Six disease-free rhesus macaques (Macaca mulatta) and six African green monkeys (Chlorocebus aethiops) were used in this study. Detailed records of these non-human primates (NHPs) are provided in Supplementary Table S1 . Each animal received an injection of 2.5 mg of IDR-053, a synthetic TLR7/8 agonist with the sequence 5’-X5UGCUGCUUGUG-X-GUGUUCGUCGUX5-5’, where X is a glycerol linker and X5 is 1,5-pentanediol. The synthesis protocol for this molecule is described in patent EP2357231A2. All animals were heathy and fully conscious before and after the injection. They were housed in accordance with the standards of the Guide for the Care and Use of Laboratory Animals, both during and after the experiment period. The study was approved by the Merck Institutional Animal Care and Use Committee (IACUC) and the Institutional Review Board (IRB) of BGI-Shenzhen (No. 13040).

Peripheral blood was collected prior to stimulation to establish a baseline. Following the administration of 25 mg of IDR-053, blood samples were collected at 8, 24, 48, and 72 hours, as well as 1-week post-stimulation. Approximately 2.6 mL of blood collected at 0, 24, 72 hours, and 1 week was analyzed by flow cytometry.

For cell surface staining, fresh PBMCs were resuspended in 100 μL of flow cytometry staining buffer (3% bovine serum albumin in phosphate-buffered saline) containing primary antibody ligated with biotin against CD3 ( Supplementary Table S2 ), incubated for 40 min at 4 °C and were washed with FACS staining buffer. Cells were then resuspended in 100 μL of flow cytometry staining buffer containing antibodies against biotin, CD20, and HLA-DR ( Supplementary Table S2 ) and were incubated for 40 min at 4 °C. For intracellular staining, cells were washed with flow cytometry staining buffer, permeabilized using the Cytofix/Cytoperm Kit (BD Biosciences, Cat. no: 554714), and then stained with anti-Ki-67 antibody. The stained cells were analyzed by a BD FACS Aria II cell sorter (BD Biosciences). B cells were gated by CD3^-/CD20^low/+ ( Supplementary Figure S1 ).

Peripheral blood mononuclear cells (PBMCs) were isolated from blood using Ficoll-Paque (GE Healthcare) centrifugation method. Total RNA was extracted from PBMCs using Trizol (Invitrogen, 15596-026). For each sample, 600 ng of RNA was used for antibody repertoire library preparation. We performed 5’-Rapid amplification of complementary DNA ends (5’RACE, Invitrogen, 18374-058) method with single-end-specific primers to specifically amplify the BCR repertoire, and then the PCR products were fragmented into 150–200 bp using Covaris E220, which described in details in our previous studies (21, 31). Primers targeting the constant regions of both heavy chain (21) and light chain ( Supplementary Table S3 ) were used. The sequencing library were quantified using Agilent 2100 Bioanalyzer system, and then were sequenced via Hiseq2000 (Illumina) using paired-end 150 bp.

Raw data were processed using IMonitor (24), as described previously. Briefly, 1) low-quality reads were filtered out, and paired-end reads were merged; 2) Merged sequences were aligned to the IMGT (http://www.imgt.org, for IgKL) or KIMDB database (32) (for IgH) reference sequences using BLAST, and followed by a re-alignment step to improve the accuracy of alignment, and then V and J genes were assigned for each sequence; 3) Complementarity-determining region 3 (CDR3) region was determined based on assigned V and J genes. Sequencing errors were corrected by sequencing qualities and frequencies. Clonotypes were defined as sequences with the same V and J gene usage and identical CDR3 animo-acid sequences. The statistics of processed data are shown ( Supplementary Table S4 ).

Lineages were defined as antibody sequences sharing the same V- and J-gene and differing by no more than one amino-acid mismatch within the CDR3; thus the similar clonotypes were clustered into a single lineage, indicating derivation from a common ancestral B cell (33). Lineages were identified separately for IgH and IgKL. Data from all timepoints of the same monkey are combined for lineage identification.

We randomly subsampled 500,000 light-chain and 400,000 heavy-chain sequences per sample for diversity estimation. There are several indices as follow.

Shannon index: an overall diversity of BCR repertoire.

where p_i is the frequency of the i th clonotype in a sample, and n is the total number of clonotypes. For the calculation of lineage diversity, lineages were used in place of clonotypes.

Gini index: an overall diversity of BCR repertoire.

where x i   and x j   denote the frequency of the ith and jth gene, clonotype or lineage.

Top 100: it represents the cumulative frequency of the top100 clonotypes or lineages.

where ​p_i is the frequency of clonotype i, sorted in descending order by clonotype frequency. For the calculation of lineage Top 100, lineages were used in place of clonotypes.

For two samples A and B, the Morisita–Horn similarity index (CH) is calculated as:

where x_i and y_i is the counts of ith CDR3 in samples A and B, respectively, X and Y are the total CDR3 counts in A and B, and S is the number of unique CDR3 sequences.

Based on sequence alignments, mismatches in the V and J regions were counted for each antibody sequence, separately for IgH and IgKL. The somatic hypermutation rate was calculated as:

where v_i and j_i are the numbers of mismatches in the V and J regions of sequence i, and V_i and J_i are the corresponding region lengths.

Lineages were assigned per monkey across all six timepoints; consequently, each lineage may contain IgH or IgK/IGL clonotypes sampled at multiple timepoints. For each lineage, clonotypes were first grouped by timepoint, and then three indices were calculated for each timepoint: cumulative frequency, number of unique CDR3 sequences, and mean mutation rate. For each index we computed the fold change (FC) relative to the pre-stimulation baseline (0 hr), and took the base-2 logarithm:

A log2​(FC)>0 indicates an increase relative to baseline. We defined a lineage as increased lineage for a given index when log2​(FC)>1 (i.e., ≥2-fold increase), and as decreased lineage when log2​(FC)<0.5. Increased and decreased lineages were identified independently for each of the three indices; thus, a given lineage could be classified as increased by one metric (e.g., cumulative frequency) but decreased by another (e.g., mutation rate).

We observed that the number of increased lineages was significantly higher at 48hr and 72hr post-stimulation compared with decreased lineages, indicating that TLR7/8 agonist–induced lineages were present at these timepoints. From the pool of increased lineages, a lineage was classified as a TLR7/8 agonist–induced expanded lineage if its log2​(FC) at either 48hr or 72hr met the following criteria: (1) IgKL: frequency log2​(FC) > 3 and absolute frequency > 0.001%; unique CDR3 number log2​(FC) > 2 and unique CDR3 number > 5; mutation rate log2​(FC) > 1.3 and mutation rate > 6%. (2) IgH: frequency log2​(FC) > 3 and frequency > 0.001%;mutation rate log2​(FC) > 1.5 and mutation rate > 5%.

PCR products yielded antibody sequences with variable V-region lengths (ranging from 10 bp to 160 bp). To obtain uniform-length sequences for tree construction, we extracted from each antibody sequence a subsequence comprising a 10-bp non-CDR3 V-region segment, the CDR3, and the J-gene region. Based on our lineage definition, all subsequences within a lineage were of equal length; therefore, multiple sequence alignment was not performed. Using these subsequences as input, phylogenetic trees were constructed by the Neighbor-Joining method implemented in MEGA7 (34).

Statistical comparisons between paired groups were performed using the Wilcoxon signed-rank test. When multiple tests were performed within a single figure, p-values were adjusted for false discovery rate (FDR), with adjusted P< 0.05 considered significant. Correlations were assessed using Spearman’s rank correlation coefficient. All analyses were conducted using in-house scripts in Perl (v5.26.2) and R (v3.6.0), employing the ggplot2 and dplyr packages.

The raw sequencing data are available at the China National GeneBank DataBase (CNGBdb) under accession number CNP0000866.

To investigate the B cell immune response, particularly the antibody repertoire, under a TLR7/8 agonist stimulation, six IRMs and six AGMs were included and each individual received an injection of TLR7/8 agonist ( Figure 1A ). Peripheral blood samples were collected at baseline (pre-stimulation) and at 8, 24, 48, and 72 hours (hr), as well as one week (1wk) post-stimulation. These samples were used to assess lymphocyte population via flow cytometry ( Supplementary Figure S1 ) and to monitor dynamic changes in the antibody repertoire through antibody sequencing. The counts of B cells, CD4+ T cells, and CD8+ T cells in peripheral blood transiently decreased at 24hr post-stimulation in both IRMs and AGMs ( Supplementary Figure S2 ), consistent with previous studies (35–37). This suggests the migration of these immune cells to lymphoid tissues, transient activation, and redistribution. To further evaluate lymphocyte activation, we measured HLA-DR expressed on B cells and CD69 expressed on T cells using flow cytometry. The mean fluorescence intensity (MFI) of HLA-DR on B cells increased at 24hr, while CD69 expression on CD4⁺ and CD8⁺ T cells was also upregulated at this timepoint ( Figure 1B ), indicating activation of both B and T cells. Furthermore, the cell proliferation marker Ki67 was examined in both B and T cells. Ki67⁺ B cells and Ki67⁺ CD4⁺ T cells increased at 24hr, and Ki67⁺ CD4⁺ and CD8⁺ T cells increased at 72hr in IRMs, whereas Ki67 expression did not change in AGMs ( Figure 1C ), suggesting the proliferation of lymphocytes in IRMs but not in AGMs. Taken together, the findings indicate that the TLR7/8 agonist induces lymphocyte redistribution at 24hr, accompanied by B and T cell activation and proliferation.

To evaluate the specific change of B cells induced by TLR7/8 agonist stimulation, the antibody repertoire, including immunoglobulin heavy-chain (IgH) and kappa and lambda light-chain (IgKL), was obtained by antibody sequencing. Due to the lack of identified germline genes in AGMs, IgH repertoire of AGMs were not presented. The gene usage and diversity of the antibody repertoire were first examined to provide a global overview. The Gini-index of IgKL V genes, J genes, and V-J gene pairings decreased markedly from 72hr to 1wk in IRMs, while in AGMs, it declined at both 48hr and 72hr but returning to baseline at 1wk ( Figure 2A ), suggesting an increase in gene-level diversity. We then illustrated the frequency change of each V gene, separated into IgK and IgL, and found that the abundance of IgL V genes increased after stimulation, with the most significantly expanded genes observed from λ chain in both IRM and AGM ( Figure 2B ; Supplementary Figure S3, S4 ), which was further confirmed by κ/λ ratio analyses ( Figure 2C ). The trends of the κ/λ ratio in both IRMs and AGMs were consistent with changes in gene diversity, indicating an increased frequency of λ genes usage following TLR7/8 agonist stimulation. To validate this, we used TRUST4 (38) to analyze RNA-seq data from B cells cultured with a TLR7/8 agonist (R848) (39). In two of three donors, we observed a decrease in the κ:λ ratio ( Supplementary Figure S5 ).

For clonotype repertoire diversity, Shannon-index increased from 48hr to 1wk in IRMs, while in AGMs, it rose at 48hr and 72hr but declined at 1wk in AGMs ( Figure 2D ). However, the cumulative frequency of the top100 clonotypes showed the opposite trends ( Figure 2D ), indicating that increased diversity was driven by a more balanced distribution of clonotypes rather than the very high-frequency ones. The diversities were further confirmed using the tool Recon (40), which calculates both observed and estimated diversity metrics ( Supplementary Figure S6 ). Since antibodies undergo proliferation, SHM, and isotype class switching for affinity maturation, lineages—representing functional units—were defined by clustering the similar antibodies derived from the same original B cell source (21). We then examined the overall diversity indices as well as the top100 lineages and found their kinetic changes following stimulation mirrored those observed in clonotypes ( Figure 2E ). To further investigate the frequency distribution of lineages contributing to the increased diversity, we compared lineage frequency distributions at 72hr to those at pre-stimulation. We found that the percentage of lineages with a frequency<0.001% decreased, while those with a frequency between 0.001% and 1% increase in both IRMs and AGMs ( Figure 2F ), suggesting that lineages with moderate frequency contributed to the observed diversity increase. Since SHM contributes to affinity maturation, we examined the mutation rate distribution for all samples in IRM. Almost all samples exhibited three mutation peaks ( Supplementary Figure S7 ), likely corresponding to naïve cells with very low mutation rates (IgKL non-mutated, mutation<0.5%), memory/effector cells with high mutation rates (IgKL mutated, mutation>4%), and mixed cell types with intermediate mutation rates. The proportion of IgKL non-mutated sequences decreased at 8hr, whereas IgKL mutated sequences increased at 48hr and maintained this trend in five individuals ( Figure 2G ).

For the IgH repertoire in IRM, V-gene usage was relatively stable, although several V genes showed significant changes from 48hr to 1wk ( Supplementary Figure S8 ). For clonotype- and lineage-based diversity, the Shannon index and Top100 metrics did not differ between timepoints; however, Recon-estimated diversity decreased at both 48hr and 72hr ( Supplementary Figure S9 ). Mutation rate distributions also exhibited three peaks across samples ( Supplementary Figure S10 ). IgM and IgD sequences displayed low mutation rates (<0.5%), whereas IgA, IgG, and IgE sequences showed high mutation rates (>3%) ( Supplementary Figure S11 ). The proportion of IgG decreased slightly at 24hr, whereas the proportions of other isotypes remained stable across timepoints ( Supplementary Figure S12A ). This decline in IgG may correspond to the reduction in absolute B-cell counts at the same time point ( Supplementary Figure S2 ). Consistent with single-cell sequencing studies reporting that naïve B cells primarily express IgM or IgD with very low mutation rates, while memory and effector B cells mainly express IgA or IgG with high mutation rates (41), we partitioned sequences into IgM/D non-mutated (mutation rate< 0.5%) and IgA/G/E mutated (mutation rate > 3%) groups as proxies for naïve and memory/effector cells, respectively. Using this classification, IgM/D non-mutated sequences decreased at 48hr, whereas IgA/G/E mutated sequences increased at 48hr ( Supplementary Figure S12B ).

Since we observed an increase in the diversity of antibody lineages, we further examined their characteristics. The definition of lineage reflects the affinity maturation process of B cells, which involves cell proliferation and SHM. These processes could correspond to the three characteristic analyses of lineage: frequency, unique CDR3 count, and mutation rate. At each timepoint of post-stimulation, lineages were classified as either increased or decreased by comparing the fold change (FC) of each characteristic with those at 0hr (see Methods). Under normal physiological conditions, without external interventions, the distributions of increased and decreased lineages would theoretically remain the same to maintain the repertoire homeostasis. However, external stimuli such as TLR7/8 stimulation may disrupt this equilibrium. First, we directly compared the distributions of the three characteristics between increased and decreased lineages. We found that lineages with either a frequency > 0.001% or a unique CDR3 count > 5 were significantly more prevalent among the increased lineages than among the decreased lineages from 48hr to 1wk post-stimulation, while the lineages with a mutation rate > 6% among the increased lineages were notably more frequent from 8hr to 1wk post-stimulation in both IRMs and AGMs ( Supplementary Figure S13 ). We further calculated the FC in frequency, unique CDR3 number, and mutation rate each lineage. For lineage frequency, the number of increased lineages with FC >3 was greater than that of decreased lineages at both 48hr and 72hr post-stimulation in both IRMs and AGMs ( Figures 3A, D ). For the unique CDR3 number per lineage, the number of increased lineages with FC >2 was significantly higher than that of decreased lineages at 48hr post-stimulation in both species ( Figures 3B, E ). Regarding mutation rate, the number of increased lineages exceeded that of decreased lineages at most timepoints including as early as 8hr post-stimulation in both species ( Figures 3C, F ), which may be attributed to the expansion of B cells originating from memory-like B cells with high-mutation rates. Additionally, in the IgH repertoire of IRMs, increased and decreased lineages exhibited similar results including the direct distribution and FC-based distributions across the three characteristics ( Supplementary Figure S14, S15 ). Collectively, these results demonstrate that TLR7/8 stimulation induces the expansion of a subset of antibody lineages through increased frequency, clonotype count and mutation rates, particularly at 48hr and 72hr post-stimulation.

Since we observed an increase in lineage distributions following TLR7/8 stimulation, we next aimed to identify the significantly expanded lineages at 48hr and 72hr timepoints as the TLR7/8 agonist-induced antibodies, based on the three predefined characteristics (see Methods). A total of 4653 expanded lineages were identified in IRMs, with 42.0% meeting the thresholds for two or three characteristics, including frequency, unique CDR3 number, and mutation rate ( Figure 4A ). Similarly, 3941 expanded lineages were identified in AGMs, with 38.7% meeting the thresholds for two or three characteristics ( Figure 4B ). Stream plots showed a consistent increase in expanded lineages over time, peaking at 72hr post-stimulation ( Figure 4C ; Supplementary Figure S16 ). To examine clonotype composition and dynamics, we constructed phylogenetic trees for several representative expanded lineages ( Figure 4D ; Supplementary Figure S17 ). The trees indicate that antibody clonotypes evolved from the 0hr sample to the 48hr and 72hr samples, underwent class switching from IgM to IgA, and accumulated somatic mutations, resulting in increased mutation rates ( Figure 4D ). To assess the identified expanded lineages, we examined the correlation between their frequencies and B cell activation, as reflected by the significant increase in HLA-DR expression in both IRMs and AGMs ( Figure 1B ). The changes of lineage frequencies were positively correlated with HLA-DR changes in IRMs ( Figure 4E ), while AGMs exhibited a similar trend of positive correlation ( Figure 4F ). The findings suggest that the TLR7/8 agonist-induced expanded lineages are associated with B cell activation.

We analyzed the characteristics of the expanded lineages following TLR7/8 agonist stimulation. First, we examined the composition of light chain. The κ/λ ratio of expanded lineages was significantly lower than that of randomly sampled lineages with an identical frequency distribution in both IRMs and AGMs ( Figure 5A ), suggesting a preferential usage of λ chain in response to TLR7/8 stimulation. Second, we assessed the similarity of expanded lineages across timepoints. Overlap indices were calculated between each pair of timepoints for each individual, revealing that expanded lineages showed the highest overlap index compared to both the all lineage pool and increased lineages ( Figure 5B ; Supplementary Figure S18 ). Furthermore, more than half of expanded lineages (> 51%) were present at all timepoints ( Figure 5C ), indicating that these lineages were consistently maintained over one week. Third, we examined the similarity of expanded lineages across individuals. Since disease and vaccine stimulation could elicit convergent antibodies across individuals to target antigens (19, 27, 42), it is important to determine whether TLR7/8 agonists, as potential vaccine adjuvants, induce convergent or divergent antibody responses. To serve as a control, we selected a set of control lineages with the same κ/λ ratio and the same number of lineages as the expanded lineage group. We then compared the expanded lineages to the control lineages based on both gene usage similarity and clonotype similarity. For gene usage, correlation coefficients between individuals were not significantly different, and IRM λ-originated expanded lineages exhibited even lower correlation values ( Figure 5D ). Regarding clonotypes, the inter-individual similarity of expanded lineages was significantly lower than that of control lineages, except for AGM λ-originated expanded lineages ( Figure 5E ). These findings demonstrate that the expanded lineages induced by TLR7/8 stimulation were highly divergent antibodies, and so the TLR7/8 agonist could provide a broad and diverse immune foundation when used as an adjuvant.

The proportion of IgG within expanded lineages decreased after stimulation, whereas the proportions of other isotypes remained stable across timepoints ( Supplementary Figure S19A ). In IgH expanded lineages, the proportion of IgM/IgD non-mutated sequences declined from 48hr to 1wk, while IgA/IgG/IgE mutated sequences showed a modest increase at 48hr ( Supplementary Figure S19B ). For IgKL expanded lineages, the proportion of non-mutated sequences decreased significantly after stimulation, whereas the proportion of mutated sequences increased significantly ( Supplementary Figure S19C ). Together, these results indicate that expanded lineages are characterized by elevated somatic mutation rates. To investigate the status of original B cells affected by TLR7/8 simulation, we tracked and analyzed the antibody sequences from the expanded lineages at the pre-stimulation (0hr), referred to as original-expanded lineages. The analysis included mutation rate and IgH isotypes, which provided insights into the status of the B cells. At 0hr, except for the original-expanded lineages, we randomly sampled the equal number of lineages from the remaining lineages as controls. Compared with the controls, the original-expanded lineages contained more IgA antibodies and fewer IgM antibodies ( Figure 6A ). Since antibodies from memory B cells have higher levels of SHM, while antibodies from naïve B cells have lower levels of SHM (43), we examined the mutation rates in different isotype of heavy chain and light chain. As a result, all isotype antibodies in original-expanded lineages exhibited significantly higher mutation rates, including IgM ( Figure 6B ). Interestingly, although IgM antibodies are generally from naïve B cells with very low mutated rates, the high-mutated IgM antibodies in these lineages may originate from IgM memory B cells (44). Within the IgK repertoire in IRM, mutation rates were higher in originally expanded lineages than in controls ( Figure 6C ). Antibodies were grouped into three origin categories—IgA/G/E-mutated, IgM-mutated, and IgM/G non-mutated—as proxies for memory/effector-like, IgM memory-like, and naïve B cells (45). IgA/G/E-mutated sequences dominated the expanded lineages (60%), significantly more than in controls; ~20% were mutated IgM ( Figure 6D ). In summary, our data suggest that TLR7/8 stimulation primarily activates memory- and effector-like B cells in peripheral blood; however, functional validation will be required in future studies.