SARS‐CoV‐2 variants of concern elicit divergent early immune responses in hACE2 transgenic mice

Knowledge about early immunity to SARS‐CoV‐2 variants of concern mainly comes from the analysis of human blood. Such data provide limited information about host responses at the site of infection and largely miss the initial events. To gain insights into compartmentalization and the early dynamics of host responses to different SARS‐CoV‐2 variants, we utilized human angiotensin converting enzyme 2 (hACE2) transgenic mice and tracked immune changes during the first days after infection by RNAseq, multiplex assays, and flow cytometry. Viral challenge infection led to divergent viral loads in the lungs, distinct inflammatory patterns, and innate immune cell accumulation in response to ancestral SARS‐CoV‐2, Beta (B.1.351) and Delta (B.1.617.2) variant of concern (VOC). Compared to other SARS‐CoV‐2 variants, infection with Beta (B.1.351) VOC spread promptly to the lungs, leading to increased inflammatory responses. SARS‐CoV‐2‐specific antibodies and T cells developed within the first 7 days postinfection and were required to reduce viral spread and replication. Our studies show that VOCs differentially trigger transcriptional profiles and inflammation. This information contributes to the basic understanding of immune responses immediately postexposure to SARS‐CoV‐2 and is relevant for developing pan‐VOC interventions including prophylactic vaccines.


Introduction
Human Coronavirus Disease 2019 (COVID- 19) is characterized by a diverse spectrum of clinical manifestations ranging from mild Correspondence: Dr. Björn Corleis and Prof. Dr. Anca Dorhoi e-mail: Bjoern.Corleis@fli.de;Anca.Dorhoi@fli.deto severe outcomes that affect multiple organs, with the highest impact in the respiratory tract [1].The causative agent is the severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) and the related variants of concern (VOCs) with the potential to escape immune responses rapidly and spread globally [2].These VOCs propelled investigations aiming to clarify their impact on disease severity and identify virus mutations [3][4][5][6][7], with a focus on the spike protein (S protein), given its role in cell entry and vaccine design [8,9].The Beta (B.1.351)VOC was first detected in South Africa in May 2020 but did not establish globally.Starting in September 2020, cases of infection with the Alpha (B.1.1.7)VOC were reported in the United Kingdom, and in October of the same year, the Delta (B.1.617.2) VOC arose in India and spread globally.For both Alpha and Delta, an increase in transmissibility was described.Subsequently, Omicron (B.1.1.529)subtypes and related variants (BA.1, BA.2, including XBB.1.5 and BQ.1) with the ability to evade immune responses became globally dominant and spread worldwide [2,10].
Insights into innate immune events, starting immediately postexposure, were mainly gained for ancestral SARS-CoV-2 in animal models [24], while less is known for humans.Comparisons of innate immune responses in the lung early after infection with Alpha, Beta, or Delta VOCs have not been established in detail.Moreover, animal models (e.g.K18-hACE2 mice) enabled the investigation of early CD4 + and CD8 + T cell infiltration into the lung after infection [24], but studies comparing the early T cell phenotype triggered by different VOCs are still limited.Knowledge about lung immune reactivity during the first days after infection, particularly within distinct compartments and anatomical regions is largely missing and nearly impossible to obtain in humans.Gaining further insights into immune responses triggered by different VOCs could contribute to the fundamental understanding of early infection and help evaluate prophylactic vaccine approaches.
In this study, we monitored immune events in the lung interstitium and systemically promptly after a challenge infection of K18-hACE2 mice with various SARS-CoV-2 VOCs.Our results indicate significant differences in the transcriptional response, immune mediators, and immune cell abundances when comparing different VOCs.Early inflammatory responses (e.g.type I IFNs, inflammatory chemokines, and cytokine as well as infiltration of myeloid cells) at 1 to 5 days postinfection (DPI) did not result in reduced viral loads in the lung.However, in animals with reduced viral load at later time points, we noted robust anti-receptor binding domain (RBD) antibody titers and cellular adaptive immune responses (e.g. S peptide-specific CD4 + and CD8 + T cells producing IFN-γ), which mutually controlled viral spread and replication.

Distinct SARS-CoV-2 variants induce varying early gene and inflammatory responses
We employed the K18-hACE2 mouse model to comparatively investigate early immune responses to ancestral SARS-CoV-2 D614G (BavPat1) and two VOCs (Beta and Delta) in different tissues and compartments (Fig. 1A), although the Delta variant was primarily employed for extended studies using low-dose challenge.Challenge infection with high dose, that is 10 5 tissue culture infection dose 50 (TCID 50 ) in this model, is known to cause lung disease resembling partially severe COVID-19 in human [46] and is associated with lethal encephalitis [47].We used, in addition, a lower infectious dose (Supporting Information Table S1), that is 10 3 to 10 4 TCID 50 , which resulted in a less severe course of infection and longer survival of the mice (Supporting Information Fig. S1).In a multipronged approach, we monitored viral loads, transcriptional responses, phenotyped immune cells, and quantified soluble mediators in the lungs, blood, and spleen over the first 10 DPI to pin down differences between distinct SARS-CoV-2 variants.
We detected relatively low lung viral loads for BavPat1 at 1 DPI, which increased towards 5 DPI, while Beta-infected mice had significantly higher viral burdens already at 1 DPI (Fig. 1B, Supporting Information Fig. S2A).Despite the lower inoculum of Delta (10 3.62 TCID 50 ), lung viral loads were comparable to BavPat1 at 1 and 5 DPI.In Delta-infected mice that did not succumb to infection, viral burdens decreased by 10 DPI (Fig. 1B).Bulk RNA sequencing (RNAseq) unveiled that BavPat1 induced differentially expressed genes (DEGs) in the murine lung (Fig. 1C) and spleen tissues (Supporting Information Fig. S2B & C) at 5 DPI.In contrast, Beta triggered transcriptional activity in the lungs already at 1 DPI (Fig. 1C).Also, SARS-CoV-2-specific reads within the sequencing data reflected the realtime reverse transcription quantitative polymerase chain reaction (RT-qPCR) results for SARS-CoV-2 genome loads (Supporting Information Fig. S2D), and correlated with the number of DEGs (Fig. 1C).We noted a drop in DEGs (Fig. 1C) and high intersample variability (Supporting Information Fig. S2E) at 3 DPI for both investigated VOCs.Lower DEGs were maintained for Beta over time.A limited number of shared DEGs were observed between BavPat1 and Beta infection (Fig. 1D).Principal component analysis (PCA) revealed that Beta-induced lung transcriptional patterns, especially at 1 DPI, were distinct from those triggered by BavPat1 (Supporting Information Fig. S2E).Analysis of the timepoints further indicated substantial enrichment for IFN-signaling pathways for Beta at 1 DPI.Upon disease progression, lung DEGs for BavPat1 at 5 DPI were associated with cardiovascular and hormone-related gene pathways (Fig. 1D, Supporting Information Fig. S2F).DEGs in the spleen (Supporting Information Fig. S2B & G), however, were immune system-related for both virus variants at 1 DPI (Supporting Information Fig. S3A & B).In summary, the transcriptional analysis revealed DEGs with minimal overlap between lung and spleen, and between SARS-CoV-2 variants across tested time points.Generally, the number of DEGs is correlated with tissue viral loads.
To further substantiate differences in early host responses to VOCs, we quantified soluble inflammatory mediators systemically and at the site of infection and focused first on Beta and Bav-Pat1, which were delivered at similar infectious doses.In line with global transcriptome findings, the highest increase of chemokines in the lungs was triggered at 1 DPI by Beta (Fig. 2A, Supporting Information Figs.S4 & 5).Matched gene reads gained from the RNAseq results supported the high abundances detected by multiplex analysis (Supporting Information Fig. S6).Overall, abundances of most chemokines and cytokines directly correlated with lung viral loads, independent of the time point or the virus variant (Fig. 2B).Since type I IFNs are critical antiviral proteins, we evaluated members of this family, that is, the cytokines IFN-α and IFNβ, as well as IFN-induced chemokines, that is, CXCL9 and CXCL10.RNA and protein levels of both chemokines were upregulated at 1 DPI in the lungs of Beta-infected mice (Fig. 2C).Supporting these findings, Beta-infected animals had elevated type I IFN levels in the lungs in comparison with their BavPat1 counterparts (Fig. 2D).However, IFN-α and IFN-β concentrations decreased upon disease progression at 5 DPI.IFN-α levels decreased in the spleens of BavPat1-infected mice by 5 DPI but maintained the same concentrations in Beta-challenged animals, indicating not only local but also systemic differences between ancestral SARS-CoV-2 and the Beta VOC (Fig. 2D).In contrast to the high-dose challenge with BavPat1 and Beta VOC, for low-dose Delta-infected mice, the chemokine and cytokine concentrations in the lungs as well as in the spleens peaked at 5 DPI.These animals presented the highest concentrations of IFN-α and IFN-β at 5 DPI, which was congruent with CXCL9 and CXCL10 pulmonary concentrations in the same animals (Supporting Information Fig. S7).Thus, transcriptional patterns and abundances of chemokines and type I IFNs indicate the development of early inflammatory responses against SARS-CoV-2 in K18-hACE2 mice.Different virus variants and viral doses induced variability in soluble inflammatory factors over time, which were partially correlated with viral loads.

Lung innate immune cell dynamics differ by SARS-CoV-2 variant
Based on the observed cytokine and chemokine patterns, we hypothesized that early infiltration of immune cells into the lungs differs with SARS-CoV-2 variants.To investigate the dynamics of cell populations over time, we employed a flow cytometry panel allowing accurate identification of lung myeloid cell populations and compared cell dynamics upon challenge with ancestral Bav-Pat1 and Beta VOC in a head-to-head approach.In addition, we employed Delta VOC at a low dose for reasons mentioned above (Supporting Information Fig. S8A).High-dose infection with Bav-Pat1 and Beta resulted in an increased number of Ly6G+ neutrophils in the lungs (Fig. 3A).Beta caused a gradual and significant increase in monocytes and macrophages within lungs by 5 DPI (Fig. 3B & D), while BavPat1 triggered significant cell recruitment of myeloid cells, including DCs and granulocytes at 1 DPI, which diminished at 3 and 5 DPI (Fig. 3A-D).Besides the late accumulation of monocytes and MDMs, low-dose Delta infection caused a decline in AMs without a detectable replenishment at 10 DPI (Supporting Information Fig. S8B).The frequencies of AMs remained largely unaltered for the other VOCs (Fig. 3B).Thus, alterations of myeloid cells in the lung were distinct between virus variants.Moderate and temporary cellular infiltrations occurred after high-dose ancestral SARS-CoV-2 and low-dose Delta challenge, whereas high-dose Beta infection resulted in the accumulation of diverse mononuclear phagocytes, notably monocytes, macrophages, and MDMs over time.

Activation and differentiation of virus-specific adaptive immunity occur later postchallenge
Analysis of viral genome copies during innate immune responses suggested that these responses might not be sufficient to provide early control of viral replication in K18-hACE2 mice.We monitored adaptive immune responses over time to gain a deeper understanding of disease pathogenesis.T cells were immunophenotyped (Fig. 4) in the lung and spleen using CD45 i. v. (CD45iv) labeling to distinguish vascular and tissue-resident lymphocytes (Supporting Information Fig. S9A).High-dose infection with Bav-Pat1 and Beta led to an early increase of CD45iv − CD4 + and naïve CD8 + T cells in the lungs (Fig. 4A, Supporting Information Fig. 9B).For Beta, the numbers of T cells gradually increased by 5 DPI (Fig. 4A).For both variants, CD45iv − T cells did not upregulate markers associated with T cell activation, differentiation, or memory phenotypes (e.g.effector memory [T em ; CD44 hi , CD62L lo ]) during the first 5 DPI.Spleen-resident CD4 + T cells, but not CD8 + T cells, selectively increased expression of differentiation and activation markers (e.g.CXCR3) postinfection (Fig. 4B & C).Thus, high-dose infection with BavPat1 or Beta resulted in early T cell infiltration into the lungs without progression toward effector phenotypes.
We assumed that prolonged survival of the K18-hACE2 mice could be accompanied by the development of antigen-specific T cell responses in the lungs, which remained undetectable in mice succumbing by 5 DPI (Fig. 4B & C).Hence, we took advantage of the dose-dependent survival of K18-hACE2 mice (Supporting Information Fig. S1) infected with SARS-CoV-2 and applied a low dose to investigate later time points after infection.We analyzed adaptive immunity in K18-hACE2 infected with a low infectious dose of either Delta (Fig. 5, Supporting Information Fig. S9C &  D) or Beta (Supporting Information Fig. S10A-C) up to 10 DPI.Similar to findings from high-dose BavPat1 and high-dose Beta challenge experiments, Delta also did not induce expression of T cell activation as well as differentiation markers at 1 and 5 DPI (Fig. 5A).However, expression of cell surface markers on lung CD45iv − CD4 + and CD8 + T cells associated with activation, differentiation, or homing (CXCR3, KLRG1, programmed cell death protein 1 [PD-1], CD103, CD69, and CD44) were increased at 7 and 10 DPI (Fig. 5A, Supporting Information Fig. S10A).The same markers were upregulated early on CD45iv − spleen T cells but did not further increase at later time points (Supporting Information Fig. S10D).Lung CD45iv − CD4 + and CD8 + effector T cells (T em , CD44 hi , CD62L lo ), as well as tissue-resident memory (T rm ; CD44hi, CD62Llo, CD103+, CD69+) T cells, were detected earliest at 7 DPI (Fig. 5B, Supporting Information Fig. S10B).An experiment using a lower dose of SARS-CoV-2 Beta (10 4.77 TCID 50 ) with prolonged survival of some animals confirmed the observation that T cell activation and differentiation in the lung occurred at 7 to 10 DPI in the K18-hACE2 mouse model (Supporting Information Fig. S10A-C).
Because virus-specific immune responses are crucial for the protection from COVID-19 in humans [30,33] and the observed development of T rm in the K18-hACE2 mice suggested differentiation of SARS-CoV-2-specific adaptive immunity, we quantified RBD-specific antibodies and S-peptide-specific T cell responses upon challenge with low-dose Delta and Beta (Fig. 5D, Supporting Information Fig. S10A-C, S11A).Anti-RBD antibodies in sera were detected starting from 7 DPI and increased until 10 DPI (Fig. 5C) upon different virus variants and isolates administered in various viral doses from 10 3.62 to 10 5.69 TCID 50 .We also observed a significant increase of IFN-γ producing lung and spleen (Fig. 5D) CD4 + and CD8 + T cells at 7 DPI upon in vitro restimulation with ancestral SARS-CoV-2 peptides.IFN-γ-producing T cells were hardly detectable during the first week after challenge (Supporting Information Fig. S11B & C).Polyfunctionality analysis revealed that the majority of SARS-CoV-2-specific T cells only produced IFN-γ.In contrast, other cytokines or antiviral proteins (e.g.granzyme B) were below the limit of detection (Fig. 5E).The observed SARS-CoV-2-specific cellular and humoral immune responses at 7 to 10 DPI were coincident with the decline of Delta viral loads, which peaked at 5 DPI (Supporting Information Fig. S11D & E).Thus, developing a SARS-CoV-2-specific host response in the context of reduced virus burdens in Delta-infected animals emphasizes a critical role of adaptive immunity in virus control in this mouse model.

SARS-CoV-2-specific T and B cell responses reduce viral spread and burdens in K18-hACE2 mice
To further substantiate our hypothesis that adaptive immunity is crucial for limiting the viral loads in the K18-hACE2 mouse model, we systemically depleted CD4 + and CD8 + T cells or CD20 + B cells and monitored the mice challenged with a low dose of the Delta (B.1.617.2) variant for up to 12 DPI (Fig. 6A).B cell abundances in the lungs (CD45iv + and CD45iv − ) and spleen (CD45iv − ) were significantly reduced through the depletion.T cell depletion efficiency was overall lower for lungresident (CD45iv − ) than for CD45iv + (circulating) or spleenresident cells (Supporting Information Fig. S12A-C).T celldepleted mice were unable to generate S-specific T cell responses in the lungs (7-12 DPI) (Fig. 6B).B-and T cell-depleted animals did not develop RBD-specific antibody responses.In contrast and congruent with our previous data (Fig. 5C), the control animals seroconverted starting at 7 DPI (Fig. 6C).Although adaptive immune cells were successfully depleted, we observed no significant differences in the average weight loss (Supporting Information Fig. S12D) or in the survival of the B or T cell-depleted animals compared with the isotype control group (Fig. 6D).However, T cell depletion resulted in the highest viral loads in the conchae and lungs (12 DPI) (Fig. 6E) and triggered increased viral loads in the brain and spleen (Supporting Information Fig. S12E).We determined Spearman's rank correlation coefficients for all measured parameters at 12 DPI to single out coregulated biological events.We observed a significant negative correlation between the lymphocyte counts (e.g.T cell counts and S-peptide-specific responses in the spleen and lungs) and viral loads in conchae, lungs, brain, and spleen.SARS-CoV-2-specific T cells and antibody abundances also negatively correlated with the conchae and brain viral burdens (Fig. 6F).Thus, the depletion studies suggest mutual roles of adaptive immune cells in protection against viral replication and spread.thereby missing the detection of virological and immunological parameters during early infection.Ongoing inflammation and unknowns regarding infection time, early immune responses, and infectious dose reduce the predictive value of human samples.Animal models can overcome such limitations by allowing controlled early and high-resolution longitudinal measurements following a defined experimental infection in various anatomical sites.Here, we identified early distinct transcriptional patterns between ancestral BavPat1 and Beta VOC in the K18-hACE2 mouse model.Beta disseminated early into lung tissue, causing an immediate release of inflammatory cytokines and steadily higher frequencies of immune cells in the lungs.Further, our comparative approach suggests that in this mouse model of severe COVID-19, innate immunity is insufficient to clear the virus.A decline in virus burden and restriction of viral spread were associated with the development of antiviral adaptive immune responses in the lung at a later time.

Discussion
Our results demonstrate divergent inflammatory responses in K18-hACE2 mice upon infection with ancestral SARS-CoV-2 (D614G, B.1), Beta, and Delta VOC, for example, a much higher inflammatory response at 1 DPI with Beta.Others have similarly reported higher cytokine amounts (e.g.IL-6 and chemokine [C-C motif] ligand [CCL-2]) in the lungs of mice infected with Beta in comparison with the B.1 variant [48], even if a 100-fold higher dose for ancestral SARS-CoV-2 was used [49].In line with our results demonstrating early differences in lung viral loads and inflammatory markers, higher viral loads in the lungs of Betainfected K18-hACE2 mice compared with the original B.1 lineage up to 7 DPI have been documented [48].However, despite comparable replication, Alpha and Delta VOC may trigger variable pulmonary inflammation [50].These observations suggest that inflammation can be modulated by viral burden, viral dissemination into the lung, or variant-specific features (e.g.evasion of antiviral pathways while activating pattern recognition receptors [51] or inducing pathological inflammatory responses).The high abundance of early type I IFNs, probably triggered by the high lung viral burden of Beta 1 DPI, was not protective in mice, possibly due to SARS-CoV-2 interference with downstream IFN-signaling pathways and viral resistance [15].Ancestral SARS-CoV-2 did not induce significant transcriptional and protein patterns of inflammation locally in the lung up to 5 DPI.However, inflammatory transcripts were detected in the spleen already at 1 DPI, suggesting differential regulation in the mucosa compared to lymphoid organs.One of the ten most regulated pathways in the lungs of mice infected with BavPat1 was associated with blood pressure at 5 DPI, suggesting a potential influence on the function of the cardiovascular system.Clinical studies demonstrated damage to endothelial cells, thrombosis, and vascular dysfunctions in humans suffering from severe COVID-19 [52,53].The differences in viral loads and inflammation we recognized between SARS-CoV-2 variants could result from altered affinity to the endogenous mouse ACE2 (mACE2).VOCs like Alpha (B.1.1.7)and Beta (B.1.351)bearing the N501Y S protein mutation are known to also bind to the murine ACE2 [54] and infect WT laboratory mice [55].Thus, viral dissemination from the upper respiratory tract to the lung in the K18-hACE2 mouse model could be influenced by the entry into mACE2-expressing cells.This potential difference in viral spread could also result in the observed differences in early inflammation.However, it was also shown that the N501Y mutation enhances binding to human ACE2 [56], which makes the mouse model for these strains even more relevant.Viral loads positively correlated with the DEG, yet at 3 DPI, few genes were differentially regulated irrespective of VOC.This observation may relate to the early activation of IFN I, which has been reported [57][58][59] and validated in our study.It may also relate to the considerable high heterogeneity in viral load and dissemination at 3 DPI, which would be indicated by limited DEG in a grouped analysis.
Lung kinetics of innate immune cell populations differed for tested VOCs in this mouse study.We observed a significant decrease in AM numbers over time following Delta infection, confirming previous findings for this VOC in K18-hACE2 mice [36].Inflammation and AM frequency changes were not associated with increased mortality in Delta-infected mice, opposingly to humans, where AM depletion was deleterious [38].In NHPs [37] and mice [24], others showed that an early reduction of AM also occurred during infection with ancestral SARS-CoV-2.While AM abundancies remained unchanged for ancestral SARS-CoV-2 and Beta, we observed an early and continuous increase of neutrophil frequencies up to 5 DPI for both VOCs.In contrast to our results, low-dose challenge infection of K18-hACE2 mice with Beta and ancestral SARS-CoV-2 led to an increase of neutrophils (3 DPI), followed by a decrease at 5 to 6 DPI in mice, suggesting that resolution of neutrophil infiltration might be dose-dependent [60].Neutrophil numbers have been positively correlated with viral loads in the upper respiratory tract of COVID-19 patients [61].Along with this, immunopathology in severely ill patients has been associated with an expansion of lung neutrophils undergoing NETosis [62,63].In our study, Delta (low dose) did not alter this population up to 10 DPI.In general, excessive infiltration of neutrophils might instead be associated with impaired viral clearance and thereby influence survival in the K18-hACE2 mice  [ 64,65].Thus, the virus variant and the virus dose may impact the dynamics of innate populations.Our study was not designed to analyze immune responses to one variant applied in different challenge doses.Titration experiments with ancestral SARS-CoV-2 in K18-hACE2 mice have demonstrated a shift of blood immune cells due to an increased viral dose [24].Nevertheless, changes in the periphery do not definitely reflect lung responses, and higher doses of distinct virus variants do not necessarily trigger a higher inflammatory response [49].
In hamsters, a replication tropism for ancestral SARS-CoV-2 (D614G, B.1) and the Alpha VOC was demonstrated for the upper respiratory tract, while Beta mainly replicated in the lungs [66], which could also be an explanation for the early increase of viral loads 1 DPI and high abundances of innate immune cells in lungs of Beta-infected K18-hACE-2 mice.In line with previous studies, the accumulation of these myeloid cells was less pronounced in mice challenged with Delta [36].Lung infiltration of DCs was described in K18-hACE2 mice infected with ancestral SARS-CoV-2 and for bronchoalveolar lavage of infected humans and was associated with the production of chemoattractants [24,67,68].Remarkably, only high-dose Beta-infected animals demonstrated a steady increase of monocytes, macrophages, and DCs in our study.MDMs with profibrotic phenotypes accumulate in the lungs of severe COVID-19 patients and induce fibrosisassociated acute respiratory distress syndrome [69], yet such phenotypes likely do not occur during the short acute disease in K18-hACE2 mice.Overall, our findings suggest that irrespective of VOC and subsequent dynamics of innate immune cells, innate immunity was not the main driver for viral clearance in the K18-hACE2 mouse model.Contrarily, in NHPs, lung-recruited monocytes and macrophages are activated by type I IFNs and endowed with viral killing capacity [70].
Beta infection, in contrast to BavPat1 infection, steadily increased T cells until 5 DPI in this study.Supporting our findings, others observed that Beta infection in K18-hACE2 mice results in increased T cell recruitment to the lungs compared with ancestral SARS-CoV-2 (WA-1) [71], indicating a direct or indirect impact of Beta viral mutations on T cell recruitment.Furthermore, our data indicated that T cell phenotypes differed between spleen and lung compartments.Expression of markers associated with activation (e.g.PD-1) [72,73] or cell recruitment (e.g.CXCR3) was present in the spleen but not the lung early on in the infection.This suggests an induction of systemic adaptive immune responses in the secondary lymphoid tissues, which did not reach infection sites within the first 5 days of infection.With prolonged survival of K18-hACE2 mice challenged with reduced viral doses of Delta or Beta VOC, we observed the development of SARS-CoV-2-specific immune response in the lungs at 7-12 DPI.Similarly, previous studies in K18-hACE2 mice found higher counts of lungresident SARS-CoV-2-specific CD4 + and CD8 + T cells expressing CD69 and CD103 in animals infected with a low dose of ancestral SARS-CoV-2 [74].However, the adoptive transfer of these cells left the survival of naïve recipient mice unchanged after challenge infection [74].Others showed a delayed viral clearance in NHPs after depletion of CD4 + and CD8 + T cells, which was not accompanied by increased mortality [75], suggesting a role of T cells in viral clearance.Investigating humoral adaptive responses, we observed seroconversion in mice as early as 7 DPI.These results support the assumption that the cellular [30] and humoral [76] adaptive immune response is also essential for virus restriction in humans.We observed prolonged survival upon developing the adaptive lung immune responses, but depletion of B or T cells did not significantly alter survival in Delta-infected K18-hACE2.Lack of virus-specific T cells and antibodies facilitated virus spread and replication.In line with our data, others have reported increased lung viral burdens after T cell depletion in AAV-hACE2 mice, another model for COVID-19 [77].Depletion of T cells marginally improved the survival of K18-hACE2 mice in our study, which may indicate potential immunopathological functions of T cells, which have already been suggested [78].T cell depletion also enhanced virus dissemination into the cerebellum, implicating a long-term effect on the survival of K18-hACE2 mice succumbing to viral brain infiltration.Increased viral loads of the nasal tissue during T cell depletion may also indicate the importance of local adaptive immunity in controlling viral spread.Future integration of histopathological investigations may clarify mechanisms underlying the immunoprotective and immunopathological roles of T cells during SARS-CoV-2 infection and dichotomous roles of T cells in COVID-19.Thus, developing an antiviral adaptive immune response characterized by Sspecific CD4 + and CD8 + T cells and SARS-CoV-2-specific antibodies is critical for the survival of K18-hACE2 mice after infection with SARS-CoV-2 VOCs.Future experimental infections using VOCs should characterize in-depth the role of adaptive immunity by investigating the influence of mutational changes within the viral genome (including coding and noncoding regions) on host immune responses to reveal pan-variant correlates of protection.Moreover, our investigations highlight differences in the postinfection adaptive immune responses between VOCs, especially in terms of T cell recruitment.These findings may contribute toward a more comprehensive understanding of prophylactic vaccine efficacy.Furthermore, our data contribute to the basic understanding of early postexposure immune responses to SARS-CoV-2 infection, which is critical to improve our knowledge about early predictors of disease severity and the prevention of hospitalization of patients [79].

T and B cell depletion
Mice were injected intravenously with 300 μg InVivoMab antimouse CD8alpha (clone 2.43; BioXCell, Hölzel Diagnostika Handels GmbH) and Ultra-LEAF Purified anti-mouse CD4 (clone GK1.5; BioLegend GmbH) in 175 μl PBS for T cell depletion or with Ultra-LEAF Purified anti-mouse CD20 (clone SA271G2; BioLegend GmbH) for B cell depletion one day before infection.A total of 300 μg InVivoMab IgG2b isotype control (clone LTF-2; BioXCell, Hölzel Diagnostika Handels GmbH) was administered to control animals.At 5 DPI, another 300 μg of depleting or isotype control antibodies were administered intraperitoneally.Depletion efficiency was monitored by flow cytometry at 5 DPI and for each mouse euthanized within the following days at 12 DPI.

Sampling and tissue processing
Oral swabs for virus detection were taken under short-term isoflurane inhalation anesthesia.Blood was collected into Z-clot activator microtubes (Sarstedt) and sera were stored at −80 °C.Heat inactivation of sera was performed for 30 min at 56 °C prior to measurements.Lung or spleen was harvested in PBS (for lung) or complete RPMI medium (cRPMI; RPMI-1640 supplemented with 1 mM L-glutamine, 0.5 mM sodium pyruvate, 10 % FCS, 100 U/mL penicillin, 0.

RNA extraction for library preparation and sequencing
RNA extraction was done on a KingFisher Flex instrument (Thermo Fisher Scientific) using the RNAdvance Tissue Kit (Beckman Coulter) and following the manufacturer's instructions.A DNA-digestion step was included in the protocol using the RNase-Free DNase Set (Qiagen).The RNA concentration was determined using a NanoDrop 1000 spectrophotometer (Peqlab).From 5 μg of the total RNA, the poly(A)+ fraction was separated using the Dynabeads mRNA DIRECT Micro Kit (Invitrogen) following the manufactures instructions.An ERCC RNA Spike-In Mix (Invitrogen) was added prior to poly(A)+ extraction as an internal control.The poly(A)+ RNA was subsequently used for library preparation with the Collibri Stranded RNA Library Prep Kit for Illumina Systems (Invitrogen) following the manufactures instructions.At each step, the extracted total RNA, poly(A)+ RNA, and final libraries were quality checked using the 4150 TapeStation System along with a High Sensitivity RNA/DNA ScreenTape and reagents (Agilent).Final libraries were quantified using a Qubit 2.0 fluorometer and the Qubit dsDNA HS Assay-Kit (Invitrogen).
Libraries were subsequently pooled and sequenced on a NovaSeq 6000 system (Illumina) running in 1 × 100 bp mode.

Sequence processing and analysis
Raw reads were quality trimmed and adapters removed using Trim Galore! (version 0.6.6)running in automatic adapter detection mode.Mouse (Mus musculus) reference GCF_000001635.27 and SARS-CoV-2 sequence NC 045512.2were used to create an index for read quantification using the salmon index (version 1.5.2).Subsequently, the trimmed reads were used for quantification with salmon quant applying 10 bootstraps as well as a sequence-and GC-specific bias correction.The quantification data were then imported into R (version 4.1.2) and RStudio (version 2021.09.1) using the function "tximport" [82].Using the expression values of the 1000 most variable genes within the dataset, a PCA was conducted using the "prcomp" function.DESeq2 [83] was used for the differential gene expression analysis.In detail, gene expressions from lung and spleen tissue samples of mice infected with either Beta (B.1.351)or BavPat1 (1, 3, and 5 DPI) were compared to a naïve noninfected con-trol.The resulting p values were adjusted using the Benjamini & Hochberg method [84] and the log2-foldchange values (LFC) were adjusted using the "apeglm" function [85].Genes were considered significantly differentially expressed when the adjusted p value was less than 0.05 and the absolute LFC was greater than 1.The "enrichGO" function from the package clusterProfiler [86] was used for further pathway analysis.The Gene Ontology (GO) enrichment analysis was conducted using a p value of less than 0.05 and a database of human biological processes.

RNA extraction for RT-qPCR
Processing and analysis were performed using established protocols [80].In brief, RNA from organ samples was extracted using the NuceloMag® VET Kit (Macherey-Nagel GmbH & Co. KG, Düren) and the Biosprint 96 platform (Qiagen GmbH).Viral RNA was quantified by real-time RT-qPCR with the CFX96 detection system (Bio-Rad Laboratories, Inc.) while targeting the viral RNAdependent RNA polymerase for amplification.

Type I IFN measurements
Mouse IFN-α ELISA Kit (Invitrogen) and mouse IFN-β LEGEND MAX TM ELISA Kit (BioLegend GmbH) were used according to the manufacturer's instructions.

Detection of RBD-specific antibody levels
RBD (ancestral SARS-CoV-2) reactive antibodies were measured using a previously described enzyme-linked immunoassay assay [87].

Cell isolation and flow cytometry
Cells were isolated and stained following previously published protocols [88].In brief, we discriminated parenchymal from vascular lymphocytes by intravascular application of anti-mouse CD45 antibodies [89].Harvested tissues were mechanically disrupted or enzymatically digested to generate a single-cell suspension.Cells were enumerated manually or using the TC20 automated cell counter (Bio-Rad Laboratories Inc.).A total of 2 × 10 6 cells were stained with viability dye (Zombie UV TM ; BioLegend GmbH), blocked (TruStain FcX TM ; BioLegend GmbH) and stained with monoclonal antibody to detect immune cell subsets (Supporting Information Table S2).Cells were fixed and permeabilized for intracellular staining with eBioscience TM Foxp3/transcription factor staining buffer set (ThermoFisher Scientific Inc.).Finally, cells were fixed for 30 min at room temperature (RT) with 4 % paraformaldehyde (PFA).Stained cells were acquired using a BD LSR Fortessa TM Cell Analyzer (BD Biosciences).The data were analyzed using BD FACSDiva and FlowJo (version 10.5.3).

In vitro restimulation
SARS-CoV-2-specific responses were investigated by in vitro restimulation of cells with SARS-CoV-2 spike peptides, followed by flow cytometry measurements of cytokines produced by lymphocytes.Single-cell suspensions were cultured in cRPMI supplemented with 0.5 μg/mL PepMix TM SARS-CoV-2 (Spike Glycoprotein) from JPT Peptide Technologies GmbH (#PM-WCPV-S-3) for 5 h at 37 °C and 5 % CO 2 .Brefeldin A (1000× Brefeldin A; BioLegend GmbH) diluted to 1× concentration was added after 1 h of stimulation.

Multiplex cytokine assay
The homogenized spleen and lung supernatants were analyzed using the Milliplex Mouse cytokine/chemokine immunology multiplex assay kit (#MCYTMAG-70K-PX32; Millipore/Merck KGaA) following the manufacturer's protocol.Premixed magnetic beads, detection antibodies, and streptavidin-phycoerythrin were diluted 1:2 in assay buffer.Beads were incubated with supernatants for 17 h at 4°C on a plate shaker.After completing the assay, samples were fixed in 4 % PFA for 30 min at RT. Afterwards, PFA was removed, and samples were resuspended in Bio-Plex sheath fluid (#171000055; Bio-Rad Laboratories Inc.) and immediately analyzed on a Bio-Plex 200 System instrument (Bio-Rad Laboratories Inc.).

Figure 4 .
Figure 4. Early kinetics of adaptive immune responses in the SARS-CoV-2 K18-hACE2 mouse infection model 1-5 DPI.(A-C) Cell counts (A) or fold change (B and C) analysis of T cells acquired by flow cytometry of the lungs (CD45iv + and CD45iv − ) and spleens (CD45iv − ) of K18-hACE2 mice infected with BavPat1 (ancestral) or Beta (B.1.351)up to 5 DPI.(B) Fold change of CXCR3 expression on CD4 + and CD8 + T cells.(C) Fold change of CD4 + and CD8 + effector memory T cells (CD44 hi , CD62L lo ).(B and C) The dashed line (-) represents the marker expression and T em abundance baseline in naïve animals.(A-C) p Values were determined by nonparametric one-way ANOVA and Dunn's multiple comparison test.Asterisks indicate statistical significance (*p < 0.05, **p < 0.005) compared with naïve animals.Each data point represents one animal.n = 4 per group and time point; n = 7 for BavPat1 5 DPI.The proportion of each CD4 + and CD8 + T cell population was used to calculate the fold change relative to naïve mice.Statistical analysis was applied to these proportions.

Figure 6 .
Figure 6.SARS-CoV-2-specific T and B cells reduce respiratory viral loads, leaving the survival of K18-hACE2 mice essentially unchanged.(A) Experimental design: B or T cells were depleted in K18-hACE2 transgenic mice prior to infection with 10 3.0 TCID 50 of the Delta (B.1.617.2) variant.The depletion was repeated at 5 DPI and the animals were monitored up to 12 DPI collecting samples for flow cytometry, receptor-binding domain ELISA, and viral loads.Created with BioRender.com(B) lung single-cell suspensions from 7 to 12 DPI were in vitro restimulated with S peptides and analyzed for IFN-γ + CD3 + T cells by intracellular flow cytometry staining.The flow cytometry plots of all animals were concatenated and normalized by downsampling to 130 000 events per plot to generate representative overviews.n = 7 for isotype control, n = 8 for T cell depletion, and n = 5 for B cell depletion.(C) Serum RBD ELISA of mice survived until 7-12 DPI.The dotted lines ( … ) represent the range between a detected ELISA reactivity (OD = 0.2) and seroconversion (OD = 0.3).(D) Survival of T or B cell-depleted versus control mice.n = 12 per group.(E) Conchae and lung viral loads at 12 DPI.(F) Correlation matrix including organ viral loads, immune cell events detected in flow cytometry, and RBD ELISA data.Spearman's rank correlation coefficient is shown.n = 15 (n = 5 for isotype control, n = 7 for T cell depletion, n = 3 for B cell depletion).(B and E) p Values were determined by nonparametric one-way ANOVA and Dunn's multiple comparison test.(B, E, and F) Asterisks indicate statistical significance (*p < 0.05, **p < 0.005, ***p < 0.0005).(C and E) Each data point represents one animal.n = 9 for each group.