Ad5-boosted COVID-19 vaccine reduces symptomatic BA.5 infection via cross-neutralizing antibodies and NK cell immunity Authors: Tingting Cui, Siyi Liu, Tingting Wang, Yarui Liu, Xinyue Liang, Faming Chen, Mingzhu Huang, Xiaoling Su, Tao Song, Jie Su, Shidong Deng, Jincun Zhao, Fulong Yu, Zhongfang Wang Published: 2025 ABSTRACT BackgroundDuring the COVID-19 pandemic, diverse vaccine strategies were applied globally, yet most relied on immunogenicity evaluations and bridging trials rather than retrospective assessment through real-world infections. This limited understanding of protective mechanisms under natural variant exposure.MethodsA retrospective cohort study of real-world BA.5 infections was used to analyze the protective efficacy of Ad5-nCoV booster vaccination (A-A). Multi-layered immune profiling was performed, including neutralizing antibodies (NAbs), single-cell BCR sequencing (BCR usage), transcriptomic profiling (RNA-seq), antibody-dependent cellular cytotoxicity (ADCC), natural killer (NK) cell cytotoxicity, and virus-specific T cell immunity. These were integrated to dissect A-A-induced protection mechanisms.ResultsThe A-A conferred 48% effectiveness against symptomatic BA.5 infection (115/239) compared to the three doses of inactivated vaccine (I-I-I), which showed 10% effectiveness (4/39). Specifically, in the A-A group, 52% (124/239) of individuals experienced symptomatic infection, 38% (90/239) had asymptomatic infection, and 10% (25/239) remained uninfected. Compared to I-I-I, A-A induced significantly higher NAbs against BA.5 (GMT = 180 vs. I-I-I GMT = 54) and comparable anti-WT NAbs (GMT = 358 vs. 226), despite lower anti-S IgG levels (GMT = 6441 vs. 12,228) at 2 months post BA.5 infection. Single-cell BCR analysis revealed broader neutralizing antibody clone types in A-A, dominated by IGHV4-39 usage with elevated somatic mutation frequency. Additionally, transcriptomic and functional assays demonstrated an enhanced NK cell-mediated ADCC response and cytotoxicity against K562 target cells in A-A.ConclusionsAd5-nCoV booster vaccination confers enhanced protection against symptomatic BA.5 infection through cross-reactive NAbs with expanded epitope breadth and NK cell-mediated innate immunity enhancement. This retrospective study provides insights into the immunogenicity and molecular mechanisms of two vaccination strategies against real-world BA.5 infection, highlighting Ad5-based technology’s efficacy. It provides guidance for future vaccination strategies against emerging Disease X, emphasizing the value of retrospective real-world analyses in vaccine development.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12916-025-04505-0. Background Since its emergence in December 2019, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has triggered a global pandemic characterized by substantial morbidity and mortality. To mitigate coronavirus disease 2019 (COVID-19), diverse vaccination strategies have been deployed worldwide. In mainland China, despite extensive administration of inactivated vaccines prior to December 2022, a surge of Omicron BA.5/BF.7 infections emerged in late 2022. A survey revealed that the self-reported infection rate reached its peak between December 19 and 21, 2022, with 82.4% of the Chinese population being infected as of February 7, 2023 [1]. Alongside inactivated vaccines, the adenovirus type 5 (Ad5) vectored COVID-19 vaccine (CanSino Biologics, Tianjin, China) has been widely utilized in China. Adenoviral vectors offer distinct advantages, including low pathogenicity, genetic safety, and non-integration into the host genome [2, 3]. These vectors not only elicit robust antibody responses but also stimulate cellular responses critical for eliminating pathogen-infected cells. A single intramuscular dose of Ad5 vaccine has demonstrated high immunogenicity, with RBD-specific IgG antibodies peaked at 656.5, and this immunogenicity remained for at least 6 months post vaccination [4, 5]. A phase 3 clinical trial of a single dose of an Ad5 SARS-CoV-2 vaccine reported 57.5% efficacy against PCR-confirmed COVID-19 and 91.7% efficacy against severe disease at 28 days post vaccination [6]. Similarly, an international phase 3 trial of the Ad26 SARS-CoV-2 vaccine showed 66.1% efficacy against moderate to severe COVID-19 at 28 days, increasing to 85.4% for critical cases [7]. To enhance immunogenicity, booster strategies have been evaluated. The heterologous prime-boost regimen of Sputnik V, employing Ad26 for priming and Ad5 for boosting, achieved 91.6% efficacy against PCR-confirmed symptomatic COVID-19, and was 100% efficacious against moderate or severe COVID-19 [8]. Furthermore, heterologous boosting with an aerosolised Ad5-nCoV following a two-dose inactivated vaccine regimen elicited significantly higher neutralizing antibodies compared to homologous three-dose inactivated vaccination [9–11]. However, the long-term protective efficacy of an Ad5-nCoV booster administered intramuscularly 1 year post-primary vaccination, particularly against BA.5 infection, remains inadequately characterized. The protective efficacy of wild-type (WT)-targeted vaccines against heterologous BA.5 infection is determined by multiple factors. From an immunological perspective, the number of cross-reactive epitopes shared between WT and BA.5 induced by distinct vaccine platforms, and the magnitude of NAbs specific to these epitopes play important roles in determining the protective efficacy. Most studies focus on comparing the level of collective NAbs against heterologous variants to evaluate the protective efficacy [12–15]. For instance, hybrid immunity—generated by prior vaccination followed by Omicron BA.1/BA.2 breakthrough infection—induces significantly higher NAb titers against BA.5 compared to vaccination or infection alone, conferring enhanced cross-protection [14, 15]. Adenovirus-based vaccines exhibit unique immunomodulatory properties. Beyond eliciting robust T cell immunity, their single-cycle replication mechanism may activate trained immunity in innate immune cells such as natural killer (NK) cells and monocytes, a feature not observed with three-dose inactivated vaccines [16–20]. This divergence in immune activation could lead to functional differences in cytotoxicity and antibody-dependent cellular cytotoxicity (ADCC) mediated by NK cells. NK cells have been shown to collaborate with vaccine-elicited antibodies to prevent infection [21, 22]. Rahman and colleagues demonstrated that SIV/HIV V1-deleted vaccines prime mucosal/systemic memory-like NK cell responses. These “trained” NK cells were associated with a reduced risk of SIV/HIV infection and also correlated with a V2-specific ADCC response, highlighting the vaccine-elicited adaptive NK cells to limit pathogen infection [23]. Furthermore, NK cells responding to booster vaccinations display a more mature phenotype and heightened cytotoxic activity compared to primary vaccination [24]. Despite these advances, the functional dynamics of NK cells post BA.5 breakthrough infection remain poorly characterized and the durability of hybrid immunity against emerging variants remains unclear. Comparative analyses of immune dynamics in breakthrough-infected populations with varying vaccine booster regimens remain scarce. Here, we initially conducted a comparative analysis of the clinical manifestations and humoral immune responses between individuals who had received a two-dose Ad5 vaccine and those who received a three-dose inactivated vaccine, focusing on monitoring the protective efficacy. To delve into the factors underlying the enhanced protection conferred by the Ad5 booster, we quantified NAbs titers and assessed antigen-specific T cell responses post BA.5 infection. Subsequently, BA.5-specific memory B cells were isolated for single B cell receptor (BCR) sequencing to characterize clonal diversity and somatic hypermutation patterns. Transcriptomic profiling via RNA-seq was employed to compare systemic immune profiles between Ad5-boosted and inactivated vaccine cohorts, while NK cell functional responses were rigorously evaluated. Our study presents an innovative investigation into the mechanistic basis of Ad5-booster mediated protection against BA.5 infection, while concurrently exploring the longevity of cross variant immunity, thereby offering a comprehensive immunological framework for optimizing heterologous vaccine strategies. Methods Cohorts and samples collection To assess the protective efficacy of the Ad5-nCoV vaccine (CanSino Biologics, Tianjin, China) against BA.5 breakthrough infection, this study enrolled 239 participants who received two intramuscular doses of the Ad5-nCoV vaccine (A-A). The median age of the A-A group was 53 (IQR: 42–66) with 95 females (40%). Meanwhile, 10 (4%) participants had diabetes and 79 (33%) participants had hypertension (Table 1). For longitudinal immunogenicity assessments, peripheral blood samples were collected at day 0 (D0), 1 month (1 M), and 3 months (3 M) post-booster vaccination. Following a longitudinal observation period, additional samples were obtained at 2 months (2 M) after confirmed BA.5 breakthrough infection (Fig. 1A). A comparator cohort of 56 participants receiving three doses of inactivated vaccines (CoronaVac, I-I-I group) was simultaneously enrolled in this retrospective study. This inactivated whole-virus vaccine was based on the wild-type strain and the third dose was boosted 6 months after the second dose. To ensure comparability with the A-A group, the I-I-I cohort was selected based on specific criteria. All participants had completed three doses of inactivated vaccines (n = 15 for I-I-I 1 M post-booster vaccination; n = 15 for I-I-I 3 M post-booster vaccination), experienced BA.5 breakthrough infection, and had samples collected at 2 months post-infection (n = 26), consistent with the sampling timepoint of the A-A group (Fig. 1A, Table 2). This multi-timepoint design enabled systematic evaluation of vaccine-induced immunogenicity and its correlation with protective immunity against SARS-CoV-2 infection. To explore the effectiveness of I-I-I, another cohort of 39 individuals was recruited in this study, containing a median age of 32 (IQR: 29–34) with 24 females (62%, Table 3). Table 1Demographic information and clinical characterization of A-A individualsA-A groupSymptomaticAsymptomaticUninfectedTotal numberTotal number (n)1249025239Age (average years)535259Range18–39282115040–49201723950–592426106060–693111749 > 702115541GenderFemale58 (47%)31 (34%)6 (24%)95 (40%)Male66 (53%)59 (66%)19 (76%)144 (60%)ComorbiditiesDiabetes5 (4%)4 (4%)1 (4%)10 (4%)Hypertension40 (32%)28 (31%)11 (44%)79 (33%)Hypertension and diabetes6 ( 5%)3 (3%)3 (12%)12 (5%)Infection identificationAntigen test + − − Clinical symptomsFever78 (63%) − − 78 (63%)Fatigue31 (25%) − − 31 (25%)Cough49 (40%) − − 49 (40%)Muscle soreness43 (35%) − − 43 (35%)Nasal congestion13 (10%) − − 13 (10%)Sore throat30 (24%) − − 30 (24%)Diarrhea9 (7%) − − 9 (7%) + : positive; − :negativeFig. 1Immunogenicity and protective efficacy of A-A versus I-I-I against BA.5 infection. A Schematic of longitudinal sampling for A-A (n = 239) and I-I-I (n = 56) groups. B Symptomatic and asymptomatic infection rates in the A-A group. C Comparison of anti-BA.5, anti-WT NAbs and anti-S IgG titers between A-A (n = 239) and I-I-I (n = 26) groups at 2 M post BA.5 breakthrough infection. D Comparison of anti-BA.5 and anti-WT NAbs between symptomatic (n = 124) and asymptomatic (n = 90) A-A individuals at 2 M post BA.5 breakthrough infection. E Geometric mean titers (GMT) of anti-BA.5 NAbs, anti-WT NAbs, and anti-S IgG at 1 M and 3 M post-booster vaccination for both groups (n = 15). 80% of I-I-I samples had anti-BA.5 titers below the detection limit (4) at 1 M post-booster vaccination, assigned a value of 2. All the samples in the I-I-I group had anti-BA.5 titers below the detection limit (4) at 3 M post-booster vaccination, assigned a value of 2. The dotted lines indicated the geometric mean with 95% confidence intervals. The Mann–Whitney test was employed for comparing central tendencies of every two groups, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001Table 2Information of I-I-I boost vaccinated and BA.5 infected individualsI-I-I groupI-I-I1 M post booster vaccineI-I-I3 M postbooster vaccineI- I-III- 2 M postBA.5 infectionTotal number (n = 56)151526Age (average years)383458GenderFemale12 (80%)11 (73%)13 (50%)Male3 (20%)4 (27%)13 (50%)Infection identificationAntigen test − − + + : positive; − : negativeTable 3Clinical information of I-I-I individuals infected with BA.5I-I-I groupSymptomaticUninfectedTotal numberTotal number (n)35439Age (average years)314032GenderFemale21 (60%)3 (75%)24 (62%)Male14 (40%)1 (25%)15 (38%)Infection identificationAntigen test + − Clinical symptomsFever31 (89%) − Fatigue8 (23%) − Cough28 (80%) − Muscle soreness15 (43%) − Nasal congestion22 (63%) − Sore throat26 (74%) − Diarrhea3 (9%) − + : positive; − : negative Table 1Demographic information and clinical characterization of A-A individualsA-A groupSymptomaticAsymptomaticUninfectedTotal numberTotal number (n)1249025239Age (average years)535259Range18–39282115040–49201723950–592426106060–693111749 > 702115541GenderFemale58 (47%)31 (34%)6 (24%)95 (40%)Male66 (53%)59 (66%)19 (76%)144 (60%)ComorbiditiesDiabetes5 (4%)4 (4%)1 (4%)10 (4%)Hypertension40 (32%)28 (31%)11 (44%)79 (33%)Hypertension and diabetes6 ( 5%)3 (3%)3 (12%)12 (5%)Infection identificationAntigen test + − − Clinical symptomsFever78 (63%) − − 78 (63%)Fatigue31 (25%) − − 31 (25%)Cough49 (40%) − − 49 (40%)Muscle soreness43 (35%) − − 43 (35%)Nasal congestion13 (10%) − − 13 (10%)Sore throat30 (24%) − − 30 (24%)Diarrhea9 (7%) − − 9 (7%) + : positive; − :negative Demographic information and clinical characterization of A-A individuals + : positive; − :negative Immunogenicity and protective efficacy of A-A versus I-I-I against BA.5 infection. A Schematic of longitudinal sampling for A-A (n = 239) and I-I-I (n = 56) groups. B Symptomatic and asymptomatic infection rates in the A-A group. C Comparison of anti-BA.5, anti-WT NAbs and anti-S IgG titers between A-A (n = 239) and I-I-I (n = 26) groups at 2 M post BA.5 breakthrough infection. D Comparison of anti-BA.5 and anti-WT NAbs between symptomatic (n = 124) and asymptomatic (n = 90) A-A individuals at 2 M post BA.5 breakthrough infection. E Geometric mean titers (GMT) of anti-BA.5 NAbs, anti-WT NAbs, and anti-S IgG at 1 M and 3 M post-booster vaccination for both groups (n = 15). 80% of I-I-I samples had anti-BA.5 titers below the detection limit (4) at 1 M post-booster vaccination, assigned a value of 2. All the samples in the I-I-I group had anti-BA.5 titers below the detection limit (4) at 3 M post-booster vaccination, assigned a value of 2. The dotted lines indicated the geometric mean with 95% confidence intervals. The Mann–Whitney test was employed for comparing central tendencies of every two groups, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 Table 2Information of I-I-I boost vaccinated and BA.5 infected individualsI-I-I groupI-I-I1 M post booster vaccineI-I-I3 M postbooster vaccineI- I-III- 2 M postBA.5 infectionTotal number (n = 56)151526Age (average years)383458GenderFemale12 (80%)11 (73%)13 (50%)Male3 (20%)4 (27%)13 (50%)Infection identificationAntigen test − − + + : positive; − : negative Information of I-I-I boost vaccinated and BA.5 infected individuals + : positive; − : negative Table 3Clinical information of I-I-I individuals infected with BA.5I-I-I groupSymptomaticUninfectedTotal numberTotal number (n)35439Age (average years)314032GenderFemale21 (60%)3 (75%)24 (62%)Male14 (40%)1 (25%)15 (38%)Infection identificationAntigen test + − Clinical symptomsFever31 (89%) − Fatigue8 (23%) − Cough28 (80%) − Muscle soreness15 (43%) − Nasal congestion22 (63%) − Sore throat26 (74%) − Diarrhea3 (9%) − + : positive; − : negative Clinical information of I-I-I individuals infected with BA.5 + : positive; − : negative This study was approved and supervised by the GMUH Ethics Committee (No.2021–78). Enzyme-linked immunosorbent assay (ELISA) To quantify spike (S) protein-specific binding antibodies in plasma from vaccinated and breakthrough-infected individuals, an indirect ELISA method was conducted. Briefly, 96-well plates were coated with wild-type (WT) SARS-CoV-2 S protein and incubated overnight at 4 ℃. On the subsequent days, the plates were blocked with a solution of 3% milk in phosphate-buffered saline (PBS). A high titer anti-S plasma sample was designated as the reference standard, assigned a value of 2000 arbitrary units (AU/mL). A twofold serial dilution of this standard was included in each assay to generate a calibration curve. Test samples were diluted 1:100, added to plates, and incubated for 1 h. Following plasma incubation, horseradish peroxidase (HRP)-conjugated secondary antibody (109–035–088, Jackson ImmunoResearch Laboratories) was added and incubated at 37 °C for 1 h. After extensive washing, 100 μL of TMB stabilized chromogen (Invitrogen, USA) was added, and the cells were incubated at 37 °C for 10 min, followed by treatment with 50 μL of stopping solution (R&D Systems, USA). Absorbance was measured at 450 nm using a Multiskan GO microplate spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Data were analyzed using a standard curve with a logistic regression model. Peripheral blood mononuclear cell (PBMC) isolation and ex vivo stimulation PBMCs were isolated from heparinized whole blood by density-gradient centrifugation using Ficoll–Paque (GE Healthcare, Cat# 17–1440–02). Cells (5 × 105) were respectively stimulated with peptide pools spanning the antigen region of the spike protein S1 subunit (171 15-mer peptides), S2 subunit (145 15-mer peptides), and a combined pool of membrane (M), nucleocapsid (N), and envelope (E) proteins (N/M/E, 171 15-mer peptides). Stimulation was performed in RPMI 1640 medium (Gibco) supplemented with 10% heat-inactivated FBS (Biological Industries, Israel Beit-Haemek), 100 U/ml penicillin (Gibco), 0.1 mg/ml streptomycin (Gibco), 10 U/ml recombinant interleukin-2 (rIL-2) and 1 µM GolgiPlug (BD Biosciences, San Diego, CA) for 24 h at 37 °C, 5% CO2. Unstimulated controls were cultured in identical medium containing 0.01% DMSO (Sigma) without peptide pools. Flow cytometry Following 24 h ex vivo stimulation, cells were stained with LIVE/DEAD® Fixable Aqua Dead Cell Stain (Thermo Fisher Scientific, Cat #L34957) for 15 min at room temperature (RT). Then cells were washed and surface-stained for 30 min at RT in the dark with the following antibodies: BV605 mouse anti-human CD3 (BD Pharmingen™, Clone SP34-2, Cat#562,994), BV786 mouse anti-human CD4 (BD Pharmingen™, Clone SK3, Cat#563,877) and AF488 mouse anti-human CD8 (BD Pharmingen™, Clone RPA-T8, Cat#557,696). After fixation and permeabilization with Cytofix/Cytoperm™ solution (BD Bioscience, Cat# 554,714) on ice for 15 min, intracellular staining (ICS) was performed on ice for 30 min with antibodies against PE-Cy7 mouse anti-human TNF-α (BD, Clone MAb11, Cat# 557,647) and APC mouse anti-human IFN-γ (BD Pharmingen™, Clone B27, Cat# 554,702). Finally, cells were resuspended in 200 µl FACS buffer, acquired using a BD LSRFortessa X-20 instrument (BD Bioscience) and analyzed with FlowJo software (Treestar). SARS-CoVo-2 conventional virus neutralization test The neutralizing capacity of plasma samples collected from vaccinated individuals and BA.5 breakthrough infection cases was assessed using a cytopathic effect (CPE)-based assay, as described previously [25]. Plasma samples were tested at an initial dilution of 1:4, and then diluted in steps of 1:2 for eight points. These twofold serially diluted plasma samples were mixed with SARS-CoV-2 WT or Omicron BA.5 viral solution respectively, containing 100 50% tissue culture infective dose (TCID50) of the virus, in a 96-well plate, and incubated for 2 h at 37 °C and 5% CO2. Thereafter, Vero E6 cell suspension was added to the virus-plasma mixture at 1.2 × 104 cells per well. Plates were further incubated for 4 days at 37 °C with 5% CO2. CPE was evaluated using a Celigo Imaging Cytometer, and the results were compared to positive control (plasma samples showing high neutralizing activity against SARS-CoV-2 in infected Vero E6 cells) and negative control (human serum samples negative for SARS-CoV-2 in ELISA and neutralization assay, and Vero E6 cells alone) wells. As plasma samples were serially diluted twofold starting from 1:4, the detection limit of the assay was 4. Neutralizing antibody titers below the detection limit were defined as 50% inhibitory dilutions (NT50) titers of 2. RNA extraction and quantification of gene expression To compare post-BA.5 infection immune responses between A-A and I-I-I groups, we performed bulk RNA sequencing on PBMCs to systematically profile the transcriptomic landscape of immune-related genes, identify differentially expressed genes (DEGs) between the two groups, and explore potential molecular pathways underlying the observed differences in immune responses. Nine samples from each group were randomly selected based on confirmed BA.5 breakthrough infection and balanced representation of key demographic characteristics (Additional file 1: Table S1). In detail, PBMC samples cryopreserved in liquid nitrogen were thawed at 37 °C and washed once with PBS. Total RNA was extracted using the Eastep® Super Total RNA Extraction Kit (Shanghai Promega, Cat# LS1040) according to the manufacturer’s instructions. RNA-sequencing was then delivered to the company (Genedenovo Guangzhou) for subsequent analysis. RNA-sequencing and differentially expressed gene analysis RNA libraries were prepared based on Eastep® Super Total RNA Extraction Kit (Shanghai Promega, catalog number LS1040) following the manufacturer’s protocol, and sequenced using Illumina Novaseq6000 in paired-end 150 bp mode. The quality of RNA-seq data (fastq files) were assessed by Babraham bioinformatics-FastQC [26]. The adaptors and reads with low quality from raw RNA-seq reads were trimmed using cutadapt software [27]. Then RNA-seq analysis was performed by the STAR1 software (version 2.7.10b) with the human reference genome GRCh38 as well as GTF file (downloaded from GENCODE release 45) [28]. After generating genome index, the clean RNA-seq reads were aligned by STAR1 with “-quantMode GeneCounts” option to count number of reads per gene. Differentially expressed genes were determined by the DESeq22 software with the read counts reported by STAR1. Only genes with the adjusted P-value below 0.05 and absolute value of log2 fold change higher than 0.5 were considered as differentially expressed genes. Gene ontology (GO) enrichment analysis GO was conducted with “g:Profiler” 3 online toolkit [29]. For visualization of the RNA-seq analysis results, we used the “SRplot” 4 online toolkit [30]. Isolation of SARS-CoV-2 spike-specific memory B cells The BA.5 S protein was biotinylated (EZ-link Sulfo-NHS-LC-Biotynlation Kit, Cat# 21,327) at a molar ratio of 1:1.3, then the biotinylated S protein was conjugated with streptavidin-phycerythrin (SA-PE; Invitrogen, Cat# S866) at a 4:1 molar ratio and incubated for 30 min at RT to form S-specific tetrameric probes. For S-specific memory B cell isolation, cryopreserved PBMCs were firstly thawed and incubated with Fixable Viability Stain 510 (BD Horizon™, Cat# 564,406) for 15 min at RT. Cells were then washed once with FACS buffer and stained with the S-specific tetrameric probe (SA-S PE) for 60 min at 4 °C in the dark. After washing once with FACS buffer, cells were further stained with surface antibodies, containing PE-Cy7 mouse anti-human CD3 (BD Pharmingen™, Clone HIT3a, Cat# 563,423), APC-Cy7 mouse anti-human CD19 (Biolegend, Clone HIB19, Cat# 302,218), PerCp/Cyanine 5.5 mouse anti-human CD27 (Biolegend, M-T271, Cat# 560,612), FITC mouse anti-human IgG (Biolegend, Clone HP6017, Cat# 409,310) for 20 min at RT. Thereafter, single S-specific memory B cells were sorted using a BD Aria III instrument (BD Bioscience) into 96-well plates through gating with CD3−CD19+CD27+IgG+S+ cells. BCR repertoire analysis Reverse transcription of total RNA from sorted single B cells was performed using SuperScript IV VILO Master Mix (Thermo Fisher, Cat# 11,754,250) according to the manufacturer’s protocol. One microgram of synthesized cDNA was used as a template for two rounds of nested PCR amplification targeting the variable region of immunoglobulin heavy and light chains. Primary amplification was conducted with PrimeSTAR Max DNA Polymerase (Takara) and gene-specific primers flanking the 5′ and 3′ ends. The last amplification products were resolved on an agarose gel to make sure the amplicons were of excised size and sent for high-throughput sequencing. Raw reads were processed using an online tool IMGT (the international ImMunoGeneTics information system) [31] to annotate IGHV/IGHJ gene usage, complementarity determining region 3 (CDR3) sequences, and somatic hypermutation (SHM) frequency. Due to certain differences in the number of antibody sequences between the two independent groups (A-A and I-I-I), the Wilcoxon test for two independent samples was used to compare the characteristics of antibody sequences between the two groups. Antibody clonotypes were defined based on criteria such as identical V and J genes, and identical HCDR3 amino acid sequences. The R package Biostrings v2.60.2 was employed to calculate the sequence identity between sequences and aligned sequences, thereby identifying antibody populations with similar immune responses. The HCDR3 of IGH sequences generated by pairing single-cell heavy and light chains was aligned with the HCDR3 of specific IGH sequences in the CoV-AbDab antibody database using Hamming distance. Antibodies with HCDR3 sequences of the same length and at least 70% amino acid sequence identity were defined as a cluster. The antibody diversity and the importance of clonotypes in the immune response were evaluated by analyzing the frequency of different clonotypes. ADCC NK cell activation assay To assess virus-specific antibody-dependent cellular cytotoxicity (ADCC) mediated by NK cells in A-A and I-I-I groups, 96-well plates were coated with SARS-CoV-2 S protein. Plasma sample from an IgG-positive donor was incubated with coated S protein for 2 h at 37 °C. Following three washes to remove unbound antibodies, 2 × 105 PBMC from two groups were added and incubated at 37 °C in 5% CO2 for 5 h. Concurrent with PBMC addition, APC mouse anti-human CD107a (Pharmingen, Cat# 560,664), brefeldin (BD Bioscience, San Diego, CA), and monensin (BD Bioscience, San Diego, CA) were added to each well. Thereafter, PBMC were stained with surface antibodies of FITC mouse anti-human CD3 (BD Bioscience, Clone HIT3a, Cat# 561,802), PE-Cy™7 mouse anti-human CD56 (BD Bioscience, Clone B159, Cat# 557,747) for 30 min at 4 °C in the dark. After fixation and permeabilization with Cytofix/Cytoperm™ solution (BD Bioscience, Cat# 554,714), the cells were further incubated with PE mouse anti-human IFN-γ (BD Bioscience, Clone B27, Cat# 559,327) for 30 min at 4 °C. Samples were acquired on a BD LSRFortessa X-20 instrument (BD Bioscience) and analyzed with FlowJo software (Treestar). NK cell cytotoxicity assay NK cells’ cytotoxicity was measured using an MHC-devoid K562 cell line as targets, modified from flow cytometry protocols. NK cells were isolated from PBMCs via CD56 positive selection using magnetic microbeads (Miltenyi Biotec, Cat# 130–050–401) according to the manufacturer’s instructions. K562 cells were labeled with 1 μl CFSE in 1 mL 1 × labeling buffer for 10 min at 37 °C in the dark. Then, CFSE-labeled K562 cells were co-cultured with NK cells at a 5:1 effector to target (E:T) ratio in a 96-well V-bottom plate. Plates were centrifuged at 200 g for 3 min and then incubated at 37 °C in 5% CO2 for 12 h. Following incubation, cells were stained with 7-AAD and PE Annexin V (BD Pharmingen, Cat# 559,763) for 15 min at RT according to the manufacturers’ instructions. Cells were acquired on a BD LSRFortessa X-20 instrument (BD Bioscience). Cytotoxicity activity of NK cells against K562 cells was calculated with the percentage of dead target cells showing CFSE-positive and Annexin V-positive subtracting the percentage of target cells that died spontaneously. Statistical analysis GraphPad Prism version 6.0 (GraphPad Software, USA) was used for analyzing antibody titers, T cell responses, and NK cell activities. R package Biostrings v2.60.2 was applied for BCR clonality analyses. Differences were considered statistically significant if P < 0.05. For comparisons of central tendencies between two independent groups, the Mann–Whitney test was employed for antibody titers, T cell responses, and NK cell activities. For BCR clonality analyses, the Wilcoxon rank-sum test was applied, following the convention of the R statistical framework. Antibody titers were reported as geometric mean titer (GMT) with 95% confidence intervals (CI). T cell responses, and NK cell activities were presented as arithmetic means with standard error of the mean (SEM). Results Ad5-boosted vaccination reduced symptomatic BA.5 infections compared to three doses of inactivated vaccine As BA.5 pandemic outbreak in China, the real-world protective efficacy of the Ad5 booster vaccine was monitored. Interestingly, our surveillance showed that 52% (124/239) of individuals in the A-A group were confirmed to have BA.5 infection with at least one characteristic symptom (Table 1) based on questionnaire reports, and these were categorized as symptomatic BA.5 infection. Next, a threshold for BA.5-specific NAb titers (NAb titer = 32) was used to distinguish asymptomatic infection from uninfection. This threshold was derived from the geometric mean ± 3 standard deviations of anti-BA.5 NAb levels in the A-A subgroup at 3 months post-boosting. We found that 38% (90/239) of individuals had anti-BA.5 NAb titers ≥ 32 without clinical symptoms, were asymptomatic infected. The remaining 10% (25/239) of participants had anti-BA.5 NAb titers < 32 and no reported symptoms, were uninfected (Fig. 1B). In comparison, 90% (35/39) of individuals in the I-I-I group experienced symptomatic BA.5 infection, while only 10% (4/39) remained uninfected (Fig. 1B, Table 3). These findings indicate that the A-A offers greater protection against symptomatic BA.5 infection compared to the I-I-I. However, in terms of effectiveness against infection, the two groups show comparable performance, with neither achieving a high level of protection. Ad5-boosted vaccination induced significantly higher anti-BA.5 NAbs, comparable anti-WT NAbs and lower anti-S IgG post BA5 infection than I-I-I To investigate whether the higher heterologous BA.5 protection of the Ad5 vaccine depended on the different induced immune responses compared to the I-I-I regimen, we assessed anti-S IgG antibody levels and NAbs against WT and BA.5 in BA.5 infected individuals who had received either inactivated or Ad5 booster vaccination. Our data demonstrated that the A-A group exhibited 3.3-fold higher anti-BA.5 NAbs than the I-I-I group (180, 95% CI 146–221 in A-A vs. 54, 95% CI 35–83 in I-I-I, P < 0.0001) at 2 M post BA.5 infection. By contrast, anti-WT NAbs were comparable between groups (358, 95% CI 307–419 in A-A vs. 226, 95% CI 138–372 in I-I-I, P = 0.0604), while anti-S IgG levels were significantly lower in the A-A group (6441 AU/ml, 95% CI 5545–7482 in A-A vs. 12,228 AU/ml, 95% CI 8420–17,759 in I-I-I, P = 0.0032) at 2 M post BA.5 infection (Fig. 1C). Notably, no significant differences in the anti-WT or anti-BA.5 NAb titers were observed between symptomatic and asymptomatic infections (Fig. 1D), among the A-A individuals. Additionally, NAb titers against WT and BA.5 in the A-A group at 2 M post infection did not vary by gender, age, or presence of underlying diseases (Additional file 1: Fig. S1). These findings confirmed that A-A induced more NAbs cross-reactive to WT and BA.5 than the I-I-I regimen at 2 M post BA.5 infection. Various WT-targeting vaccines have been proven to induce very low levels of NAbs against the BA.5 strain at 6 months post vaccination [32–35]. However, based on the above observation that A-A induces more cross-reactive neutralizing antibodies than I-I-I post BA.5 infection, we sought to investigate whether subtle differences existed in NAbs after vaccination but before BA.5 exposure. Our results indicate that A-A indeed induced higher anti-BA.5 NAb titers at 1 M and 3 M post vaccination than I-I-I (Fig. 1E). Specifically, anti-BA.5 NAb in A-A was 11 (95% CI 7–24) at 1 M, which was 3 (95% CI 2–5) in I-I-I. At 3 M post vaccination, anti-BA.5 NAb of all the individuals in the I-I-I group was below the detection limit, while in the A-A group, it remained at 5 (95% CI 1–21). A-A also induced significantly higher anti-WT Nab titers, with 4.7-fold and 8.1-fold higher titers at 1 M and 3 M post-booster vaccination, respectively, compared to I-I-I (Fig. 1E). No significant differences in the anti-S IgG levels were observed between the A-A and I-I-I groups post vaccination (Fig. 1E). Although anti-BA.5 NAb differences at 3 M post vaccination were modest, when combined with the post vaccination NAb data (Fig. 1C), these findings collectively demonstrate that A-A more effectively induces NAbs cross-reactive to WT and BA.5 than I-I-I. Ad5 booster vaccine elicits enhanced neutralizing antibody responses through increased BCR diversity, CDR3 length, and somatic hypermutation post BA.5 infection To determine whether A-A enhances anti-BA.5 NAb clonal diversity beyond elevating collective NAb titers, we quantified B cell receptors (BCRs) clonality and compared the neutralizing antibody repertoire between A-A and I-I-I individuals at 2 M post BA.5 infection. Using BA.5 S-protein tetramers, the single S-specific memory B cells were sorted with CD3−CD19+CD27+IgG+S+ cells (Fig. 2A). Then single cell BCR multiplex PCR was performed to analyze the BCR repertoire usage in S-binding B cells between the two groups. The IGHV gene frequency of the two groups represented by the blue (I-I-I) and red (A-A) curves was shown in Fig. 2B. Both groups cover a range of IGHV gene subgroups, including IGHV1 (IGHV1-3, IGHV1-69), IGHV3 (IGHV3-7, IGHV3-30), and IGHV4 (IGHV4-39, IGHV4-59). The A-A group shows a more extensive and scattered distribution across the different IGHV genes compared to the I-I-I group. Specifically, IGHV4-39 was the main component, accounting for 9% of detected sequences in A-A; IGHV3-30 dominated (16% of sequences) in I-I-I (Fig. 2B). These results demonstrate characteristics consistent with higher BCR diversity of the A-A group compared to the I-I-I group through its more diverse IGHV gene usage profile. Additionally, CDR3 length was marginally longer in the A-A group than in the I-I-I group (P = 0.049, Fig. 2C), indicating a modestly broader distribution of heterogeneous BCR repertoires that may enable wider coverage of SARS-CoV-2 antigenic targets.Fig. 2BCR repertoire diversity and clonal characteristics in A-A and I-I-I individuals at 2 M post BA.5 breakthrough infection. A Gating strategy for sorting S-specific B cells. B Immunoglobulin heavy chain variable (IGHV) gene usage between A-A and I-I-I groups. C Distribution of complementarity-determining region 3 (CDR3) lengths in B cells secreting anti-BA.5 antibodies between A-A and I-I-I groups. D Representative CDR3 amino acid sequences from dominant neutralizing clones. E Quantification of SHM in BCR showed significantly higher mutation frequency in A-A compared to I-I-I. Wilcoxon rank-sum test was used for comparing mutation frequencies, *P < 0.05 BCR repertoire diversity and clonal characteristics in A-A and I-I-I individuals at 2 M post BA.5 breakthrough infection. A Gating strategy for sorting S-specific B cells. B Immunoglobulin heavy chain variable (IGHV) gene usage between A-A and I-I-I groups. C Distribution of complementarity-determining region 3 (CDR3) lengths in B cells secreting anti-BA.5 antibodies between A-A and I-I-I groups. D Representative CDR3 amino acid sequences from dominant neutralizing clones. E Quantification of SHM in BCR showed significantly higher mutation frequency in A-A compared to I-I-I. Wilcoxon rank-sum test was used for comparing mutation frequencies, *P < 0.05 Furthermore, specific CDR3 sequences associated with neutralizing antibodies were analyzed using the CoV-AbDab database. Data showed the A-A group post BA.5 infection harbored a significantly greater number of CDR3 sequences matching known NAbs in the database compared to the I-I-I group (Fig. 2E). This indicates the Ad5 vaccine induces a more complex and broadly reactive antibody repertoire, enhancing both antigen binding and anti-BA.5 neutralization. SHM in BCRs is critical for antibody affinity maturation. Notably, the A-A group exhibited elevated SHM frequencies (Fig. 2D), highlighting the Ad5 vaccine’s capacity to drive mature, high-affinity antibody responses. The higher SHM density in A-A reflects more extensive B cell maturation, directly linked to the generation of antibodies with potent neutralizing activity against SARS-CoV-2 variants. These findings mechanistically explain the Ad5 vaccine’s superior efficacy, attributing protection to both increased NAb titers and enhanced clonal diversity/affinity. Ad5 booster vaccine with BA.5 infection enhances innate immunity through upregulation of NK cell-related genes To compare post-BA.5 infection immune responses between A-A and I-I-I groups, we performed bulk RNA sequencing on PBMCs from A-A (n = 9) and I-I-I (n = 9) individuals at 2 M post BA.5 infection. GO enrichment analysis was performed using the online bioinformatics platform “Wei Sheng Xin” [30]. This enrichment analysis revealed that up-regulated genes in A-A were predominantly associated with innate immunity pathways, including NK cell-mediated immunity, cytotoxicity, and leukocyte-mediated immune responses (Fig. 3A). Focused analysis of NK cell-related gene expression showed significantly higher levels of NCR1, NCR3, PFR1, GZMB, KIR3DL1, KIR3DL2, and KIR3DL3 in A-A compared to I-I-I (Fig. 3B, Additional file 1: Fig. S2). These findings indicate that Ad5 booster vaccination enhances innate immune signaling, particularly NK cell activation and cytotoxic potential, which may contribute to its protective efficacy against BA.5 infection.Fig. 3A-A individuals presented enhanced innate immune response profiling via RNA-seq. A Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) in A-A versus I-I-I groups. B Heatmap of NK cell activation and cytotoxicity-related gene expression A-A individuals presented enhanced innate immune response profiling via RNA-seq. A Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) in A-A versus I-I-I groups. B Heatmap of NK cell activation and cytotoxicity-related gene expression Ad5 booster vaccine with BA.5 infection triggers an enhanced NK cell-mediated ADCC response and cytotoxicity To investigate whether NK cells contribute to the enhanced protective efficacy of the A-A regimen, we assessed NK cell-mediated ADCC response in vitro by measuring the percentages of CD107a+ and/or IFN-γ+ NK cells (Fig. 4A). Comparative analysis of NK cell-mediated ADCC between A-A and I-I-I groups revealed no significant difference in CD107a⁺ NK cell frequencies (20.0%, 95% CI 15.0–25.0% vs 14.3%, 95% CI 11.1–17.5%, Fig. 4B, left panel). However, the A-A group demonstrated a significantly higher proportion of IFN-γ+ NK cells compared to I-I-I (8.3%, 95% CI 4.3–12.3% vs 2.3%, 95% CI 1.2–3.4%, Fig. 4B, middle panel). Notably, the percentage of IFN-γ+CD107a+ NK cells was 2.4-fold higher in A-A than in I-I-I (Fig. 4B, right panel).Fig. 4A-A individuals exhibited elevated NK cells mediated ADCC responses and cytotoxicity capacity. A Schematic summary of NK cells stimulation with SARS-CoV-2 S protein and the percentage of CD107a+ and IFN-γ+ NK cells was analyzed. B Frequency of IFN-γ+ NK cells, CD107a+ NK cells and CD107a+ IFN-γ+ NK cells in A-A and I-I-I groups (n = 17 in A-A, n = 17 in I-I-I). C Representative flow plots showing Annexin V+/7-AAD.+ dead target K562 cells at a 5:1 (E:T) ratio. D Quantification of cytotoxicity revealed a trend toward higher killing in A-A versus I-I-I (n = 17 in A-A, n = 17 in I-I-I). The Mann–Whitney test was employed for comparing central tendencies of every two groups, not significant; **P < 0.01 A-A individuals exhibited elevated NK cells mediated ADCC responses and cytotoxicity capacity. A Schematic summary of NK cells stimulation with SARS-CoV-2 S protein and the percentage of CD107a+ and IFN-γ+ NK cells was analyzed. B Frequency of IFN-γ+ NK cells, CD107a+ NK cells and CD107a+ IFN-γ+ NK cells in A-A and I-I-I groups (n = 17 in A-A, n = 17 in I-I-I). C Representative flow plots showing Annexin V+/7-AAD.+ dead target K562 cells at a 5:1 (E:T) ratio. D Quantification of cytotoxicity revealed a trend toward higher killing in A-A versus I-I-I (n = 17 in A-A, n = 17 in I-I-I). The Mann–Whitney test was employed for comparing central tendencies of every two groups, not significant; **P < 0.01 Additionally, direct cytotoxicity of NK cells against K562 target cells was assessed. NK cells from the A-A group demonstrated a non-significant trend toward slightly elevated cytotoxicity against K562 cells compared to I-I-I at 2 M post BA.5 infection (A-A: 37.8% vs. I-I-I: 32.5%, Fig. 4C–D). This trend, while not statistically significant, aligns with enhanced NK cell activation observed in the ADCC response (Fig. 4B) indicating that Ad5 booster vaccination in the context of BA.5 infection enhances both NK cell-mediated ADCC responses and intrinsic cytotoxicity against targets, underscoring the multifaceted role of innate immunity in A-A-mediated protection. BA.5 infection in Ad5 booster vaccinated individuals elicited reduced virus-specific CD4+ and CD8+ T cell responses As a vector vaccine, the Ad5 boost vaccine is expected to induce a T cell response. Given that the Ad5 vaccine is formulated with full-length S protein, whereas the inactivated vaccine contains all SARS-CoV-2 proteins, we separately assessed virus-specific T cell responses against S1, S2, and non-structural (N/M/E) proteins. The virus-specific T cell responses were characterized with IFN-γ+CD4+ or CD8+ T cells (Fig. 5A). Data showed that the numbers of S1/S2-specific IFN-γ+CD4+ T cell responses in the I-I-I group were significantly higher than those in the A-A group, with 1.9-fold and 2.8-fold increases observed for S1 and S2, respectively. Additionally, a marginal upward trend was noted in their percentages. However, the I-I-I group exhibited markedly higher N/M/E-specific IFN-γ⁺ CD4⁺ T cell responses compared to A-A (0.09, 95% CI 0.03–0.15 vs 0.02, 95% CI 0.01–0.03, Fig. 5B). The numbers of S1-, S2-, and N/M/E-specific IFN-γ+ CD4+ T cells in the I-I-I group were 1.9-fold, 2.8-fold, and 6.8-fold higher than in A-A, respectively (Fig. 5B), indicating limited virus-specific T cell stimulation by A-A following BA.5 infection.Fig. 5Omicron BA.5 infection in Ad5 booster vaccinated individuals elicited reduced virus-specific CD4 + and CD8 + T cell responses. A Gating strategy and schematic of ex vivo stimulation of PBMCs with S protein subunits (S1, S2) and N/M/E protein peptide pools respectively, followed by flow cytometry to assess cytokine-producing CD4+ and CD8+ T cells. B The percentages (left panel) and number/million PBMC (right panel) of IFNγ+ CD4+ T cells between A-A and I-I-I at 2 M post BA.5 infection were analyzed (n = 35 in A-A, n = 20 in I-I-I). C The percentages (left panel) and number/million PBMC (right panel) of IFNγ+ CD8.+ T cells between A-A and I-I-I at 2 M post BA.5 infection were analyzed (n = 35 in A-A, n = 20 in I-I-I). Mann–Whitney test was involved to compare differences between different groups for T cell response, ns, not significant; **P < 0.01, ***P < 0.001 Omicron BA.5 infection in Ad5 booster vaccinated individuals elicited reduced virus-specific CD4 + and CD8 + T cell responses. A Gating strategy and schematic of ex vivo stimulation of PBMCs with S protein subunits (S1, S2) and N/M/E protein peptide pools respectively, followed by flow cytometry to assess cytokine-producing CD4+ and CD8+ T cells. B The percentages (left panel) and number/million PBMC (right panel) of IFNγ+ CD4+ T cells between A-A and I-I-I at 2 M post BA.5 infection were analyzed (n = 35 in A-A, n = 20 in I-I-I). C The percentages (left panel) and number/million PBMC (right panel) of IFNγ+ CD8.+ T cells between A-A and I-I-I at 2 M post BA.5 infection were analyzed (n = 35 in A-A, n = 20 in I-I-I). Mann–Whitney test was involved to compare differences between different groups for T cell response, ns, not significant; **P < 0.01, ***P < 0.001 Similarly, no significant differences were observed in S1/S2-specific IFN-γ⁺ CD8⁺ T cell percentages or numbers between groups. However, I-I-I induced significantly higher N/M/E-specific IFN-γ⁺ CD8⁺ T cell responses (3.3-fold higher in percentage and 3.2-fold higher in cell number; Fig. 5C). These findings suggest that vaccination-induced memory T cell responses to conserved viral antigens (e.g., N/M/E) are preferentially recalled by inactivated vaccines upon re-exposure to BA.5. Discussion Although adenovirus vaccines have undergone immunogenicity evaluations, retrospective assessments using real-world infection data remain limited. This study demonstrated that 52% of individuals in the A-A group experienced symptomatic BA.5 breakthrough infection, 38% had asymptomatic infection, and 10% remained uninfected. The symptomatic infection rate was lower than that observed in the I-I-I group (90%) and China’s estimated BA.5 infection rate (over 80%), where 73% of cases were reported as symptomatic [1, 36]. This finding underscores the efficacy of the Ad5 vaccine in reducing SARS-CoV-2 transmission. Compared with three doses of inactivated vaccine, two Ad5 booster doses induced higher BA.5-specific NAb titers both immediately post-vaccination and after breakthrough infection. Critically, Ad5 elicited antibodies targeting broader neutralizing epitopes and enhanced NK cell cytotoxicity and ADCC responses. These synergistic adaptive and innate immune enhancements collectively explain the Ad5 platform’s superior protective efficacy against BA.5. Our findings demonstrate that A-A elicited significantly higher anti-BA.5 NAbs than I-I-I. The A-A regimen utilizes E1/E3-deleted Ad5 vectors to deliver the SARS-CoV-2 S protein [37], a design that enhances both humoral and cellular immunity [38]. Structural studies reveal that Ad5-derived S proteins are densely displayed on cell membranes [39], a presentation strategy that enhances the exposure of cross-reactive antigenic epitopes shared between WT and BA.5 variants. The limited replicative capacity of the Ad5 vector generates more endogenous and soluble antigens than the I-I-I regimen, facilitating efficient uptake by antigen-presenting cells (APCs), including dendritic cells and macrophages. This process promotes robust activation of MHC class I and II pathways, enabling cross-presentation of S-derived peptides [40] and priming CD8+ and CD4+ T cells to support B cell maturation and affinity refinement. Consequently, this mechanism drives the production of high-affinity antibodies targeting conformation-dependent neutralizing epitopes [41]. These hypotheses are supported by the significantly higher anti-BA.5 NAb titers observed in the A-A group. Specifically, 1 M post-vaccination, A-A recipients exhibited detectable anti-BA.5 NAbs (GMT = 11) above the assay detection limit [4], whereas NAbs in the I-I-I group remained below the detection limit (GMT = 3). This disparity was further corroborated by 3.3-fold higher BA.5-specific NAb titers in A-A at 2 M post breakthrough infection, alongside a higher percentage of BA.5 S-specific memory B cells and a greater frequency of IgG-subtype antibodies (Additional file 1: Fig. S3). The neutralizing antibody profiles induced by A-A and I-I-I immunization differ not only in their titers against BA.5 and WT strains (Fig. 1) but also in their BCR repertoire usage and the frequency of somatic hypermutations associated with antibody affinity maturation. A-A further enhances NAb potency through elevated somatic hypermutation (SHM) frequencies in B cells, a process critical for refining antibody affinity toward conserved neutralizing epitopes [41, 42]. Notably, A-A preferentially expanded B cell clones utilizing the IGHV4-39 germline, which encodes antibodies targeting the RBD domain with high neutralizing activity against SARS-CoV-2 variants [43]. Additionally, IGHV1-69, a germline associated with recognition of the L452R mutation prevalent in Omicron subvariants [44], was enriched in A-A individuals. In contrast, I-I-I responses were dominated by IGHV3-30-encoded antibodies. While IGHV3-30 is a major germline in convalescent patients and binds strongly to the RBD, its limited SHM and preferential enrichment in S2-targeting antibodies render it less effective against evolving variants [45–47]. Structural studies confirm that IGHV3-30-derived antibodies lose neutralizing activity against Omicron subvariants due to epitope alterations [48]. Consequently, despite inducing high anti-S IgG titers (GMT = 12,228 vs. 4694 in A-A), the I-I-I regimen failed to generate robust BA.5-specific NAbs. These findings collectively demonstrate that A-A’s superior BA.5 neutralization stems from its unique capacity to drive BCR diversity, extended CDR3 lengths, and SHM-enriched antibody maturation. NK cells serve as critical early responders during viral infections, exerting their functions through direct cytotoxicity and cytokine signaling, as demonstrated against herpesviruses and Hantavirus [49, 50]. Beyond their immediate effector functions, NK cells orchestrate antiviral immunity via secretion of IFN-γ, IP-10, MIP-1α, and MIP-1β that suppress viral replication and recruit adaptive immune cells [51]. Emerging evidence highlights “trained immunity” in NK cells, an innate immune memory mechanism that enhances functional responses to subsequent challenges [52]. The Ad5-boosted vaccine exemplifies this phenomenon, as its adenoviral backbone activates innate immune pathways via Toll-like receptors (TLRs) and cytosolic DNA sensors, triggering cytokines that drive persistent NK cell reprogramming [53]. In the A-A group, this trained immunity is evidenced by sustained upregulation of NK activation markers (NCR1, NCR3, CD160) and cytotoxic effectors (PRF1, GZMB), which correlate with enhanced cytotoxicity against target cells. Ad5 vaccination induces epigenetic rewiring and metabolic reprogramming in NK cells, priming them for rapid effector responses upon viral re-exposure. These trained NK cells exhibit elevated ADCC responses, evidenced by increased FCGR3A (CD16) expression and degranulation. Unlike inactivated vaccines which elicit transient NK activation, the Ad5 vaccine establishes durable NK memory, with cytotoxic and cytokine-producing subsets persisting more than 6 months post-vaccination [54]. Together, Ad5-boosted NK cells are critical contributors to the broad protective efficacy against BA.5 infection. Limitations Our study has several limitations. First, the cohort size for the I-I-I group was relatively small, potentially diminishing the statistical power and external validity of our comparative analyses. This imbalance in sample sizes between the I-I-I and A-A groups may restrict the robustness of our findings, as smaller cohorts are more susceptible to random variation and may not fully capture the diversity of immune responses in the population. Second, the binding antibody was only conducted with anti-WT S protein antibody; multiplex binding antibodies, such as subdomains of the spike protein and different variants’ binding potential should be further compared between A-A and I-I-I groups. To address these gaps, future studies should prioritize multicenter cohorts with larger sample sizes and standardized, head-to-head comparisons of vaccine platforms in real-world settings to enhance the robustness and generalizability of findings. Conclusions This retrospective cohort study demonstrates that Ad5-boosted SARS-CoV-2 vaccine confers enhanced protection against symptomatic BA.5 breakthrough infection, mediated by synergistic adaptive and innate immune responses. This vaccine’s delivery of native S protein correlates with elevated cross-reactive neutralizing antibodies between WT and BA.5, while simultaneously inducing NK cell functional states with “trained” immunity phenotypes. These findings highlight Ad5 vaccine’s superiority over inactivated platform in eliciting multi-layered immune responses against BA.5. Future vaccine strategies should prioritize platforms that integrate humoral breadth with innate immune training to combat emerging variants and novel Disease X threats. Supplementary Information Additional file 1. Table S1, Figure S1-S3. Table S1: Demographic information of A-A and I-I-I individuals selected for RNA-seq. Figure S1: Comparison of neutralizing antibody titers against WT and BA.5 variants across demographic and clinical subgroups in A-A group at 2 M post BA.5 breakthrough infection. Figure S2: Comparison of a set of genes associated with the GO-enriched NK cell cytotoxicity. Figure S3: Characterization of BA.5-specific memory B cell responses and antibody isotype distribution. Additional file 1. Table S1, Figure S1-S3. Table S1: Demographic information of A-A and I-I-I individuals selected for RNA-seq. Figure S1: Comparison of neutralizing antibody titers against WT and BA.5 variants across demographic and clinical subgroups in A-A group at 2 M post BA.5 breakthrough infection. Figure S2: Comparison of a set of genes associated with the GO-enriched NK cell cytotoxicity. Figure S3: Characterization of BA.5-specific memory B cell responses and antibody isotype distribution.