Antibody and transcription landscape in peripheral blood mononuclear cells of elderly adults over 70 years of age with third dose of COVID-19 BBIBP-CorV and ZF2001 booster vaccine

A total of 71 participants who had completed two doses of primary immunization with the inactivated vaccine for more than 6 months were recruited to receive a third dose of either the BBIBP-CorV or ZF2001 vaccine. All participants were divided into four groups according to age and booster dose, including group A (boosted by BBIBP-CorV, < 70 years old), group B (boosted by BBIBP-CorV, ≥ 70 years old), group C (boosted by ZF2001, < 70 years old), and group D (boosted by ZF2001, ≥ 70 years old) (Fig. 1). At days 7 and 28 after vaccination, we collected a total of 131 serum and 57 PBMC samples. The detailed characteristics of the participants are shown in Table S1 and Table S2.

First, we estimated the titers of binding antibodies from sera collected at 7 and 28 days after vaccination of the participants. After 7 days of homologous boost with BBIBP-CorV, the mean value of IgG antibody in the younger group was 221.09 BAU/ml, and that in the elderly group was 104.35 BAU/ml. The elderly group was significantly lower than the younger group (P = 0.0282). A similar phenomenon was observed in different age groups heterologously boosted by ZF2001, with 366.63 BAU/ml in the younger group and 89.24 BAU/ml in the elder group, and the difference was statistically significant (P = 0.0014) (Fig. 2a). After 28 days of the third booster dose, the mean IgG antibody levels in the four groups increased to 610.6, 733.9, 1631, and 1617 BAU/ml, respectively. Remarkably, there was no age-associated difference in antibody responses(P = 0.8237, P = 0.0644) (Fig. 2b).

To determine the effect of aging on the transcriptional program of the third dose of COVID-19 vaccine, we performed a transcriptome systematic scan of 57 PBMC samples from different groups. After 7 days of homologous boost with BBIBP-CorV, the number of up-regulated and down-regulated DEGs in the elder group compared with the younger group was 213 and 65, respectively(Fig. 3a). After 7 days of ZF2001 heterologous boost, 136 and 83 genes were up-regulated and down-regulated in the elder group compared with the younger group, respectively(Fig. 3b). The up-regulated DEGs of the elder group after 28 days with the BBIBP-CorV booster were 100, and the down-regulated DEGs were 134(Fig. 3c). There were 131 up-regulated DEGs and 84 down-regulated DEGs in the elder group after 28 days of the ZF2001 booster(Fig. 3d). We used the heatmap to show the expression of the top 20 up-regulated and down-regulated differentially expressed genes in each sample(Fig. 4a-d).

In order to analyze the interaction relationships between DEGs, the STRING database was used to create a PPI network. Using the MCODE plug-in of Cytoscape software, we extracted 4 subnetworks from the PPI network at the nodes and edges of all DEGs, including the characteristic induced up-regulated and depleted down-regulated DEG sets identified from the third dose of BBIBP-CorV(Figure S1a, b) and ZF2001(Figure S1c, d) in elderly individuals. These characteristic gene subnetworks contained 21, 25, 34, and 42 DEGs, respectively.

For the purpose of clarifying the functional pathways of the signature DEGs mediated by the third booster vaccine in the elderly group, we conducted GO-term analysis, focusing on the biological process entries of these DEGs. First, we showed the top 20 GO-terms enriched by BBIBP-CorV and ZF2001 in the elderly group in the GO dot-plot (Fig. 5a, b). Coincidentally, DEGs were found to be involved in multiple innate immune pathways, such as lymphocyte proliferation, mononuclear cell proliferation, and leukocyte proliferation. The connection of GO functions with specific DEGs was shown in Figures S1e, f. The chordal diagram was used to further demonstrate the degree of differential expression changes of genes involved in each functional project (Figs. 5c, d). We found that most of the DEGs involved in immune-related functional items were down-regulated, implying that these immune-related pathways were impaired to some extent in the elderly group. This may be a potential reason of early humoral immune deficiency in elderly people.

We similarly used the STRING database and MCODE plug-in of Cytoscape software to prioritize DEGs 28 days after the booster dose. A total of 4 characteristic up-regulated and down-regulated DEG sets were identified induced by the third dose of BBIBP-CorV and ZF2001 in the elderly on Day 28 (Figs. 6a, b and 7a, b).

These characteristic DEGs were also enriched in immune-related biological process GO-terms. Specifically, 28 days after the BBIBP-CorV and ZF2001 booster vaccine, DEGs in the senile group were enriched in several pathways such as the chemokine-mediated signaling pathway and humoral immune response (Figs. 6c and 7c). In the BBIBP-CorV booster vaccine group, these functional entries were mainly executed by CD83, CR2, CXCL1, CXCL2, CXCL3, CXCL8, IL1B, JCHAIN, PF4, PF4V1, PPBP, and TNF, etc. (Fig. 6d). In the ZF2001 booster vaccine group, C2, CXCR1, CXCR2, CX3CR1, CXCL5, CXCR1, CXCR2, LTF, PF4, PF4V1, and PPBP participated in immune regulation (Fig. 7d). Moreover, the chordal diagram further demonstrated that most of the DEGs involved in immune-related functional items were up-regulated, meaning that these immune-related pathways were widely activated 28 days after the booster vaccine in the elderly group (Figs. 6e and 7e).

To explore the impact of age characteristics on the transcriptome, we performed an analysis of the transcriptome using weighted gene co-expression network analysis(WGCNA). In order to ensure the construction of a scale-free network, the most suitable β was determined as a soft threshold parameter. A total of four co-expression modules were identified through hierarchical clustering, and a dendrogram of all DEGs was clustered based on a dissimilarity measure (1-TOM) at day 7. These four co expression modules based on differences in topological overlap were displayed in the 3D clustering map and assigned as brown, blue, turquoise, and grey module colors(Fig. 8a, b). Heatmap depicted the Topological Overlap Matrix (TOM) of genes selected for weighted co-expression network analysis(Fig. 8c). For each module, the gene co-expression was summarized by the eigengene and we calculated the correlations of each eigengene with clinical traits, such as vaccine type, gender and age. The correlation between the vaccination signatures and the co-expression module is shown in Fig. 8d, of which the brown module (eigengene value = -0.75, p = 0.03 × 10^–6) was significantly negatively correlated with the age of the participants, but not significantly correlated with vaccine type and gender. The brown module includes 56 genes involved in many immune function-related genes, such as AK5, BTLA, CCR5, CD24C, CR2, DPP4, GCSAM, IL7R, IL23R, LEF1, LTB, RORC, and STAP1, etc. The GO enrichment analysis showed that these genes were mainly involved in "lymphocyte costimulation", "T cell differentiation involved in immune response", "CD4-positive, alpha–beta T cell differentiation involved in immune response", and "T-helper cell differentiation" (Fig. 10a). The above co-expression modules analysis, which is significantly negatively correlated with age, once again indicates that the immune response of the elderly to booster vaccines at 7 days was impaired and delayed.

A similar WGCNA was performed on transcriptome data at 28 days. Five co-expression modules were identified on day 28, of which the brown module (eigengene value = 0.47, p = 0.01) and yellow module (eigengene value = 0.46, p = 0.01) were significantly positively correlated with the age of the participants (Fig. 9a-d). The co-expressed genes of these two modules are mainly involved in "fever generation", "chemokine activity", "chemokine receptor binding", "response to chemokine", and "chemokine-mediated signaling pathway" (Fig. 10b). This means that after 28 days of the booster vaccine, the immune transcriptional landscape associated with the elderly significantly improved.

In November 2021, we recruited healthy SARS-CoV-2-naïve individuals who had completed two doses of inactivated vaccine for more than 6 months to participate in this study in Liaocheng City, Shandong Province. All participants were assigned according to age characteristics to a younger group (age < 70) and an elder group (age ≥ 70) for a third homologous boost with BBIBP-CorV or heterologous boost with ZF2001. The anticoagulant and procoagulant venous blood samples were collected 7 and 28 days after booster vaccination, and serum and peripheral blood mononuclear cell (PBMC) samples were isolated for further study.

The IgG antibodies were detected using an indirect ELISA kit (Vazyme, China) based on the spike protein of SARS-CoV-2. All serum samples were diluted to a concentration gradient of 1:200, and optical density (OD) was read at 450 nm and 630 nm after incubation, enzyme labeling, chromogenic reaction, and termination steps. The specific operation was carried out in strict accordance with the instructions. The standard curve was prepared with 6 standard substances with known antibody concentrations provided in the kit, and the OD value of the sample to be tested was converted to the antibody concentration. The antibody concentration of the IgG antibody was calculated after three repetitions. The mean with SD was used to describe antibody titers and statistical significance was analyzed by unpaired t-tests with log-transformation using GraphPad Prism 8.0.

The transcriptome high-throughput sequencing of PBMC samples from participants was performed according to our previous research methods [35]. Total RNA was extracted from PBMCs by using the RNeasy Mini Kit (Qiagen, Germany) according to the manufacturer’s instructions. The concentration and integrity of total RNA were checked using the Qubit RNA Assay Kit in a Qubit 4.0 Fluorometer (Life Technologies, USA) and the RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 System (Agilent Technologies, USA) respectively. A total amount of 100 ng total RNA per sample was used to prepare the rRNA-depleted cDNA library by Stranded Total RNA Prep Ligation with Ribo-Zero Plus Kit (Illumina, USA). The final library size and quality were evaluated using an Agilent DNA 1000 Kit (Agilent Technologies, USA), and the fragments were found to be between 250 and 350 bp in size. The library was sequenced using an Illumina NextSeq 2000 platform to generate 100 bp paired-end reads.

The quality control(QC), trimming, and mapping of the RNA-seq raw fasta data to the human reference genome hg38 were performed in the CLC Genomics Workbench. The gene expression level was measured based on the transcripts per million (TPM). We calculated normalization factors using iterative edgeR [36] and limma [37] packages, and the standardization and filtering of gene expression were accomplished by the voom, lmFit, and eBayes functions. DEGs were filtered out according to p-value < 0.05 and 2^logFC_cutoff criteria, and visualized as volcanoes and heatmaps by the pheatmap package in R (version 4.1.0).

In order to explore the relationship between proteins encoded by identified up and down-regulated DEGs, the initial PPI networks for the protein products of DEGs were constructed using the STRING Database (version 11.5) [38], and then the network was visualized and analyzed with Cytoscape software (version 3.8.28) [39]. The molecular complex detection (MCODE) plugin in Cytoscape software was used to screen the characteristic DEGs from the initial PPI networks.

To be aware of the prospective functions of characteristic DEGs identified by the PPI network analysis, Gene Ontology (GO) terms were identified using the clusterProfiler 4.0 package [40]. P < 0.05, subjected to Bonferroni adjustment, was defined as the cut‑off criterion. Data visualization was performed using the ggplot2, enrichplot, GOplot, topGO, circlize, and ComplexHeatmap packages in R (version 4.1.0).

To identify gene modules related to the age characteristics in the COVID-19 booster vaccine, we performed weighted gene co-expression network analysis based on transcriptional profiles and sample characteristics through the WGCNA package [41] in R software. First, the best β value was confirmed with a scale-free fit index larger than 0.85 as well as the highest mean connectivity by performing a gradient test from 1 to 30. Subsequently, the topological overlap matrix (TOM) transformed by the adjacency matrix was then clustered by dissimilarity between genes, and we performed hierarchical clustering to identify modules. Finally, the co-expressed genes were determined by calculating the module membership (MM) and gene significance (GS) of the genes in the target modules. The immune landscape of important modules was analyzed using the ClueGO plugin [42] of Cytoscape software (version 3.8.2). P < 0.05, subjected to Bonferroni adjustment, was defined as the cut − off criterion.