Vaccine-induced protection against SARS-CoV-2 requires IFN-γ-driven cellular immune response

BNT162b2 induces robust cellular responses in B lymphocyte-deficient mice

To evaluate the cellular responses in μMT mice, we collected spleen tissues from vaccinated and unvaccinated C57BL/6 J wild type (WT) and μMT mice. The splenocytes were then stimulated with spike protein of SARS-CoV-2 (Fig. 1a). We first quantified antibodies titer against SARS-CoV-2 spike with ELISA and confirmed that antibody response was absent in BNT162b2-vaccinated μMT mice (Fig. 1b). Our flow cytometry analysis demonstrated that 13.2% of CD4⁺ and 18.5% of CD8⁺ T memory cells in spleen tissue of μMT mice immunized with BNT162b2 secreted IFN-γ upon stimulating with spike proteins of Alpha variant (Fig. 1c). In parallel, spike protein of Omicron BA.1 induced IFN-γ production in 17.3% of CD4⁺ and 16.1% of CD8⁺ T memory cells, suggesting robust induction of cellular responses upon stimulation of spike proteins from different SARS-CoV-2 variants (Fig. 1d). Importantly, the mean percentage of CD4⁺ and CD8⁺ T memory cells producing IFN-γ is higher in vaccinated μMT mice than in WT mice (Fig. 1c, d), suggesting a higher level of activation of cellular responses in the vaccinated μMT mice when compared to WT mice. In contrast, IL-4 expression, which reflects the differentiation of B cells into plasma cells¹³, was not stimulated by the spike proteins in either WT or μMT mice (Fig. 1e, f). Besides, it is reported that CD8⁺ T cells in B cell-deficient mice were skewed more toward effector phenotype during viral infection^14,15. Interestingly, our data revealed that stimulation of spike protein of SARS-CoV-2 could also induce the skewed differentiation of CD8⁺ T cells into highly cytotoxic terminal effector cells in the spleen of vaccinated µMT mice (Fig. S1a). The proportion of long-lived CD8⁺ T memory cell precursors is, however, dramatically reduced in vaccinated µMT mice compared to that in vaccinated WT mice (Fig. S1a). Taken together, our results demonstrated that robust cellular immunity was elicited in μMT mice immunized with BNT162b2 in the absence of humoral immunity.

a Schematic diagram of the vaccination scheme, ELISA, and flow cytometry. b Antibodies titer specific for spike protein of SARS-CoV-2 was measured by ELISA (n = 5). c–f The splenocytes were stimulated by spike proteins of SARS-CoV-2 Alpha or Omicron BA.1 variant and subjected to flow cytometry analysis for IFN-γ and IL-4 responses (n = 3). Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = not significant). Figure 1a was created with BioRender.com.

BNT162b2 elicits significant protective immunity against SARS-CoV-2 Alpha and Omicron BA.1 in B lymphocyte-deficient mice

To assess the requirement of antibody-mediated immunity in the protection against SARS-CoV-2, we vaccinated C57BL/6 J WT and μMT mice with the BNT162b2 mRNA vaccine and infected the vaccinated mice with SARS-CoV-2 Alpha or Omicron BA.1 at 21 days post boost (Fig. 2a). We found that BNT162b2 induced protective immunity against SARS-CoV-2 Alpha and Omicron BA.1 variant in both WT and μMT C57BL/6 J mice. BNT162b2 vaccination significantly inhibited Alpha and Omicron BA.1 replication in the lung tissues and nasal turbinate (NT) tissues of both WT and μMT mice at 2 dpi or 4 dpi (Fig. 2b, c). Despite the lack of B cells, BNT162b2 vaccination reduced Alpha replication in the lungs of μMT mice by 11.5-fold (P = 0.0274) and 11.6-fold (P = 0.0151) on 2 and 4 dpi, respectively (Fig. 2b). Similarly, BNT162b2 vaccination reduced Omicron BA.1 replication in the lungs of μMT mice by 3.1-fold (P = ns) and 50-fold (P = 0.0188) on 2 and 4 dpi, respectively (Fig. 2c).

a Schematic diagram of vaccination, viral challenge, and pathological studies in C57BL/6 J model. b, c The viral loads in the lung and nasal turbinate (NT) of BNT162b2 vaccinated and unvaccinated C57BL/6 J WT/μMT mice at 2 d.p.i. and 4 d.p.i. (n = 6). d, e Representative images of the H&E-stained lung tissues of BNT162b2 vaccinated and unvaccinated C57BL/6 J WT/μMT mice challenged with Alpha (d) or Omicron BA.1 variant (e). Scale bar = 200 μm. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = not significant). Figure 2a was created with BioRender.com.

SARS-CoV-2 pathogenesis is associated with the accumulation of proinflammatory myeloid cells within the lung tissue in mouse model^16,17. To identify correlates of vaccine-induced immunity in C57BL/6 J mice, we assessed the histopathological changes in the lung tissues of vaccinated and unvaccinated mice after SARS-CoV-2 Alpha and Omicron BA.1 challenge. For mice infected with Alpha, lungs in vaccinated WT mice had the least severe pulmonary pathology with relatively intact structure at 2 dpi or 4 dpi (Fig. 2d). More interstitial pneumonia and immune cell influx were observed in unvaccinated WT mice. Similarly, reduced in vivo pathology was detected in vaccinated μMT mice when compared with the unvaccinated ones. For mice infected with Omicron BA.1, the pathology of virus-induced lung tissues was milder than that of counterparts infected with Alpha, in keeping with recent findings¹⁸ (Fig. 2d, e). In addition, the degree of inflammatory cell infiltration in the lungs of vaccinated μMT mice was lower than that observed in the lungs of un-vaccinated μMT mice (Fig. 2d, e). Together, these findings indicated that BNT162b2 vaccination is capable of inducing protective immunity in μMT mice in the absence of the antibody response, suggesting cellular immunity may serve to protect against SARS-CoV-2 infection.

Cytokines are key coordinators of the cellular immune response and essential to the activation of T cells. We next evaluated the impact of BNT162b2 vaccination on the cytokine milieu in mice infected with SARS-CoV-2 Alpha (Fig. 3a, b). Cytokine profiling of blood samples from mice challenged with Alpha revealed that IFN-γ, CXCL1, MCP-1 and IP-10 protein concentrations were significantly higher in vaccinated μMT mice than those in the unvaccinated mice (Fig. 3c). In particular, Alpha challenge drove a 253-fold (p < 0.0001) increase of IFN-γ secretion in vaccinated μMT mice when compared to un-vaccinated μMT mice (Fig. 3c), suggesting strong activation of the cellular immunity upon viral challenge in vaccinated μMT mice. On the contrary, BNT162b2 vaccination is associated with a marked decrease of these key cytokines in WT mice when compared with unvaccinated ones (Fig. 3c).

a Schematic diagram of vaccination, viral challenge, and multiplex cytokine/chemokine profiling. b Cytokine/chemokine profiles in blood samples collected from BNT162b2 vaccinated and unvaccinated C57BL/6 J WT/μMT mice at 2 d.p.i. of Alpha variant determined by multiplex profiling (For WT mice, n = 10; μMT mice, n = 12). c Schematic diagram of vaccination, viral challenge, and cytokine/chemokine measurement by q-RTPCR. d–i The IFN-γ, IP-10, and MCP-1 expression was determined by q-RTPCR using RNA samples extracted from lung and NT from BNT162b2 vaccinated and unvaccinated C57BL/6 J WT/μMT mice at 2 d.p.i. and 4 d.p.i. of Alpha (d, f, h) or Omicron BA.1 variant (e, g, i) (n = 6). Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = not significant). Figure 3a, c were created with BioRender.com.

We next measured the cytokine levels in lung and NT tissue of vaccinated and unvaccinated mice upon Alpha or Omicron BA.1 infection by quantitative reverse transcription PCR (RT-qPCR). In line with the data in blood samples, BNT162b2 vaccination increased the expression of IFN-γ, MCP-1, and IP-10 in lung and NT tissues of μMT mice challenged with Alpha or Omicron BA.1 (Fig. 3d–i).

The K18-hACE2 mouse model has been frequently used to assess SARS-CoV-2 pathogenesis^12,19. To evaluate the impact of B cell depletion in the K18-hACE2 mice, we obtained K18-hACE2 μMT mice by breeding K18-hACE2 mice with μMT mice. In the K18-hACE2/μMT mice, BNT162b2 vaccination similarly offered protection against Alpha and Omicron BA.1 infection. Alpha and Omicron BA.1 replication in the lung and NT tissues were lower in vaccinated K18-hACE2/μMT mice when compared with un-vaccinated ones (Fig. 4b, c). In line with the results in C57BL/6 J model, we observed ameliorated lung pathology in vaccinated WT or μMT mice when compared with unvaccinated ones (Fig. 4d, e). For cytokine profiling, we quantified IFN-γ, MCP-1, and IP-10 expression in lung tissue and NT in the mice by RT-qPCR. We found that BNT162b2 vaccination increased the expression of IFN-γ, MCP-1, and IP-10 in lung and NT tissues of K18-hACE2 μMT mice challenged with Alpha or Omicron BA.1 (Fig. 4f-k), similar to what we observed in C57BL/6 J μMT mice (Fig. 3d–i). Taken together, these findings indicate that BNT162b2 elicits significant protective immunity against SARS-CoV-2 Alpha and Omicron BA.1 in B lymphocyte-deficient mice.

a Schematic diagram of vaccination, viral challenge and cytokine/chemokine measurement, and pathological studies in K18-hACE2 model. b, c The viral loads in the lung and NT of BNT162b2 vaccinated and unvaccinated K18-hACE2 WT/μMT mice at 2 d.p.i. and 4 d.p.i. (n = 6). d, e Representative images of the H&E-stained lung tissues of BNT162b2 vaccinated and unvaccinated K18-hACE2 WT/μMT mice challenged with Alpha (d) or Omicron BA.1 variant (e). f–k The IFN-γ, IP-10, and MCP-1 level was quantified by q-RTPCR using RNA samples extracted from lung and NT from BNT162b2 vaccinated and unvaccinated K18-hACE2 WT/μMT mice at 2 d.p.i. and 4 d.p.i. of Alpha (f, h, j) or Omicron BA.1 variant (g, i, k) (n = 6). Scale bar = 200 μm. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = not significant). Figure 4a was created with BioRender.com.

T cell-mediated immunity is highly activated in B lymphocyte-deficient mice upon SARS-CoV-2 infection

To characterize the vaccine-induced protective immunity in μMT mice, we performed bulk RNA-seq of WT and µMT C57BL/6 J mice lung and NT tissues with or without vaccination on 2 dpi (Fig. 5a). Gene functional enrichment analysis based on the significantly differential expressed genes showed prominent immune characters between different conditions (Fig. 5b). We found that BNT162b2 vaccination boosted both humoral and cellular immunity in both NT and lung tissues in WT mice (Fig. 5b). In contrast, due to the lack of mature B cells, only genes involved in T cell activation and T cell receptor signaling pathways were significantly upregulated in vaccinated µMT mice comparing to unvaccinated µMT mice, suggesting that cellular immunity can be effectively activated by BNT162b2 vaccination in B cell-deficient mice (Fig. 5b). Cell type functional enrichment also indicated that the function of CD8 + T cell is highly enriched in vaccinated µMT mice when compared to the unvaccinated µMT mice (Fig. S2). According to the differential gene expression analysis, we found that T cell-related genes/pathways were significantly upregulated in the vaccinated µMT mice in both NT and lung tissue when compared to unvaccinated µMT mice (Fig. 5c, d). Among these genes, integral membrane glycoprotein CD8a serves as a coreceptor of class I MHC molecular in CD8⁺ T cells, and cell adhesion molecular CD6 that regulates T cell responses was both upregulated in NT and lung tissues (Fig. 5c, d). Besides, the cxcr6 gene, which controls the localization of resident memory T lymphocytes to different compartments of the lung and maintains airway resident memory T lymphocytes²⁰, was also significantly upregulated in vaccinated µMT mice (Fig. 5c, d).

a Schematic diagram of the design of the RNA-seq. C57BL/6 J µMT and WT mice were either unvaccinated or vaccinated with BNT162b2 three weeks before SARS-CoV-2 Alpha challenge. NT and lung tissues were then collected 48 h after the challenge for further RNA-seq library construction and sequencing. b Heatmap plot showing Gene ontology and KEGG pathway where significantly differentially expressed genes are enriched. The colored block indicated genes enriched terms showed in the right with corresponding Benjamini-Hochberg adjusted P values. (UV: unvaccinated) c Volcano plot showing differential expression genes from DESeq2 comparing BNT162b2 vaccinated transcriptome to unvaccinated ones in both NT and lung tissues of µMT mice. Genes with orange labels were significantly upregulated in NT or lung (Benjamini-Hochberg adjusted P value < 0.05 and gene expression fold change > 2), scatters with grey labels indicated gene upregulated in both NT and lung (Benjamini-Hochberg adjusted P value < 0.05 and gene expression fold change > 2). d Heatmap plot showing the expression levels of upregulated genes labeled in (b) across different conditions, expression of each gene was scaled in NT and lung, respectively. Figure 5a was created with BioRender.com.

IFN-γ facilitates vaccine-induced immunity in B lymphocyte-deficient mice

As mentioned earlier, IFN-γ secretion was substantially upregulated in serum, NT, and lung tissues of vaccinated μMT mice. To further evaluate the role of IFN-γ in the vaccine-induced protective immunity in μMT mice, we injected IFN-γ depleting antibody into vaccinated WT and μMT mice before Omicron BA.1 challenge (Fig. 6a). Our results demonstrated that the IFN-γ depleting antibody significantly increased Omicron BA.1 replication by 19.3-fold (P = 0.0029) in the lung of vaccinated μMT mice. In contrast, the IFN-γ depleting antibody did not modulate Omicron BA.1 replication in the lungs of vaccinated WT mice (0.7-fold, P = NS) (Fig. 6b). In the NT, the IFN-γ depleting antibody has a trend to promote virus replication in vaccinated WT and μMT mice, albeit no statistical significance was reached (Fig. 6b). In keeping with these observations, treatment of IFN-γ depleting antibody resulted in more severe lung pathology in vaccinated μMT mice while no significant change in lung pathology was observed in vaccinated WT mice when IFN-γ depleting antibody was administered (Fig. 6c). A recent study reported that supplementation of IFN-γ reverses the age-dependent COVID-19 phenotype in mice²¹. Therefore, we next treated naïve μMT mice daily with 10 μg IFN-γ from −1 to 1 d.p.i. to investigate whether administration of IFN-γ could alleviate the disease severity upon SARS-CoV-2 infection. Supplementation of IFN-γ resulted in lower viral load and milder lung pathology in NT and lung tissues of naïve μMT mice than PBS control group when infected with SARS-CoV-2 Alpha (Fig. 6e, f). Together, our results indicate that IFN-γ contributes to the vaccine-induced cellular immunity that offers protection against SARS-CoV-2 in B lymphocyte-deficient mice.

a Schematic diagram of vaccination, IFN-γ depletion, viral challenge, and pathological studies. b The viral loads in the lung and NT of BNT162b2 vaccinated and unvaccinated C57BL/6 J μMT mice at 2 d.p.i. with IFN-γ depletion (n = 10). (mIFN-γ, anti-IFN-γ monoclonal antibody). c Representative images of the H&E-stained lung tissues of BNT162b2 vaccinated WT/μMT mice treated with mIFN-γ or PBS. Scale bar = 200 μm. d Schematic diagram of IFN-γ administration in naïve μMT mice upon Alpha challenge. e The viral loads in the lung and NT of naïve μMT mice supplemented with IFN-γ or PBS at 2 d.p.i. (n = 6). f Representative images of the H&E-stained lung tissues of naive μMT mice treated with IFN-γ or PBS. Scale bar = 200 μm. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA test or unpaired two-tailed Student’s t-test. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = not significant). Figure 6a, d were created with BioRender.com.

CD4⁺ and CD8⁺ T cells are required for the vaccine-induced clearance of SARS-CoV-2 in NT and lung tissue in B lymphocyte-deficient mice

As previously mentioned, CD4⁺ and CD8⁺ T memory cells in vaccinated μMT mice produce IFN-γ upon spike protein stimulation (Fig. 1c, d). We then investigated if CD4⁺ and CD8⁺ T cells contribute to the vaccine-induced cellular immunity in μMT mice by using CD4-, CD8-, or CD4/CD8-depleting antibodies. In lung tissues, CD4-, CD8-, and CD4/CD8-depleting antibodies increased Omicron BA.1 replication by 10-fold (P = NS), 10-fold (P = NS), and 42-fold (P = 0.0038), respectively (Fig. 7b). In the NT tissues, CD4-, CD8-, and CD4/CD8-depleting antibodies increased Omicron BA.1 replication by 0.6-fold (P = NS), 15.6-fold (P = NS), and 7.6-fold (P = 0.028), respectively (Fig. 7b). In keeping with the virus replication findings, simultaneous depletion of both CD4⁺ and CD8⁺ T cells resulted in the most severe lung pathology in vaccinated μMT mice among all evaluated groups (Fig. 7c). To further dissect the role of CD4⁺ and CD8⁺ T cells, we characterized tissue-resident memory (TRM) and effector cells in lung tissues of vaccinated μMT mice by flow cytometry analysis. Consistent with the bulk RNA-Seq data, our results demonstrated that 26-fold (P = 0.015) higher percentage of CD44⁺ CD69⁺ CD8⁺ T cells were found in lung tissues of vaccinated μMT mice than in unvaccinated ones after SARS-CoV-2 Alpha infection (Fig. 7e). In addition, the percentage of CXCR6⁺ lung tissue-resident memory (TRM) CD8⁺ T cells was also significantly higher (67.6-fold, P = 0.0007) in vaccinated μMT mice (Fig. 7e, S7a). More IFN-γ⁺ and Granzyme B⁺ CD8⁺ T cells were detected in vaccinated μMT mice (5.5-fold, P = 0.0159 and 8.3-fold, p < 0.0001, respectively) when compared with unvaccinated μMT mice, suggesting that robust cellular immunity was induced in vaccinated μMT mice upon SARS-CoV-2 challenge (Fig. 7f, g). Likewise, the percentage of CD44⁺ CD69⁺ CD4⁺, CXCR6⁺ CD4⁺, and IFN-γ⁺ CD4⁺ T cells were also higher in vaccinated μMT mice than unvaccinated mice, but less significant differences were found (Fig. 7e–g, S7a, b). Taken together, these data indicate that while both CD4⁺ and CD8⁺ T cells are required for optimal vaccine-induced protection against SARS-CoV-2 in lung tissues of μMT mice, CD8⁺ T cells play a more predominant role than CD4⁺ T cells in the NT of μMT mice.

a Schematic diagram of vaccination, CD4⁺ or CD8⁺ T cell depletion, viral challenge, and pathological studies. b The viral loads in the lung and NT of BNT162b2 vaccinated and unvaccinated C57BL/6 J μMT mice at 2 d.p.i. with CD4⁺, CD8⁺, or both T cell depletion (n = 10). (aCD4, anti-CD4 monoclonal antibody; aCD8, anti-CD8 monoclonal antibody). c Representative images of the H&E-stained lung tissues of BNT162b2 vaccinated μMT mice treated with aCD4, aCD8, both depleting antibodies or PBS. Scale bar = 200μm. d Schematic diagram of flow cytometry analysis of CD4⁺ and CD8⁺ T cells in lung tissues of vaccinated and unvaccinated C57BL/6 J μMT mice after Alpha infection. e Percentage of CXCR6⁺ and CD69⁺ CD4/8⁺ T cells in lung tissues of vaccinated and unvaccinated C57BL/6 J μMT mice after Alpha infection (n = 5). f Percentage of CD44⁺ IFN-γ⁺ CD4/8⁺ and CD44⁺ Granzyme B⁺ CD8⁺ T cells in lung tissues of vaccinated and unvaccinated C57BL/6 J μMT mice after Alpha infection (n = 5). g Representative dot plots showing IFN-γ- and Granzyme B-producing CD44⁺ CD4/8⁺ T cells in lung tissues of vaccinated and unvaccinated C57BL/6 J μMT mice after Alpha infection. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA test or unpaired two-tailed Student’s t-test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = not significant). Figure 7a, d were created with BioRender.com.

BNT162b2 induces protective immunity against SARS-CoV-2 Omicron BA.5.2 in B lymphocyte-deficient mice

SARS-CoV-2 Omicron BA.5, which has substantially higher immune escape capacity than BA.1, are the dominant strain in many countries as of January 2023^22,23,24. To test whether BNT162b2 could induce protective immunity against SARS-CoV-2 Omicron BA.5.2 in μMT mice, we infected vaccinated C57BL/6 J and K18-hACE2 mice with SARS-CoV-2 Omicron BA.5.2 and evaluated the viral replication in these mice. The quantitative RT-PCR showed that the viral load in lung tissues of vaccinated μMT mice is 34-fold (P = 0.0283) and 188-fold (P = NS) lower than C57BL/6 J and K18-hACE2 unvaccinated μMT mice at 2 d.p.i., respectively (Fig. 8b). Although much lower viral load was detected in NT of both C57BL/6 J and K18-hACE2 vaccinated mice at 2 and 4 d.p.i., the statistical difference was not seen between the vaccinated and unvaccinated groups (Fig. 8b). In line with the viral replication, the lung pathology was also alleviated in vaccinated mice (Fig. 8c). Consistent with the therapeutic effect of IFN-γ in naïve μMT mice against SARS-CoV-2 Alpha, supplementation of IFN-γ could also mitigate the viral replication and lung pathology in naïve μMT mice upon Omicron BA.5.2 challenge (Fig. 8d, e).

a Schematic diagram of vaccination, viral challenge, and pathological studies in C57BL/6 J and K18-hACE2 model. b, c The viral loads in the lung and NT of BNT162b2 vaccinated/unvaccinated C57BL/6 J and K18-hACE2 WT/μMT mice at 2 d.p.i. and 4 d.p.i. (n = 6). d The viral loads in the lung and NT of naïve μMT mice supplemented with IFN-γ or PBS at 2 d.p.i. upon Omicron BA.5.2 infection (n = 6). e, f Representative images of the H&E-stained lung tissues of BNT162b2 vaccinated/unvaccinated C57BL/6 J (e) and K18-hACE2 (f) WT/μMT mice challenged with Omicron BA.5.2 Scale bar =  200  μm. g Representative images of the H&E-stained lung tissues of naive μMT mice treated with IFN-γ or PBS. Scale bar = 200 μm. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns = not significant). Figure 8a was created with BioRender.com.