Human Cord Blood B Cells Differ from the Adult Counterpart by Conserved Ig Repertoires and Accelerated Response Dynamics

Neonates and infants suffer from enhanced susceptibility to infections (1), and four out of 10 childhood deaths worldwide are due to infectious diseases (2, 3). Birth is a dramatic transition for the neonate, exposing it to an environment with a high microbial burden compared with that in utero, and the infant immune system itself is considered immature to meet these challenges (1, 3, 4). Low neutralizing Ab titers are a critical limitation in infant immunity, and current vaccination strategies aim at generating protective seroconversion (3, 5). T cell–independent (TI) polysaccharide vaccines either fail (6) or elicit short-lived responses with quickly waning IgG titers (7) and do not induce B cell memory (8) but rather hyperresponsiveness (immune exhaustion) (9). Similarly, T cell–dependent (TD) conjugate vaccinations are generally of low magnitude and do not persist well in early infancy (5, 10). Immature infant B cell immunity is associated with imbalances in T cell support (4, 11) and are further evident by the importance of maternal-derived Abs in neonatal immunity (1).

Several observations suggest that human umbilical cord blood (UCB) B cells (neonatal B cells) are intrinsically different from adult peripheral blood (PB) B cells (adult B cells) [e.g., by expressing lower levels of important surface molecules, including CD21 (12), TNFRSF13B (13), CD73 (14), CD22 (15), CD62L and CCR7 (16), CD40, CD80, and CD86 (13), all of them associated with a lower magnitude of B cell responses]. Differences in TLR expression influence neonatal immunity (17), and BCR signal transduction may differ significantly from adult B cells, depending on the stimulus (15, 18). It was shown that neonatal B cells can, in principle, respond to stimulation (19) but require lower TD stimulus doses (20), synergistic stimulation to overcome partial immaturity (21), or lack adequate surface triggering receptors (22). Moreover, IgV gene mutations are infrequent in neonatal B cells (23–25), affinity maturation is delayed (3), and IgG/IgA class switching is principally possible, but impaired (18), even upon costimulation with IL-21 (17, 26). Recent analyses further support the notion that the BCR repertoires of human fetal B cell precursors (27), or bulk neonatal B cells (28–32) differ from those of the adult B cell pool.

Environmental factors also contribute to the lower magnitude of infant responses compared with adults (3, 5). In mice, the interaction of neonatal lymphocytes and dendritic cells is immature, resulting from delayed maturation of follicular dendritic cell networks (33). Human infants also show limited germinal center (GC) reactions (34). Newly generated plasma cells (PC) compete for limited access to survival niches (35–37) and limited survival signals (38). Finally, single vaccine doses at birth can fail to elicit seroconversion while priming for secondary responses (5). The low IgV gene mutation load in neonatal B cells was proposed to favor memory B cell over PC differentiation (3, 39). In light of these multiple limitations of infant immunity, B cell–intrinsic differences are considered unlikely by themselves to provide an insurmountable obstacle to strong responses (1).

Current knowledge of the infant B cell system is also inferred from mouse models. In mice, neonatal B cells include developmentally and functionally distinct B cell lineage (B-1 cells) (40) whereas adult hematopoiesis produces conventional B-2 cells (41). Among other markers, CD5 is used to distinguish murine B-1a (CD5⁺) from B-1b (CD5⁻) cells. B-1 and B-2 cell transcriptome patterns diverge (42–44). BCR specificity and signaling strength influence fetal B cell development (45) and can instruct B-1 phenotypes in adult B cells (46). Fetal, but not adult B-1 cell Ig repertoires show a proximal-biased V-gene usage (47), and limited addition of nongermline-encoded (N) nucleotides (48, 49). Neonate mice exhibit preimmune clonal B-1 cell expansions (50). B-1 cells display a primitive, innate-like functionality and produce natural Abs, independent of T cell help (51). In humans, the existence of a B-1 cell counterpart is debated (23, 24, 52–60). CD5 expression in humans is not confined to a fetal B cell lineage (61–64) and human neonatal B cells share only a few of the defects observed in mice (1, 16).

Our understanding of human infant B cell immunity is limited. We assume that this is partly because previous studies did not separate mature from transitional B cell subsets and also mixed naive and memory B cells when comparing neonatal with adult CD19⁺ B cells. As these B cell subsets significantly differ in numbers between infants and adults (63, 65, 66), the outcome of BCR repertoire or functional analyses in vitro may have been blurred. Moreover, most previous studies performed end-point analyses, disregarding response dynamics. We present, in this study, a comprehensive molecular and functional characterization of purified human neonatal and adult mature B cells.

We show that cord blood includes a major fraction of mature B cells that is phenotypically most similar (but not identical) to adult naive B cells, including a large number of mature CD5⁺ B cells. These B cells persist in the PB during childhood but decrease with age. Mature neonatal B cells significantly differ in development, BCR repertoire, and response dynamics from their adult naive B cell counterpart. UCB mature B cells respond faster and more efficiently to autologous T cells but also to innate stimulation compared with adult naive B cells. They show strong IgA switching capacity, and their differentiation potential is mostly confined to Ab-secreting cell (ASC) fate. Their molecular and functional distinctness is supported by a conserved UCB Ig repertoire, as up to 8% of neonatal BCRs among the 180,000 BCR rearrangements analyzed are clonotypic (interindividually shared). Finally, we distinguish UCB mature CD5⁺ and CD5⁻ B cell subsets and show both similarities and differences to murine B-1a and B-1b cells, suggesting similar developmental pathways for ontogenetically early B cell subsets in humans and mice and a propensity to IgM secretion but overall limited comparability between both species.

The objective of this study was to analyze and compare mature B cells from UCB and adult PB to clarify whether UCB B cells are impaired on molecular or functional level. Sample sizes were chosen according to the stochastic requirements of each test, granting statistical interpretability of all experiments. No outliers were removed for statistical analyses. PB samples from healthy adults and UCB samples from healthy term neonates (n > 200, ratio male/female 0.97) were obtained after informed consent according to the Declaration of Helsinki and approval by the ethics committee of the Medical Faculty at the University of Duisburg-Essen (Essen, Germany) (BO-10-4380). For samples from PB of children aged 1.5–6 y, the study was approved by the Human Research Ethics Committee of the Medical Faculty at University of Gothenburg (Gothenburg, Sweden). Informed written consent was obtained from the parents of the children.

Human lymphocytes or hematopoietic precursor cells were isolated by Ficoll density centrifugation (density 1.077 g/ml; PAN-BioTech, Aidenbach, Germany) followed by selection of CD19-, CD3- or CD34-expressing cells by MACS (Miltenyi Biotec, Bergisch Gladbach, Germany).

Lymphocytes were analyzed on a CytoFLEX flow cytometer (Beckman Coulter, Krefeld, Germany) using the CytExpert or sort purified on a FACSAria III cell sorter (BD Biosciences, Heidelberg, Germany)/FACSAria Fusion cell sorter (BD Bioscience) equipped with BD FACSDiva software (BD Biosciences) after staining with fluorescence-conjugated Abs. If not indicated otherwise, mature CD5⁺ B cells were defined by the marker constellation CD5⁺CD23⁺CD27⁻CD38^lowIgD⁺, and CD5⁻ (naive) B cells were defined as CD5⁻CD23⁺CD27⁻CD38^lowIgD⁺ cells. Autologous T cells were isolated from CD19-MACS flow through by sort-purification of CD4⁺CD25⁻ lymphocytes. Intracellular IgM staining (IgM FITC, BD Biosciences) was performed after the indicated incubation periods using the Fixation/Permeabilization Kit (BD Biosciences) in the presence of GolgiStop and GolgiPlug (both BD Biosciences). BHLHE41 expression in human B cell subsets was assessed by a PrimeFlow RNA Assay (Thermo Fisher Scientific, Oberhausen, Germany) with a high-sensitivity Alexa Fluor 647 probe targeting human BHLHE41 normalized on a CD8A probe.

ELISPOT membranes (Mabtech, Macleod, Australia) were coated with anti-IgM Abs. B cells were sort purified and distributed at 5,000 cells in 100 µl of RPMI 1640 (PAN-BioTech) per well. After 16 h of incubation (37°C and 5% CO₂), cells were discarded and secreted IgM was visualized (Mabtech, Macleod). The developed spots were quantified by an ELISpot reader (ELISpot Reader System, Autoimmun Diagnostika, Strassberg, Germany).

Mature B cell subsets were cultured in RPMI 1640 medium with 20% FBS (PAN-BioTech), 100 U/ml penicillin and 100 μg/ml streptomycin at 37°C and 5% CO₂. TD stimulation was performed using 0.03 μg/μl anti-Ig (Jackson ImmunoResearch Laboratories, Newmarket, UK) and 1 μg/ml CD40-ligand-HA with 5 ng/ml anti-HA Abs (R&D Systems, Minneapolis, MN) and separately in combination with IL-4 and IL-21 (100 IU each). TI type 1 stimulation was mimicked by incubation with CpG OND Type B or R848 (InvivoGen, San Diego, CA). TI type 2 simulation was performed by 0.03 μg/μl anti-Ig treatment. For coculture experiments, B cells and autologous T cells were cocultivated in equal numbers (250,000 cells). T cells were left unstimulated or activated via anti-CD2/anti-CD3/anti-CD28 beads, according to the manufacturer’s instructions (Treg Suppression Inspector; Miltenyi Biotec).

Survival was determined by assessing the fraction of vital cells at defined time points during culture by flow cytometric exclusion of DAPI (Merck, Darmstadt, Germany) and allophycocyanin Annexin V–positive cells (BD Biosciences). Proliferation was assessed using isolated B cells that were pulsed with Cell Proliferation Dye eFluor 670 (5 μM; Invitrogen). The fraction of proliferated (eFluor 670^low) cells was assessed after 48 and 96 h of in vitro stimulation. The single peak formed by B cells after 48 and 96 h of unstimulated in vitro cultures, respectively, served as control to identify nonproliferated cells in the gating strategy (eFluor 670^high cells).

NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice (https://www.jax.org/strain/005557), were bred and housed at the University Hospital Essen animal care facility. All animal experiments were carried out following institutional guidelines and with protocols approved by the Animal Care Committee of the University Hospital Essen. All mice used in the experiments were kept single housed under pathogen-free conditions. In total, 24 NSG mice (12 male and 12 female, age 8–14 wk, weight 31 ± 5 g) were treated i.v. with 30 mg/kg busulfan (Busilvex; Pierre Fabre, Sankt Georgen, Germany) 24 h prior to cell transfer that were transplanted by injection into tail vein with ∼200,000 human CD34⁺ progenitor cells derived from UCB or adult PB. Between 2 and 3 mo after injection, blood serum, spleen, and bone marrow (BM) were harvested, and a peritoneal cavity (PerC) lavage was performed. Harvested tissues were homogenized, splenic erythrocytes were lysed, and mononuclear cells were extracted by density gradient centrifugation (Ficoll).

For each stimulation condition, 1 × 10⁶ CD3⁻CD10⁻ cells (anti-CD3–allophycocyanin [(BD Biosciences] and anti-CD10–FITC [BD Biosciences]) were sort purified and stimulated with either anti-Ig or CpG at 37°C and 5% CO₂ for 30 min and subsequently stained with anti-CD5–PE-Cy5 (BD Biosciences, Heidelberg, Germany). The stained cells were fixed with Fixation/Permeabilization Solution (BD Biosciences), followed by intracellular staining with anti-IgM FITC (BD Biosciences, Heidelberg, Germany) and Hoechst 33342 (Thermo Fisher Scientific). Confocal microscopy was performed with a Leica SP8 gated STED Super Resolution Microscope at the Imaging Center Essen (66).

IgM memory (CD5⁻CD23⁻CD27⁺CD38^lowIgM⁺IgD^low) and IgA or IgG class-switched memory (CD5⁻CD23⁻CD27⁺CD38^lowIgG⁺ or IgA⁺) B cells from PB of adult donors, and CD5⁻ (CD5⁻CD23⁺CD27⁻CD38^lowIgD⁺) and CD5⁺ (CD5⁺CD23⁺CD27⁻CD38^lowIgD⁺) B cells from UCB and adult PB, were isolated by cell sorting (purity > 99%) from six adult donors and 6 UCB. RNA was isolated by the RNeasy Micro Kit (Qiagen, Hilden, Germany). For the generation of RNA-sequencing (RNAseq) libraries, the NuGEN Trio RNA-Seq System (NuGEN, Redwood City, California) was used. Samples were split equally and processed in independent sequencing steps to allow for correction of batch effects. Sequencing was performed with paired-end sequencing and two times 100-bp length. Sequences were aligned with HiSAT2 version 2.1.0. (67) to the human genome hg38 and analyzed with DESeq2 (68) and R (www.R-project.org/). Hierarchical tree analysis was performed using GeneSpring GX 11 (Agilent Technologies, Waldbronn, Germany).

Real-time quantitative PCR (qRT-PCR) was performed using TaqMan probes (Applied Biosystems) targeting PRDM1 in 9 × 10⁴ UCB and adult mature B cell subsets per condition after stimulation for 48 h (unstimulated control, CpG, anti-Ig, anti-Ig/CD40L). LIN28B, LET-7B, and ARID3A-specific TaqMan probes (Applied Biosystems) were analyzed among >1 × 10⁶ human hematopoietic precursors. Total RNA was extracted using an RNeasy Micro Kit (QIAGEN), and cDNA synthesis was performed according to manufacturer’s instructions using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) for PRDM1, LIN28B, and ARID3A and the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems) for LET-7B. The qRT-PCR conditions consisted of 2 min at 50°C, 10 min at 95°C, and 45 cycles of 15 s at 55°C and 1 min at 60°C. Every experiment was performed three times, and the samples were tested in triplicate. The average of ΔCT values for the amplicon of interest was normalized to that of GAPDH. IgM constant-specific primer was 5′-GCCCTGCCCAACAGGGTCA-3′. IgM transcript splice variants for membrane expression or secretion were amplified with 5′-AGCAAAGCAGTGTGGGGTAGA-3′ and 5′-ACACACAGAGCGGCCAGC-3′ primers (all Integrated DNA Technologies, Leuven, Belgium), respectively. After agarose gel electrophoresis, products were visualized by GelRed (Biotium, Fremont, CA), and product intensity was measured by ImageJ software (69) and normalized to ACTIN amplicon products (5′-GACGACATGGAGAAAATCTG-3′ and 5′-ATGATCTGGGTCATCTTCTC-3′ [Sigma-Aldrich, St. Louis, MO]).

Genomic DNA was extracted from B cell populations sorted in duplicates (Gentra Puregene Core Kit; QIAGEN). Rearranged IGHV genes were sequenced by massive parallel sequencing. By PCR using IGHV1, -3, and -4 subgroup primers for IGHV framework region 2 and a mixture of all IGHJ primers (70), DNA was amplified to generate deep sequencing libraries on the MiSeq Illumina platform (Illumina, San Diego, CA). In the first two PCR cycles the template (rearranged IGVH gene) is copied only once, whereas 12 nucleotides unique molecular identifier are introduced into the amplicon by the first set of primers. Following removal of those primers, the constructs are amplified by a second set of primers that, in turn, tag the construct with sequences essential for hybridization on the flow cell. Sequencing was performed with custom primers and the rapid run program with paired-end sequencing and two times 300-bp length. Reads generated by MiSeq were included only when the average quality was ≥25. Ambiguities between forward and reverse reads were replaced by “N.” Sequences with identical unique molecular identifier were classified as PCR duplicates and reduced to the longest detected sequence. Each N-nucleotide in a given sequence was replaced by the most frequent (majority vote) nucleotide. Only in-frame sequences detected more than once were further processed. Differences in the amount of sequences between mature CD5⁺ and naive B cells were balanced by random drawing from the population with the higher amount of sequences. Sequences were considered clonally related when using the same IGHV gene and sharing 100% CDRIII nucleotide sequence identity. All statistical and bioinformatical evaluations were performed in R (www.R-project.org/) and based on the international ImMunoGeneTics information system database (www.imgt.org/). Mutation frequencies were calculated based on the number or relative position of nucleotide exchanges in the IGHV region of each sequence in comparison with the most similar allelic variant present in the respective donor (determined from unmutated sequences). IGHV identification was carried out with the Basic Local Alignment Search Tool. IGHD and IGHJ identification was performed by calculating the best letter-wise matching gene. For IGH gene usage, the fraction of a given IGH gene among total sequences (all subgroups) detected was calculated. Every nucleotide between and including the first and the last N-nucleotide between IGHV and IGHD or IGHD and IGHJ gene with highest germline identity was counted as N1 (V-D) or N2 (D-J) N-nucleotide, respectively. IGHV usage was determined by the R package iggeneusage (71).

Statistical parameters, including the description of each data point (n value), the number of mice or human samples per experiment, the number of replicates, the meaning of bars, and the statistical tests used are contained in the figure legends. Statistics were calculated using R (www.R-project.org/). Other R packages or software used are given in the related section. For comparisons between two groups, a two-sided paired Wilcoxon test was used. For comparison between more than two groups, statistical analysis was performed using one-way ANOVA with the post hoc Tukey test. In the figures, asterisks represent p values as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Human neonates were reported to have a high frequency of CD5⁺CD10⁺CD24⁺CD38^high transitional B cells with immature functionality (63, 65). To investigate whether human neonatal B cell responses are limited by mature B cell numbers, we determined the average fraction of mature and transitional B cells in a cohort of 43 umbilical UCB samples. Our analysis showed that, on average, the B cell proportions among neonatal and adult are similar (10% of mononuclear cells [range 3–18%] compared with 11% [range 4–20%] in PB of adults >20 y).

Although the fraction of transitional B cells (CD5⁺CD10⁺CD24^highCD38^high) is higher in UCB than in adult PB (median 22.1% versus 5.9%, respectively), most UCB B cells (64%, range 48–79%) show a mature naive phenotype (CD19⁺CD10⁻CD24⁺CD27⁻CD38^low IgM⁺IgD⁺) (Fig. 1A–D, Supplemental Fig. 1A, 1B). Our initial staining panel indicates that the B cell pool in UCB includes a major fraction of IgM expressing (likely foreign Ag-naive) mature B cells. To allow for comparison of age-related changes (and to assess similarities to murine B-1a and B-1b cells), we analyzed mature CD5⁻ (CD5⁻CD23⁺CD27⁻CD38^lowIgD⁺) and CD5⁺ (CD5⁺CD23⁺CD27⁻CD38^lowIgD⁺) B cells derived from human UCB and adult PB separately. Notably, the phenotype of UCB B cells persists until childhood (Fig. 1B, 1E). On average UCB mature B cells are composed of equal-sized fractions of mature CD5⁻ and CD5⁺ B cells (Fig. 1A, Supplemental Fig. 1C). Both subsets are also abundant (average 46%, 32–62% among B cells) in infants and young children [10 individuals below 24 mo tested (Fig. 1B)], albeit the CD5⁺ fraction decreases with age (Fig. 1E). UCB B cells express higher levels of surface IgM compared with adult naive B cells (Fig. 1A, 1C; Supplemental Fig. 1A).

We further quantified memory B cells (CD27⁺; (Fig. 1A–C), plasmablasts (CD27^highCD43⁺), and the recently proposed CD27⁺CD43⁺ human B-1 cells (Supplemental Fig. 1A, 1B, 1D) (54). Mature UCB B cells barely include memory B cells and plasmablasts or CD27⁺CD43⁺ B cells, the numbers of which increase with age until adulthood (Supplemental Fig. 1D), in line with previous publications (65, 72). We conclude that a major fraction of human umbilical UCB B cells is mature.

Although the majority of UCB B cells shows a mature phenotype, neonates and infants have an elevated susceptibility to infections compared with adults, which could be attributed to a functional impairment of the B cell system. Transcriptome-wide comparison is a reasonable approach to estimate functional differences between B cell subsets (62, 73). To this end, we generated RNAseq profiles of sort-purified (Fig. 1A, 1B, sorting gates marked in red) CD5⁻ (CD5⁻CD23⁺CD27⁻CD38^lowIgD⁺) and CD5⁺ (CD5⁺CD23⁺CD27⁻CD38^lowIgD⁺) mature B cells from UCB and adult naive CD5⁻ (CD5⁻CD23⁺CD27⁻CD38^lowIgD⁺) and CD5⁺ (CD5⁺CD23⁺CD27⁻CD38^lowIgD⁺) mature B cells from PB for comparison. Human adult IgM memory (CD5⁻CD23⁻CD27⁺CD38^lowIgM⁺IgD⁺) and IgA or IgG class-switched memory (CD5⁻CD23⁻CD27⁺CD38^lowIgG⁺ or IgA⁺) B cells from PB and tonsil served as analytical root for a robust determination of the relative similarity between adult and UCB B cell subsets.

Human UCB B cell transcriptomes are unique and deviate from adult B cell samples, according to a t-distributed stochastic neighbor embedding analysis of the top 10,000 expressed genes of each subset analyzed (Fig. 2A) and a hierarchical clustering of the 4400 most variable transcripts (Fig. 2B). CD5⁺ and CD5⁻ B cell subsets in adult and UCB differ in expression of 87 (out of 17,000) transcripts (Supplemental Fig. 2A, 2B). The biological relevance of these marginal differences is questionable.

To investigate the transcriptional differences between UCB and adult mature B cell subsets, we performed gene set enrichment analysis. Enriched signatures included multiple gene sets associated with mitogen responses, BCR, and CD40 signaling (Fig. 2C), suggesting that UCB B cells are transcriptionally prepared to respond to CpG, anti-Ig, and anti-Ig/CD40L stimulation, respectively. Moreover, the TGF-β signaling pathway as a prerequisite for IgA switching (74) and multiple gene sets associated with TNFR, IL, and chemokine signaling suggested that UCB B cells have particular response capabilities in both TI and TD immunity (Fig. 2C). A supervised analysis of central molecules in BCR signaling, B cell activation, or T cell–B cell interaction, showing up to 4-fold changes in expression, underlines the transcriptional distinctness of UCB from adult B cell subsets (Fig. 2D). In some instances, selected pathway components are not in line with the overall pathway enrichment (e.g., despite the BCR signaling pathway being overall enriched among UCB B cells, LYN and PLCG2 are increased and SYK and BTK, and BLNK decreased in comparison with adult B cells).

Arid3a and Bhlhe41 expression are characteristic for murine B-1 cells (75–77). However, the RNAseq profiles of UCB B cells did not show a consistent enrichment of typical murine B-1 cell expression patterns (Supplemental Fig. 2C, 2D) (42–44).

We conclude that human UCB B cells show unique transcriptome patterns, suggesting efficient responsiveness to CpG, anti-Ig, and anti-Ig/CD40L stimulation but barely share signatures with murine B-1 cells.

Our transcriptome analysis supports the idea of many differentially expressed genes with high impact on B cell responsiveness. To validate differential expression, which does not directly translate to protein expression levels (78, 79), we selected 29 surface molecules and performed flow cytometric analysis on a total of 30 UCB and 21 adult PB samples.

Flow cytometric analysis did reflected neither the transcriptional distinctness of resting mature UCB B cells nor support their previously reported lower levels in T cell–B cell-interaction molecules or CD21 expression (Fig. 3A, and compare, for example, CD79b, ICOSLG, CD86, CD21, and IL-4R in (Fig. 2D). In contrast, we observed mild phenotypical differences between resting adult and UCB subsets, including significantly different surface expression of PTPRJ, CR2/CD21, CD79b, FCGR2A, IFNGR1 and FCER2 (Fig. 3A, Supplemental Fig. 3). Typical murine B-1 cell markers (80) (CD11b, CD43, or a low level of IgD, CD21, or CD23) were not observed (Fig. 3, Supplemental Fig. 3).

We next investigated the responsiveness of CD19⁺ UCB and adult naive B cells in vitro, stimulated with CpG, anti-Ig, and anti-Ig/CD40L, and quantified the mean fluorescence intensity (MFI) for 29 markers on mature CD5⁺ and CD5⁻ B cell subsets (Supplemental Fig. 3B). Compared with steady state, we observed changes in surface expression upon stimulation, unidirectional (e.g., CD69, CD86, FCGR2A, CD21), as well as differential (e.g., ITGAM, ICOSLG, OX40L, FCER2, IFNGR, IL4R) between UCB and adult B cells. The type of stimulus often had a strong impact on response intensity, with a few exceptions (FCER2, IFNAR2) (Fig. 3B, Supplemental Fig. 3). Note that in this analysis CD5⁺ and CD5⁻ B cells were not sort purified; thus, CD5 up- or downregulation during culture cannot be excluded, but CD5⁺ and CD5⁻ B cells mostly responded in a similar manner. We conclude that, with few exceptions, UCB and adult mature B cells are phenotypically similar in the resting state but have different response potentials.

As human UCB B cells quickly adopt an activated phenotype upon stimulation, we analyzed UCB and adult B cell response kinetics in more detail and observed further qualitative and quantitative differences. UCB B cells proliferated earlier, as the first cell division was already detectable on day 2 upon CpG or anti-Ig/CD40L stimulation, whereas adult B cells proliferated at the earliest on day 4 (Fig. 4A). Anti-Ig stimulation showed weak induction of proliferation in general, in contrast to anti-Ig/CD40L stimulation (Fig. 4B), particularly in combination with IL-4 and IL-21 (Supplemental Fig. 4A). In contrast, adult B cell division was more delayed and barely detectable upon TI stimulation (Fig. 4A, 4B). Unstimulated B cells did not proliferate within 4 d of culture. With regard to the known differences in neonatal and adult T cell composition (2, 4), we performed coculture assays of mature B cells and autologous Th cells (CD4⁺CD25⁻) stimulated with anti-CD2/anti-CD3/anti-CD28 beads. The coculture assays confirmed an anti-Ig/CD40L proliferation potential among both UCB and adult B cells, although UCB B cell division was slower than that by stimulation with selected ligands (Fig. 4B).

We investigated Ig secretion potentials by ELISPOT and showed that UCB B cells are capable of secreting Ig earlier than adult B cells (Fig. 4C). Please note that the ELISPOT membrane covered with IgM-capture Abs likely generates BCR-stimulating signals during the incubation time. Earlier Ig secretion was in line with an increased IgM production rate (Fig. 4D) and increased IgM secretion (Fig. 4E), which was already seen in unstimulated UCB B cells, but was particularly increased under stimulation with CpG. These findings were supported by accelerated ASC differentiation as determined by FSC/SSC^highCD27^highCD38^high phenotype (Fig. 4F, 4G), PRDM1 induction (Supplemental Fig. 4B), ratio of secreted versus membrane IgM transcripts (Supplemental Fig. 4C), and cell polarization (“IgM capping”) after 30 min of stimulation (Supplemental Fig. 4H) in UCB versus adult B cells. In the first 4 d, UCB B cells homogenously showed ASC differentiation fate, independent of in vitro stimulation conditions (Fig. 4C, 4E, 4F), whereas only a fraction of adult B cells differentiated. This result was consistent with autologous T cell coculture (Fig. 4B, 4D, 4F). Neither in vitro stimulation nor T cell coculture induced preferential survival among UCB over adult B cells but rather increased apoptosis over time (Supplemental Fig. 4F, 4G), suggesting a transient nature of UCB B cell responses.

Finally, in line with the efficient upregulation of TGFBR1 (Fig. 3B), we observed enhanced IgA class switching among UCB but not adult B cells treated with anti-Ig and TGFβ already at day 3 (Fig. 4G, 4H; Supplemental Fig. 4D). In contrast, class switching to IgG was more efficient among adult naive B cells yet delayed until day 5 (Supplemental Fig. 4D).

Our data suggest that UCB B cells are not impaired but show rapid TI and TD responsiveness (mimicked in vitro), whereas adult B cell proliferation is mostly confined to anti-Ig/CD40L stimulation. Second, ASC differentiation is the predominant fate of UCB B cells. These accelerated and likely transient dynamics may cause the quickly waning Ig titers and hyperresponsiveness in vivo. The quick and preferential response to TI stimulation, efficient IgM secretion, and enhanced IgA class switching are also central to murine B-1 cells (57, 80, 81).

Human UCB B cells respond fast and efficiently to various types of stimulation; however, their responses may be hampered by reduced Ig repertoire diversity or skewed IGHV gene usage. We performed BCR repertoire deep sequencing of mature CD5⁺ and CD5⁻ B cells from four UCB and three adult PB samples. We sequenced rearranged IGHV genes of at least 100,000 B cells per population, split into two biological replicates, and used unique molecular identifiers to distinguish clonal expansions from Ig sequence amplification during library generation. The vast majority of BCR rearrangements were unique and unmutated, demonstrating that the UCB and adult naive BCR repertoires are similarly variable. Clonal expansions were rare but more frequent in UCB than adult PB samples (Fig. 5A). The UCB-derived clones were larger, with up to 40 members and an average of four members per clone (Fig. 5B). The clonal expansions among both UCB and adult B cells were devoid of Ig mutations, indicating GC-independent generation. The comparison of CD5⁺ and CD5⁻ B cell populations within an individual UCB or adult sample revealed that both subsets were highly related. In any given sample, the majority of clonal expansions were distributed between both subsets. Whereas clones derived from CD5⁺ B cells alone were frequent, particularly in UCB, clones that consisted only of CD5⁻ B cell–derived sequences were barely detectable (Fig. 5C, 5D), indicating that CD5 is not stably expressed on a distinct subset.

We compared the UCB BCR repertoire on an interindividual level and therefore analyzed the distribution of clonotypes [i.e., BCR rearrangements with identical IGHV gene and 80–100% identical amino acid CDRIII sequence between individuals (Fig. 5E)]. At 90% or more CDRIII identity, UCB samples (n = 4) included up to 8% (Ig unmutated) clonotypes among all BCR rearrangements, but adult naive B cells (n = 3) showed none (Fig. 5F). The selection of conserved specificities is not confined to neonatal B lymphopoiesis but is also evident in adult CD5⁺ B cells, as these are significantly enriched for similar BCR specificities. Clones and clonotypes barely intersected.

Finally, with regard to murine B-1 cells, the Ig repertoire of human UCB B cells did not show an enrichment of proximal versus distal IgV genes or major biases in IgV gene usage. However, we confirmed a significantly reduced amount of N-nucleotide insertions at D_H-J_H-junctions among UCB B cells (Fig. 5G), in line with previous publications (82).

We conclude that the human UCB mature B cell repertoire is selected to a large extent by inherent mechanisms that may be similar but not identical to rodents.

The functional differences between mature B cell subsets from UCB and adult PB raise the question whether these are simply instructed by, for example, microenvironmental stimulation, or do UCB B cells derive from a developmentally distinct lineage, as in mice. We analyzed the expression of LET7-B, LIN28B, and ARID3A by qRT-PCR among lymphoid precursor cells (CD34⁺) isolated from UCB and adult PB and observed significant differences, particularly the absence of LIN28B expression in adult precursors (Fig. 6A), as noted previously (75). To determine whether these transcription patterns correlate with their lymphopoietic potential, we generated humanized mice by injecting UCB (n = 16) or adult PB-derived hematopoietic precursor cells (n = 8) into NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice. UCB precursors showed a higher engraftment capability compared with adult-derived cells in the BM microenvironment (Fig. 6B). However, the composition of the B cell compartment differed, as UCB precursors gave rise to a higher frequency of mature CD5⁺ versus CD5⁻ B cells (Fig. 6C, 6D), as it is also observed in human UCB and infant PB (Fig. 1E, Supplemental Fig. 1D). Moreover, when normalizing for engraftment efficiency, all mice showed comparable reconstitution of mature B cells in the spleen, but only mice injected with UCB precursors showed mature (CD5⁺) B cells in the peritoneal cavity (Fig. 6E, 6F).

Human UCB and adult PB-derived hematopoietic precursor cells differ in their expression of key molecules regulating murine B-1 lineage development, and they provide different engraftment, B cell composition, and localization in humanized NSG mice.

Infants suffer from enhanced susceptibility to infections and limited responsiveness to vaccination. Short-lived and quickly waning titers of neutralizing Ab represent a critical factor, particularly during the first months of life. Multiple reasons for this have been described, including immaturity of immune cells or lymphoid tissues. In contrast to previous reports (1–5), our analysis revealed the presence of a large and mature B cell population in UCB that responds fast and efficiently in vitro. We assume that this population of mature B cells (up to 60% of total UCB and infant PB B cells) is the major source for neonatal and infant B cell responses. Recently, it was shown that the neonatal immune system changes rapidly in the first weeks after birth (83), which is likely driven by the extensive host-microbe interactions and potentially also affects the B cell pool composition. Our data support this view, as UCB B cells show a strong plasmablast differentiation capacity, thereby favoring changes in the composition of the neonatal immune system in response to a novel environment. However, we show in this study that the fraction of mature cells is already high at birth and remains largely stable throughout life. We assume that the transition from the neonatal to an adult B cell system occurs during early childhood, which is in line with the observed gradual decrease of CD5⁺ B cell numbers in children and reflected by the increasing responsiveness and memory formation to vaccination.

Our study aimed at a comprehensive characterization of UCB B cells to reveal putative differences to adult naive B cells by phenotypical, molecular, and functional analysis. UCB B cells show a mature phenotype but express higher levels of IgM and CD38, which may also contribute to their fast responsiveness, even superior to that of memory B cells (73, 84, 85). Of note, the majority of surface molecules tested did not show significant differences between UCB and adult B cells, contrasting with previous studies that did not separate B cell subsets, in particular transitional from mature B cells (12–16). On a transcriptional level, a major distinctness can be observed, spanning thousands of differentially expressed transcripts and gene sets, confirming previous observations that genotype and phenotype can deviate substantially (78, 79). Our data suggest that UCB B cells are similarly resting B lymphocytes as their adult counterparts in steady state. However, their distinct transcriptome profiles convey their unique differentiation capacities and distinct response dynamics in comparison with adult B cells, particularly their responsiveness to CpG or anti-Ig versus anti-Ig/CD40L stimulation. Indeed, functional analyses revealed that UCB B cells are highly sensitive to CpG, anti-Ig, and anti-Ig/CD40L stimulation, whereas adult naive B cells required T cell help for best performance. The responsiveness to CpG is in line with the previously reported elevated TLR9 expression in cord blood B cells (17). The reported reduction in class switching (17, 18) is putatively limited by cytokine levels in neonates (86) but is not intrinsic to UCB B cells, as these showed fast and efficient class switch recombination, in particular to IgA. We suggest that the high propensity to IgA (over IgG) class switching facilitates the quick generation of mucosal immunity in neonates. This again reflects that, already at birth, mature B cells exist in humans that actively and quickly help the immune system to adapt to the environmental challenges.

The human UCB BCR repertoire is unmutated but highly variable, and in this regard, no major differences to the adult naive BCR repertoire could be observed. However, clonal expansions lacking intraclonal diversification are detectable, indicating that early-in-life selection processes exist that do not require GC diversification. Such clonal expansions were also observed among adult B cells and then always related to the mature CD5⁺ B cell subset, showing autoimmune specificities (87, 88), associated with autoimmune disorders (89, 90) and chronic lymphocytic leukemia pathogenesis (62). It was previously reported that neonates show few similar Ig rearrangements (clonotypes) shared by three neonates (29). Our dataset differs in study design and data analysis from Soto et al. (29), in particular as we sort purified neonatal mature and adult naive CD5⁺ and CD5⁻ B cells, thereby excluding transitional and memory B cell subsets. We show in this study that up to 8% of UCB B cells retrieved from four neonates express clonotypic VDJ rearrangements. It is tempting to speculate on selection of autoantigenic specificities during neonatal B lymphopoiesis, similar to mouse models (91) and potentially contributing to removal of cell debris in the periphery of human neonates (57). Presumably, conserved clonotypes in the UCB B cell repertoire cause the consistent presence of IgM autoantibodies in the serum of neonates (92). The rarity of UCB clonotypes among adult naive and also memory B cells (66) raises the question whether these rare clones are long-lived or adult-derived, and whether de novo–generated BCR specificities instruct a neonatal B cell fate (46). We propose that the UCB mature B cell repertoire is substantially selected by inherent, evolutionary conserved mechanisms, likely to a large extent including selective autoimmunity as detectable in serum Ab specificities (92).

Our study shows that these specifically selected mature UCB B cells are highly sensitive to stimulation and quickly differentiate into ASCs. They respond efficiently to TI stimulation, thereby possibly avoiding the time-consuming search for cognate T cells, GC-dependent maturation, and efficient generation of memory B cells. Considering the immature infrastructure of lymphoid tissues (33, 34, 93), the rarity of PC survival niches in the BM (35, 36), and the deprivation of survival signals (38), it is easily anticipated why the rapid UCB B cell responses are transient and wane quickly (3, 5). Finally, the differentiation into likely short-lived ASCs (in contrast to adult naive B cells, UCB B cells underwent apoptosis after several cell divisions upon stimulation in vitro), is a further reason for the observed vulnerability of neonates to infection. Based on the highly diverse BCR repertoire among UCB B cells, neonates are likely prepared to encounter multiple different pathogens, therefore granting a low and broad level of primary B cell immunity. However, when the pathogen load is overwhelming by excess or repeated infections, reactive UCB B cells are quickly depleted, running the risk to fail in protection and even develop hyperresponsiveness (2, 3, 5).

The reasons why UCB B cells behave so distinctly from their adult counterpart are difficult to address, given the limitations of studies on humans. It could be that the distinctness is simply imposed (e.g., by a unique microenvironment in neonates and to some extent also in infants). However, it has been shown that human hematopoietic progenitor cells from UCB exhibit higher differentiation potential and repopulation capacity compared with adult precursors (94), and it was suggested that human UCB progenitor subsets diverge in B cell lineage development potential from that of adults (95–97). Previous studies showed that NSG mice reconstituted with human adult or UCB CD34⁺ precursors give rise to mature CD5⁺ B cells in the spleen (98, 99). We made use of such humanized mice to evaluate differences in engraftment efficiency and reconstitution potential of the B cell compartment. Our data suggest that under similar microenvironmental conditions, B cell development from UCB precursors is qualitatively (and quantitatively) different from that of adult precursors. In line with the characteristic expression pattern in murine B-1 cell development, human UCB, but not adult precursors express LIN28b (75), and we confirmed that the downstream expression of the inhibitory microRNA LET-7b shifted significantly in favor of the B-1 lineage-associated transcription factor ARID3A. We suggest that, in addition to the developmental program, the conserved (autoimmune) selection of the BCR repertoire may also be a molecular reason for this. Thus, our study supports the view that the human B cell system is layered as in mice, in which B cell lineages with intrinsic differences develop in separate pathways.

Although many aspects of our knowledge on infant B cell immunity is based on mouse models, the human B-1 lineage counterpart is a long-standing matter of controversy (23, 24, 52–60). In particular, a small subset of Ig mutated CD20⁺CD27⁺CD43⁺CD70⁻ B cells has been proposed as a human B-1 cell counterpart (54). However, this finding is controversially discussed (55, 56), and we show in this study that most UCB B cells express neither CD43 nor CD27 and do not carry mutated Ig genes. We compared our findings on human UCB B cells to the murine B-1 lineage. We observed similarities in function (e.g., the fast and high responsiveness to TI stimulation and the fast and efficient IgM secretion of UCB B cells argue for overlapping functions). As we observed IgM secretion among UCB B cells (and to a lower extent also among mature CD5⁺ B cells from adults) already under unstimulated conditions, this may argue for a natural IgM secretion potential. Note, however, that sort-purified and in vitro–cultured human B cells are not entirely “unstimulated cells.” Moreover, the murine B-1 cell–typical transcriptional patterns were barely detectable in humans, and the UCB BCR repertoire did not show the biased Ig gene usage observed in mice. Thus, human neonatal B cells can only partially be explained by murine B-1 lineage characteristics.

We conclude that human UCB and adult naive B cells are phenotypically similar but differ markedly in their response potentials. This major distinction exists in mice and humans, suggesting that a fast and independent IgM and IgA Ab response early in life proved beneficial in evolution. However, the molecular patterns underlying these neonatal B cell responses vary between species. Our study contrasts with many aspects of the current view on human neonatal and infant B cell responses. This has important implications not only for our understanding of infant vulnerability to microbes but also may explain why certain vaccination strategies cause hyperresponsiveness rather than long-lasting protection. Consequently, systematic vaccination strategies encompassing the neonate’s environment (elder siblings and insufficiently immunized adults) are mandatory to generate herd immunity. Moreover, passive immunization and treatment strategies to sustain ASC survival should be taken into research focus.

We thank Julia Jesdinsky-Elsenbruch for excellent technical assistance and Klaus Lennartz for valuable engineering support. We thank Ludger Klein-Hitpass and the staff members at the Biochip-laboratory at the Institute of Cell Biology in Essen. We thank the staff members at the flow cytometry and fluorescence core facility Imaging Center Essen.

The authors have no financial conflicts of interest.