A single C-terminal residue controls SARS-CoV-2 spike trafficking and incorporation into VLPs

Abstract

The spike (S) protein of SARS-CoV-2 is delivered to the virion assembly site in the ER-Golgi Intermediate Compartment (ERGIC) from both the ER and cis-Golgi in infected cells. However, the relevance and modulatory mechanism of this bidirectional trafficking are unclear. Here, using structure-function analyses, we show that S incorporation into virus-like particles (VLP) and VLP fusogenicity are determined by coatomer-dependent S delivery from the cis-Golgi and restricted by S-coatomer dissociation. Although S mimicry of the host coatomer-binding dibasic motif ensures retrograde trafficking to the ERGIC, avoidance of the host-like C-terminal acidic residue is critical for S-coatomer dissociation and therefore incorporation into virions or export for cell-cell fusion. Because this C-terminal residue is the key determinant of SARS-CoV-2 assembly and fusogenicity, our work provides a framework for the export of S protein encoded in genetic vaccines for surface display and immune activation.

Fig. 1. The conserved C-terminus determines S protein trafficking and coatomer interactions.

A Upper panel: Schematic of bidirectional trafficking of S in the secretory pathway. The newly synthesized S protein (yellow) in the ER (endoplasmic reticulum) is packaged in vesicles and trafficked to the ERGIC (endoplasmic reticulum-Golgi intermediate complex; steps 1, 2), and delivered to the cis-Golgi (steps 3, 4). COPI-coated vesicles (pink) retrieve S to the ER and ERGIC by retrograde trafficking (steps 4-6). S not retrieved by COPI is exported to the PM (step 5’) via the furin-containing trans-Golgi by anterograde trafficking. Middle panel: The coatomer is a hetero-heptamer of seven distinct gene products. N-terminal WD40 domains on α and β‘ subunits form the binding site for client proteins such as S. The illustration in panel A was created using Microsoft PowerPoint. Lower panel: Alignment of S residues in the extra-membrane domain of the cytosolic tail (1255-1273) showing the dibasic motif, extended motif, and clientizing substitutions at the C-terminus. B Multiple mutational hotspots in the N-terminal two-thirds of the full-length SARS-CoV-2 S, which corresponds to the ectodomain. C Few mutations occur in the S tail. D Clientized GST-S tail fusion protein pulls down ~17-fold more coatomer than wild-type S tail-GST fusion. This represents one of three biologically independent samples. E Immunofluorescence imaging of full-length wild-type and clientized S in permeabilized and non-permeabilized HeLa cells. Wild-type S localizes to the PM and early secretory compartments. Clientized S protein is absent from the PM and restricted to early secretory compartments. White outline, intra- and inter-cellular space devoid of S; white arrowheads, early secretory compartments. This represents one of two biologically independent samples. F Western blots of HeLa cells expressing wild-type or clientized full-length S. Wild-type S yields intact, unproteolyzed fractions from early secretory compartments and proteolyzed, furin-cleaved fragments from PM (left). Clientized S yields only an intact fraction from early secretory compartments (right). S₀: intact S; S₂, S₂’: proteolyzed fragments. This represents one of two biologically independent samples.

Fig. 2. Clientized S binds to a basic cluster on the coatomer WD40 domain.

A–C Co-crystal structures of SARS-CoV-2 S tail heptapeptides (residues 1267-1273) in complex with β‘WD40. A Thr1273 in wild-type S shows no side-chain interactions with the basic cluster. B Glu1273 in clientized S forms complementary electrostatic interactions with the β‘WD40 basic cluster. C The β‘WD40 surface provides electrostatic complementarity to the S dibasic motif and the C-terminus. Color code: red, acidic; blue, basic. D, E BLI assays show that αWD40 Arg13Ala mutant has weaker affinity for D wild-type S heptapeptide than for, E Thr1273Glu clientized S heptapeptide. F Superposition of wild-type (white) and Arg13Ala mutant (yellow) crystal structures shows minimal perturbation near the mutation site. G, H BLI assays show that αWD40 Arg300Ala mutant has similar affinity for, G wild-type S heptapeptide and, H Thr1273Glu clientized S heptapeptide. I Superposition of wild-type (white) and Arg300Ala mutant (yellow) crystal structures shows minimal perturbation near the mutation site. J, K BLI assays show that αWD40 Lys15Ala mutant shows no interaction with,(J wild-type S heptapeptide or, K Thr1273Glu clientized S heptapeptide. L Superposition of wild-type (white) and Lys15Ala mutant (yellow) crystal structures shows substantial perturbation and inward rotation of Arg13 and His31 side chains in the mutant to block the S heptapeptide binding site. M, N BLI assays for αWD40 shows loss of binding with, M amidated wild-type S heptapeptide but not, N amidated Thr1273Glu clientized S heptapeptide. Color key indicates αWD40 concentrations for BLI assays.

Fig. 3. The clientized S tail has broad selectivity for αWD40 and β’WD40 domains.

A, B BLI assays show direct binding of β‘WD40 to, A Thr1273Glu clientized S heptapeptide but not to, B clientized S heptapeptide with a scrambled sequence. C Co-crystal structure shows the β‘WD40 Tyr33 side chain pushing the Tyr1272 side chain in the Thr1273Glu clientized S heptapeptide away from the domain surface. D Co-crystal structure shows that substitution of the β‘WD40 Tyr33 side chain for Ala causing minimal movement of the S tail Tyr1272 side chain towards the domain surface. E, F BLI assays show that β‘WD40 Tyr33Ala mutant does not bind to either, E wild-type or F clientized S heptapeptide. G BLI assay shows loss of binding of β‘WD40 Lys17Ala mutant to Thr1273Glu clientized S heptapeptide. H Superposition shows a similar conformational arrangement of juxtaposed Arg residues near the substitution sites in αWD40 Lys15Ala (white-blue) and β‘WD40 Lys17Ala (orange-blue) crystal structures. Color key indicates WD40 concentrations for BLI assays.

Fig. 4. Clientization alters the S tail conformation and strengthens binding to WD40 domains.

A Two-dimensional ¹H-¹⁵N HSQC spectrum of wild-type S tail 21-mer peptide at 25^oC with backbone amide assignments. Assignments in italics are putatively due to a small population ( ~ 10%) of peptide with cis-Pro1263. B Plot of ¹³C secondary chemical shifts (experiment-calculated random coil) in the wild-type (top) and Cys1253Ala-Thr1273Glu clientized (bottom) S tail peptides for Cα (black) and CO (red) resonances. Weak consensus β-propensity is observed in residues Glu1258-Asp1260 for both the wild-type and clientized S-tails and in the dibasic motif-containing Val1268-His1271 for the wild-type S tail but not the clientized S tail. The error bars represent an estimate for the error in measuring the ¹³C chemical shifts based on two independent experiments with wild-type S tail samples. C, D BLI assays showing αWD40 interactions of, C wild-type and, D clientized S tails, the latter being substantially stronger but slower. The concentration of WD40 domains used in BLI assays are shown on the right. E PRE plots of Iox/Ired versus residue for Cys1253Ala S tail 21-mer peptide, Cys1253Ala-Thr1273Glu clientized S tail, and a control for intermolecular effects. PRE experiments were performed in duplicate for the Cys1253Ala A tail peptide and intermolecular control, and in triplicate for the Cys1253Ala-Thr1273Glu clientized S tail. The plots show the mean values (bars, points) and error bars are for ±1 standard deviation for n = 2 (SARS-CoV-2 wild-type S tail), n = 3 (SARS-CoV-2 Cys1253Ala-Thr1273Glu clientized S tail), and n = 2 (inter-peptide PRE) biologically independent samples. The position of the MTSL spin label is indicated (orange arrow).

Fig. 5. Clientization inhibits S protein fusogenicity and incorporation in VLPs.

A Schematic of S protein fusogenicity assay. If fusion-competent S proteins traffic to the surface of S-expressing “effector” HEK293T cells (co-express S, M, E, and N-HiBiT), they engage with “target” HEK293T cells expressing the human receptor for SARS-CoV-2 (hACE2) and activate cell-cell fusion. This results in multi-nucleate syncytia, allowing DSP (dual split protein) reconstitution of nano-luciferase from DSP_1-7 and DSP_8-11. Reconstitution does not occur if S is fusion incompetent or not trafficked to the PM (plasma membrane). The illustration in panel A was created using Microsoft PowerPoint. B Nano-luciferase luminescence for wild-type, and Thr1273Glu or Thr1273Asp clientized S in protein fusogenicity assays. Clientized S shows 10-fold lower nano-luciferase luminescence than wild-type S. Individual data points, mean, and standard error are shown for n = 4 biologically independent samples. C Western blots of whole cell lysates (WCLs) from effector cells used for protein fusogenicity assays show intact S (S₀) from early secretory compartments and cleaved S₂ fragments from PM when wild-type S is expressed. In contrast, no proteolyzed fragments are observed when Thr1273Glu or Thr1273Asp clientized S is expressed. This represents one of two biologically independent samples. D Schematic of SARS-CoV-2 VLP assembly assay. S, M, E, and HiBiT-N plasmids co-transfected in HEK293T cells generate VLPs, which are purified from the extracellular medium. VLPs displaying functional S proteins will engage vesicles displaying hACE2 fused to a cytosolic LgBiT fragment, resulting in VLP-vesicle fusion, nano-luciferase reconstitution, and a luminescence signal. The illustration in panel D was created using Microsoft PowerPoint. E Luminescence signals from VLP assembly assay. Only wild-type S results in fusogenic activity and enhanced luminescence. Individual data points, mean, and standard error are shown for n = 3 biologically independent samples. F Western blots of purified VLPs show the abundant presence of wild-type S and the absence of Thr1273Glu or Thr1273Asp clientized S. This represents one of two biologically independent samples.

Fig. 6. Coatomer binding facilitates incorporation of S into VLPs and inhibits cell-cell fusogenicity.

A Luminescence signals from VLP assembly assay. An S protein double mutant (Lys1269Ala/His1271Ala) that does not interact with coatomer shows a substantially smaller increase in luminescence than wild-type indicating poor incorporation into VLPs. Individual data points, mean, and standard error are shown for n = 3 biologically independent samples. B Western blots of purified VLPs confirm poor incorporation of the non-interacting S protein. C Western blots of WCLs (whole cell lysates) show no expression differences between the wild-type and non-interacting S protein. This represents one of two biologically independent samples. D Nano-luciferase luminescence for non-interacting S protein is 2.3-fold higher than wild-type. Individual data points, mean, and standard error are shown for n = 3 biologically independent samples.