SARS-CoV-2 Omicron variant replication in human respiratory tract ex vivo

Emergence of SARS-CoV-2 variants of concern (VOC) with progressively increased transmissibility between humans is a threat to global public health. Omicron variant also evades immunity from natural infection or vaccines 1 . It is unclear whether its exceptional transmissibility is due to immune evasion or inherent virological properties. We compared the replication competence and cellular tropism of the wild type (WT) virus, D614G, Alpha, Beta, Delta and Omicron variants in ex vivo explant cultures of human bronchus and lung. Dependence on TMPRSS2 for infection was also evaluated. We show that Omicron replicated faster than all other SARS-CoV-2 in the bronchus but less eciently in the lung parenchyma. All VOCs had similar cellular tropism as the WT. Delta was more dependent on serine protease than other VOCs tested. Our ndings demonstrate that Omicron is inherently able to replicate faster than other variants known to date and this likely contributes to its inherently higher transmissibility, irrespective of its ability to evade antibody immunity. The lower replication competence of Omicron in human lung may be compatible with reduced severity but the determinants of severe disease are multifactorial. These ndings provide important biological clues to the transmissibility and pathogenesis of SARS-CoV-2 VOCs.

which cause enhancement of ACE2 binding, transmission and immune evasion 9,10 . The Omicron variant was identi ed in Botswana and South African in November 2021 and was designated as VOC by the WHO in the same month. It has 37 amino acid substitutions in the S protein, 15 of these being in the receptor binding domain. Thus, it was likely that the phenotype of the virus would be markedly affected in transmission, escape for prior immunity of both. As of 14 December 2021, the Omicron variant has been detected in 76 countries globally 11 . Differences in disease severity between variants has been more subtle. A modest increase in disease severity has been reported with Delta VOC with higher hospital admissions compared with the alpha VOC 12,13 . The Omicron variant contains 37 amino acid substitutions in the spike protein 14 and these are likely to have major implications on transmissibility, disease severity and immune evasion from serum neutralizing antibodies from both infection-, vaccineelicited and therapeutic monoclonal antibodies. However, there is limited information on the biological or virological characteristics of this variant. Omicron virus replication is ACE-2 dependent and it is able to markedly evade neutralizing antibody elicited by past infection or vaccination 1 .
Correlation of the phenotype of a SARS-CoV-2 variant in vitro with epidemiology is well illustrated by the experimental observations with the D614G mutation. An isogenic wild-type virus engineered to carry the spike D614G substitution had enhanced viral replication in human lung epithelial cells and primary human airway tissues by increasing the infectivity and stability of virions 15 . Hamsters infected with this virus produced higher infectious titres in nasal washes and the trachea, but not in the lungs, supporting clinical evidence showing that the mutation enhances viral loads in the upper respiratory tract of COVID-19 patients and may increase transmission. But there are no studies using physiologically relevant experimental models investigating the phenotype of the Omicron variant in comparison with previous variants of SARS-CoV-2.
We have previously used ex vivo explant cultures of the human bronchus and lung parenchyma to study virus replication competence and cellular tropism for avian in uenza viruses 16,17 , MERS [18][19][20] and SARS-CoV-2 21 . This provided a suitable platform to rapidly compare the replication pro le and tropism of the Omicron variant with others and to provide insight into the observed epidemiology of this variant. We initially compared the virus replication kinetics of wild type SARS-CoV-2 (WT) with the D614G variant and VOCs from B.1.1.7 (Alpha) and B.1.351 (Beta) lineages in ex vivo explant cultures of human bronchus and lung. In a subsequent set of experiments, we compared viral replication pro les of WT with B.1.617.2 (Delta) and B.1.1.529 (Omicron) lineages. We also investigated the virus tropism and innate immune responses elicited by these variants. Finally, we compared the dependency of WT, Delta and Omicron variants on TMPRSS2 for their replication.

Higher replication of Omicron in bronchus
We rst compared the replication kinetics of WT strain, D614G, Alpha and Beta virus strains isolated from returning travellers or the community in Hong Kong in ex vivo cultures of human bronchus and lung by titrating infectious virus using 50% tissue culture infectious dose (TCID 50 ) titrations. The only signi cant difference observed was a higher replication Beta variant in the human bronchus at 72 hpi; no signi cant difference being noted between viruses at 24 and 48 hpi (Figure 1a). When WT, Delta and Omicron variants were compared, Omicron variant replicated to signi cantly higher titres than WT or Delta at 24 and 48 hpi; the difference being over 70-fold (Figure 1b). At 72 hpi, both Delta and Omicron viruses replicated signi cantly more than WT in human bronchus but there was no signi cant difference between Delta and Omicron viruses. When these experiments were done at 33 o C, the viral titres were similar to those at 37 o C for each virus (data not shown). In ex vivo cultures of human lung, the only signi cant difference observed was reduced replication of Omicron compared to WT at 24, 48 and 72 hpi ( Figure  1c,d). These ndings in bronchus and lung were con rmed in area-under-curve (AUC) analysis of aggregate virus titres at 24-72 hpi (Figure 1e

ACE2 and TMPRSS2 expression in respiratory tract
Entry of SARS-CoV-2 into cells requires both angiotensin-converting enzyme 2 (ACE2) and transmembrane serine protease 2 (TMPRSS2) which cleaves and activates the spike at the S1-S2 junction 22 . ACE-2 expressed at cell surface exists as short and long forms, with the short form lacking the domain that binds to SARS-CoV-2 spike 23 . We showed that there was signi cantly higher expression of both long and short form ACE2 in human bronchus than in the lung ( Figure 4a). Immunohistochemical staining also showed clear evidence of more extensive ACE2 staining in bronchus than lung (Figure 4b and 4c). We next investigated the TMPRSS2 expression and showed signi cantly higher expression of TMPRSS2 in bronchus than lung by mRNA expression levels ( Figure 4a).

Variants and dependence of TMPRSS2
We investigated WT, Delta and Omicron virus replication for their dependence on TMPRSS2. We compared the viral replication kinetics of wild-type, Delta and Omicron in Vero E6 and Vero E6-TMPRSS2 (E6/T2) cells. While virus titres at 24 hpi was higher in E6/T2 cells in all three viruses, the magnitude of the difference was 100-1000 fold in Delta, 5-6 fold for Omicron and marginal for WT ( Figure 4c). The AUC levels of the viral titres at 24-48 hours con rmed these ndings (Figure 4d). We next infected Vero E6/T2 cells with WT, Delta and Omicron in the presence of a serine protease inhibitor camostat mesylate which blocks TMPRSS2 activities 22 . Delta and Omicron variants were propagated in Vero E6/T2, in order to exclude selection of TMPRSS2-dependent virus during propagation, Beta which was used as a control virus was also propagated in Vero E6/T2 cells. The viability of Vero E6/T2 was above 80% when treated with camostat mesylate at concentrations up to 300 µM (Extended Data Figure 1). In the presence of camostat mesylate treatment, only Delta was sensitive to the inhibitor and the half-maximal inhibitory concentration (IC50) was 10.66 µM (Figure 4f). In contrast, the IC 50 of WT, Beta and Omicron was higher than 300 µM.

Discussion
Preliminary epidemiological data from the United Kingdom suggests that the risk of household transmission with Omicron is 3.2 (95% con dence interval 2.0-5.0) higher than with Delta, which in turn was a highly transmissible virus, and that Omicron has a Reproduction number (R) of 3.7 24 . It has been unclear whether this increased e ciency of transmission is driven by mutations that confer immune escape from antibody in the population elicited by prior infection or vaccination, by intrinsic viral factors or a combination of the two. It is increasingly clear that high neutralizing antibody titres elicited by prior infection or immunization are markedly compromised by Omicron virus 1 . Our results suggest that the Omicron variant has a substantial (over 70-fold increase) and signi cantly higher replication competence in the human bronchus compared to both wild-type and Delta viruses at 24 hpi. Higher infectious virus load in conducting airways is likely to result in increased amounts of infectious virus released while breathing or speaking thus enhancing transmission by the airborne route. Examples of such airborne transmission of Omicron have been documented 14 . Infectious SARS-CoV-2 has been detected in ne aerosol particles in air exhaled by patients with COVID-19 25 . Thus, our data indicates that Omicron has intrinsic biological properties that potential mediate enhanced transmission, which may also synergise with its ability for evading prior immunity in leading to enhanced overall transmissibility in the community. The mechanistic reasons underlying this increased replication competence remains to be elucidated. The Omicorn variant has 37 amino acid substitutions in the spike protein, 15 of which are in the receptor binding domain 14 . Infection with Omicron is ACE-2 dependent 1 and that the binding of the spike of Omicron spike to ACE-2 is enhanced when compared to the wild-type virus 26 . We showed that there is more extensive ACE2 expression in human bronchus than lung which may explain the enhanced replication SARS-CoV-2 in the bronchus. Omicron also has amino acid substitutions in the nucleocapsid protein (R203K and G204R) that have been associated with enhanced virus replication 27 .
Epidemiological studies suggested that Delta variant was substantially more transmissible than Alpha variant 28 , which was itself was more transmissible than the earlier virus stains 4 . Thus, our data indicating that Delta variant has higher titres of infectious virus than wild-type virus in the bronchus is also compatible with the epidemiological observations. It has been shown that pseudoviruses expressing Delta spike are better able to infect ACE2 low human bronchial epithelial cells compared to previous variants 29 . Moreover, the spike of Delta is in a predominantly cleaved state which may promote its replication e ciency in human airway 30 . P681R enhanced the cleavage of furin-cleavage sequences, which attributes to the enhanced tness of Delta over Alpha in a competition assay in Calu-3 and human airway epithelium in vitro models 31,32 .
Our data showing that the Omicron variant had lower viral replication competence in the lung, compared to the bronchus is of particular interest. This difference in virus tropism is also con rmed by the immunohistochemistry studies showing less virus infected cells in human lung explant cultures ex vivo. The biological determinants of this divergence in comparative replication competence of Omicron and Delta variants in the bronchus and lung remains to be explored. These observations may suggest that Omicron may have reduced clinical severity but such interpretations need to be quali ed because the disease severity of COVID-19 is determined not only by virus replication but also by dysregulated innate immune responses. Reliable epidemiological assessments of severity of disease caused by Omicron is still awaited. Although reports from South Africa suggest that hospitalization and intensive care admissions appears to be less common during the recent Omicron outbreak 33 , preliminary assessment of hospitalisation and asymptomatic infection indicators in the UK did not show differences in hospitalization rates with Delta and Omicron variants 34 .
Our nding that Omicron is less dependent on TMPRSS2 activation compared with Delta suggests that Omicron may have a wider spectrum of target cells comparing to Delta and also that the use of therapeutic inhibitors of TMPRSS2 may be of limited bene t in the management of clinical infections with the Omicron variant. Though there have been single cell studies on the distribution of TMPRSS2 in cell cultures, the extent of distribution is hampered by sensitivity and speci city of antibodies in xed tissues.
One of the limitations of this study is that only one virus strain from each lineage was tested. Furthermore, all six virus variants were not tested in parallel in the same experiment, but this is very di cult to carry out logistically because of limited tissue available to test six viruses with replicates.
In summary, our ndings reveal faster and enhanced viral replication e ciency of Omicron in the human bronchus compared with the precedent lineages, suggesting an intrinsic capacity for enhanced transmission which is compatible with current epidemiological data. The lower replication competence of Omicron in human lung suggests that it may cause less severe COVID-19 than Delta. Even if disease severity is modestly reduced, the very e cient transmissibility of Omicron will pose a major threat to global public health and health care systems. MCWC: study design, overall coordination, interpretation of results and writing of the manuscript.

Con ict of interest
There is no con ict of interest for all authors.

Role of the funding source
The sponsors had no role in study design, data collection, analysis, or interpretation of the data, or in the writing of the report. The corresponding author had full access to all of the data and the nal responsibility to submit for publication.

SARS-CoV-2 isolation
Vero E6 cells were used for virus isolation and propagation of wild-type virus, D614G strain and Alpha variant, and Vero E6-TMPRSS2 overexpressed cells 35 were used for Beta, Delta and Omicron variant. Both cell-lines were cultured in DMEM with 10% FBS. The original clinical specimens were collected from SARS-CoV-2 con rmed patients in Hong Kong from January 2020 to November 2021 (Extended Data   Table 1) and isolated as previously described 36

Ex vivo cultures and infection of human respiratory tract
Fresh non-tumour bronchus (n=12) and lung (n=12) tissues were obtained from patients with age 51-78 years undergoing elective surgery in Department of Surgery at Queen Mary Hospital (Pok Fu Lam, Hong Kong, China) from May to December 2021 and were removed as part of routine clinical care but surplus for routine diagnostic requirements as detailed previously. 37,38 Sampling of tissues was de ned by obtaining informed consent and convenience. The donor information was listed in the Extended Data Table 2. The virus infection procedures were performed as previously described 36 . Brie y, pieces of human bronchus and lung tissues were infected with each virus at 5x10 5 TCID 50 /ml for 1 h at 37°C. Each tissue fragment was washed three times in culture medium to remove residual virus inoculum, topped up with fresh medium and incubated at 37 o C as indicated. Mock-infected tissues served as negative controls. Aliquots of culture medium were removed at times indicated and stored at -80 o C until titration. Infectious viral titres in culture supernatants were assayed by TCID 50 in Vero E6 or Vero E6-TMPRSS2 cells respectively, depending on the cells used for virus isolation and passage. Infected tissues were xed in 10% formalin and processed for immuno-staining at 72 hours post infection.

Viral titration by TCID 50 assay
A con uent 96-well tissue culture plates of Vero-E6 or Vero E6-TMPRSS2 cells was prepared one day before the virus titration (TCID 50 ) assay. Cells were washed once with PBS and replenished with DMEM (Gibco) with 2% foetal bovine serum (Gibco) supplemented with 100 units/ml penicillin and 100 µg/ml streptomycin (Gibco). Serial dilutions of virus supernatant, from 0.5 log to 7 log, were performed and each virus dilution added to the plates in quadruplicate. The plates were observed for cytopathic effect daily. The end-point of viral dilution leading to CPE in 50% of inoculated wells was estimated using the Karber method 39 . Area under the curve (AUC) was calculated from the viral titers from different time points indicated in the Y-axis.
Immuno-staining of para n-embedded tissues Human respiratory tract tissues (lung and bronchus) were xed with 10% formalin overnight at 4 °C and the xed tissues were embedded in para n blocks. For immunohistochemistry, the 4µm sliced sections were microwaved for 15 min for antigen retrieval. Endogenous peroxidase activity was stopped by quenching the tissue sections with 3% H 2 O 2 for 20 min. The slides were then blocked with 10% normal horse serum at room temperature (RT) and incubated with primary antibodies (SARS-CoV-2 Nucleoprotein (NP) (Sino Biological) or ACE2 (abcam) antibodies for 90 min at RT followed by peroxidase (HRP)conjugated anti-rabbit antibody (Vector Laboratory). The sections were developed using NovaRED Substrate Kit (Vector Laboratory). The cell nuclei were counterstained with Mayer's Hematoxylin.
To characterize the SARS-CoV-2-infected cells, double-antibody immuno uorescence staining of SARS-CoV-2 NP antibody with different cellular markers was performed. The tissue sections were rst stained with SARS-CoV-2 NP antibody similarly as mentioned above, except after SARS-CoV-2 NP antibody incubation, the sections were incubated with alkaline phosphatase conjugated anti-rabbit or anti-mouse antibody (Vector Laboratory) and developed using Vector® Red (VR) Substrate Kit (Vector Laboratory). The sections were then microwaved, incubated with SCGB1A1/CC10 (Protein-tech), acetylated α Tubulin (Santa Cruz), MUC5AC (Thermo-Fisher), for 90 min at room temperature followed by goat anti rabbit-AF488 or goat anti mouse-AF488. The cell nuclei were counterstained with DAPI (blue). The sections were imaged using Nikon Eclipse Ti-S microscope.
Real-time PCR assay RNA was extracted from homogenized tissues at 72 h post infection (hpi) using RNeasy Micro Kit (Qiagen). The viral RNA in culture supernatants were extracted at 48 hpi using QIAamp Viral RNA Mini Kit (Qiagen). RNA was reverse-transcribed by using random 6-mer primers with PrimeScript RT reagent Kit (Takara). mRNA expression of target genes was detected with the corresponding primers (Extended Data  Table 3) using an ABI ViiA 7 real-time PCR system (Applied Biosystems). All procedures were performed according to the manufacturers' instructions. The viral gene and gene expression pro les of cell lysates were quanti ed and normalized with b-actin as previously described [40][41][42][43] .
Replication of SARS-CoV-2 on Vero-E6 and Vero-E6-TMPRSS2 and camostat mesylate treatment Vero-E6 and Vero-E6-TMPRSS2 cells in 48-well plates were infected with SARS-CoV-2 at a multiplicity of infection (MOI) of 0.1 for viral replication kinetics. Viral titers in culture supernatants (1-72 hpi) were determined using TCID 50 assay (see above). To assess impact of TMPRSS2 inhibitors, the Vero-E6-TMPRSS2 cells were infected with SARS-CoV-2 at a MOI of 0.05. Camostat mesylate (Sigma), 0.41-300µM, was added 1 h before, during and after infection. Vehicle was used as a negative control. Antiviral activities were evaluated by quanti cation of SARS-CoV-2 ORF1b copy number in the culture supernatant by using qPCR at 48 hpi 43 .

Quanti cation of cell viability
Cell viability of Vero-E6-TMPRSS2 following the treatment of camostat mesylate was evaluated using Cell Counting Kit 8 (WST-8 / CCK8) (Abcam). In brief, cells grown to 70% con uency in 48-well plates were incubated for 48 h in the absence or presence of various concentrations (0.41-300µM) of camostat mesylate. Then, the culture medium was aspirated and washed once with PBS, incubated with CCK8 working solution (CCK8: medium = 1:10) at 37 °C, 5% CO 2 for 2 h. Subsequently, OD values at 450 nm were measured by a multiplate reader (BMG FLUOstar OPTIMA). Cell viability was expressed as a percentage of that of the control cells.

Statistical analysis
Experiments with the human ex vivo cultures were performed independently with six different donors. Results shown in gures are geometric mean (+/-SD). Area-under-curve (AUC) was calculated by integrating infectious virus titres at 24-72 hpi in ex vivo bronchus, lung tissues or at 24-48 hpi in cell lines. The differences in log10-transformed viral titres and quantitative cytokine and chemokine mRNA between viruses and over time were compared using two-way ANOVA followed by a Tukey's multiple-comparison test using GraphPad Prism version 9.0.0. Comparison of AUC and quantitative cytokine and chemokine mRNA between viruses were calculated using one-way ANOVA followed by a Tukey's multiple-comparison test. Differences were considered signi cant at a p value less than 0.05. horizontal dotted line denotes the limit of detection in the TCID 50 assay. Bar-charts show the geometric mean (n=6) (SD). Statistics were performed using Two-way ANOVA followed by a Tukey's multiplecomparison test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. e, f. Viral titers from a to d are depicted as area under the curve (AUC). Bar-charts show the geometric mean (n=6) (SD). Statistics were performed using One-way ANOVA followed by a Tukey's multiple-comparison test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 2
Tissue tropism of SARS-CoV-2 variants in ex vivo cultures of human respiratory tract. Ex vivo cultures of a. human bronchus and b. human lung were infected with various lineages of SARS-CoV-2 with 5x10 5 TCID 50 /mL at 37°C and the tissues were xed with formalin at 72 hpi. Para n-embedded sections were subjected to immunohistochemical staining with a monoclonal antibody against the SARS-CoV-2 nucleoprotein. Positive cells are red-brown. Scale bar, 100 mm. The images are representative of three individual donors.