T cell–independent development and induction of somatic hypermutation in human IgM⁺IgD⁺CD27⁺ B cells

IgM⁺IgD⁺CD27⁺ B cells can be detected in neonatal cord blood, suggesting that these cells developed before birth. To understand when and where these cells developed during fetal life, we performed an extensive flow cytometry analysis for IgD⁺CD27⁺ B cells in the human fetus. The fetal tissue used in these experiments ranged from week 14 to 18 of gestation. Adult spleen and neonatal cord blood were used as a positive control. IgD⁺CD27⁺ B cells were found in fetal cord blood, liver, MLNs, spleen, and BM (Fig. 1 A). Additional B cell stainings for surface IgD and CD27 expression in fetal tissue are shown in Fig. S1. The percentages of the IgD⁺CD27⁺ B cells in the different organs were on the average between 3 and 4.8% of IgD⁺ B cells (Fig. 1 B). We did not find significant differences between the different fetal organs analyzed. In addition, we did not observe significant differences in the frequencies of IgD⁺CD27⁺ B cells between fetuses from gestational week 14 to 18 (Fig. S2). The IgD⁻CD27⁺ B cells found in the fetal liver, MLNs, and BM were pro–/pre–B cells based on the expression of CD34 and the lack of BCR expression (Fig. S3). This confirms the findings of Vaskova et al. (25) who described transient expression of CD27 on early B cells that lack light chain rearrangements in human BM.

We next determined whether the observed phenotype of fetal IgD⁺CD27⁺ B cells corresponded to the phenotype of MZ B cells using flow cytometry analysis of different fetal organs and fetal cord blood (Fig. 1 C). In neonatal cord blood, the IgD⁺CD27⁺ B cells are sIgM⁺CD21^highCD1c^high. In fetal cord blood and MLNs, the expression of CD21 was comparable to that in control neonatal cord blood. However, expression of CD21 was lower in the fetal liver, spleen, and BM. CD1c expression in neonatal cord blood was similar to that of fetal cord blood, MLNs, and spleen. In fetal liver and BM, IgD⁺CD27⁺ B cells expressed low levels of CD1c. This suggests that the fetal spleen, liver, and BM contain IgD⁺CD27⁺ B cells with a more immature phenotype as compared with fetal cord blood and MLNs. The IgD⁺CD27⁺ B cell population in the fetal spleen appears to be in between these two stages. Collectively, our data indicate that the phenotype of IgD⁺CD27⁺ B cells in fetal cord blood and fetal MLNs is similar to that in neonatal cord blood. Although sIgM, CD21, and CD1c are expressed on IgD⁺CD27⁺ B cells in both the fetal and adult spleen, the expression levels are different (Fig. 1 C). The biological significance of the difference in protein expression is presently unknown.

A proportion of the CD19⁺IgD⁺CD27⁺ B cells derived from adult and neonatal cord blood contains AID-driven SHMs (1, 13). We examined whether AID was expressed in the different fetal organs. AID expression was detected by RT-PCR in mRNA from the fetal liver and MLNs but not from the spleen and BM (Fig. 1 D). As expected, AID was detected in tonsil mRNA. We extended these findings by performing a semiquantitative RT-PCR for AID on sorted IgD⁺ B cells from the fetal liver, spleen, and BM (Fig. S4). Both fetal spleen and fetal BM mature B cells were negative for AID expression, whereas fetal liver B cells were positive, suggesting that the fetal liver and MLNs are the organs where SHMs take place.

The presence of AID in the fetal liver and MLNs may imply that the Ig-encoding genes in CD19⁺IgD⁺CD27⁺ cells contain SHMs. Therefore, CD19⁺IgD⁺CD27⁺ and CD19⁺IgD⁺CD27⁻ cells derived from the fetal spleen were sorted, mRNA was isolated, and an RT-PCR specific for V_H3-Cμ was performed. Products were cloned into a vector and sequenced. Approximately 20% of the sequences contained SHMs (Table I). Collectively, these data show that SHMs are present on fetal IgD⁺CD27⁺ B cells.

To determine whether development of IgD⁺IgM⁺CD27⁺ B cells and induction of SHMs are dependent or independent of T cells, we used an in vivo model for human lymphocyte development. We and others have set up a human mouse model in which development of the human immune system (HIS) can be studied (26, 27). We used alymphoid Rag2^−/− γc^−/− mice (28) to monitor the development of human IgD⁺IgM⁺CD27⁺ B cells. To this end, we injected sorted CD34⁺CD38^–lineage^– fetal liver HSCs into sublethally irradiated newborn BALB/c Rag2^−/− γc^−/− mice. After 6–8 wk, the resulting HIS (BALB-Rag/γ) mice showed a robust reconstitution by human cells, including T cells, B cells, and plasmacytoid DCs in the different lymphoid organs (27, 29). The level of human reconstitution, as indicated by the frequency of huCD45⁺ cells, ranged from 30 to 90% in peripheral blood of HIS (BALB-Rag/γ) mice, with a large majority of human CD19⁺ B cells (Fig. S5, A and B). Most reconstituted Rag2^−/− γc^−/− mice (>10% huCD45⁺) contained both human IgM and IgG in their serum (Fig. S5 C), with titers similar to previously published observations (26, 27). IgD⁺CD27⁺ B cells could be detected in the spleen and the blood of HIS (BALB-Rag/γ) mice (Fig. 2 A). A more detailed phenotype of splenic IgD⁺CD27⁺ B cells showed that they were sIgM⁺CD21^highCD1c⁺ (Fig. 2 B), i.e., they were similar to MZ B cells found in the MZ and peripheral blood of human beings. The frequency of CD27⁺ cells among IgD⁺CD19⁺ B cells was around 2.5 and 1% in the spleen and blood, respectively (Fig. 2 C). The percentages found in the reconstituted mice are similar to the percentages found in young children (13).

A detailed analysis of the anatomical structure of the spleen of reconstituted Rag2^−/− γc^−/− mice (n = 12) failed to reveal MZ structures (not depicted) despite the presence of IgM⁺IgD⁺CD27⁺ B cells in the circulation and peripheral organs. This suggests that there is a disconnection between the anatomical MZ structure in the spleen and the presence of IgD⁺CD27⁺ B cells. Similarly, in children under the age of 2 yr, IgM⁺IgD⁺CD27⁺ B cells can be found in the peripheral blood despite the absence of anatomical MZ structure (13, 16).

To determine whether the IgD⁺IgM⁺CD27⁺ B cells, which developed in the repopulated HIS (BALB-Rag/γ) mice, also had undergone SHMs we sequenced V_H3-Cμ heavy chain regions derived from these cells. Splenic CD19⁺IgD⁺CD27⁺ and CD19⁺IgD⁺CD27^– cells were sorted from three independent HIS (BALB-Rag/γ) mice, mRNA was isolated, and an RT-PCR specific for V_H3-Cμ was performed. Sequencing of the cloned PCR products revealed that approximately half of the sequences from the CD19⁺IgD⁺CD27⁺ cells contained SHMs (Table II). These data demonstrate that the human IgD⁺CD27⁺ B cells in the HIS (BALB-Rag/γ) mice have undergone SHMs.

To obtain further evidence that IgD⁺CD27⁺ B cells that developed in the HIS (BALB-Rag/γ) mice were MZ-like B cells, we made use of a genetic approach. In normal mice, MZ B cell development is dependent on NOTCH2, as NOTCH2^−/− mice lack this B cell subset (30). Therefore, we investigated whether NOTCH2 was required for human IgD⁺CD27⁺ B cell development in HIS (BALB-Rag/γ). We made use of a lentiviral vector to inhibit NOTCH2 expression through short hairpin RNA (shRNA) interference. The NOTCH2-specific shRNA lentiviral vector induced a 60% reduction of NOTCH2 mRNA levels as determined by RT-PCR in the B cell line Raji (Fig. S6). We compared the effects of the NOTCH2-specific shRNA lentiviral-transduced cells with the control shRNA lentiviral-transduced cells within the same HIS (BALB-Rag/γ) mouse. We used two different fluorescent markers, YFP and GFP, to distinguish between the two vectors. Human FL CD34⁺CD38⁻ HSCs were transduced with either a YFP control vector or with a GFP vector expressing the shRNA cassette to knock down NOTCH2. After lentiviral transduction, control and NOTCH2 knockdown-transduced cells were mixed and injected in neonatal Rag2^−/− γc^−/− mice. 8 wk later, repopulated mice were killed and the percentage of CD19⁺IgD⁺CD27⁺ cells was determined in the GFP and the YFP population of the spleen. The ratio of YFP versus GFP was equal 8 wk after reconstitution in the HIS (BALB-Rag/γ) mice as compared with the original inoculum, demonstrating that there was no selection due the shRNA-mediated knockdown of NOTCH2. The shRNA-mediated knockdown of NOTCH2 resulted in a 50% reduction of CD19⁺IgD⁺CD27⁺ cells as compared with the control-transduced cells (Fig. 3 A) and the nontransduced cells (not depicted). Importantly the shRNA-mediated NOTCH2 knockdown did not result in decreased frequency of other immune cell subsets, e.g., CD3⁺ T cells (Fig. 3 B), showing the specificity of the developmental block. These data strongly suggest that in humans, NOTCH2 is important for the development of IgM⁺IgD⁺CD27⁺ B cells.

To determine whether development of IgD⁺CD27⁺ B cells is strictly dependent on the presence of T cells, we made use of Rag2^−/− γc^−/− mice on a Nude background. These mice have no functional thymus, and therefore T cell development cannot take place (31, 32). HIS (BALB-nude-Rag/γ) mice showed a similar level of human engraftment in peripheral blood than control HIS (BALB-Rag/γ) mice (Fig. S5, A and B). In control HIS (BALB-Rag/γ) mice, a functional mouse thymus was present and fully supported human T cell development, leading to colonization of peripheral lymphoid organs by mature human T cells (Fig. 4 A). In contrast, no T cells could de detected in the different lymphoid organs of HIS (BALB-nude-Rag/γ) mice, indicating that no thymus-independent T cell development occurs in this humanized mouse model. Nonetheless, IgD⁺CD27⁺ B cells were present in the blood and spleen of HIS (BALB-nude-Rag/γ) mice (Fig. 4 B). A more detailed phenotypic analysis of the splenic IgD⁺CD27⁺ B cells revealed that the cells were sIgM⁺CD21^highCD1c⁺ (Fig. 4 C). The percentages of IgD⁺CD27⁺ B cells in the spleen and blood were similar between HIS (BALB-nude-Rag/γ) mice (Fig. 4 D) and control HIS (BALB-Rag/γ) mice (Fig. 2 C). From these data, we conclude that the development of human IgD⁺CD27⁺ B cells can occur in a T cell–independent manner in vivo.

To determine the impact of the absence of T cells on the induction of SHMs, we determined the frequency of clones carrying SHMs in the VH3-Cμ region in three repopulated HIS (BALB-nude-Rag/γ) mice. Sequence analysis of the RT-PCR product showed that ∼25% of these cells had undergone SHMs as compared with the 50% of IgD⁺CD27⁺ B cells derived from the control HIS (BALB-Rag/γ) mice (Table III). Importantly, the IgD⁺CD27⁺ B cells from the HIS (BALB-nude-Rag/γ) mice that had undergone SHMs had similar numbers of mutations, indicating that the induction of SHMs can occur in a T cell–independent manner.

We report here that postrearrangement Ig diversification occurs in utero in human B cells via induction of SHM. We describe that in the human fetus during weeks 14–18 of gestation, IgM⁺IgD⁺CD27⁺ B cells that have undergone SHM are present. AID expression, which is a prerequisite for the induction of SHM, could be detected only in the fetal liver and MLNs, suggesting that the induction of SHM occurs in these organs. In both organs, IgM⁺IgD⁺CD27⁺ B cells are detected that are CD21^high and CD1c^high. One could speculate that these are more mature IgM⁺IgD⁺CD27⁺ B cells as compared with fetal BM IgM⁺IgD⁺CD27⁺ B cells, which express low levels of surface CD21 and/or CD1c. However, there are obvious limitations when working with human material, making it difficult to address the maturation status of human IgM⁺IgD⁺CD27⁺ B cells. To address the question of whether development and induction of SHM of IgM⁺IgD⁺CD27⁺ B cells is a T cell–independent process, we made use of a recently developed in vivo–humanized mouse model. We observed IgM⁺IgD⁺CD27⁺CD21^highCD1c^high B cells in such HIS (BALB-Rag/γ) mice. These cells contained SHMs and strongly resembled human MZ B cells phenotypically.

It is well documented that there are differences between rodent and human MZ B cells because human MZ B cells recirculate and contain SHMs, unlike rodent MZ B cells (5, 6, 13–15). Moreover, on an anatomical level, the spleens of humans and rodents are slightly different (7). In contrast to mice, humans have an inner and outer MZ surrounded by a perifollicular zone, which can be directly irrigated by blood vessels. However, such MZ structures were not found in HIS (BALB-Rag/γ) mice (not depicted). This observation is in line with findings that in children under the age of 2 yr, no CD27⁺ MZ B cells are present in histological sections of the spleen, despite the fact that IgM⁺IgD⁺CD27⁺ B cells can already be found in the blood of newborns (15) and, as shown here, in the fetus. This clearly indicates that the development of IgM⁺IgD⁺CD27⁺ B cells does not depend on the presence of anatomical MZ structure.

In mice it is established that NOTCH2 is essential for MZ B cell development, as NOTCH2 knockout mice do not have MZ B cells (30). Combining the in vivo human–mouse model with shRNA-mediated knockdown for NOTCH2, we established that development of IgM⁺IgD⁺CD27⁺ B cells in HIS (BALB-Rag/γ) mice is NOTCH2 dependent.

Using Rag2^−/− γ_C^−/− mice on a Nude background, in which no T cell development is possible due to the lack of a functional thymus, we demonstrated that development and induction of SHMs of human IgM⁺IgD⁺CD27⁺ B cells is a T cell–independent process. However, the frequency of IgM⁺IgD⁺CD27⁺ B cells that had undergone SHM decreased from 40–67% in control HIS (BALB-Rag/γ) mice to 22–26% in HIS (BALB-nude-Rag/γ) mice. Importantly, IgM⁺IgD⁺CD27⁺ B cells in HIS (BALB-nude-Rag/γ) mice that had undergone SHM showed a similar level of mutation frequency per cell as compared with control HIS (BALB-Rag/γ) mice. Collectively, this indicates that T cells are probably involved in increasing the numbers of these cells, but they are not essential to induce SHM in IgM⁺IgD⁺CD27⁺ B cells.

The fact that IgM⁺IgD⁺CD27⁺ B cells develop during fetal life raises the question whether IgM⁺IgD⁺CD27⁺ B cells still develop and undergo induction of SHM during adult life. An answer to this question might come from adult patients with rheumatoid arthritis, who have been depleted of B cells using an anti-CD20 antibody (33). Even 2 yr after the anti-CD20 treatment, the percentages of the IgD⁺CD27⁺ B cells in these patients are still severely reduced as compared with healthy controls, suggesting that the development of IgD⁺CD27⁺ B cells is not efficient in adults. On the other hand, numbers of CD27⁺ class-switched B cells are also maintained at low levels after anti-CD20 treatment, suggesting a general developmental defect in CD27⁺ memory B cells not necessarily restricted to the IgD⁺CD27⁺ B cell population.

We hypothesize that the development and the induction of SHM in IgD⁺CD27⁺ B cells occurs mainly during fetal development and in young children. It has been shown that upon vaccination, IgD⁺CD27⁺ B cells from adults can undergo additional SHM (5). It will be of interest to determine whether the mechanism of vaccination-induced SHM is different from the mechanism involved in utero.

Postrearrangement Ig diversification has been reported in several species including birds and mammals, resulting in B cells that are able to recognize T-independent antigens and mount a high affinity immune response (34, 35). This process takes place in utero and early in life in the gut-associated lymphoid tissue by SHM and/or gene conversion. Our data indicate that a similar process occurs in humans via induction of SHM during fetal life. This process most likely takes place in the fetal liver and fetal MLNs, as AID expression could be detected in these organs. The MLN is considered to be part of the gut-associated lymphoid tissue (36). In this respect, it is interesting to note that in mice with a highly restricted BCR repertoire, SHM can be detected in B cells with an immature phenotype. In the absence of T cells a similar mutation frequency was observed (37). These data are in line with our findings that induction of SHM can occur in a T cell–independent manner; however, in this mouse model SHM only occurred when the BCR repertoire was severely restricted.

Igs that recognize autoantigen are removed from the repertoire at two checkpoints during B cell development in the BM and the periphery (38, 39). Recently, a third checkpoint against self-reactive IgM during IgM⁺CD27⁺ B cell development in humans was described (40). This novel checkpoint for B cell tolerance is active between the naive and IgM⁺CD27⁺ B cells that have undergone SHM. Interestingly, the data suggest that self-reactive IgM antibodies are removed before the onset of SHM.

In conclusion, here we show that in utero postrearrangement Ig diversification via SHM occurs in human IgM⁺IgD⁺CD27⁺ B cells. Using a humanized mouse model, we furthermore show that the IgM⁺IgD⁺CD27⁺ B cells strongly resembled MZ B cells and that development and induction of SHM can occur in the absence of T cells. The T cell–independent IgM⁺IgD⁺CD27⁺ B cell development and induction of SHM strongly suggest that classical switched memory B cells and IgM⁺IgD⁺CD27⁺ B cells develop along separate pathways.

BALB/c Rag2^−/− γc^−/− mice and BALB/c Rag2^−/− γc^−/− mice on a Nude background were bred and maintained in isolators and fed autoclaved food and water under a protocol approved by the AMC-UVA. Repopulated Rag2^−/− γc^−/− mice with cells of the HIS-(BALB-Rag/γ) mice were generated as described previously with minor changes (27). In brief, newborn (<1-wk-old) Rag2^−/− γc^−/− mice received sublethal (3.5 Gy) total body irradiation with an x-ray source and were injected intra-hepatically with 0.5–10⁶ CD34⁺ human fetal liver cells or 0.5 to 2 × 10⁶ CD34⁺CD38^–lineage^– human fetal liver cells. For the lineage depletion, we used antibodies against CD3, CD19, CD56, and BDCA2. All manipulations with the mice were performed under a laminar flow. Cell suspensions were prepared in RPMI medium with 2% fetal calf serum.

After the preparation of single cell suspension and isolation of mononuclear cells by density gradient centrifugation over Lymphoprep Ficoll-Hypaque (Nycomed Pharma), a magnetic enrichment of fetal liver CD34⁺ cells or B cells (>98% pure) was performed by using the CD34 Progenitor Cell Isolation kit or the CD19 MACS kit, respectively (Miltenyi Biotec). This was followed by cell sorting on a FACS Aria (Becton Dickinson). Human fetal liver was obtained from elective abortions, with gestational age ranging from 14 to 20 wk. The use of this tissue was approved by the Medical Ethical Committee of the AMC-UVA and was contingent on informed consent.

Cell suspensions were labeled with anti–human mAb targeting the following cell surface markers: CD1a (T6-RD1), CD21 (BL13; Beckman Coulter), CD3 (SK7), CD4 (SK3), CD8 (SK1), CD19 (HIB19), CD45 (2D1), IgM (G20-127), IgD (IA6-2; BD Biosciences), CD27 (LT27; AbD Serotec), CD27 (LG.7F9; eBioscience), BDCA1/CD1c, (AD5-8E7), and BDCA2 (AC144; Miltenyi Biotec). Stained cells were analyzed with an LSR II (Becton Dickinson), and flow cytometric data were processed with CELLQuest software (Becton Dickinson).

Total mRNA was isolated from cells using the RNeasy mini kit (QIAGEN). mRNA was reverse-transcribed in a volume of 20 μl containing 5× first-strand buffer, 500 μM dNTPs, 25 μg/liter oligo(dT), and 200 U superscript II RT (Invitrogen). A portion of the cDNA solution (1 μl) was amplified by PCR in a 50-μl solution containing 20 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl₂, 5 mM dNTPs, 2.5 U Taq DNA polymerase (Invitrogen), and 30 pmol of each primer. PCR conditions were as follows: a 7-min denaturing step at 94°C followed by 40 cycles of 30 s at 94°C, 30 s at 60°C, and 30 s at 72°C, and a final 7-min extension at 72°C. Actin and AID primers were similar to those described previously (41).

The PCR conditions for the VH3-Cμ genes were similar to those described above, including a 7-min denaturing step at 94°C followed by 35 cycles of 1 min at 94°C, 1 min at 60°C, and 1 min at 72°C, and a final 10-min extension at 72°C. The oligonucleotide used for RT-PCR was as follows: VH3 forward, 5′-GGGGGTCCCTGAGACTCTC-3′.

Quantitative real-time PCR analysis was performed with an iCycler PCR (Bio-Rad Laboratories). The iQ SYBR Green Supermix (Bio-Rad Laboratories) was used for amplification. After incubation at 95°C for 6 min, 40 cycles of amplification were performed. Each cycle consisted of 30 s at 95°C, 30 s at 60°C, and 30 s at 72°C. The oligonucleotides used for real-time RT-PCR were as follows: Notch2 forward, 5′-TGTGAATGCAGTGGATGACC-3′ and Notch2 reverse, 5′-GGGCAGCAAGAAACAGAGG-3′; Actin forward, 5′-GGATGCAGAAGGAGATCACTG-3′ and Actin reverse, 5′-CGATCCACACGGAGTACTTG-3′.

PCR products were cloned into Topo TA cloning vectors (Invitrogen). Subsequently, mini-preps were made of each individual clone and both strands of the inserts were sequenced to obtain the sequence. Sequencing was performed with an ABI sequencer (PerkinElmer) using the dye-terminator cycle-sequencing kit (PerkinElmer).

The sequences found were compared with published germline sequences using the National Center for Biotechnology Information Immunoglobulin blast on the Internet (http://www.ncbi.nlm.nih.gov/igblast/) to identify mutations. The last nucleotide position of the V gene was excluded from the mutational analysis in view of possible nucleotide deletions at the joining sites.

Replication-defective self-inactivating HIV vectors were produced by transient transfection of 293T cells using FUGENE (Roche) and three different plasmids. The plasmids used were the VSV-G envelope coding plasmid pMD.G; the packaging plasmid pCMVDR8.91, designed to provide the Gag, Pol, Tat, and Rev proteins to produce the virus particle; and the transfer vector pTRIPU3-E1 (42). pTRIPU3-E1 has been modified to include the siRNA cassette from the pSUPER vector containing the human NOTCH2-targeting sequence 5′ (43, 44), and it carries the enhanced GFP or YFP gene driven by the elongation factor-1α promoter. 48 h after transfection, medium was replaced and two samples of virus were collected at 24 and 48 h. The virus-containing supernatants were centrifuged for 4 min at 1,800 rpm to remove cells and then were passed through a 0.22-mm filter and kept at −80°C until use.

Transduction of CD34⁺ human cells was performed by one cycle of overnight exposure to viral supernatant on retronectin (Takara Shuzo, Co.) -coated 24-well plates in the absence of cytokines. The next day, cells were washed and injected intra-hepatically in the newborn mice.

To measure Ig levels in the serum of the reconstituted mice, we performed an ELISA for IgM and IgG using the serum of the mice. Microtiter plates (9018; Costar) were coated with 100 μl/well of anti-IgM (10 μg/ml goat anti–human polyclonal; Jackson ImmunoResearch Laboratories) or anti-IgG (10 μg/ml goat anti–human polyclonal; Jackson ImmunoResearch Laboratories) in PBS. ELISA was developed with horseradish peroxidase–labeled goat anti–human IgM and IgG antibodies (Dako) and horseradish peroxidase substrates (Sigma-Aldrich). OD₄₀₅ was measured with a microplate reader (MDS Analytical Technologies).

Fig. S1 shows four representative IgD/CD27 dot plots of analyzed fetal tissue. Fig. S2 shows the gestational age of the analyzed fetal tissue and the percentage of IgD⁺CD27⁺ B cells within the IgD⁺ B cell population. Fig. S3 shows that CD19⁺sIgD^–CD27⁺ B cells express CD34 and lack BCR expression. Fig. S4 shows that AID is expressed in the IgD⁺ B cells of fetal liver but not in fetal BM and fetal spleen IgD⁺ B cells. Fig. S5 shows the reconstitution efficiency of repopulated HIS mice, the percentage of CD45⁺, CD3⁺, and CD19⁺ cells, and the IgG and IgM levels in the serum. Fig. S6 shows the knockdown efficiency for the NOTCH2 shRNA-mediated knockdown construct.

B. Hooijbrink is acknowledged for his help with FACS sorting and maintaining the FACS facility. The Bloemenhove Clinic in Heemstede, Netherlands, is thanked for providing fetal tissue, and the department of gynecology, Academic Medical Center, University of Amsterdam, the Netherlands for providing newborn cord blood. H. Yssels is thanked for providing adult spleen. D. Amsen is acknowledged for critical reading of the manuscript. N. Papazian is acknowledged for technical assistance. C.J.M. van Noesel is acknowledged for scientific input. J. Volmer and J. Meerding are acknowledged for enriching the fetal material and handling the mice.

T. Cupedo was supported by VENI grant 916.66.018 from the Netherlands Organization for Scientific Research (ZON-MW) and H. Spits was supported by a NWO grant 901-08-093. K. Weijer, N. Legrand, and H. Spits were supported by the Human Vaccine Consortium “Grand Challenge in Global Health #4: Devise Reliable Testing Systems for New Vaccines” (http://www.hv-consortium.org). H. Spits is currently an employee of Genentech, a company that develops and markets drugs.

The authors have no conflicting financial interests.