ORF8 interacts with an array of ER chaperone proteins [7], suggesting that ORF8 is 98 subcellularly localized to the ER. Computational analysis (Protter) of the amino acid 99 sequence of ORF8 predicts that the first 16 N-terminal amino acids are an ER signal 100 peptide (Fig 1A), suggesting that, upon de novo synthesis, ORF8 is translocated into protein PDI (red signals) (Fig 1B), as manifested by a high degree of similarity between 106 the two signal intensities (Fig 1C) along the cross-sectional arrow ( Fig 1B). The 107 possibility that ORF8 is an ER protein was further evaluated by biochemical studies.

108
HEK293T cells transfected with a plasmid encoding C-terminal Flag-tagged ORF8 109 (ORF8-Flag) were subcellularly fractionated by differential centrifugation, yielding major 110 cellular compartment fractions (e.g., ER, mitochondria, and cytosol) (Fig 1D), and those 111 fractions were evaluated by immunoblot analyses for ORF8-Flag signals. The ORF8 112 signal was observed only in the ER fractions (characterized by Calnexin), but not in 113 mitochondria (COX4) or cytosol (β-actin), indicating that ORF8 is predominantly 114 localized to ER within cells. 115 Lacking a transmembrane domain (Fig 1A), we predicted that ORF8 is a luminal 116 protein after translocating to the ER. To test the prediction, the ORF8-containing ER Three SARS-CoV-2 proteins (i.e., Spike, ORF7a, ORF8) contain an ER signal peptide, 129 and Spike is a key viral component highly implicated in the viral infectivity. With ORF8 130 and Spike existing in the same subcellular space of the ER, as manifested by the 131 colocalization of ORF8 and Spike signals (Fig 2A), we investigated the possibility that 132 ORF8 alters Spike levels. HEK293T cells co-transfected with plasmids encoding C-133 terminal Flag-tagged Spike (Spike-Flag), and ORF8-Strep or eGFP-Strep (negative 134 control) were lysed for immunoblot analyses (Fig 2B and 2C) with an antibody targeting 135 the Spike S2 or S1 region ( Fig 2D). Two immunoblot bands were detected for Spike (Fig  136   2B), corresponding to uncleaved nascent Spike (220 kDa), and the Spike that is cleaved 137 (90 kDa in αS2 blot, and 130 kDa in αS1 blot) at the furin-cleavage site, a reaction 138 thought to occur at the ER-Golgi intermediate complex (ERGIC) or Golgi [8], resulting in 139 S1 and S2 fragments (Fig 2B and 2D). The total Spike levels (calculated by combining 140 uncleaved and S2 signals) decreased (> 50%) in an ORF8-dependent manner ( Fig 2C). 141 Moreover, the band intensities corresponding to S2 or S1 fragments decreased to a 142 greater extent (> 95% decrease) in an ORF8-dependent manner (Fig 2C). No other 143 immunoblot bands were detected under our experimental conditions (Fig S1A), 144 validating our quantitative measurement of Spike protein levels. The ORF8-dependent 145 modification of Spike protein levels was reproduced using non-tagged Spike and ORF8 146 (Fig S1B), validating the use of the C-terminal tagged constructs for our investigation. their own SARS-CoV Spike (Fig D). These findings suggest that the ORF8 modulation 157 of cellular Spike levels is a SARS-CoV-2-specific mechanism.
158 159 Next, we investigated whether ORF8 and Spike in the ER interact by creating an ORF8-

162
Flag construct with an I9P mutation (ORF-Flag I9P) that disrupts the α-helix structure of 163 the ER signal peptide by introducing a proline kink. Loss of ability to translocate to the 164 ER was validated by immunofluorescence microscopy analysis, as manifested by the 165 cytosolic distribution of ORF8-Flag I9P or ORF8-Flag lacking the entire ER signal 166 peptide (ORF8-Flag Δ1-17) (green signals) (Fig S2A), as well as the loss of ER (red 167 signals) colocalization with ORF8-Flag I9P or ORF8-Flag Δ1-17 ( Fig S2A). Furthermore, 168 immunoblot analysis under non-reducing conditions (to preserve disulfide bonds) 169 showed the non-mutated ORF8-Flag as multiple bands (Fig S2B), which is attributed to 170 intermolecular disulfide bonds that form within the oxidizing ER lumen environment [11], 171 whereas ORF8-Flag I9P and ORF8-Flag Δ1-17 were observed as a single band.

172
To evaluate the importance of the ORF8 localization to the ER for its effect on Spike 173 protein levels, cells were co-transfected with a plasmid encoding non-tagged Spike, and 174 a plasmid encoding GFP-Flag (negative control), ORF8-Flag, or ORF-Flag I9P. The 175 cells were lysed, the lysates were incubated with anti-Flag magnetic beads, and the 176 immunoprecipitated proteins were analyzed by western blotting. We observed a loss of 177 cleaved S2 fragment of Spike in cells co-expressing ORF8-Flag (Fig 3A, input), but not 178 in cells co-expressing GFP-Flag control or ORF8-Flag I9P. Moreover, Spike was 179 detected in the immunoprecipitated samples collected from cells co-expressing ORF8-

180
Flag, but nor GFP-Flag control or ORF8-Flag I9P (Fig 3A), indicating that Spike co-181 immunoprecipitated with ORF8-Flag but not with ORF-Flag I9P. These studies support the model that ORF8 interacts with Spike at the ER, and that ORF8 translocation to the 183 ER is required for the ORF8-Spike interaction and for altering Spike protein levels.

184
More cleaved Spike-Flag was lost (> 95%) than total Spike (> 50%) (Fig 2B and 2C), 185 and the Spike cleavage rate was lower (Fig 2B and 2C) in cells co-expressing ORF8- 186 Strep, suggesting that furin cleavage of Spike is inhibited by ORF8. The furin-dependent 187 Spike cleavage ( Fig 2D) is a post-ER event that occurs at the ERGIC or Golgi. Thus, we 188 hypothesized that ORF8 interaction with Spike at the ER inhibits Spike translocation to 189 Golgi, preventing furin-cleavage. In support of this model, the Spike species that 190 interacts with ORF8 is uncleaved (Fig 3A). To further investigate whether Spike  To directly test whether the ORF8-Spike interaction is mostly associated with 223 disulfide bonds, cells co-transfected with plasmids encoding Spike or ORF8-Flag were 224 lysed and the cell lysates were pre-incubated at 95°C for 5 min in 2% SDS (to break up 225 non-covalent protein-protein interactions) and in the absence or presence of 0.2% β-ME 226 (to break up intra-and inter-molecular disulfide bonds). After the pre-incubation, the 227 lysates were immunoprecipitated with anti-Flag magnetic beads and analyzed by immunoblotting under non-reducing or reducing conditions (Fig 3F and 3G). We 229 observed co-immunoprecipitation of Spike even after the denaturation (lane # 9), at a 230 level that is not significantly different from the same lysates that were not pre-incubated

239
Host protein synthesis is inhibited within cells expressing ORF8 240 We further tested this proposed mechanism with decanoyl-RVKR-CMK (or simply 241 CMK), a furin inhibitor ( Fig 4A). However, the decrease in total Spike-Flag levels   significantly decreased (> 90%) by ORF8 co-expression ( Fig 5E). These studies 302 demonstrated that levels of N-terminal S1 fragment of cell-surface Spike also decreases 303 in an ORF8-dependent manner.

304
Lastly, the SARS-CoV Spike, which was expressed to a similar level as SARS-CoV-2 305 Spike ( Fig 5F), was detected in much lower levels at the cell surface (normalized by the 306 total Spike levels) (Fig 5G), and no significant reduction of cell-surface SARS-CoV 307 Spike levels was detected in cells co-expressing the SARS-CoV-derived ORF8 308 genotypes (Fig 5H), demonstrating that reduction of cell-surface Spike levels is a 309 SARS-CoV-2 ORF8-specific phenomenon.  with S-VSV (Fig 7D), but reduced in cells infected with S(+ORF8)-VSV (Fig 7D). These production of a viral component [22]. Consistently, we found that ORF8 restricts Spike 409 incorporation into viral particles (Fig 7B), and in turn, the virions were less infectious 410 ( Fig 7C). However, our studies also revealed the ancestorial ORF8 and VOC-derived ORF8 limit reactivity of anti-SARS-CoV-2 human sera to infected cells. Therefore, our 412 studies represent a SARS-CoV-2 strategy to control Spike antigen levels, retained 413 through the course of evolution.

474
Computational prediction of ORF8 subcellular localization 475 The whole ORF8 amino acid sequence (WA1/2020) [7] was analyzed using Protter 476 (ETH, Zürich).  Aldrich), in a humidified environment at 37 °C with 5% CO 2 . Cells were detached by 506 incubating with trypsin-EDTA (0.05%) (Thermo Fisher) and seeded in well plates at an 507 appropriate cell density not exceeding 90%. When firm cellular attachment is required 508 with HEK293T cells, plates were pre-coated with rat-tail purified collagen (Gibco) as 509 described by the manufacturer. Cells were subcellularly fractionated using the ER isolation kit (Sigma, ER0100) as 547 instructed by the manufacturer's protocol. Briefly, cells plated on two 15-cm plates were 548 suspended in 1 X hypotonic extraction buffer, and incubated at 4 °C for swelling. After 549 20 min, the cells were centrifuged at 600 x g for 5 min and resuspended in 1 X isotonic 550 extraction buffer. The cells were mechanically homogenized using a 7-mL Dounce 551 homogenizer (10 strokes), and the lysate was centrifuged at 1,000 x g 10 min at 4 °C for 552 removal of nuclear fractions. The supernatants were further centrifuged at 12,000 x g for 553 15 min at 4 °C, resulting in mitochondria-enriched pellet (washed two times with PBS 554 before analysis). For isolation of the ER, the supernatant was ultracentrifuged at 555 100,000 x g at 4 °C for 60 min, and the ER-enriched pellet were resuspended in 100 μL 556 of isotonic extraction buffer (ER fraction), which was analyzed by immunoblot, or further 557 incubated in the presence of freshly prepared 0.035 or 0.2% digitonin (Sigma) for 45 558 min at 4 °C for evaluation of the differential solubility. Signaling, Cat #: 7076, 1:5,000 dilution) prepared in the WB blocking buffer. After 1 h, 588 the blot was washed with TBS-T with gentle shaking in TBS-T five times (5 min each).

594
Monolayers of mammalian cells were briefly washed with phosphate-buffered saline 596 (PBS) (Corning) and then lysed by incubating at 4 °C with IP lysis buffer (20 mM NEM, 597 1% NP-40 alternative (Millipore) in IP buffer base (50mM Tris-HCl (pH 7.4), 150 mM 598 NaCl), supplemented with 1 X Halt™ Protease and Phosphatase Inhibitor Cocktail 599 (Thermo Fisher). After 5 min, the lysates were collected and the cell debris were 600 removed by centrifuging at 300 x g for 3 min. The clear supernatants were collected and 601 prepared for immunoblot by mixing with the equivalent volume of 2 X WB lysis buffer (or 602 2 X non-reducing WB lysis buffer), or, further incubated with anti-Flag magnetic beads 603 (Sigma-Aldrich, Cat #: M8823) at room temperature. For pre-treatment, the lysates were 604 incubated with 2 % SDS (for denaturation) at 95 °C for 5 min, in the absence or 605 presence of 0.2 % β-ME (for reduction). After 1 h, the mixture was separated using a 606 magnetic separator and the beads were washed with IP wash buffer (0.05 % NP-40 607 substitute in IP buffer base) three times, and incubated in the WB lysis buffer (or the 608 non-reducing WB lysis buffer) at 95 °C. After 5 min, the proteins eluded from the beads 609 were analyzed by immunoblot.