Unique Transcriptome Changes in Peripheral B Cells Revealed by Comparing Age Groups From Naive or Vaccinated Mice, Including snoRNA and Cdkn2a

Abstract

Aging is associated with a decline in immune function that is not fully understood including vaccine failure. Here we report transcriptomic analysis on B cells from naive or influenza-vaccinated mice of 3 ages: young (15–23 weeks), middle-aged (63–81 weeks), and old mice (103–119 weeks). Our goal was expression profiling of B cells by age and history of vaccination to identify novel changes at the transcriptome level. We observed waning vaccine responses with age. In B cell transcripts, age and vaccination history were both important with notable differences observed in conducted analyses (eg, principal component, gene set enrichment, differentially expressed [DE] genes, and canonical pathways). Only 39 genes were significantly DE with age irrespective of vaccine history. This included age-related changes to box C/D small nucleolar (sno) RNAs, Snord123 and Snord1a. Box C/D snoRNAs regulate rRNAs through methylation and are linked to neurodegenerative, inflammatory, and cancer diseases but not specifically B cells or age. Canonical pathway changes implicated with age irrespective of vaccination history included EIF2, mTOR signaling, p53, Paxillin, and Tec kinase signaling pathways as well as cell cycle checkpoint. Importantly, we identified DE genes and pathways that were progressively altered starting in middle-age (eg, signaling by Rho family GTPases) or only altered in middle-age (eg, sphingosine-1-phosphate signaling), despite minimal differences in the ability of these mice to respond to vaccination compared to younger mice. Our results indicate the importance of vaccination or immune stimulation and analyses of multiple age ranges for aging B cell studies and validate an experimental model for future studies.

B cells are key cells of the adaptive immune system, with distinct lineage and gene transcriptional profiles (1,2). During infection or vaccination, clonal expansion of naive cells with B cell receptor affinities for microbial antigens leads to differentiation into antibody-secreting cells as observed for influenza, cytomegalavirus, torquetonovirus, and the coronavirus causing COVID-19 infections (3,4). Alterations or failures in this process during aging lead to poor control of infections and risk of severe disease or even death. Aging is characterized by a progressive degradation of biological processes that impairs immunologic function, as well as inflammaging. The latter is characterized by chronic low-grade inflammation and reduction in appropriate responses to immune stressors (5). Accordingly, old age is linked to higher susceptibility to infections (eg, COVID-19) and poor responses to vaccination (eg, influenza vaccines) with noted deficiencies in B cells (6–8).

How exactly the aging process affects the B cell population remains an active research area. The B cell repertoire is altered with age, including naive and memory populations and B cell receptor specificities (8). Several age-related autonomous B cell defects have been reported in humans, including downregulation of activation-induced cytidine deaminase, the enzyme necessary for somatic hypermutation and IgG production (9), and increased expression of senescent-associated secretory phenotype in memory B cell subsets (10). As with many aging-related cellular studies, one issue is the potential for bias based on current subset definitions. With age this can be particularly difficult as antigen experience is dramatically different between young and older individuals (7), and the antibody responses to the same stimulus (eg, influenza vaccination) are influenced by age and history of original antigen exposure (11). This is pronounced for mice housed in standard laboratory conditions, most commonly used in immunologic and aging studies, versus free-living community or pet store mice (12). In addition, B cell subpopulations have been reported in older mice and humans that do not exist in significant numbers in younger individuals, yet seem to play important roles in interaction with other immune cells or responding to antigenic threats (13–17).

Transcriptome analysis by RNA-sequencing has the potential to address some of these issues, by unbiased, in-depth analysis of gene transcripts. Evaluation of peripheral aged B cells has never been performed, although B cell populations from young mice (2,18–21) and B cells by age in the thymus (22) have been evaluated by this method. One recent study evaluated age-related changes in antibody-secreting cells with microarray analyses and observed changes in genes involved with immune regulation, cell metabolism, and endoplasmic reticulum (ER) stress response (23). The objective of the current study was to identify genes and pathways altered in aging B cells as well as to inform future analyses of B cell subpopulations or single-cell analyses. To do so, we evaluated splenic B cells by RNA-sequencing using mice that were 3 different ages (young, middle-age, old). As transcriptome analyses would not address issues of recent antigen experience from infection or vaccination, we also explored how B cells from naive mice compared to vaccinated mice, and how vaccinated mice of different ages compared to each other.

Method

Animals and B-Cell Isolation

All animal studies were approved by the Tulane University Institutional Animal Care and Use Committee. For immunizations, some BALB/C mice were injected intramuscularly with 50 μL Flulaval 2014-15 (GlaxoSmithKline) and serum was collected 3 weeks later for anti-hemagglutinin IgG1 or IgG2a ELISAs. Splenocytes were collected and processed by negative magnetic selection using CD4 Microbeads kit, then flow-through (CD4⁻ fraction) by Pan B Isolation Kit II (Mitinyi Biotec).

RNA Analyses

After extraction by RNeasy Protect Mini Kit (Qiagen), transcriptome analysis was performed by RNA-sequencing using Ion Torrent semiconductor technology on Ion Proton system (ThermoFisher) and aligned to Ensembl Mus musculus gene annotation v79 and expression normalized to transcript length (reads per kilobase per million reads [RPKM]). Differentially expressed (DE) genes were identified by false discovery rate (FDR) ≤ 0.05 and a posterior fold change of ≥2 or ≤0.5. Gene set enrichment analysis (GSEA) v3.0 for Hallmark and Immunologic gene sets was performed. Pathway analysis of DE genes was performed with Ingenuity Pathway Analysis software.

Reverse transcription-quantitative polymerase chain reaction

Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) was performed using iScript cDNA Synthesis Kit followed by mouse TaqMan Non-Coding RNA Assay for Snord123 (Applied Biosystems) on a CFX96 Real-Time PCR detection system (Bio-Rad).

p16INK4a Cytometry

Splenocytes were stained with anti-CDKN2A/p16INK4a DY488 (LSBio), CD3 PE-Cy7 (BD Biosciences), and CD19 PE-Texas Red (Life Technologies), run on an LSRFortessa and analyzed using BD FACSDiva software.

For complete list of materials and methods, see Supplementary Files.

Results and Discussion

Age and Vaccination History Alter Gene Expression in Splenic B Cells of Young, Middle-Aged, or Old Adult Mice

Aging is a dynamic process that occurs throughout adult life and prior to any immunosenescence in older individuals. Thus, to evaluate age-related changes in B cells, we used 3 ages: young (15–23 weeks), middle-aged (63–81 weeks), and old mice (103–119 weeks; Figure 1A). Select mice also received a flu vaccination prior to analyses (groups designated with “v.”). Vaccinated mice displayed age-dependent serum flu-specific IgG1 or IgG2a antibodies, with low or no development of antibodies in v.Old group or naive mice, typical of immunosenescence in older individuals or animals (Figure 1B). These changes were minimal in the v.Middle-age group compared to v.Young except for IgG2a, consistent with the decrease in Th1 immunity noted with age (7,8).

Figure 1.

Age and vaccination history alter gene expression in splenic B cells of Young, Middle-aged, or Old adult mice. (A) Schematic of experimental design. BALB/c mice ages shown were either left naive or immunized with the flu vaccine, then 21 days later serum and B cells collected, the latter for RNA-sequencing analysis. (B) Serum anti-hemagglutinin IgG1 or IgG2 responses in vaccinated or naive mice, yellow dots indicate samples chosen at random for B cell analyses. *p < .05 or **p < .01. (C) (top) Posterior fold change of age and vaccination groups compared to Young group by violin plot. Posterior fold change ≥2 or ≤0.5 (which has a negative reciprocal of −2) indicated with dashed lines. (Bottom) Principal component analysis of gene expression for all genes with an average RPKM > 1 across all replicates by age and vaccination status. Percentages in parenthesis indicate data variation explained by PC1 or PC2. (D) Gene set enrichment analysis (GSEA) analysis for Hallmark gene sets by indicated group comparisons and (E) GSEA analyses for select Immunologic B cell subset gene sets. All gene sets performed using RPKM with indicated normalized enrichment score (NES), FDR q-value (color), and number of genes in set (dot size).

Splenic B cells were isolated using negative magnetic selection and analyzed by RNA-sequencing to estimate individual gene transcripts. Gene expression changes in B cells compared to the naive Young group were observed with vaccination (v.Young) as well as age (Middle-age, v.Middle-age, Old, v.Old; Figure 1C, top). Clustering of samples within each group by principal component analysis (PCA; Figure 1C, bottom) confirmed that both vaccination and age contributed to the variance of transcriptome changes. Interestingly, in PC1, all vaccinated groups clustered to the right of unvaccinated groups, suggesting a major effect of prior vaccination, even in the oldest vaccinated group (v.Old), which did not have a robust antibody response to vaccination.

To identify differences between well-defined biological states or processes, GSEA was performed using Hallmark gene sets (Figure 1D; Supplementary Table 1). Normalized enrichment score (NES) indicates up- or downregulation of the set. Three gene sets (NOTCH signaling, WNT/β-catenin signaling, and KRAS signaling downregulated [DN], indicated in red) were significantly downregulated in Old versus Young groups. However, this increased to 19 significantly different gene sets when v.Old and v.Young groups were compared, including both upregulated and downregulated sets indicative of signaling (eg, TNFA, TGF_BETA, IL2_STAT5, MTORC1), immune or protein response (inflammatory response, interferon_alpha or gamma, E2F targets), proliferation (G2M checkpoint), and metabolism (eg, oxidative phosphorylation, glycolysis). Coagulation gene set was the only difference observed between v.Old and Old. Five gene sets were different between v.Young and Young, though only one of these (TNFA signaling) was also observed in the v.Old and v.Young comparison. Thus, the most observable GSEA Hallmark gene set changes occurred in age-group comparisons with a prior history of vaccination (v.Old vs v.Young).

To relate our findings to those of previous immunologic studies, GSEA was performed using 52 Immunologic gene sets specific for B cell subset changes (Figure 1E; Supplementary Table 1). Seven downregulated gene sets were significantly altered between Old versus Young groups, with higher levels of memory, germinal center, or light/dark zone B cells than naive cells observed within the Old group. A similar trend was observed for these gene sets when v.Old and v.Young groups were compared; however, a number of additional gene sets were also significant (34 in total) including reduced B2 versus B1 cells in v.Old and other upregulated gene sets identifying changes with memory and germinal center cells. Many gene sets were also altered between v.Young and Young, partially overlapping with previously mentioned comparisons (with 12 shown and 17 additional compiled in Supplementary Table 1). In contrast, minimal differences were observed between v.Old and Old, with only follicular versus early germinal center being significant. Thus, many changes involving B cell activation with antigen stimulation or vaccination are lost with age, as expected. Yet, as before, the most observable GSEA B cell subset gene set changes occurred in age group comparisons with a prior history of vaccination.

A Small Number of Genes Including Cdkn2a and Snord123 Are Consistently Expressed in Aging B Cells Irrespective of Vaccination History

Next, we examined DE genes significantly altered in older mice compared to young (Figure 2A). In the naive groups, we observed 499 DE genes (302 upregulated, 197 downregulated) significantly altered in B cells from both Middle-age and Old groups compared with Young; with additional unique expression in 391 genes (262 up, 129 down) in the Middle-age group and 1328 genes (729 up, 599 down) in the Old group. In the vaccinated groups, we observed a smaller number of shared transcripts significantly altered in B cells from both v.Middle-age and v.Old mice versus v.Young mice: 129 total DE genes (65 up, 64 down). Curiously, we saw many more DE genes altered in v.Middle-age versus v.Young groups than with naive groups: 1967 genes (1459 up, 508 down); and fewer in the v.Old versus v.Young: 435 genes (256 up, 179 down).

Figure 2.

A small number of genes including Cdkn2a and Snord123 are consistently altered in aging B cells irrespective of vaccination history. (A) Venn diagrams of significantly differentially expressed (DE) genes from comparison groups as shown. (B) Hierarchical clustering by expression levels of the 39 genes DE from all age/vaccination comparisons. (C) Percentage of CD19⁺ B cells from vaccinated mice expressing Cdkn2a-encoded protein p16IN4a intracellular by cytometric staining. *p < .05. (D) Snord123 levels quantified by real-time PCR from B cells.

We observed a total of 39 genes related to aging (17 up, 22 down; Figure 2A) shared by all age comparisons within naive or vaccinated groups. Hierarchical clustering by expression levels clearly distinguished between young groups but not older groups, as might be expected with our DE gene selection strategy (Figure 2B). Only 1 DE gene, Snord123 (red box), was shared in additional age comparisons made across vaccination status (eg, Middle-age vs v.Young). Several of the identified genes have been associated with age and/or senescence (Cdkn2a, Edaradd, Pde4c), cell cycle and apoptosis (Kdm6b, Bmf), or immune cell activation (Clcf1, Nlrc3). However, many have not previously been associated with age and therefore could be considered novel targets for future study (eg, Snord123). Box C/D small nucleolar (sno) RNAs regulate rRNAs through methylation or pre-mRNA splicing and are linked to malignancy but not specifically B cells or age (24).

To confirm our results, we analyzed the Cdkn2a-encoded protein p16INK4a in splenic B cells from vaccinated mice (Figure 2C). Expression of this protein was significantly higher levels in the v.Old group compared with v.Young and v.Middle-Age mice. In addition, we quantified Snord123 levels by real-time PCR and observed a similar age-related trend (Figure 2C), though additional follow-up using specific tools for small RNA analyses is warranted.

Canonical Pathway Changes in B Cells Are Dependent Upon Age and Vaccination History

For canonical pathways analyses, we observed 7 pathways significantly altered in B cells from 3 out of 4 comparisons between our older groups compared to young irrespective of vaccination history, including: EIF2 signaling, cell cycle: G2/M DNA damage checkpoint regulation, mTOR signaling, p53 signaling, molecular mechanisms of cancer, Paxillin signaling, and Tec kinase signaling (Table 1; Supplementary Table 1). Several of these are implicated in aging or age-related malignancies, including cell growth and proliferation (eg, EIF2 signaling, cell cycle) and both p53 and mTOR signaling are major pathways altered with aging (25). Tec kinase signaling is critical to B cell development and was decreased in older groups (26). Though Paxillin signaling has not been evaluated in aging B cell populations, it is critical for cell adhesion and migration and can be altered in malignancies (27) or wound healing and tissue repair in aging skin (28).

Table 1.

Significant Changes in Canonical Pathways and Gene Ratios in B Cells Based on Age and Vaccination History*

	Significant Gene Ratios for Each Group Comparison†
Pathway	Middle-age vs Young	Old vs Young	v.Middle-age vs v.Young	v.Old vs v.Young	Middle-age vs Old	v.Middle-age vs v.Old
Old and Middle-Age vs young (irrespective of vaccination history)
EIF2 signaling	20/194	20/194	14/194	ns	ns	ns
Cell cycle: G2/M DNA damage checkpoint Regulation	5/492	9/49	ns	7/49	ns	ns
mTOR signaling	9/199	15/199	13/199	ns	ns	ns
p53 signaling	7/111	10/111	ns	5/111	ns	ns
Molecular mechanisms of cancer	15/347	25/374	ns	13/374	ns	ns
Paxillin signaling	6/113	9/113	ns	5/113	ns	ns
Tec kinase signaling	7/170	ns	16/170	6/170	ns	ns
Old and Middle-Age vs Young (not observed with prior vaccination)
Cell cycle: G1/S checkpoint regulation	4/64	8/64	ns	ns	ns	ns
Regulation of eIF4 and p70S6K signaling	10/157	12/157	ns	ns	ns	ns
Middle-age vs Young or Old (irrespective of vaccination history)
Signaling by Rho family GTPases	14/247	ns	19/247	10/247	ns	12/247
RhoGDI signaling	7/173	ns	13/173	6/173	ns	6/173
Leukocyte extravasation signaling	9/210	ns	ns	7/210	1/210	ns
Middle-age vs Young (irrespective of vaccination history)
Role of PRR in recognition of bacteria and viruses	6/137	ns	14/137	ns	ns	ns
Role of tissue factor in cancer	6/122	ns	10/122	ns	ns	ns
Sphingosine-1-phosphate signaling	8/122	ns	9/122	ns	ns	ns
Mitochondrial L-carnitine shuttle	2/17	ns	3/17	ns	ns	ns
Aryl hydrocarbon receptor signaling	6/140	ns	11/140	ns	ns	ns
Old vs young (irrespective of vaccination history)
Cyclins and cell cycle regulation	ns	12/78	ns	6/78	ns	ns
Mitotic roles of polo-like kinase	ns	12/66	ns	7/66	ns	ns
DNA damage-induced 14-3-3σ signaling	ns	4/19	ns	3/19	ns	ns
GADD45 signaling	ns	3/19	ns	3/19	ns	ns
Estrogen-mediated S-phase entry	ns	4/24	ns	3/24	ns	ns
ATM signaling	ns	7/80	ns	4/80	ns	ns
Other
Phagosome formation	ns	10/122	13/122	ns	ns	ns

Notes: ns = not significant.

*Pathways significant (p < .05) for ≥2 comparisons to Young or v.Young included.

^†Ratios defined as # of DE genes/# of genes in pathway with only significant ratios shown.

In addition, we observed 2 pathways related to cell growth and protein transcription only in comparisons between naive age groups: cell cycle: G1/S checkpoint regulation and regulation of eIF4 and p70S6K signaling. Three other pathways were significantly altered between middle-age groups and young or old groups irrespective of vaccination history, including: signaling by Rho family GTPases, RhoGDI signaling, and leukocyte extravasation signaling. Rho family GTPases are key signal-transducing proteins regulating B cell development, migration, and activation with ties to Paxillin (27). Likely, these pathways are progressively altered with increasing age and could be playing major roles in poor and delayed responses to infections, like COVID-19, or vaccinations in older individuals.

We also observed canonical pathway changes that appeared to be intrinsic to either middle- or old-age B cells. Five pathways were altered only in middle-age versus young groups irrespective of vaccination history that were not observed in older mice, including role of PRR recognition of bacterial and viruses, role of tissue factor in cancer, sphingosine-1-phosphate signaling, mitochondrial L-carnitine shuttle pathway, aryl hydrocarbon receptor signaling (Table 1). In addition, 5 pathways were only altered in old versus young groups irrespective of vaccination history that were not observed in middle-aged mice, including cyclins and cell cycle regulation, mitotic roles of polo-like kinase, DNA damage-induced 14-3-3σ signaling, GADD45 signaling, estrogen-mediated S-phase entry, and ATM signaling. Such age-specific changes, particularly in the middle-age group when changes in B cell responses are only starting to emerge (eg, Figure 1B) would be intriguing to study further, particularly how they contribute to immunosenescence, development of unique age-associated B cell subpopulations, and as early targets of therapeutic intervention.

Taken together, our results indicate the importance of evaluating several ages and previous antigen exposure for aging B cell studies, validating an experimental model for future studies (eg, single-cell sequencing, B cell subpopulations). We also identified several novel pathways and DE genes for further investigation, particularly in the context of mounting immune responses to vaccinations or new infections. Future studies using both genders and repeated or early-life antigen exposure are also suggested to expand these results.

Funding

This work was supported by the National Institutes of Health (Aging COBRE pilot project P20GM103629 and R01AI114697 to E.B.N.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflict of Interest

None declared.