Several groups have reported novel nucleotide insertions in the S gene of SARS-CoV-2, as indicated by a multiple sequence alignment for the S protein sequences of different coronaviruses, such as a polybasic site “PRRA” produced by a 12-nucleotide acquisition at the S1-S2 boundary through multiple host-species adaptations (Andersen et al., 2020; Walls et al., 2020). However, S protein sequence alignments between SARS-CoV-2 and SARS-CoV showed the possibility of the insertions “NSPR” (Zhang C. et al., 2020) and “SPRR” (Xie et al., 2020) at the S1-S2 boundary. It has previously been reported that the sequence insertion at the S1-S2 boundary constitutes a furin cleavage site (Ord et al., 2020; Peacock et al., 2021). A comprehensive understanding of the consequence of the sequence insertion at the S1-S2 boundary is still missing, possibly because research focused on understanding the differences in the pathogenicity of the different SARS-CoV-2 variants and subvariants, which emerged rapidly. To determine whether the earlier SARS-CoV-2 isolate (USA/WA-CDC-WA1/2020 isolate, GenBank accession no. MN985325) has multiple novel sequence insertions in the S protein compared to SARS-CoV (Urbani strain, GenBank accession no. AY278741), we aligned the S protein sequences of both viruses using a constraint-based alignment tool for multiple protein sequences (COBALT) (Papadopoulos and Agarwala, 2007). We did not use MER-CoV for comparison because there is only 40% similarity between SARS-CoV-2 and MERS-CoV (Hu et al., 2020). Similar to a previous report (Zhang C. et al., 2020), we found sequence insertions (IS) in the SARS-CoV-2 S protein at four independent positions: IS1 “GTNGKTR,” IS2 “YYHK,” IS3 “HRSY,” and IS4 “NSPR” (Figures 1A, B). To determine whether any of these sequence insertions constituted or resembled any protein motifs, such as an NLS, we analyzed the SARS-CoV-2 S protein in silico with the PSORT II web portal for NLS prediction (Nakai and Horton, 1999). We found that the SARS-CoV-2 glycoprotein contained an NLS of the “pat7” motifs, one of the three NLS motifs described above (Supplementary Figure S1). To our surprise, the NLS motif “PRRARSV” was present at the proposed polybasic site and was due to the fourth sequence insertion “NSPR” (Figures 1A, B; Supplementary Figure S1). As expected, the NLS in the S protein was unique to SARS-CoV-2 among human pathogenic beta-coronaviruses, as neither the SARS-CoV S protein nor the MERS-CoV S protein has an NLS (Supplementary Figure S1).

Although viral glycoprotein nuclear translocation is rare, NLS-driven protein nuclear translocation has already been established in different viral infections (Ozawa et al., 2007; Boisvert et al., 2014). Thus, it is important to determine whether the SARS-CoV-2 S protein translocates into the nucleus in addition to its canonical cell surface localization through the ER-Golgi pathway. We hypothesized that the S protein could translocate into the nucleus in SARS-CoV-2-infected cells via the identified NLS motif (Rowland et al., 2005; Ozawa et al., 2007; Boisvert et al., 2014). We infected highly differentiated pseudostratified airway epithelial cells (which mimics in vivo human airway epithelium) with SARS-CoV-2 at a multiplicity of infection (MOI) of 0.1 for 4 days. First, we confirmed the presence of S mRNA and S protein in a 5 μm section of formalin-fixed paraffin-embedded SARS-CoV-2-infected cells by RNAscope and immunofluorescence analysis. Despite the rarity of viral mRNA (or even positive-strand RNA virus genome) to be nuclear (Whittaker and Helenius, 1998; Vargas et al., 2005), a recent study showed that SARS-CoV-2 mRNA accumulates in the nucleus of infected cells (Addetia et al., 2021). Our results showed that SARS-CoV-2 S mRNA was nuclear (Figure 2 left panel and merged images in the right panel). To confirm the physical apposition between S mRNA and the nucleus by comparing their distributions in fluorescent images, we used the spot-to-spot colocalization function in Imaris image analysis software (Oxford Instruments). We found that S mRNA was nuclear and abundant in the cytoplasm (Supplementary Figure S2, top panel, three donors). To avoid image artifacts, we imaged multiple independent slides of SARS-CoV-2-infected airway epithelium (from three independent donors) using at least two different high-end confocal microscopes. Additionally, we used at least two different image processing strategies to determine nuclear localization. Based on high-resolution imaging, we determined the subcellular distribution of S mRNA at near single-molecule and single-cell levels (Figures 2, 3A; Supplementary Figures S2, S3). Importantly, we were able to determine S mRNA nuclear translocation not only inside the nucleus but also on the nuclear surface (Figures 3A, C; Supplementary Figure S3). The determination of S mRNA distribution and abundance showed that S mRNA subcellular localization spans from the inside and outer surface of the nuclear membrane to everywhere in the cytoplasm. We found that almost 90% of S mRNA was distributed in the cytoplasm, which was expected, as SARS-CoV-2 transcription and replication occur in the cytoplasm (Figures 3A, C; Supplementary Figure S3). Interestingly, less than 10% of S mRNA was detected at the nuclear surface, which could explain the transitionary stage of S mRNA before it enters the nucleus or the novel transnuclear-membrane translocation of S mRNA, which was examined and described later (Figures 3A, C; Supplementary Figure S3). In approximately 1% of instances, S mRNA successfully translocated into the nucleus (Figures 3A, C; Supplementary Figure S3). The nuclear translocation of S mRNA is highly unusual because there have been few previous reports of S mRNA nuclear translocation and no information on the mechanism of nuclear translocation. However, we have explored how S mRNA could translocate into the nucleus.

We investigated whether the S protein translocated into the nucleus in the SARS-CoV-2-infected airway epithelium. Consistent with the S mRNA data, we found that the S protein translocated into the nucleus and was abundant on the cellular surface through the cytoplasmic ER-Golgi pathway (Figure 2, middle panel and merged images in the right panel). Based on high-resolution imaging, we determined that S protein nuclear translocation included both the inside of the nucleus and the nuclear surface (Figures 3B, C; Supplementary Figure S3). Similar to S mRNA quantification, we were able to quantify the distribution and abundance of S protein in the infected airway epithelium. We found that the S protein was distributed inside and in the outer membrane of the nucleus, the cytoplasmic ER-Golgi and the cell surface. We did not determine what portion of the S protein was localized inside and outside the cell surface. However, we quantified the subcellular distribution of the S protein inside the nucleus, outside the surface of the nucleus, and in the cytoplasm, which included cell surface expression because the S protein is a type 1 transmembrane glycoprotein. We found that approximately 75% of the S protein was distributed in sites other than the nucleus, including the cell surface and the cytoplasm, which was expected, as S protein translation and protein processing occur in the cytoplasm and via cytoplasmic ER-Golgi pathway, respectively (Figures 3B, C; Supplementary Figure S3). Interestingly, approximately 15% of the S protein was detected at the nuclear surface, which could explain the S protein transitionary stage before entering the nucleus or a novel transnuclear-membrane translocation of S protein, which was examined and described later (Figures 3A, C; Supplementary Figure S3). Interestingly, we found that a higher percentage of total S protein translocated into the nucleus than S mRNA (Figures 3A–C; Supplementary Figure S3). Although viral type-1 transmembrane glycoprotein translocation into the nucleus is rare, the NLS in the S protein is responsible for nuclear translocation. It was apparent that NLS-driven S protein nuclear translocation was SARS-CoV-2 specific, and a side-by-side infection experiment with both viruses showed that the S protein of SARS-CoV did not translocate into the nucleus (Supplementary Figure S4). As both S mRNA and S protein translocated into the nucleus, it is important to determine whether S mRNA and S protein colocalize in different subcellular sites.

While we can explain S protein nuclear translocation due to the presence of an NLS motif in the amino acid sequence, we can only hypothesize that S mRNA nuclear translocation is possible due to a direct interaction between S protein and S mRNA, which can be explained by the colocalization between them. The SARS-CoV-2 N protein is an abundant RNA binding protein that is essential for viral genome packaging (Cubuk et al., 2021). While the structural basis of N protein binding to single-or double-stranded RNA is known (Dinesh et al., 2020), there is no information about whether S protein binds to S mRNA. As we found similar intracellular distribution of both S mRNA and S protein, we hypothesized that S protein interacts with S mRNA to translocate the protein–mRNA complex to different subcellular locations, including the cytoplasm and nucleus, but not the cell surface. By examining the colocalization between the S protein and S mRNA, we could confirm the presence of the protein–mRNA complex in the SARS-CoV-2-infected airway epithelium. Here, we refer to colocalization as an association between S mRNA and S protein at different intracellular locations. Technically, two separate fluorescence molecules that emit different wavelengths of light are superimposed within an indeterminate microscope resolution. To determine whether S mRNA and S protein colocalize, we used a high-resolution imaging strategy. We quantified the colocalization on a percentage scale. We found that approximately 85% of the colocalization, which was the highest, was observed outside the nucleus (Figure 3D). These data are consistent with the previously described spatial expression data of both S protein and S mRNA (Figures 3A–C). As expected, lower and the lowest percentages of colocalization between the S protein and S mRNA were observed on the nuclear surface and inside the nucleus, respectively (Figure 3D). We were able to pinpoint the colocalization site by using high magnification confocal imaging followed by image processing. Representatives of S mRNA and S protein colocalization in the cytoplasm (Figure 4, top two panels), on the surface of the nuclear membrane (Figure 4, middle two panels), or inside the nucleus (Figure 4, bottom two panels) are shown. We observed S protein and S mRNA colocalization in three subcellular locations, which was confirmed at the single-cell level in SARS-CoV-2-infected cells (Supplementary Figure S5). We also observed that S mRNA inside and on the nuclear surface was associated with the S protein, in contrast to the cytoplasmic S mRNA distribution with or without colocalization with the S protein (Figure 4; Supplementary Figure S5). Thus, S mRNA translocates into the nucleus through the S protein-S mRNA complex and is driven by the S protein (Figure 4; Supplementary Figure S5).

To determine whether SARS-CoV-2 S protein’s nuclear translocation is independent of other viral protein’s involvement, we transfected primary normal human bronchial epithelial (NHBE) cells with a recombinant plasmid containing codon-optimized full-length DNA sequence of SARS-CoV-2 S protein (Leventhal et al., 2021) for 48 h. We used an immunofluorescent assay (IFA) to detect the S protein in the transfected cells. We found a broader subcellular distribution of S protein (Supplementary Figure S6A). Both vertical and horizontal cross-sections of the epifluorescence images showed that the fluorescence signal of the S protein overlapped the nuclear stain. However, it was prudent to determine whether the observed S protein nuclear distribution is independent of its abundant endoplasmic reticulum (ER) distribution, as it is processed there before translocating to the cell surface. Thus, we further investigated the S protein’s nuclear translocation in the lung epithelial A549 cells. We first co-stained the S protein and a common ER protein, calnexin. We found that S protein sub-cellular distribution beyond ER, which complemented our previous finding of its nuclear translocation (Supplementary Figure S6B). However, we need to show S protein translocated into the nucleus in the transfected A549 cells. We co-stained the S protein and lamin B1, a nuclear inner membrane protein. We found that S protein fluorescence overlapped with the lamin B1 suggesting S protein’s nuclear translocation in the transfected A549 cells (Supplementary Figure S6C). S protein nuclear distribution was distinct from the eGFP’s diffusive expression throughout the cells (Supplementary Figure S6C). In addition to a fluorescent-based detection strategy, we used a biochemical approach to confirm the S protein’s nuclear translocation. We hypothesized that the SARS-CoV-2 S protein should be detected in nuclear and cytoplasmic fractionating to the SARS-CoV-2 S plasmid transfected cells. We transfected A549 cells with the same plasmid for 72 h. For control, we similarly mock-transfected A549 cells without the plasmid. We collected cytoplasmic and nuclear fractions of the S protein plasmid transfected or mock-transfected using ReadiPrep Nuclear/Cytoplasmic Fraction kit (AAT Bioquest) according to the manufacturer’s instructions. We found S protein in both the nuclear and cytoplasmic fractions of the transfected A549 cells (Figure 5). We detected lamin A and C and cell division cycle protein 42 (Cdc42) on the same gel to confirm no mix-up between nuclear and cytoplasmic portions. While lamin A and C were detected in the nuclear portion, Cdc42 was detected only in the cytoplasmic portion (Figure 5). Our results suggested that transient expression of the SARS-CoV-2 S protein caused its wider subcellular distribution, including the nucleus in the transfected cells.

There is no comprehensive information on SARS-CoV-2 N protein NLS motifs, and we already know that other pathogenic coronaviruses, particularly SARS-CoV (Rowland et al., 2005; Timani et al., 2005) and MERS-CoV (Yang et al., 2013) N proteins, have NLSs. Therefore, we searched for NLSs in the SARS-CoV-2 N protein by using the PSORT II. We found that the SARS-CoV-2 N protein has 7 NLSs covering all three types of NLS motifs (pat4: 2; pat7: 3, and bipartite: 2); however, the SARS-CoV N protein has 8 NLS motifs (pat4: 2; pat7: 4, and bipartite: 2) (Supplementary Figure S7). N protein translocation into the nucleus or at least the perinuclear region was confirmed (Figure 6A; Supplementary Figure S8). However, SARS-CoV-2 N protein nuclear translocation was not as robust as SARS-CoV nuclear translocation (Figure 6B; Supplementary Figure S8). The reduced nuclear translocation of the SARS-CoV-2 N protein is probably due to the absence of one pat7 NLS motif in the SARS-CoV-2 N protein compared to that in SARS-CoV (Supplementary Figures S8, S9).

Primary normal human bronchial epithelial (NHBE) cells from healthy adults and high-risk adults (deidentified) were obtained from Dr. Kristina Bailey at the University of Nebraska Medical Center (UNMC) (Omaha, NE) under an approved material transfer agreement (MTA) between the University of North Dakota (UND) and UNMC (Omaha, NE). The protocol for obtaining cells was reviewed by the UNMC IRB and was determined to not constitute human subject research (#318-09-NH). In this study, we used cells from three donors: nonsmoker healthy adults (donors #1 and #2) and adult with chronic obstructive pulmonary disease (COPD) (donor #3). The protocols for subculturing primary NHBE cells were published previously (Osan et al., 2020, 2021, 2022; Talukdar et al., 2022). Human lung epithelial A549 cells was maintained in F-12 medium (Thermo Fisher Scientific) as described previously (Mehedi et al., 2016). SARS-CoV-2 (USA/WA-CDC-WA1/2020 isolate, GenBank accession no. MN985325; kindly provided by CDC), SARS-CoV (Urbani strain, GenBank accession no. AY278741; kindly provided by Rocky Mountain Laboratories (RML), NIAID, NIH), and MERS-CoV (GenBank accession no. NC_019843.3; kindly provided by the Department of Viroscience, Erasmus Medical Center, Rotterdam, The Netherlands) were used for in vitro infections described below.

We have used open-source web portals for different in silico analyses. 1. Constraint-based alignment tool for multiple protein sequences (COBALT)¹ was used for multiple sequence alignment. 2. PSORT II² was used for NLS prediction. 3. The RPIseq web portal³ was used for RNA–protein interactions.

The protocols for differentiating primary NHBE cells to form a pseudostratified bronchial airway epithelium were published previously (Osan et al., 2020, 2021, 2022; Talukdar et al., 2022). Briefly, Transwells (6.5 mm) with 0.4-μm-pore polyester membrane inserts (Corning Inc.) were coated with PureCol (Advanced BioMatrix) for 20 min before cell seeding. NHBE cells (5×10^4) suspended in 100 μL of complete airway epithelial cell (cAEC) medium [AEC medium (Promocell) + SupplementMix (Promocell) + 1% penicillin–streptomycin (V/V) (Thermo Fisher Scientific) + 0.5% amphotericin B (V/V) (Thermo Fisher Scientific)] were seeded in the upper chamber of the Transwell. Then, 500 μL of cAEC medium was added to the lower chamber of the Transwell. When the cells formed a confluent layer on the Transwell insert, the cAEC medium was removed from the upper chamber, and in the lower chamber, the cAEC medium was replaced with complete ALI medium [PneumaCult-ALI basal medium (Stemcell Technologies Inc.) + with the required supplements (Stemcell Technologies) + 2% penicillin–streptomycin (V/V) + 1% amphotericin B (V/V)]. The complete ALI medium in the lower chamber was changed every day. The upper chamber was washed with 1x Dulbecco’s phosphate-buffered saline (DPBS) (Thermo Fisher Scientific) once per week initially but more frequently when more mucus was observed during later days. All cells were differentiated for at least 4 weeks at 37°C in a 5% CO₂ incubator. We observed motile cilia in the differentiated airway epithelium similar to previously described (Osan et al., 2020).

All viral infection experiments were conducted in the high biocontainment facility at RML, NIAID, NIH, Hamilton, MT. After approximately 3 weeks, the differentiated airway epithelium on Transwells was shipped to RML in an optimized transportation medium (Osan et al., 2020, 2022), and the recovered cells were maintained in complete ALI medium for approximately 1 week before infection. For infection, the airway epithelium on Transwells was washed with 200 μL of 1x PBS to remove mucus and were infected on the apical site with SARS-CoV-2, MERS-CoV, or SARS-CoV at a MOI of 0.1 in 100 μL 1x PBS for 1 h (at 37°C with 5% CO₂). For mock infection, the Transwells were similarly incubated with 100 μL 1x PBS without virus. The viral inoculum was then removed, and the epithelium on the Transwell was washed twice with 200 μL of 1x PBS. Complete ALI medium (1,000 μL) was added to the lower chamber of each Transwell, and the upper chamber was kept empty. Mock-infected and virus-infected Transwells were incubated for 4 days at 37°C in an incubator with 5% CO₂ (Osan et al., 2022).

At 4 days postinfection (DPI), 200 μL of 1x PBS was added to the apical site of the Transwell for washing before PFA fixation. In the basal side of the Transwell inserts was 200 μL of 1x PBS. For PFA fixation, 200 μL of 4% PFA (Polysciences) was added to the upper chamber of the Transwells and incubated for 30 min, and the Transwells were further maintained overnight in 4% PFA prior to removal from high biocontainment. The PFA fixation protocol was approved as an inactivation method for coronaviruses by the RML Institutional Biosafety Committee. The PFA-fixed airway epithelium was paraffin-embedded and sectioned at a thickness of 5 μm for slide preparation as previously described (Osan et al., 2021).

Slides with 5 μm sections were first deparaffinized by incubation in a Coplin jar as follows: 1. Histo-Clear for 5 min, two times; 2. 100% ethanol for 5 min, three times; 3. 95% ethanol for 5 min, 4. 70% ethanol for 5 min, and 5. distilled water for 5 min. The deparaffinized slides were immediately incubated in 0.5% Triton X-100 in 1x PBS for 30 min. The slides were washed three times with 1X PBST (1x PBS with Tween 20) or 1x PBS for 5 min. A hydrophobic barrier was drawn around the 5 μm section on the slides by using an Immedge Hydrophobic Barrier Pen. To reduce nonspecific antibody binding, the section was blocked with 10% goat serum (Vector Laboratories) in 1x PBST for 2 h at 4°C. The slides were then incubated with viral protein-specific primary antibody solution in 1x PBST (e.g., SARS-CoV/SARS-CoV-2 S protein-specific rabbit polyclonal antibody at a 1:100 dilution, SARS-CoV/SARS-CoV-2 N protein-specific mouse monoclonal antibody at a 1:100 dilution, or MERS-CoV N protein-specific mouse monoclonal antibody at a 1:100 dilution) (Osan et al., 2021) overnight at 4°C. The slides were then incubated with the corresponding secondary antibody solution (anti-mouse or anti-rabbit AF488 or AF647, Thermo Fisher Scientific) in 1x PBST for 2 h at room temperature. We then stained the nuclei with DAPI reagent (Advanced Cell Diagnostics) or used RNAscope multiplex V2 to detect SARS-CoV-2 S mRNA (Probe V-nCoV2019-S) according to the manufacturer’s instructions (Advanced Cell Diagnostics). The sections were mounted on Tech-Med microscope slides (Thomas Fisher Scientific) using ProLong-Gold antifade mounting medium (Thermo Fisher Scientific).

For the immunofluorescence (IFA)-based detection of SARS-CoV-2/ SARS-CoV S and N protein were detected by using S-specific rabbit polyclonal antibody (Thermo Fisher Scientific) and N-specific mouse monoclonal antibody (Thermo Fisher Scientific), respectively. Similarly, MERS-CoV N protein was detected using N-specific mouse monoclonal antibody (Sino Biological). Similarly, calnexin and lamin B1 were detected using calnexin-specific mouse monoclonal antibody (Antibody) and lamin B1 specific mouse monoclonal antibody (Abcam), respectively. The slides were then incubated with the corresponding secondary antibody solution (anti-mouse or anti-rabbit AF488 or AF647, Thermo Fisher Scientific). For primary antibody we used 1:1000 dilution, whereas we used 1:200 dilution for secondary anybody incubation. Rhodamine Phalloidin (Cytoskeleton Inc.) (1:500 dilution) and DAPI (Thermo Fisher Scientific) (1 drop in 1,000 μL) were used for staining F-actin and nucleus, respectively. For mounting coverslip on a cover glass, we used ProLong Gold Antifade Msounting medium (Thermo Fisher Scientific). The images were taken under an Olympus FluoView laser scanning confocal microscope (Olympus FV3000) enabled with a 60X objective (Olympus), a Leica Stellaris confocal microscope (Leica Microsystem) using a 63x oil objective or a Leica DMI8 epifluorescence microscope (Leica Microsystem). The images were then deconvolved using Huygen Essential deconvolution software (Scientific Volume Imaging). The surface rendering function of Imaris image processing software (Oxford Instruments) was used. The images were also analyzed for spot-to-spot colocalization by Imaris. Where applicable, images taken under a Leica DMI8 microscope were processed using 3D deconvolution and 3D view modules in LASX software (Leica Microsystem). For figure preparation, Prism version 9 (GraphPad) and Adobe Photoshop (Creative Cloud) software were used.

NHBE cells (two independent donors #1 and #2) were seeded on 24 well plate in cAEC medium at 37°C in a 5% CO₂ incubator. Cells were then transfected with SARS-COV-2 Spike (S) expressing plasmid (generous gift from Shanna S Leventhal and David M Hawman) (Leventhal et al., 2021) (0.5 μg) using TransfeX transfection reagent (ATCC) according to the manufacturer instructions. Similarly, A549 cells were seeded on a 24 well plate in F-12 medium at 37°C in a 5% CO₂ incubator. For imaging purpose, a coverslip was previously added onto the well of 24-well plate before seeding cells. The cells were transfected with SARS-COV-2 Spike (S) expressing plasmid (0.5 μg) using TransLT-1 transfection reagent (Mirus Bio LLC) according to the manufacturer instructions. For control, eGFP in pCAGGS plasmid (generous gift from Dr. Thomas Hoenen) was transfected similarly. For mock-transfection, cells were treated similarly without any plasmid.

Old medium was removed, and cells were washed at 1x PBS. Both cytoplasmic and nuclear portion from the S-transfected or mock-transfected A549 cells were collected using nuclear fractionating kit according to the manufacturer instructions. Around 30 ul of cytoplasmic and nuclear portions were run 4–12% bis-tris gel (Thermo Fisher Scientific) and followed by transferred on polyvinylidene fluoride (PVDF) membrane (Thermo Fisher Scientific) using iblot (Thermo Fisher Scientific). The membrane was blocked using Li-Cor blocking buffer (LI-COR Biosciences). For SARS-CoV-2 S protein was detected using SARS-CoV/SARS-CoV-2 S protein-specific rabbit polyclonal antibody (Thermo Fisher Scientific) at a 1:2500 dilution. Lamin A and C were detected protein-specific rabbit polyclonal antibody (Abcam) (molecular weight around 75 kDa and lower, respectively) at a 1:5000 dilution. Cdc42 were detected protein-specific rabbit polyclonal antibody (Cell Signaling Inc) (molecular weight around 20 kDa) at a 1:1000 dilution. A sequential different primary and secondary antibodies (e.g., anti-rabbit goat IRDye 680RD or ant-rabbit goat IRDye 800CW incubation for different protein detection on the same gel. For accuracy of protein detection, individual protein specific gel was also generated.