Ultrastructural insight into SARS-CoV-2 entry and budding in human airway epithelium

SARS-CoV-2 rapidly caused a global pandemic, with high infectivity linked to the spike (S) glycoprotein’s affinity for the human ACE2 receptor. ACE2 is expressed at the apical surface of airway epithelial cells, and the serine protease TMPRSS2 facilitates spike activation and plasma membrane fusion. Alternative endosomal entry routes exist in TMPRSS2-low contexts via cathepsin cleavage. Replication occurs in ER-derived double-membrane vesicles (DMVs), and virions are thought to bud at ERGIC, with possible egress via lysosomes. Multiple variants, including B.1.1.7 (VOC 202012/01), carry spike mutations potentially affecting transmissibility. This study investigates, at ultrastructural resolution in primary human airway epithelium (HAE), where SARS-CoV-2 virions attach, enter, and bud, and whether these processes differ among three isolates (B.1.1.7, B.1.258, B.1.117.19).

Prior work established ACE2 and TMPRSS2 as key host factors for SARS-CoV-2 entry, with TMPRSS2-mediated S2′ cleavage driving plasma membrane fusion, while cathepsin-dependent endosomal pathways can substitute in TMPRSS2-deficient settings. Coronavirus replication organelles include DMVs, convoluted membranes, and spherules, enriched for double-stranded RNA. Budding was localized to ERGIC for SARS-CoV and potentially SARS-CoV-2, with evidence that coronaviruses may exploit lysosomes for egress. Furin precleavage at S1/S2 can enhance entry in respiratory tissue, and variants lacking the furin site favor endosomal entry with reduced respiratory transmission. The B.1.1.7 lineage gained rapid dominance in the UK, with spike mutations including N501Y and P681H. Reports on target cell tropism suggest preferential infection of ciliated cells over goblet cells in human airway epithelia, though some studies have observed goblet cell infection under certain conditions.

Study design: Primary human nasal airway epithelial (HAE) cells were differentiated at an air–liquid interface (ALI) to generate ciliated, goblet, and basal cells. Cells were infected with three SARS-CoV-2 isolates (B.1.1.7, B.1.258, B.1.117.19) at MOI 0.01 pfu/cell, and fixed at 72 h post-infection for ultrastructural analysis and correlative imaging. Human samples: Nasal brushings were obtained from a healthy donor (used to derive HAE cultures for in vitro infections) and from a PCR-confirmed SARS-CoV-2–infected donor (embedded for EM; single experiment). Ethical approvals and informed consent were in place. Cell culture: HAE cultures were established from nasal brushings, expanded, and seeded onto 6.5 mm, 0.4 µm-pore transwell inserts. After confluence, cells were transitioned to ALI with PneumaCult-ALI medium; ciliation developed over 4–6 weeks. Virus stocks and infection: Clinical isolates hCoV-19/England/204661721/2020 (B.1.1.7), hCoV-19/England/204501206/2020 (B.1.258), and hCoV-19/England/204501194/2020 (B.1.117.19) were passaged twice in Vero cells. Prior to infection, apical mucus was removed by 200 µl DMEM incubation (10 min, 37 °C, 5% CO2). Inoculum was applied apically for 1 h (37 °C, 5% CO2), removed, and cultures incubated a further 72 h. Electron microscopy (TEM): Samples were fixed in 2.5% glutaraldehyde (0.05 M sodium cacodylate, pH 7.4) ≥24 h, post-fixed in 1% osmium tetroxide (1 h, RT), en bloc stained (UA-Zero, 30 min), dehydrated (ethanol series), transitioned via propylene oxide, embedded in araldite, and polymerized (60 °C, 48 h). Ultrathin sections were stained with Reynolds’ lead citrate. Imaging used JEOL 1400Plus TEM with AMT XR16 or Gatan Orius SC1000B CCD. Electron tomography (ET): Dual-axis tilt series (±60° over two perpendicular axes) were acquired using SerialEM and reconstructed with IMOD. Subtomographic averaging used PEET; surface rendering used UCSF Chimera. Immunofluorescence (IF): HeLa, HAE, and nasal brushing sections (∼15 µm cryostat sections for HAE/brushings) were fixed (4% PFA, 1 h), permeabilized with saponin, blocked (1% BSA). Primary antibodies included: ACE2 (extracellular and intracellular domain), TMPRSS2, tubulin (cilia), phalloidin (actin/microvilli), Ezrin, S glycoprotein, and nucleocapsid. Secondary Alexa Fluor-conjugated antibodies were used. Imaging on Leica SP8 confocal; analysis in Fiji. Immunoelectron microscopy (iEM): Cryostat sections of HAE and fixed HeLa cells were blocked and incubated with antibodies to S glycoprotein, nucleocapsid, or HA (for spike-transfected HeLa), followed by 5-nm protein A-gold; post-fixed and processed for TEM without propylene oxide in dehydration. HeLa transfections: HeLa cells were transfected with myc-ACE2, ACE2, HA-TMPRSS2, or ss-HA-Spike constructs using Lipofectamine 3000; fixed 2 days post-transfection for IF/iEM validation and subtomography comparisons. Quantification and statistics: Quantified numbers, sizes, and locations of VCs and DMVs; counts and densities of virions within VCs; virion diameters, S-projection counts, and distances from plasma membrane; frequency of virion attachment to microvilli vs cilia; and VC prevalence in goblet vs ciliated cells. Sample sizes included N=28 infected cells for VC counts, N=16 for DMV counts, and N=30 VCs/DMVs for size/density measurements; virion structural metrics with N≈23–26. Paired, two-tailed Student’s t-tests; data reported as mean ± SEM.

- Across three SARS-CoV-2 isolates (B.1.1.7, B.1.258, B.1.117.19), primary outcomes were highly similar by conventional TEM and tomography; no isolate-specific ultrastructural differences were detected in virion morphology, size, or cell attachment. - Cell tropism: Ciliated epithelial cells showed robust infection, evidenced by abundant viral-containing compartments (VCs), double-membrane vesicles (DMVs), and extracellular virions. Goblet cells rarely displayed VCs/DMVs or surface virions. - Receptor/protease localization: IF revealed ACE2 and TMPRSS2 enriched at the apical plasma membrane, including microvilli, but largely excluded from cilia. This pattern was corroborated in nasal brushing biopsies and antibody validation in transfected HeLa cells. - Virus attachment: TEM showed extracellular virions predominantly attached to microvilli/apical plasma membrane, rarely to cilia. Quantification (N=14 cells per isolate) showed significantly more virions attached to microvilli than cilia: B.1.1.7 P=5.3×10^-13; B.1.258 P=3.44×10^-19; B.1.117.19 P=7.19×10^-18 (paired two-tailed t-tests). - Variant comparisons: No significant differences among isolates in VC counts, VC size, virions per VC, virion density within VCs (N=28 cells for counts; N=30 VCs for size/density), nor in DMV counts/sizes (N=16 cells; N=30 DMVs). Virion diameter, distance to plasma membrane, and number of S projections per virion did not differ among variants (diameter N=23; distance N=26; S projections N=25; paired two-tailed t-tests). - Goblet vs ciliated cells: VCs per µm² cytoplasm were significantly lower in goblet cells versus ciliated cells for all isolates (single sample per isolate, N=28 cells; B.1.1.7 P=3.48×10^-13; B.1.258 P=6.07×10^-11; B.1.117.19 P=1.81×10^-11). - Plasma membrane fusion: ET captured virions with membranes continuous with the host apical plasma membrane and diluted intraviral nucleocapsids, consistent with fusion and nucleocapsid release into the cytoplasm; a neck-like fusion structure was observed. - Budding in single-membrane VCs: Intracellular virions displayed budding-like profiles from the limiting membrane of VCs bearing S glycoprotein facing the lumen. Nucleocapsids concentrated at the budding virion periphery; microtubules ran into or near budding sites, suggesting trafficking of components. - S glycoprotein at infected cell surfaces: S-like protrusions were observed on microvilli and plasma membranes of infected ciliated cells (and in patient nasal brushing), but absent on cilia and on goblet cells or non-infected cells. Subtomographic averages of plasma membrane and VC protrusions resembled viral S; iEM localized S to these protrusions. - Overall pathway: Data support microvilli-mediated attachment, plasma membrane fusion-based entry in ciliated cells, and virion budding into S-coated single-membrane VCs as a mechanism for acquiring the envelope.

These findings clarify the spatial biology of SARS-CoV-2 entry and assembly in human airway epithelium. Contrary to some prior reports suggesting ciliary ACE2 localization, ACE2 and TMPRSS2 were enriched on the apical plasma membrane including microvilli but excluded from cilia, aligning with the observed preferential virion attachment to microvilli. Tomographic evidence of membrane continuity and nucleocapsid dilution confirms plasma membrane fusion as a route of entry in ciliated cells, consistent with TMPRSS2-dependent activation. The scarcity of infection markers in goblet cells suggests limited susceptibility at early stages, potentially due to receptor/protease distribution and mucin barriers. Intracellularly, budding-like profiles from S-bearing single-membrane VCs with peripheral nucleocapsid enrichment propose a mechanism by which virions assemble and acquire spike-bearing envelopes. The presence of S on VC membranes and infected cell surfaces provides both a marker of infection and a mechanistic link: S may be deposited at the plasma membrane via viral fusion and/or VC–plasma membrane fusion during egress, with possible implications for syncytia formation. The lack of detectable differences among isolates at this resolution suggests conserved entry and budding pathways, notwithstanding sequence variation in spike. These insights reinforce targets such as TMPRSS2 and pathways governing VC formation and trafficking for therapeutic intervention.

In primary human airway epithelium, SARS-CoV-2 preferentially infects ciliated cells, attaching to microvilli where ACE2 and TMPRSS2 are enriched, and enters via plasma membrane fusion. Virions assemble and likely bud into single-membrane viral-containing compartments bearing S glycoprotein, providing a route to acquire the envelope, while goblet cells are largely refractory under early infection conditions. No ultrastructural differences were observed among three clinical isolates including B.1.1.7. These data refine our understanding of the virus life cycle in airway tissue and highlight microvilli-localized entry machinery and VC-mediated budding as potential therapeutic targets. Future work should define the origin and molecular composition of VCs (ERGIC vs lysosomal pathways), delineate egress routes, clarify the relative contribution of fusion versus budding at VC membranes, and assess temporal dynamics and cell-type susceptibility over extended infection periods using higher-resolution cryo-EM/ET approaches.

Experiments were performed primarily at a single 72 h post-infection timepoint and on HAE cultures derived from a single healthy donor for in vitro infections, with single samples per isolate for TEM/IF analyses (staining reproducibility assessed across multiple sections). Conventional TEM provides limited molecular resolution; definitive assignment of VC origin (ERGIC vs lysosome) and discrimination between budding vs occasional fusion at VC membranes cannot be fully resolved. Antibody-based localization may include diffuse background; accessibility issues (e.g., cilia) can affect IF signal. Variant comparisons may lack power to detect subtle molecular differences resolvable only by higher-resolution cryo-EM. Goblet cell infection may emerge at later stages not captured here. Viral isolates were passaged in Vero cells before HAE infection, which could introduce culture adaptations.