All animal protocols were approved by the Animal Care and Use Committee at the University of Virginia. TET2-KO mice (24) were provided by Dr. Kenneth Walsh (University of Virginia). The model was generated by Ko et al. and targeted the endogenous TET2 locus to create a conditional allele that enabled the deletion of exons 8, 9, and 10, the catalytic region of TET2 (24). Mice were maintained on a 12-h light/dark schedule in a specific pathogen-free animal facility and given food (standard chow diet, Tekland 7012) and water ad libitum. The number of mice included in each study is indicated in the figures or the associated legends.

Bone marrow, spleen, and peritoneal cavity cells were processed for flow cytometry as previously described (48). Briefly, following sacrifice by CO₂ overdose, peritoneal cells were harvested by flushing the peritoneal cavity with 10 mL FACS buffer (PBS containing 1% BSA, 0.05% NaN3). The spleen and one femur and tibia were removed. Spleens and flushed bone marrow were filtered through a 70 μm cell strainer. Red blood cells were lysed from single-cell suspensions of bone marrow and spleen using a lysis buffer containing 155 mM NH₄Cl, 10 mM KHCO₃, and 0.1 mM EDTA. Cell surface Fc receptors were blocked using anti-CD16/32 (clone:93, 4 eBioscience), then cells were stained with fluorescently conjugated antibodies against cell surface markers. Cells were stained with fixable Live/Dead Zombie NIR (Life Technologies) for dead cell discrimination, then fixed in 2% PFA in PBS. For FAC sorting, cells were resuspended in modified FACS buffer (PBS with 1% BSA) and 4’,6-Diamidino-2-Phenylindole (DAPI) live/dead stain then immediately taken to the University of Virginia Flow Cytometry Core for sorting. B-1a and B-1b cells were sorted to better than 99% purity from their parent gate. Clone and fluorophore information for the flow cytometry antibodies used in murine experiments to immunophenotype or FAC-sort B cell subsets are given in Tables 1 , 2 respectively. All flow cytometry was conducted at the University of Virginia Flow Cytometry Core Facility. Immunophenotyping was performed on an Aurora Borealis 5-laser (Cytek) cytometer. FAC-sorting was performed on an Influx Cell Sorter (Becton Dickinson). Data analysis and flow plots were generated using OMIQ software (Dotmatics). Representative flow plots were chosen based on the samples whose population frequencies were closest to the mean for that group. Gates on flow plots were set using fluorescence minus one (FMO) controls.

Total IgM in mouse plasma was measured using colorimetric ELISA as described previously (48). Briefly, EIA/RIA high-binding microplates were coated with goat anti-mouse IgM, capture antibody (Southern Biotech, 1020-01). Mouse IgM standards (Southern Biotech, 0101-01), or plasma samples were detected with alkaline phosphatase-conjugated goat anti-mouse IgM secondary antibody (Southern Biotech, 1020-04); and pNPP substrate (Southern Biotech 0201-01). Absorbance measurements were analyzed with a SpectraMAX 190 microplate reader (Molecular Devices) at 405 nm. The standard curve was determined using a 4-parameter function and concentration measurements were extrapolated using Softmax Pro 3.1.2 software. Only samples with CV<15% and within the standard curve were included in the analysis.

Peritoneal B-1a, B-1b, and B-2 cells obtained from TET2-KO and TET2-WT C57BL/6 mice were sort-purified directly into RLT Plus Buffer (Qiagen). RNA and DNA were extracted using the Qiagen AllPrep kit. The purified RNAs were stored at −80°C before being sent to Novogene for sequencing. Total RNA was stored in RNase-free water to directly synthesize first strand, followed by the whole-length LD-PCR amplification. The amplified ds-cDNA(double-stranded DNA) was purified with AMPure XP beads and quantified with Qubit. The cDNA samples were sheared by the Covaris system, and then the sheared fragments were end-repaired, A-tailed, and ligated to sequencing adaptors. A size selection of about 200 bp was performed before the PCR enrichment. Library concentration was first quantified using a Qubit 2.0 fluorometer (Life Technologies), and then diluted to 2 ng/µl before checking insert size on an Agilent 2100 and quantifying to greater accuracy by qPCR. Ultra-low input bulk RNA sequencing was performed on the NovaSeq 6000 PE150 (Illumina).

The quality trimming was performed using fastp (49) with default settings. Mapping to the GRCm39 genome was performed with STAR (50), followed by featureCounts (51) to count reads mapped to genes. DESeq2 (52) was used for differential analysis, followed by pathway analysis using clusterProfiler (53) with the Gene Ontology (GO) (54) database.

For BCR analysis, quality trimming was performed using fastp (49) and TRUST4 (55) was subsequently used to identify BCR repertoire in paired sequencing reads using the international ImMunoGeneTics (IMGT) information system database as a reference. Results were analyzed using R and circus plots were made using circos Bioconductor package (56). The code developed for these analyses will be available on the following Github page: https://github.com/mariamurach/TET2 and https://github.com/mariamurach/bcr_R.

In Figures 1 , 3D , 4C , comparisons were conducted between the TET2-KO and WT strains using Prism 10.0 with unpaired, two-tailed Mann-Whitney U-tests. Values shown are mean ± SD. In Figures 3B , 4A Wilcoxon Rank Sum and Signed Rank Tests were used to determine the significance of differences in proportions of unique CDR3 sequences, isotypes, and usages of specific V, D, and J chains between TET2-KO and WT groups. In Figure 5B , chi-squared test was performed to assess the significance of the association between the number of unique CDR3 amino acid sequences in B-1a and B-1b cells from TET2-KO and WT mice.

To determine the impact of the loss of TET2 on major immune cell subtypes in the peritoneal cavity, BM, and spleen of TET2-KO and littermate control mice, spectral flow cytometry was performed ( Figure 1 ). B cells were defined as CD45+ CD19+; T cells were defined as CD45+ CD5+ CD19-; Macrophages (Macs) were defined as CD45+ CD5- CD19- F4-80+ CD11b+; and NK cells were defined as CD45+ CD5- CD19- NK1.1+ ( Supplementary Figure 1 ). We found that there was a higher B cell frequency and number in the peritoneal cavity ( Figures 1A, D ), but not in the BM ( Figures 1B, E ) or spleen ( Figures 1C, F ) of TET2-KO mice compared to controls. Numbers of T cells (p-value = 0.0043), and NK cells (p-value = 0.0152) from TET2-KO mice in the peritoneal cavity ( Figure 1D ) were also greater than controls. There were no significant differences in immune cell numbers from TET2-KO mice in the bone marrow ( Figure 1E ), while in the spleen there was a trending increase in B cells (p-value = 0.0649) with a significant increase in T cells (p-value = 0.0260) and a trending increase in NK cells (p-value = 0.0649) compared to WT ( Figure 1F ). Upon examination of B cell subsets specifically, we found that in the peritoneal cavity, all B cell subsets were significantly increased in frequency (B-1a p-value = 0.0152, B-1b p-value = 0.0411, B-2 p-value = 0.0260) and in number (B-1a p-value = 0.0022, B-1b p-value = 0.0043, B-2 p-value = 0.0022) in TET2-KO mice compared to WT ( Figures 1G, J ). However, in the BM ( Figures 1H, K ) only the B-1a cell subset frequency was significantly increased in TET2-KO compared to WT mice (p-value = 0.0260). In the spleen ( Figures 1I, L ) B-1b cells but not B-1a cells were elevated in both frequency and number (B-1a p-value = 0.0173, B-1b p-value = 0.0043). There was no difference in TET2-KO B-2 cell frequency in the spleen, although the total number of B-2 cells was significantly increased (p-value = 0.0260).

To identify genes differentially expressed in B cell subtypes in mice with TET2-KO compared to WT control, we performed RNA-sequencing (RNASeq) on sort-purified peritoneal B-1a and B-1b cells from TET2-KO and WT mice ( Figures 2A, B ; Supplementary Figure 2 ). We utilized peritoneal B-1 cells due to their abundance in this specific niche, as well as due to the phenotypic changes we observed in Figure 1 . We found that the global knockout of TET2 had a more significant impact on gene expression within B-1a cells compared to B-1b cells. Specifically, we observed a downregulation in the expression of several immunoglobulin genes in B-1a cells (and to a lesser extent in B-1b cells) from TET2-KO mice compared to their WT counterparts ( Figures 2C, D ; Supplementary Figure 3 ). Consistent with this finding, Gene Set Enrichment Analysis (GSEA) revealed that TET2 loss markedly affects pathways linked to immunoglobulin production and immune response activation, primarily within B-1a cells ( Figure 3A ). Indeed, the expression of numerous V genes (from both heavy and light chains) was decreased in B-1a cells from TET2-KO mice ( Figures 3B, C ). Similarly, the expression of genes involved in the activation of molecular mediators of the immune response was also decreased in these cells ( Supplementary Figures 3, 5 ). Interestingly, significant enrichment in neurotransmitter and synapse-related pathways was seen in both B-1a and B-1b cells from the TET2-KO animals compared to the control animals ( Supplementary Figure 3B , These GSEA results provide potential avenues for further hypothesis-driven studies of the role of TET2 in sensory-neural control of B cells, an emerging area of potential significance, recently also connected to cardiovascular disease development via other immune cells (57–59). Notably, while there are too many differentially expressed genes (DEGs) to test all at the protein level, one of the proteins encoded by our DEG, Sell, also known as CD62L, was also in our flow panel, allowing us to determine if the change in gene expression was also accompanied by changes in the protein level. Indeed, consistent with the decrease in CD62L RNA, we also saw a decrease in CD62L on the surface B-1a and B-1b cells in TET2-KO mice ( Figures 3C, D ).

Using TRUST4, a tool for analyzing the B cell receptor (BCR) using bulk RNASeq (57) and the IMGT (58) database, we were able to identify Ig isotype transcripts present in sequencing data and their distribution across B-1a and B-1b cells from TET2-KO and WT mice ( Figure 4A ). We found that there was a statistically significant increase in the expression of IgM in the B-1a cells from TET2-KO mice, but we do not see that effect in B1-b cells which is in accordance with the increase expression of AIDCA, a gene involved in class-switch recombination ( Figure 4B ; Supplementary Figure 5 ). In contrast, there was not a significant change in the distribution of IgD, IgG, or IgA isotypes expressed by the different B-1 cells in TET2-KO and WT mice ( Figure 4A ). Consistent with the increase in B-1 cells that we observed in niches that support antibody production, such as the spleen and BM ( Figures 1E, F ), and the increase in the IgM transcript in B-1a cells, circulating plasma IgM levels were higher in the TET2-KO compared to WT mice (p-value = 0.0075) ( Figure 4C ). Marginal zone B-2 cells (MZB) are another source of IgM and we did observe an increase in MZB cell number in TET2-KO mice compared to controls, which could contribute to the overall increase in circulating IgM ( Supplementary Figure 6 ). We observed no change in circulating IgG levels.

To assess differences in the heavy chain BCR repertoire in B-1 cells from TET2-KO and WT mice, we performed an analysis of the CDR3 sequences using our bulk RNASeq data and TRUST4 (57). Results demonstrated that in both B-1a and B-1b cells from TET2-KO mice, CDR3 diversity was reduced compared to WT mice ( Figure 5A ). The reduction in CDR3 sequence diversity in the B-1a cells from TET2-KO mice compared to WT was statistically significant (p-value = 0.02857) ( Figure 5A ), while the reduction in unique CDR3 sequences in TET2-KO B-1b cells compared to WT was trending (p-value = 0.05714) ( Figure 5A ). It is not feasible to establish the presence of clonal expansion based on bulk RNASeq data, due to the inability to determine the absolute number of cells and their level of expression of each Ig transcript at a single-cell resolution in a given B cell population. However, a high proportion of replicated sequences suggests the presence of clonally dividing, or self-renewing B-1a cells, as they are known to do. Here we define replicated sequences as those whose frequency is greater than 1% of all sequences. In B-1a cells from TET2-KO mice, we observed that 72% of the CDR3 sequences were replicated, compared to B-1a cells from WT mice which only had 15% of the total CDR3 sequences replicated ( Figures 5B , 6A ). Thirteen unique CDR3 sequences covered 72% of all CDR3s in B-1a cells from TET2-KO mice, while 4972 CDR3 sequences made up the other 28% of the total number of identified CDR3s ( Figures 5B , 6A ). In B-1b cells from TET2-KO mice, 25% of all CDR3 sequences are made up of 12 unique CDR3 sequences, while 11107 CDR3 sequences made up the rest of the 75% ( Figures 5B , 6B ). Differences in the number of replicated unique CDR3 sequences were significant based on Chi-squared tests for B-1a (p-value = 3.2x10^-13) and B-1b cells (p-value = 3.1x10^-6) from TET2-KO mice compared to WT ( Figure 5B ). An analysis of the commonality of replicated CDR3 sequences revealed that there was minimal overlap in the CDR3 sequence between B-1a cells from TET2-KO and WT mice or in B-1b cells from TET2-KO and WT mice, respectively ( Figure 5C ). Interestingly, there was a greater degree of commonality in the CDR3 sequence comparing B-1a and B-1b cells from TET2-KO mice. We visualized the proportion of replicated CDR3 sequences across B-1 cell subsets in TET2-KO and WT mice using pie charts ( Figures 6A, B ). Consistent with our findings in Figure 5C , the CDR3 sequences that were most abundantly represented in B-1a cells from WT mice were represented in B-1a cells from TET2-KO mice at different proportions, and there were more similarities in replicated CDR3 sequences between B-1a and B-1b cells from TET2-KO mice ( Figures 6C–D ).

Since antigen binding specificity is not just determined by the heavy chain CDR3, we performed an analysis of the light chain BCR repertoire from our bulk RNASeq data with TRUST4. Results demonstrated that in both B-1a and B-1b cells from TET2-KO mice, CDR3 diversity was reduced compared to WT mice ( Supplementary Figure 7 ). The reduction in CDR3 sequence diversity in the B-1a cells from TET2-KO mice compared to WT was statistically significant (p-value = 0.029) ( Supplementary Figure 7A ). The reduction in CDR3 sequences in B-1b cells from TET2-KO mice compared to WT was trending (p-value = 0.133) ( Supplementary Figure 7A ). Similar to what we observed in the heavy chain CDR3 sequences, the number of replicated light chain CDR3 sequences was over 2-fold greater in B-1a cells from TET2-KO compared to WT mice ( Supplementary Figures 7B-D ). The number of unique CDR3 sequences from the light chain accounted for a similar percentage of total CDR3 sequences as seen in the heavy chain results in B-1a cells from TET2-KO mice ( Supplementary Figures 7B-D ). Differences in the number of replicated unique CDR3 sequences were significant based on Chi-squared tests for B-1a (p-value = 0.01) and B-1b cells (p-value = 0.02) from TET2-KO mice compared to WT. While the role of the light chain in antigen binding and specificity remains less well-known compared to the heavy chain, it still contributes to those functions (59). These results in the light chain CDR3 provide additional support that B-1a cells are more profoundly impacted by loss of TET2 than B-1b cells, and the diversity of antigen-specific IgMs may be affected as a result.

Analysis of specific V_H–D_H–J_H gene region usage in B-1a and B-1b cells from TET2-KO and WT mice revealed high usage of V_H1, V_H11, and V_H12 in B-1 cells consistent with prior findings ( Supplementary Figure 8 ) (18, 60, 61). The B-1a cells from the TET2-KO mouse appeared to have greater usage of these regions. There were also several reductions in V_H region usage in the B-1a cells from the TET2-KO mice, but these were regions of minimal usage and of unclear significance. We also analyzed differences in the specific V_K/L–J_K/L gene regions of the light chain CDR3 sequence in B-1a and B-1b cells from TET2-KO or WT mice ( Supplementary Figure 9 ) and similarly found differences in V_K and J_K usage predominantly in B-1a compared to B-1b cells. An analysis of kappa and lambda ratio revealed that there is more lambda light chain utilization in B-1b cells from TET2-KO mice compared to WT despite not reaching significance, while showing no difference in kappa/lambda ratio in B-1a cells ( Supplementary Figure 9E ).

Circos plots ( Figures 7A–H ), measuring the relative frequency of each V-J pairing revealed a greater abundance of V_H1-J _H1, V_H11-J _H1, and V_H12-J _H4 in B-1a cells from TET2-KO mice compared to control ( Figure 7I ), suggesting that TET2 has an important role in specific V-J recombination of B-1a cells. These specific recombination events could be important for creating the over-representation of the specific CDR3s in B-1a cells from TET2-KO mice. The increase in V_H12-J _H4 pairing in TET2-KO mice was also seen in the B-1b cells but only constituted 5% of all pairings compared to over 20% in the B-1a cells ( Figures 7I–J ). These data are consistent with the loss of TET2 generating a more pronounced effect on the BCR in B-1a cells compared to B-1b cells.