SARS-CoV-2 Delta variant-specific ORF3a mutations destabilize lysosomal homeostasis to trigger lysosomal damage-mediated cell death

Abstract

The variants of concern (VOCs) of SARS-CoV-2 emerged independently and became dominant globally at different times. Despite the emergence of multiple VOCs, the Delta variant of SARS-CoV-2 showed heightened pathogenicity and unprecedented mortality. However, the Delta variant specific mechanisms underlying its increased pathogenicity are unclear. Here, we show that variations in SARS-CoV-2 ORF3a correlate with the pathogenic potential of VOCs and drive Delta variant-specific lysosomal damage that activates inflammatory cell death. ORF3a from the Delta variant shows unique mutational patterns distinct from other SARS-CoV-2 VOCs. ORF3a-specific phylogenetic analysis reveals noticeable differences in the evolutionary trajectories of VOCs, likely reflective of their pathogenic relatedness. Unlike the Omicron variant and the ancestral Wuhan strain, Delta-specific ORF3a mutations promotes robust lysosomal damage, peripheral distribution, and membrane localization with no apparent effect on viral titers. Furthermore, Delta ORF3a mutations-induced lysosomal damage promotes both apoptosis and necroptosis activation in human cells. Structurally, these Delta variant-specific mutations appear to stabilize the ORF3a oligomers through helical packing and formation of a non-native disulfide bond, possibly facilitating their lysosomal association and damage. Overall, our observations indicate that ORF3a disrupts lysosomal homeostasis and triggers cell death, suggesting Delta-variant-specific regulation of cell fate and inflammation-associated pathogenesis.

SARS-CoV-2 ORF3a variations from the Delta variant differentiate it from other variants of concern; these mutations promote lysosome damage and multimodal cell death activation associated with disease severity.

Introduction

During the course of the COVID-19 pandemic, the SARS-CoV-2 Wuhan strain acquired a wide variety of mutations^1–3. This led to the emergence of variants of concern (VOCs), which were designated based on changes in transmissibility, immune evasion, and clinical severity^1,3,4. The Delta variant emerged in Maharashtra, India, in late 2020⁵, and its high transmissibility made it the dominant VOC worldwide. It contributed to high hospitalization rates and severe clinical outcomes^2,6–10. Omicron showed higher transmissibility compared to the Delta variant; however, epidemiological studies revealed that Omicron infections were routinely associated with lower risks of hospitalization, ICU admissions, and mortality^6–8,11. The difference in the pathogenicity profiles of VOCs has been largely attributed to population immunity¹. However, the precise mechanisms underlying the distinct pathological outcomes remain poorly understood. Although mutations in the spike protein are strongly associated with the transmissibility and immune escape of SARS-CoV-2 VOCs^1,12, the pathogenic potential of these variants cannot be explained by spike evolution alone and is likely determined by additional viral and host factors^1,13. Also, phylogenetic trees reconstructed from whole-genome SARS-CoV-2 sequences largely recapitulate patterns of spike evolution and changes in viral transmissibility¹⁴. However, whether variation in non-spike regions reflects distinct evolutionary trajectories linked to pathogenic potential remains unclear.

Regulated cell death activation is an innate immune response that facilitates the clearance of viral niches^15,16. However, aberrant activation of cell death and inflammation can exacerbate tissue damage and mortality during viral infections^17,18. Excessive cell death, inflammation, and necrotic tissue damage are associated with the disease severity in COVID-19 patients, particularly with the Delta variant infections^6,8,17–20. Whether the new variations in SARS-CoV-2 non-spike proteins drive host regulation to promote cell death and inflammation needs to be established.

SARS-CoV-2 replicates within membrane-bound replication organelles, called double-membrane vesicles derived from host intracellular membranes, and remodels the endolysosomal network during infection to support its replication²¹. In addition, SARS-CoV-2 infection significantly alters the cellular autophagy. SARS-CoV-2 has been demonstrated to disrupt autophagic flux and lysosomal function, to mitigate viral component degradation^22–25. Also, accumulating evidence suggests that SARS-CoV-2 utilizes a lysosome-dependent trafficking pathway to facilitate viral egress from infected cells²⁶. Inhibition of autophagosome-lysosome fusion and the peripheral distribution of lysosomal vesicles enable virus egress. Loss of lysosomal homeostasis and integrity trigger activation of cell death and inflammatory signaling, contributing to tissue damage and mortality^19,27–30. Understanding how SARS-CoV-2 viral proteins modulate autophagy and lysosomal homeostasis is critical for elucidating mechanisms of viral pathogenesis and disease severity across SARS-CoV-2 variants.

ORF3a is the largest accessory integral membrane protein of SARS-CoV-2; it contains three transmembrane helices and a cytosolic domain enriched in β-sheets^31,32. ORF3a exists in oligomeric states, predominantly as dimers and to a lesser extent as tetramers^31,33,34. ORF3a interacts with multiple trafficking proteins, is enriched at the plasma membrane and in the endocytic pathway, and impacts host autophagy and endolysosomal pathways³⁴. The late endosome-lysosome localized ORF3a sequesters VPS39 of the homotypic fusion and vacuolar protein sorting (HOPS) complex and prevents its interaction with the autophagosomal SNARE protein, STX17^23,33–35. This inhibits the fusion of autophagosomes and lysosomes^23,24,35. SARS-CoV-2 ORF3a promotes the lysosomal targeting of BORC-ARL8b complex, which promotes the membrane targeting of lysosomes, which in turn contributes to viral egress through lysosomal exocytosis²⁶. However, whether ORF3a directly induces lysosomal damage and damage-mediated lysosomal exocytosis is unclear. A recent study showed that depletion of ORF3a reduced virulence in SARS-CoV-2 infection in mice compared to WT SARS-CoV-2 or mutants deficient in other accessory proteins³⁶. Thus, SARS-CoV-2 ORF3a plays an important role in viral fitness and pathogenicity^26,36,37. Yet, how the ORF3a sequence varies across VOCs and the implications of the associated mutations for virulence and pathogenicity remain to be understood.

Here, we report that mutational patterns in ORF3a of SARS-CoV-2 are associated with the pathogenic potential of VOCs. We unravel the role of Delta variant-specific mutations in ORF3a in triggering robust lysosomal damage, in contrast to Omicron variant-specific mutations. We further demonstrate that variant-specific mutations in ORF3a mediate differential lysosomal damage and damage-dependent cell death. These ORF3a mutations did not affect viral titers, suggesting their impact on pathogenicity rather than on viral transmission. Our observations suggest that Delta mutations in ORF3a promote excessive lysosomal damage, leading to enhanced activation of cell death pathways and hence immunopathology and disease severity. Also, these observations provide molecular insights into differences in variant biology.

Results

SARS-CoV-2 ORF3a associates with the distinct pathogenic evolution of Delta and Omicron variants

To understand why the Delta variant displayed enhanced pathogenesis over other SARS-CoV-2 VOCs, we investigated all the proteins of SARS-CoV-2 VOCs except the Spike protein to establish a unique subset of mutations in the Delta virus compared to other VOCs. Our analysis revealed that the SARS-CoV-2 proteins Nsp2, Nsp3, Nsp13, Nucleocapsid, ORF3a, and ORF8 showed unique mutational patterns in the Delta variant compared to other VOCs, in particular the Omicron variant (Figs. 1A and S1A–E). ORF3a is the largest and most expressed accessory protein of SARS-CoV-2; it also differs significantly in its sequence and function from the SARS-CoV-1 ORF3a^32–35. Unlike SARS-CoV-1 ORF3a, the SARS-CoV-2 ORF3a enables inhibition of autophagosome-lysosome fusion and promotes viral egress through lysosomal targeting³⁵.

Fig. 1. SARS-CoV-2 ORF3a shows a unique set of mutations in the Delta variant compared to the Omicron variant.

A The heatmap illustrates amino acid residues in ORF3a which have undergone mutations across VOCs. The purple cells are annotated with the introduced mutations. The pink cells represent the absence of mutations at that position. B Maximum likelihood phylogenetic tree constructed for SARS-CoV-2 whole genome sequences representing the relatedness between VOCs. Nodes at the end represent a unique sequence. C Maximum clade credibility phylogenetic tree constructed for consensus SARS-CoV-2 orf3a gene sequences representing the phylogenetic subgrouping of VOCs. Values presented at the nodes are posterior probabilities. D Root-to-tip divergence plot indicating selection pressure on SARS-CoV-2 orf3a and spike genes in comparison to whole genome. E The annotation of domains and VOC mutations in the SARS-CoV-2 ORF3a structure (PDB ID: 6XDC). S26 residue position in the ORF3a structure was not annotated as the coordinates for this position are missing in the structure. NTD: N-terminal domain; TM1-3: transmembrane helices 1-3. F The heatmap illustrates the average fitness effects of amino acid mutations at positions of interest in SARS-CoV-2 ORF3a protein. Each cell with a blue gradient is annotated with estimated fitness values for the particular substitution. All grey cells represent the absence of the substitution in natural population. All white cells represent the WT amino acid in the SARS-CoV-2 Wuhan clade. The mutations of interest are outlined in a coral-pink box.

To assess whether ORF3a contributes to the pathogenic ability of SARS-CoV-2 variants, we performed viral phylogeny studies of SARS-CoV-2 variants to identify potential evolutionary signatures linking ORF3a sequence variations across VOCs to their pathogenicity. We analyzed gene sequence data of six variant lineages: Alpha (B.1.1.7), Gamma (P.1), Kappa (B.1.617.1), Delta (B.1.617.2), and Omicron (BA.1 and BA.2). We constructed maximum likelihood (ML) phylogeny trees for SARS-CoV-2 whole genome sequences and spike sequences (Figs. 1B and S1F). In agreement with previous studies, both the ML trees were similar in topology, with a clear distinction between sequences from the several variant lineages³⁸. Omicron lineages, BA.1 and BA.2, consistently grouped together (Figs. 1B and S1F). The Delta and Kappa lineages formed a closely related cluster that diverged from other variants early in evolution. Root-to-tip divergence analysis indicated that the orf3a gene evolved under evolutionary pressure similar to that of the whole genome sequence, in contrast to spike, which showed higher divergence, suggesting stronger selective pressure possibly from spike-targeting vaccines³⁹ (Fig. 1D)^40–42. Due to the limited diversity in orf3a sequences, we generated the maximum clade credibility tree for orf3a sequences (Fig. 1C). The topology of this tree differed from that of the trees for whole-genome and spike sequences. orf3a of the Alpha and Gamma variants clustered together and emerged much closer to Delta and Kappa, all sharing a common ancestor. This was in sharp contrast to the whole genome and spike trees, which predicted a common ancestor for the Alpha, Gamma, and Omicron variants. The orf3a phylogeny showed that Delta variant sequences did not cluster solely with the Kappa variant but also grouped with Alpha and Gamma, suggesting a common ancestor with SARS-CoV-2 variants that have become more pathogenic than Omicron (Fig. 1C)⁴. In contrast, Omicron remained separated from this clade in the orf3a tree. Notably, the orf3a tree correctly segregates the VOCs into two major groups (Alpha, Gamma, Delta, and Kappa variants versus Omicron variants) with VOCs containing similar within-group but different between-group pathogenicities. This highlights the possible importance of ORF3a mutations in defining the distinct evolutionary trajectories of Delta and Omicron and in reflecting their differences in pathogenic profiles.

In line with the distinct evolutionary trajectories of the Delta and Omicron variants, ORF3a showed a distinct mutational pattern in Delta and Omicron. The S26L mutation in the ORF3a N-terminal domain appears in nearly all sequences of the Delta variant (Fig. 1E)^43,44. The other ORF3a mutations, L108F and W131C, spanning transmembrane domain 3, and D155Y, spanning beta-barrel region, appear to be less frequent than S26L in the Delta variant (Fig. 1E)^38,44–46. Only the D155Y appeared in some of the Omicron strains, but not any other Delta mutations. A recent study established the fitness effects of mutations in SARS-CoV-2 proteins that appeared in each variant of concern⁴⁷. The fitness score was calculated based on the mathematical assumption that neutral mutation rate during SARS-CoV-2 evolution led to the occurrence of non-deleterious single-nucleotide mutations independently multiple times. The expected number and observed count of independent occurrences of each mutation are compared to estimate the effect of each mutation on virus fitness. The ratio of actual to expected counts for each mutation is directly correlated to its effect on viral fitness. We observed based on estimates from that study that amino-acid substitutions in the Delta variant (W131C, D155Y, and L108F) conferred greater fitness to the protein than Omicron variant mutations, although none of the mutations of interest had a deleterious functional effect (Fig. 1F)⁴⁷. We examined next the functional implications of these ORF3a variations.

Delta-variant-specific ORF3a mutations promote lysosomal damage

SARS-CoV-2 ORF3a is localized to the lysosomes due to its interaction with lysosomal surface-localized HOPS complex. ORF3a binding to HOPS complex restricts autophagosome-lysosome fusion and promotes lysosome-dependent viral exocytosis^23,26,35. To understand the functional effects of ORF3a Delta mutations, we examined their impact on lysosomal dynamics. We ectopically expressed Delta mutants, S26L, L108F, and D155Y and W131C in combination (hereafter called DW). We generated a combinatorial mutant, D155Y/W131C, because structural studies of SARS-CoV-2 ORF3a revealed that D155 and W131, among other amino acids, may enable tetramerization of the ORF3a protein³¹. To assess whether the Delta variant of ORF3a differs from other variants in its function, we included Omicron mutations (L106F and T223I) in this analysis^38,45. Ectopic expression of Delta and Omicron mutants of ORF3a did not alter ORF3a protein expression (Fig. S2A). To assess whether the Delta and Omicron mutations in ORF3a affect ORF3a mediated lysosome integrity, we monitored the localization of lysosomes and their membrane integrity by probing lysosomal membrane surface proteins (LAMP1) and Galectin-3. Lysosomal membrane damage leads to the exposure of the lysosomal glycocalyx, which is recognized by multiple Galectins such as Galectin-3 and Galectin-8^22,48,49. Galectin-3 puncta formation is a direct measure of lysosomal damage²². The cells that did not express ORF3a showed LAMP1-positive lysosomes lacking Galectin-3 signals, suggesting intact lysosomes without damaged membranes. Ectopic expression of ORF3a-WT led to Galectin-3 puncta formation that colocalized with LAMP1-positive lysosomes, suggesting ORF3a mediated lysosomal damage (Fig. 2A–C). The Delta-specific mutations in ORF3a (S26L, DW, and L108F) significantly augmented the Galectin-3 puncta colocalized with LAMP1, whereas the Omicron-specific ORF3a mutations (L106F and T223I) showed significantly less Galectin-3 puncta and more intact lysosomes. These observations indicate that Delta- and Omicron-specific mutations in ORF3a alter lysosomal damage dynamics.

Fig. 2. Delta-specific mutations in SARS-CoV-2 ORF3a show altered lysosomal dynamics.

A Confocal microscopy imaging of HeLa cells co-expressing ORF3a-WT and mutants with LAMP1-mRFP, stained for Galectin-3 puncta, ORF3a (Strep-Tag). Scale bar, 15 μm. Quantification of percentage of Galectin-3 negative (intact) (B) and Galectin-3 positive (damaged) (C) lysosomes (LAMP1-mRFP) in HeLa cells expressing ORF3a-WT and mutants. ****P < 0.0001, ^####P < 0.0001, ***P = 0.0002, *P = 0.0129 (ORF3a-WT vs DW), *P = 0.0164 (ORF3a-WT vs L108F), ^#P = 0.0381 (ORF3a-WT vs L106F) (one way ANOVA, n = 24 cells across three independent experiments). Data shown are mean ± SEM. ‘*’ indicates a significant increase and ‘#’ indicates a significant decrease.

Our observations of increased Galectin-3-positive lysosomes in response to Delta variant mutations prompted us to assess the effects of Delta- and Omicron-specific mutations on lysosomal damage. Expression of Galectin-3-GFP alone in HeLa cells showed cytosolic distribution and did not form Galectin puncta, indicating that Galectin-3 expression does not promote lysosome damage (Fig. 3A). Co-expression of ORF3a-WT and Galectin-3-GFP led to the appearance of Galectin-3 puncta. The Delta-specific mutations in ORF3a significantly increased the number of Galectin-3 puncta formed, indicating robust lysosomal damage and membrane permeabilization (Fig. 3A, B). Notably, the ectopic expression of the Omicron mutant, T223I, did not show the formation of Galectin-3 puncta, and the expression of the L106F mutant induced very few Galectin-3 puncta. Immunostaining of endogenous Galectin-3 in ORF3a-expressing cells further corroborated the Delta variant-specific (in particular, DW and L108F) lysosomal damage compared to the Omicron mutants (Figs. 3C and S2B). Enhanced lysosomal damage by Delta-specific mutations in ORF3a was further validated by increased Galectin-8 puncta formation (Fig. S2C, D). These observations indicate that Delta variants of ORF3a robustly induce lysosomal damage.

Fig. 3. Delta-specific mutations in SARS-CoV-2 ORF3a promote enhanced lysosomal damage.

A Confocal microscopy imaging of HeLa cells co-expressing ORF3a-WT and mutants with Galectin-3-GFP, stained for ORF3a (Strep-Tag). Scale bar, 10 μm in both complete image and zoomed insets. B Quantification of the number of Galectin-3-GFP puncta per cell for the microscopy images of HeLa cells co-expressing ORF3a-WT and mutants with Galectin-3-GFP. ****P < 0.0001, ^####P < 0.0001,**P = 0.0014 (ORF3a-WT vs S26L), **P = 0.0046 (ORF3a-WT vs DW),**P = 0.0049 (ORF3a-WT vs L108F), ^#P = 0.0261 (one way ANOVA, n = 60 cells across three independent experiments). Data shown are mean ± SEM. ‘*’ indicates a significant increase and ‘#’ indicates a significant decrease. C Quantification of the number of endogenous Galectin-3 puncta per cell for the microscopy images of HeLa cells expressing ORF3a-WT and mutants. ****P < 0.0001, **P = 0.0040, and ^#P = 0.0198, ns – not significant (one way ANOVA, n = 60 cells across three independent experiments). Data shown are mean ± SEM. ‘*’ indicates a significant increase and ‘#’ indicates a significant decrease. D Confocal microscopy imaging of A549-ACE2 cells infected with SARS-CoV-2 (HKU, Delta, and Omicron strains) at 72 h of infection, stained for SARS-CoV-2 nucleocapsid protein and Galectin-3. Scale bar, 10 μm. E Quantification of the number of endogenous Galectin-3 puncta per cell for the microscopy images of A549-ACE2 cells infected with SARS-CoV-2 (HKU, Delta, and Omicron strains) at 72 h of infection. **P = 0.0027, ***P = 0.0003, ns – not significant (one way ANOVA, n = 30 frames across three independent experiments). Data shown are mean ± SEM.

The SARS-CoV-2 Delta variant shows increased lysosomal damage

Whether SARS-CoV-2 infection promotes lysosome damage and Galectin-3 puncta formation is unknown. To further determine whether SARS-CoV-2 infection induces VOC-specific lysosomal damage, we infected SARS-CoV-2 permissive A549-ACE2 cells with the ancestral Wuhan-like SARS-CoV-2 (Hong Kong), Delta, and Omicron isolates. Expression of the SARS-CoV-2 nucleocapsid (N) protein indicated productive SARS-CoV-2 infection by these strains. Immunostaining for endogenous Galectin-3 showed Galectin-3 puncta formation in A549-ACE2 cells after SARS-CoV-2 infection (Fig. 3D, E). Upon Delta variant SARS-CoV-2 infection, A549-ACE2 cells showed a significantly higher number of Galectin-3 puncta, indicating increased lysosomal damage and membrane permeabilization, corroborating ORF3a-ectopic expression experiments. However, infection with the Omicron variant induced lesser Galectin-3 puncta formation.

Delta variant-specific ORF3a mutations show robust inhibition of autophagosome-lysosome fusion

ORF3a interacts with and sequesters VPS39 in the HOPS complex localized on the endolysosomal membrane and prevents its fusion with STX17 of autophagosomes, leading to inhibition of autophagosome fusion with lysosomes^23,35. To assess whether lysosomal damage mediated by Delta-variant-specific ORF3a mutations disrupts autophagic flux, we monitored ORF3a-mediated autophagy inhibition using an mCherry-GFP-LC3 assay^50,51. Before fusion of autophagosomes with lysosomes, mCherry-GFP-LC3 on autophagosomes is marked by yellow puncta, indicating signals from both fluorophores. After the fusion of autophagosomes with lysosomes, mCherry-GFP-LC3 in the autolysosomes shows only the mCherry (red) signal, as the GFP signal is quenched because of the acidic pH of the compartment^23,50,51. As previously reported, the mCherry-GFP-LC3 assay showed that ectopic expression of ORF3a inhibited autophagosome fusion with lysosomes as measured by the increased percentage of yellow puncta when compared to control cells (Fig. 4A, B). Delta variant-specific mutations in ORF3a significantly augmented autophagosome-lysosome fusion inhibition and accumulation of autophagosomes (yellow puncta) compared to ORF3a-WT (Fig. 4A–C). However, Omicron mutants of ORF3a did not enhance autophagosome-lysosome fusion inhibition compared to ORF3a-WT. Upon autophagy inhibition, lipidated LC3 (LC3-II) accumulates in the cells. Immunoblotting analysis of LC3 revealed increased levels of LC3-II upon expression of ORF3a-WT and Delta mutants. In contrast, Omicron mutants of ORF3a expression did not promote LC3-II accumulation (Fig. 4D). Bafilomycin-A1-mediated autophagosome-lysosome fusion inhibition is further exacerbated in the presence of ORF3a-WT and Delta mutants, as evidenced by elevated accumulation of LC3-II in comparison to Omicron mutants (Fig. 4D). These observations indicate that the Delta variant ORF3a mediated lysosomal damage impairs autophagic flux.

Fig. 4. Delta-specific mutations in SARS-CoV-2 ORF3a lead to enhanced autophagosome-lysosome fusion inhibition.

A Confocal microscopy imaging of HeLa cells co-expressing ORF3a-WT and mutants with mCherry-GFP-LC3, stained for ORF3a (Strep-Tag). Scale bar, 15 μm. B Quantification of percentage of unacidified autophagosomes for the microscopy images of HeLa cells co-expressing ORF3a-WT and mutants with mCherry-GFP-LC3. ****P < 0.0001, *P = 0.0284 (ORF3a-WT vs S26L), *P = 0.0310 (ORF3a-WT vs DW), *P = 0.0185 (ORF3a-WT vs L108F), ns – not significant (one way ANOVA, n = 45 cells across three independent experiments). Data shown are mean ± SEM. C Quantification of number of yellow puncta (mCherry+GFP+) per cell for the microscopy images of HeLa cells co-expressing ORF3a-WT and mutants with mCherry-GFP-LC3. **P = 0.0100, *P = 0.0179 (ORF3a-WT vs DW), *P = 0.0161 (ORF3a-WT vs L108F), ns – not significant (one way ANOVA, n = 45 cells across three independent experiments). Data shown are mean ± SEM. D Immunoblot analysis of lysates of HeLa cells of ORF3a-WT and mutants after treatment with Bafilomycin A1, showing expression of ORF3a (Strep-Tag), LC3 and GAPDH.

Delta variant-specific mutations stabilize ORF3a structure by transmembrane helical packing and non-native oligomerization

SARS-CoV-2 ORF3a exists primarily as a dimer and to a lesser extent in a tetrameric conformation and displays an unstructured loop critical for VPS39 binding, which is not seen with SARS-CoV-1 ORF3a³⁴. To understand how Delta variant mutations reshape the ORF3a structure and trigger lysosomal damage, we monitored the assembly of ORF3a dimer and tetramer complexes. DSP-crosslinking of ORF3a-expressing cells indicated that Delta and Omicron mutants of ORF3a did not alter dimer formation (Fig. 5A). These dimers were also observed in lysates without a crosslinker, suggesting that the Delta- and Omicron-specific ORF3a mutations do not affect dimer formation. Structural examination suggests that SARS-CoV-1 and SARS-CoV-2 ORF3a have a conserved cysteine-rich region (C130, C133, C148, and C157) known to promote disulfide-mediated dimerization of SARS-CoV-1 ORF3a^31,34. However, these cysteines are reduced in the SARS-CoV-2 ORF3a structure and are dispensable for ORF3a dimer formation³¹. In SARS-CoV-2 ORF3a, the residues W131 and D155 are localized towards the cytosolic portion of the ORF3a structure and critically contribute to stabilizing the tetrameric conformation (but not dimeric) of ORF3a³¹. In the Delta variant, W131 is mutated to cysteine (W131C), which is in close proximity to the cysteine-rich region, indicating a potential possibility for non-native disulfide formation to stabilize the tetramer conformation (Fig. 5B). Although the cross-linked lysates did not capture the ORF3a tetramer species, the lysates without crosslinker and the disulfide reducing agent (β-mercaptoethanol) showed a prominent tetrameric band in the DW mutant of ORF3a (Fig. 5A). This observation suggests that the Delta mutations of ORF3a, W131C, and D155Y appear to stabilize dimer and tetramer conformations to possibly promote their binding to VPS39 on lysosomes.

Fig. 5. SARS-CoV-2 ORF3a Delta mutations stabilize dimer and tetramer conformation.

A Immunoblot analysis of lysates (with or without crosslinker) of HeLa cells expressing ORF3a-WT and mutants, in non-reduced (without BME) and reduced (with BME) conditions. B Structure analysis of tetramers of ORF3a WT (W131 and D155) and DW mutant (C131 and Y155) after alignment. M-monomer, D-dimer, T-tetramer. C Structure analysis of transmembrane helices of ORF3a WT (L106 and L108), L106F, and L108F mutants after alignment (ORF3a WT PDB ID: 6XDC).

Although L106F of the Omicron variant and L108F of the Delta variant are on the same transmembrane helix 3 (TM 3) of ORF3a, they show contrasting effects on ORF3a-mediated lysosomal damage and distribution to the cell membrane. ORF3a structural analysis showed that L108 is located in the helical interface of a heptad repeat that tolerates non-polar residues, whereas L106 is opposite to the helical interface (Fig. 5C). The AlphaFold predicted structure of TM 3 suggests that L108F appears to stabilize helical packing through enrichment of hydrophobic surfaces, and L106F does not contribute to the helical packing. This indicates that Delta-variant-specific L108F contributes to helical packing-based stabilization of ORF3a, unlike the Omicron-specific L106F mutation.

Peripheral targeting of lysosomes due to Delta variant-specific SARS-CoV-2 ORF3a

Loss of lysosomal integrity and damage are associated with lysosomal targeting to the cell membrane and exocytosis^29,52–54. SARS-CoV-2 ORF3a promotes peripheral targeting of lysosomes through the BORC-ARL8b complex, contributing to viral egress through lysosomal exocytosis. The differential impact of Delta and Omicron mutations in ORF3a on lysosomal damage prompted us to monitor their effects on lysosomal distribution and cell membrane targeting. To monitor lysosomal distribution, we quantified the dispersal of LAMP1-positive lysosomes in the cells using a semi-automated shell analysis (Fig. 6A)^55,56. The LAMP1-GFP-based immunostaining and shell analysis showed that expression of ORF3a led to the dispersion of LAMP1-positive lysosomes from the perinuclear space, as observed in the absence of ORF3a expression (Fig. 6B, C). The Delta mutants significantly augmented the dispersion of the lysosomal population towards the cell periphery when compared to ORF3a-WT. However, Omicron-specific mutations did not increase lysosomal dispersion compared to ORF3a-WT, and the T223I mutation in the Omicron lineage reduced lysosomal dispersion. To assess whether Delta variant-specific robust lysosomal distribution confers lysosomal targeting to the PM, we performed immunostaining of ORF3a-expressing cells with a LAMP1 antibody, without permeabilizing the cells, to visualize the cell-surface localization of LAMP1 (Fig. 6D, E). Expression of ORF3a-WT promoted PM localization of LAMP1 compared to the mock cells. The Delta mutants of ORF3a (S26L, DW, and L108F) significantly increased the cell-surface localization of LAMP1 compared to ORF3a-WT and Omicron mutants of ORF3a (T223I and L106F), suggesting the robust membrane targeting of lysosomes by Delta variant ORF3a. These observations further suggest that Delta variant-specific ORF3a mutations enable robust lysosomal distribution to the periphery compared to Omicron mutants. Also, we examined the effect of the variant-specific mutations on the recruitment of BORC complex to LAMP1 positive lysosomes. We observed that the expression of ORF3a-WT and its mutants modulated the positioning of BORC associated lysosomes but did does not affect the BORC complex recruitment to lysosomes (Fig. S3).

Fig. 6. Peripheral targeting of lysosomes is enhanced by SARS-CoV-2 ORF3a Delta mutations.

A Schematic diagram showing the workflow of shell analysis. Cell outlines were traced and then iteratively shrunk by 1.2 μm to generate five shells per cell. The lysosomal thresholded area was calculated after generation of each shell. B Confocal microscopy imaging of HeLa cells co-expressing ORF3a-WT and mutants with LAMP1-GFP, stained for ORF3a (Strep-Tag). Scale bar, 15 μm. C Quantification of peripheral targeting of lysosomes (LAMP1-GFP) in HeLa cells expressing ORF3a-WT and mutants. ***P = 0.0007, **P = 0.0017 (Mock vs ORF3a-WT), **P = 0.0011 (ORF3a-WT vs DW), *P = 0.0155, ^#P = 0.0284, ns – not significant (one way ANOVA, n = 60 cells across three independent experiments). Data shown as box and whiskers. Whiskers indicate minimum to maximum range and ‘+’ denotes the mean. ‘*’ indicates a significant increase and ‘#’ indicates a significant decrease. D Confocal microscopy imaging of HeLa cells expressing ORF3a-WT and mutants, stained for membrane LAMP1 without permeabilization. Scale bar, 20 μm in both complete image and zoomed insets. E Quantification of membrane LAMP1 mean intensity in HeLa cells expressing ORF3a-WT and mutants. ****P < 0.0001, ns – not significant (one-way ANOVA, n = 45 frames across three independent experiments). F SARS-CoV-2 titers in cell supernatants of SARS-CoV-2 infected A549-ACE2 cells (2 MOI) ectopically expressing ORF3a WT and mutants, measured by plaque assay. ^####P < 0.0001, ns – not significant (one-way ANOVA, n = 3). Data shown are mean ± SEM. G Immunoblot analysis of SARS-CoV-2 nucleocapsid (N) protein and GAPDH in SARS-CoV-2 infected A549-ACE2 cells expressing ORF3a-WT and mutants.

Delta-VOC specific ORF3a variations marginally affect SARS-CoV-2 viral titers

SARS-CoV-2 ORF3a mediates extracellular viral release in infected cells through the lysosomal exocytosis pathway²⁶. We investigated whether the Delta mutations in ORF3a confer efficient viral release by targeting lysosomes to the plasma membrane. SARS-CoV-2 infection in A549-ACE2 cells ectopically expressing ORF3a showed reduced viral titers compared to mock transfected cells. SARS-CoV-2 infection in cells expressing Delta and Omicron mutants of ORF3a did not significantly alter viral release compared to ORF3a-WT, as measured by plaque assay-based viral titers (Fig. 6F). Immunoblotting analysis of SARS-CoV-2 nucleocapsid protein indicated no apparent difference in intracellular viral replication (Fig. 6G). These observations suggest that Delta- and Omicron-specific SARS-CoV-2 ORF3a mutations did not significantly alter viral release, indicating possible host-directed effects.

Delta-specific mutations aggravate ORF3a-induced lysosomal damage-mediated cell death

Excessive respiratory epithelial damage and lung tissue necrosis underlie the Delta variant-associated pathological outcome^6,8,9. We asked whether the Delta variant mediated lysosomal damage promotes host cellular damage associated with inflammation. Lysosomal damage triggers the activation of either apoptosis or programmed necrosis as a cellular homeostasis mechanism^{27,28,57–59}, and thus, we tested whether Delta-variant-specific lysosomal damage induces cell death activation. Real-time analysis of cell death using cleaved caspase-3/7 specific fluorophores (apoptosis marker) showed that LLOMe treatment, a lysosomal damage inducer⁶⁰, preferentially triggered apoptotic cell death, and ORF3a expression augmented LLOMe-induced apoptosis (Fig. 7A, B). Lysosomal damage releases lysosomal luminal Cathepsins (Cathepsins B and D), which promote cell death^28,30,58,61. Cathepsin inhibitor cocktail treatment inhibited LLOMe- and LLOMe+ORF3a-mediated cell death, further confirming ORF3a-mediated lysosomal damage in promoting cell death (Fig. 7C). Immunoblotting analysis showed cleaved caspase-3 after LLOMe treatment, and the levels were increased upon ORF3a expression (Fig. 7B). ORF3a expression induces marginal necroptosis upon LLOMe treatment. Pan-caspase inhibitor (zVAD) treatment significantly reduced LLOMe and LLOMe+ORF3a induced apoptosis. zVAD treatment promoted MLKL phosphorylation in LLOMe and LLOMe+ORF3a cells, suggesting activation of necroptosis upon caspase inhibition (Fig. 7A, B). Interestingly, after zVAD treatment, ORF3a-expressing cells treated with LLOMe showed higher levels of Sytox green-positive (lytic cell death indicator) cells than LLOMe-treated cells alone, suggesting ORF3a-mediated necroptosis activation (Fig. 7A). Delta variants of ORF3a significantly enhanced ORF3a-mediated apoptotic cell death, and Omicron mutants did not show apparent differences in cell death levels compared to the ORF3a-WT (Fig. 7D). Upon treatment with necroptosis trigger (TNF+SMAC mimetic+zVAD), Delta variant-specific ORF3a mutations, particularly DW and L108F, induced significantly increased necroptosis compared to Omicron mutations (Fig. 7E, F). In addition, Delta variant-specific ORF3a mutations showed significantly increased LLOMe-mediated cell death, whereas Omicron mutations did not (Fig. 7G). Cathepsin inhibitor cocktail treatment inhibited Delta variant-induced LLOMe-mediated cell death (Fig. S4). These observations indicate that ORF3a promotes lysosome damage-mediated apoptosis and necroptosis. Lysosomal damage induced by Delta variant-specific ORF3a mutants promotes apoptotic and necrotic cell death, whereas Omicron-specific ORF3a mutations do not.

Fig. 7. Delta-specific SARS-CoV-2 ORF3a mutants promote lysosomal damage-mediated apoptotic and necrotic cell death.

A Real-time cell death measurement by Caspase 3/7 and Sytox green staining of HT-29 cells after treatment with LLOMe, LLOMe + zVAD, or zVAD in the presence or absence of ORF3a-WT expression. ****P < 0.0001, ^####P < 0.0001, and **P = 0.0030, ns – not significant (two-way ANOVA, n = 3). Data shown are mean ± SEM. ‘*’ indicates a significant increase and ‘#’ indicates a significant decrease. B Immunoblot analysis of ORF3a-WT(Strep-Tag), full-length Caspase-3, cleaved Caspase-3, pMLKL, MLKL and GAPDH in lysates of HT-29 cells after treatment with LLOMe, LLOMe + zVAD, or zVAD in the presence or absence of ORF3a-WT expression. ‘fl’ indicates full-length and ‘cl.’ Indicates cleaved. C Real time cell death measurement by Caspase 3/7 staining of HT-29 cells expressing ORF3a-WT and treated with LLOMe, in absence or presence of Cathepsin inh (Cathepsin inhibitor cocktail), ^####P < 0.0001, (two-way ANOVA, n = 3). Data shown are mean ± SEM. ‘#’ indicates a significant decrease. D Real time cell death measurement by Caspase 3/7 staining of HT-29 cells expressing ORF3a WT and mutants. **P = 0.0085, *P = 0.0109, ns – not significant (ORF3a-WT vs Mock, DW, T223I, L106F) (RM one-way ANOVA, n = 3). Data shown are mean ± SEM. E Real time cell death measurement by Sytox green staining of HT-29 cells expressing ORF3a-WT and mutants after treatment with TNF, zVAD and SMAC mimetic. **P = 0.0018 (ORF3a-WT vs DW), **P = 0.0023 (ORF3a-WT vs L108F), ns – not significant (two-way ANOVA, n = 3). Data shown are mean ± SEM. F Immunoblot analysis of lysates of HT-29 cells expressing ORF3a-WT and mutants after treatment with TNF, zVAD and SMAC mimetic, showing expression of ORF3a (Strep-Tag), pMLKL, MLKL and GAPDH. G Real time cell death measurement by Caspase 3/7 staining of HT-29 cells expressing ORF3a WT and mutants in presence and absence of LLOMe, ****P < 0.0001 (LLOMe vs ORF3a-WT+LLOMe), **P = 0.0035 (ORF3a-WT+LLOMe vs DW+LLOMe), *P = 0.0495 (ORF3a-WT+LLOMe vs S26L+LLOMe), ns – not significant (two-way ANOVA, n = 3). Data shown are mean ± SEM.

Discussion

The COVID-19 pandemic led to the evolution of SARS-CoV-2 VOCs. Amongst the VOCs, the SARS-CoV-2 Delta variant (B.1.617.2) emerged as one of the most pathogenic lineages with severe clinical outcomes^1,2,6,8,9. The Delta-dominant COVID-19 wave showed high hospitalization rates and mortality as compared to subsequent variants. In contrast, the Omicron variant (BA.1 and BA.2) and its sublineages have been associated with milder clinical symptoms and reduced mortality rates despite higher transmissibility^1,3,4,11. While enhanced transmissibility and immune evasion have been studied, the virologic factors contributing to the high pathogenicity of the Delta variant compared to the other VOCs remain incompletely understood. Our observations in this study show that the distinct mutational patterns in SARS-CoV-2 ORF3a are associated with pathogenic VOCs evolution. The Delta variant-specific host cellular lysosomal disruption is associated with inflammation-associated pathogenicity. ORF3a is the largest accessory protein conserved in human and bat Sarbecoviruses⁶². Phylogenetic analysis suggests that mutations in the Orf3a gene define distinct evolutionary trajectories in the Delta and Omicron variants, potentially indicating the differential host mechanisms associated with disease severity.

Coronaviruses, including SARS-CoV-1 and SARS-CoV-2, exit cells through lysosomal exocytosis⁶³. However, unlike SARS-CoV-2 ORF3a, SARS-CoV-1 ORF3a does not inhibit autophagosome fusion with lysosomes³⁵. This difference suggests that the SARS-CoV-2-mediated disruption of autophagy and lysosomal dynamics might have additional roles, along with viral egress, possibly regulating host cellular physiology without affecting viral titers. In line with this explanation, we find that Delta variant-specific ORF3a mutations promote robust lysosomal damage and their targeting to the cell membrane, however, with no differences in viral titers. Instead, this lysosomal damage led to cellular damage through apoptosis and necroptosis, which might contribute to the robust necrotic tissue phenotypes observed during Delta variant infection. Notably, the Omicron-specific ORF3a mutations caused less lysosomal damage and did not affect the SARS-CoV-2 viral release.

Our observations suggest that Delta-specific mutations in ORF3a enhance lysosomal damage, whereas Omicron-specific mutations do not. Our structural analysis and oligomerization-based studies of ORF3a with Delta- and Omicron-specific mutations indicate an unexpected dimer- and oligomer-stabilizing role for Delta-specific mutations. Notably, the ORF3a transmembrane helical packing and formation of a potential non-native disulfide at the dimer-dimer interface underlie the ORF3a stabilization due to Delta mutations. This stabilization likely facilitates increased accessibility and a stable interaction of ORF3a with the HOPS complex at the lysosomal membrane³⁴, leading to ORF3a-mediated lysosomal damage and cathepsin release.

Uncontrolled activation of cell death is the leading cause of dysregulated inflammation and tissue damage in patients with SARS-CoV-2 infection^8,16–19,64. Lysosomal damage leads to lysosomal membrane permeabilization (LMP), releasing hydrolytic cathepsins into the cytosol, which promote the activation of multi-modal cell death pathways^{27,28,30,57–59,61}. Our studies reveal that ORF3a-mediated lysosomal damage is associated with the activation of regulated cell death, suggesting that SARS-CoV-2 evolved to orchestrate host cell death. The Delta-specific mutations in ORF3a exacerbated lysosomal damage, leading to enhanced apoptotic and necroptotic cell death, compared with WT and Omicron. The heightened pathogenicity of the SARS-CoV-2 Delta variant is attributed to enhanced cytopathic effects and excessive cell death activation^6–10. Another study demonstrates that Delta variant infection enhances Z-RNA-ZBP1-mediated necroptosis, potentially contributing to disease severity⁸. Our findings have identified a distinct and possible parallel pathway for cell death activation through ORF3a-mediated lysosomal damage, which is enhanced by Delta-specific mutations. This might be one of the molecular cues that explain the altered pathogenicity profiles of Delta and Omicron. Although Spike-mediated tropism and host immune responses determine the disease outcome during SARS-CoV-2 infection, our observations suggest that the non-spike proteins, particularly ORF3a, might potentially determine tissue damage and disease pathogenesis. Overall, our work reveals a possible mechanistic link between SARS-CoV-2 ORF3a variations and aberrant cellular alterations by pathogenic variants. Future studies using animal infection models are necessary to substantiate the effect of ORF3a mutations on pathogenic profiles at the physiological level.

Methods

Cell culture

All the cell lines used in this study (A549-ACE2, HT-29, HeLa, and Vero-E6) were cultured at 37 °C and 5% CO₂ inside humidified incubators. The cells were maintained in DMEM medium supplemented with 10% v/v FBS, antibiotic-antimycotic, and non-essential amino acids (NEAA). The cells were subcultured using PBS and trypsin. STR profile analysis was used to authenticate all cell lines obtained from NCCS (National Centre for Cell Science, Pune, Maharashtra, India). Cell lines were tested to be negative for mycoplasma contamination.

HT-29, HeLa, HEK-293T and Vero-E6 were obtained from NCCS cell repository. A549-ACE2 cells were kind gifts from Dr. Shashank Tripathi’s Lab (Indian Institute of Science, Bengaluru)

Mutational heatmaps

The information about mutations in SARS-CoV-2 proteins across VOCs Alpha, Beta, Gamma, Kappa, Delta and Omicron, was obtained from Covariant (Emma B. Hodcroft. 2021. “CoVariants: SARS-CoV-2 Mutations and Variants of Interest.” https://covariants.org/) and Nextstrain (https://nextstrain.org/). The heatmaps were generated using GraphPad Prism 10.1.1 software.

Phylogenetic analysis

Whole genome (WG) sequences were downloaded from the public repository BV-BRC submitted from four countries—France, Germany, India, and the UK—for six variant lineages: Alpha (B.1.1.7), Gamma (P.1), Kappa (B.1.617.1), Delta (B.1.617.2) and Omicron (BA.1 and BA.2). Ten wholly annotated sequences for each variant from each country were downloaded. After removing sequences with gaps larger than 10% of the sequence length, these sequences were aligned using the MUSCLE algorithm in AliView v1.28 against the parent Wuhan sequence (GenBank accession: NC_045512.2)⁶⁵. The WG alignment was trimmed according to the annotation of the Wuhan sequence to derive spike and orf3a gene alignments. This yielded 94 WG and spike sequences, and 26 unique orf3a sequences, respectively, including the Wuhan sequence.

The phylogeny between the aligned WG, spike, and orf3a sequences was inferred separately as follows. Using IQ-TREE v2.2, the maximum likelihood tree was constructed with 1000 bootstrap replicates⁶⁶. To identify the nucleotide substitution model for each case, jModelTest tool of IQ-TREE was used, and the one with the lowest Akaike Information Criterion was used to construct the tree. The trees for WG and spike sequences were generated with this approach. However, this method failed to identify a robust phylogenetic relationship between orf3a sequences because there was little diversity in this gene between variants. So, the phylogenetic relationship between orf3a sequences was inferred as follows.

Firstly, the molecular clock model that could be used was identified. For that, the root-to-tip divergence of the sequences in the maximum likelihood trees of WG, spike, and orf3a were obtained using TempEst v1.5.3⁶⁷. For the WG and spike sequences, the correlation coefficient was high with a large R², offering strong support for the strict molecular clock model for WG and spike. The intercept on the time axis was very close to the year 2020, which is the time around which SARS-CoV-2 first emerged. These observations provided confidence in our analysis. The fit to orf3a data had wide variability. However, since the fit lines for orf3a and WG sequences are similar, a strict molecular clock model was employed to obtain the orf3a phylogeny.

The lack of diversity across gene sequences of orf3a would result in a topology with low statistical support. To avoid that problem, the phylogeny was inferred using consensus sequences. We obtained a consensus orf3a sequence for each variant using EMBOSS Cons (https://www.ebi.ac.uk/jdispatcher/msa/emboss_cons). Six unique orf3a sequences were obtained, each representative of a variant. These sequences were aligned using MUSCLE and the phylogenetic tree was constructed using a strict molecular clock model in BEAST v1.10.4 from the alignment⁶⁸. GTR+I nucleotide substitution model was employed, and 10⁸ samples were generated using the Markov Chain Monte Carlo algorithm of BEAST, with 10% of them discarded as burn-in samples. The maximum clade credibility tree was identified using the retained samples.

Structural modelling of ORF3a mutants and protein visualization

SARS-CoV-2 ORF3a structure (PDB ID: 6XDC) (40-275 residue position) was visualized using UCSF ChimeraX 1.8 protein model visualization tool⁶⁹. The full three-dimensional structure of the SARS-CoV-2 ORF3a was predicted using the Structure Prediction module in UCSF ChimeraX 1.8, which uses the full-length amino acid sequence as input in FASTA format. Structure prediction was performed using the ColabFold backend, which uses AlphaFold2 neural networks to perform accelerated multiple sequence alignments and generated predicted structures with reduced computational requirements. Default parameters of the module were used. The top-ranked model was selected based on the predicted local distance difference test (pLDDT) confidence scores. The structures of the full length ORF3a mutants were predicted using the same approach. The structures were aligned using the Matchmaker module of UCSF ChimeraX 1.8. The predicted structures were visualized, analyzed and annotated in UCSF ChimeraX 1.8. A tetrameric ORF3a model was constructed in UCSF ChimeraX 1.8 by duplicating the dimer structure and arranging the second dimer relative to the first one using rotation and translation.

Plasmids and constructs

pLVX-EF1alpha-SARS-CoV-2-ORF3a-2xStrep-IRES-Puro plasmid was a kind gift from Prof. Nevan Krogan’s lab (University of California, San Francisco). All the variant-specific mutations (S26L, D155Y, W131C, L108F, L106F, T223I, G188C and Y211H) were introduced in the backbone of pLVX-EF1alpha-SARS-CoV-2-ORF3a-2xStrep-IRES-Puro by site-directed mutagenesis using Inverse PCR approach. LAMP1-GFP, LAMP1-mRFP, Galectin-3-GFP, mCherry-GFP-LC3, Galectin-8-GFP reporter plasmids were kind gifts from Dr. Prasad Tammineni’s lab (University of Hyderabad).

Transient transfection and cell stimulations

For transient expression of proteins, plasmids were transfected into specific cell lines using Xfect (TakaraBio; 631318) or Lipofectamine 2000 (Thermo Fisher Scientific; 11668-019) transfection reagents in reduced serum media (Opti-MEM). We used 0.4 μl of Xfect for 1 μg of DNA and 1.5 μl of Lipofectamine 2000 for 1 μg of DNA (1:1.5). The transfection efficiency was 80%. All the treatments were performed 6 h post transfection.

For autophagy inhibition, HeLa cells were treated with 20 nM Bafilomycin A1 (Sigma Aldrich; B1793) for 12 h. For apoptosis and necroptosis activation, HT-29 cells were treated with 30 μM Z-VAD(OMe)-FMK (Cayman Chemical; 14463) and 500 nM SMAC mimetic SM-164 (MedChemExpress; HY-15989). Two hours after treatment, 100 ng/mL recombinant human TNF-α (Abclonal) was added. For LLOMe-induced cell death and lysosomal damage, both HeLa and HT-29 cells were treated with 2 mM LLOMe (Cayman Chemicals; 16008). For inhibition of LLOMe-induced cell death, HT-29 cells were treated with a combination of 10 μg/ml Pepstatin A (Sigma Aldrich; P5318) and 20 μM E-64d (Sigma Aldrich; E8640). To induce LLOMe mediated necroptosis, LLOMe treated HT-29 cells were treated with 30 μM Z-VAD(OMe)-FMK (Cayman Chemical).

Immunoblotting analysis

The cells were lysed using NP40-lysis buffer and sample-loading buffer containing SDS and BME was added to the lysates. The lysates were resolved in SDS-PAGE gels (8%, 10% and 12%), and the gels were then transferred onto PVDF membrane. The membranes were blocked in 5% skimmed milk at room temperature for 1 h and then incubated with primary antibodies at 4 °C overnight. The membranes were washed using Tris-buffered saline with tween 20 (TBST), incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature, and then washed thrice with TBST. The indicated antibodies were diluted in skimmed milk at indicated ratios (primary antibodies: 1:1000; secondary antibodies: 1:5000). Blots were developed using Immobilon Forte Western HRP substrate (Millipore; WBLUF0500) and visualized using iBright1500 (Invitrogen, Thermo Fisher Scientific) or Image Quant 800 (Cytiva, Amersham).

The following primary antibodies were used: Mouse monoclonal Anti-Strep-Tag (Qiagen; 34850), Rabbit monoclonal Anti-p-MLKL (Human-specific) (Cell Signaling Technology, 91689), Rabbit polyclonal Anti-Caspase-3 (Cell Signaling Technology, 9662), Mouse monoclonal Anti-GAPDH (Invitrogen, MA5-15738), Mouse monoclonal Anti-SARS-CoV-2 N protein (Cell Signaling Technology; 33717), Rabbit polyclonal Anti-LC3A/LC3B (Invitrogen, PA1-16931)

The following secondary antibodies were used: Peroxidase AffiniPure™ Goat Anti-Rabbit IgG (Jackson ImmunoResearch; 111-035-047) and Peroxidase AffiniPure™ Rabbit Anti-Mouse IgG (Jackson ImmunoResearch; 315-035-047).

Immunofluorescence and imaging analysis

For immunofluorescence assays, HeLa cells were transfected with ORF3a-WT and mutant plasmids alone or in combination with LAMP1-GFP, LAMP1-GFP, LAMP1-mRFP, Galectin-3-GFP, mCherry-GFP-LC3 depending on the assay. In virus-based experiments, A549-ACE2 cells were infected with SARS-CoV-2. Twenty to twenty-two hours after transfection or 72 h post infection, the cells were fixed with 4% paraformaldehyde in 1× PBS (ChemCruz; sc-281692) for 10 min at room temperature. Following fixation, cells were permeabilized using 0.1% Triton X-100 for 10 min at room temperature. Three percent BSA (GBiosciences; RC1021) was used to block non-specific binding. Cells were incubated with following antibodies for 2 h at room temperature: anti-Strep Tag (1:100; Qiagen; 34850), anti-Galectin-3 (1:250; Cell Signaling Technology; 87985), anti-SARS-CoV-2-Nucleocapsid protein (1:200; Cell Signaling Technology; 33717), anti-LAMP1 antibody (1:300; Merck; MABC1108), anti-LAMP1 antibody (1:500; Cell Signaling Technology; 9091S). The cells were then subjected to immunostaining with Alexa Fluor 488-conjugated anti-mouse IgG (Invitrogen; A-11001), Alexa Fluor 568-conjugated anti-mouse IgG (Invitrogen; A-11004), Alexa Fluor 647-conjugated anti-mouse IgG (Invitrogen; A-21235) for 1 h at room temperature. After each step, the cells were washed thrice with 1× PBS. The cells were counterstained with DAPI. All images were acquired using Olympus FV 300 Confocal microscope system. Galectin-8 imaging was performed using EVOS^TM M7000 Imaging System (Thermo Fisher Scientific). BORC-S5 imaging was performed using Stellaris Confocal microscope system (Leica). For plasma membrane targeting assay, the same protocol was followed without the permeabilization step.

To quantify object count, images were subjected to thresholding and ‘Analyze particles’ module of Fiji was used. Mean fluorescence intensity was also obtained from ‘Measure’ module in Fiji. For mCherry-GFP-LC3 assay, ‘AND’ operation of image calculator in Fiji was used to colocalize images and the colocalized objects were counted using the above-mentioned approach. Percentage autophagy inhibition was calculated as the percentage of non-acidified LC3-positive vesicles (mCherry+GFP+ puncta/total mCherry+ puncta × 100). To quantify the peripheral distribution of LAMP1 structures, a shell-based outline was used. Cell outlines were traced in Fiji by boosting the LAMP1 signal to visualize cytosolic fluorescence. Optimum threshold was applied to the traced cells to eliminate the background and the total cellular LAMP1 signal was measured. The cell outline was iteratively reduced by 1.2  μm to produce a total of five shells, shell 1 demarcating the perinuclear region and shell 5 representing the outermost edge of the cell. The LAMP1 signal intensity within each shell was measured. Since in mock-transfected cells, the largest fraction of their lysosomes remained within shells 1 and 2, the area under shell 1–2 was marked as perinuclear and the area under shells 3–5 was marked as peripheral. The peripheral/total LAMP1 signal was computed as the fraction of LAMP1 signal present in outer shells 3–5 relative to total LAMP1 signal. A custom Fiji macro for semi-automation of shell analysis to measure lysosomal positioning was used as previously described and made publicly available by the authors⁵⁶ (https://github.com/AmraSaric/LysoPosMacro).

SARS-CoV-2 infection and virus titer assays

For viral release assay, constructs expressing ORF3a-WT, DW, L108F, T223I and L106F mutants were transfected into A549-ACE2 cells. Twenty-four hours after transfection, cells were infected with the SARS-CoV-2 virus (Isolate Hong Kong/VM20001061/2020; NR-52282; BEI Resources, NIAID, NIH) at an MOI of 0.5 and 2. Virus inoculum was prepared in 2% FBS supplemented DMEM and was added to the cells, followed by incubation of the cells for 1 h at 37 °C for allowing virus adsorption and entry into cells. After 1 h, DMEM supplemented with 2% FBS was added to cells. Forty-eight hours post infection, supernatant was collected to perform plaque assay. Whole cell lysates were collected for immunoblotting analysis of SARS-CoV-2 N-protein.

Viral titers were estimated plaque assay in Vero-E6 cells. At 90–95% confluency after seeding, 150 μL of increasing dilutions of virus supernatant collected from SARS-CoV-2 infected A549-ACE2 cells transiently expressing ORF3a-WT and mutant proteins were added to the cells and incubated at 37 °C. After 1 h of incubation, the virus inoculum was removed and the cells were overlaid with 0.6% Avicel (Dupont, RC-591) prepared in DMEM containing 2% FBS. The cells were fixed with 4% paraformaldehyde after 48 h of incubation at 37 °C. Following fixation, the cells were stained with Crystal Violet (Sigma Aldrich; C0775) for 20 min, and plaques were visualized.

For VOC induced lysosomal damage assay, infection was performed in A549-ACE2 cells at 3 MOI using SARS-CoV-2 virus (Isolate Hong Kong/VM20001061/2020; NR-52282; BEI Resources, NIAID, NIH), Delta (Isolate hCoV-19/USA/PHC658/2021, Lineage B.1.617.2, Delta variant) and Omicron (Isolate hCoV-19/USA/MD-HP20874/2021, Lineage B.1.1.529; Omicron variant NR-56461, BEI Resources, NIAID, NIH) viruses using the same protocol as mentioned above. Seventy-two hours post-infection, the infected cells were fixed with 4% paraformaldehyde in 1× PBS (ChemCruz) and subjected to immunostaining with anti-Galectin-3 and anti-SARS-CoV-2 nucleocapsid antibody using above-mentioned protocol.

This study was performed according to the guidelines and in compliance with institutional biosafety guidelines (IBSC/IISc/KVS/25/2023). SARS-CoV-2 experiments were conducted at viral Biosafety level 3 facility at the Indian Institute of Science (IISc).

Oligomerization assay

HeLa cells transiently transfected with indicated clones (ORF3a WT and mutants) were lysed with NP-40 lysis buffer (1.0% NP40, 50 mM HEPES, pH 8.0, 150 mM NaCl with protease inhibitor and phosphatase inhibitor cocktails) containing 2 mM Pierce Premium Grade DSP (Thermo Fisher Scientific; PG82081). Whole cell lysates were collected and incubated at room temperature for 1 h. 1 M Tris-HCl was added to stop the reaction and samples were incubated for 10 min. Whole cell lysates were centrifuged at 8000 rpm at 4 °C for 15–20 min to separate insoluble fractions (pellet) and soluble fractions. Insoluble fractions in the form of pellets were dissolved in NP-40 lysis buffer. Soluble and insoluble lysates were mixed with 4× sample buffer with or without β-mercaptoethanol (BME) and immunoblotting analysis was performed to monitor the oligomeric status of ORF3a-WT and mutants.

Real-time cell death analysis

Real-time cell-death assays were performed using a two-color IncuCyte S3 Live-Cell Analysis instrument (Sartorius). HT-29 cells were seeded in 12-well plates and transfected with ORF3a WT and mutants or treated with lysosomal damage triggers, cell death-stimulating or inhibiting agents. Dead cells were stained with 20 nM Sytox Green (Thermo Fisher Scientific; S7020), a cell-impermeable DNA-binding fluorescent dye that enters dying cells after membrane permeabilization and emits green, fluorescent signal. For quantification of apoptotic cells, 2.5 μM Caspase-3/7 Green Dye (Sartorius; 4440) was used. The Incucyte Caspase-3/7 dye is a DNA intercalating dye that is activated upon cleavage by Caspase3/7. The resulting images were analyzed for quantification of fluorescent signal using IncuCyte S3 software, which provides a count for the Sytox or Cleaved Caspase-3/7 Green-positive cells. This data was plotted using GraphPad Prism 10.1.1 software.

Statistics and reproducibility

Statistical significance was determined by two-way ANOVA or one-way ANOVA followed by post hoc multiple-comparisons test using GraphPad Prism 10.1.1 software. All analyses were performed for a minimum of n = 3. One-way ANOVA was followed by Dunnett’s multiple comparisons test and Fisher’s LSD. Two-way ANOVA was followed by Tukey’s multiple comparisons test and Šidák’s multiple comparison test. Exact sample sizes for the analyses are provided in the figure legends. Detailed P values for each statistical analysis can be found at the end of the respective figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplemental Information ^{(6.7MB, pdf)}

42003_2026_10801_MOESM2_ESM.pdf ^{(27.9KB, pdf)}

Description of Additional Supplementary Files

Supplementary data 1 ^{(126.1KB, xlsx)}

Reporting summary ^{(3.1MB, pdf)}

Peer Review File ^{(982.1KB, pdf)}

Author contributions

S.K. conceptualized the study. A.A.D., P.T., and S.K. designed the experiments and methodologies. A.A.D., S.M., V.R.A., and A.C. performed cell and viral culture, biochemical experiments, and microscopy-based experiments and analyses. S.K., A.A.D., and D.J. annotated SARS-CoV-2 protein variations across VOCs. B.V. and N.M.D. performed phylogenetic analyses and assisted in the interpretation of the data. S.K. and A.A.D. wrote the manuscript; all authors contributed to manuscript editing. S.K. supervised the study. S.K. and P.T. provided guidance and funding.

Peer review

Peer review information

Communications Biology thanks Rajendra Karki and the other anonymous reviewer(s) for their contribution to the peer review of this work. Primary handling editors: Si Ming Man and Dario Ummarino. A peer review file is available.

Funding

S.K. research work is supported by funding from the Scheme for Transformational and Advanced Research in Sciences (STARS)-MoE (MoE-STARS/STARS-2/2023-0464), the Indian Council of Medical Research (ICMR) (EMDR/SG/9/2023-2019), (IIRPSG-2024-01-01943), the Department of Biotechnology (DBT) (BT/PR45145/COT/142/24/2022), and the Anusandhan National Research Foundation (ANRF) (ANRF/ARG/2025/007263/LS), for supporting research in his laboratory. P.T. thanks the DBT/Wellcome India Alliance intermediate fellowship (IA/I/23/1/506758), for supporting research in his laboratory. A.A.D. acknowledges the Hindustan Unilever Limited women in STEM fellowship.

Data availability

The numerical Source data for the plotted graphs are provided in the Supplementary data file (Supplementary Data 1). Supplementary Figs. S5–S10 show the uncropped blots for Figs. 4D, 5A, 6G, 7B, F, and S2A. All other data are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10801-z.