Research ArticleHuman Immunology

Timely and spatially regulated maturation of B and T cell repertoire during human fetal development

See allHide authors and affiliations

Science Translational Medicine  25 Feb 2015:
Vol. 7, Issue 276, pp. 276ra25
DOI: 10.1126/scitranslmed.aaa0072

Developing Immunity

The adaptive immune response plays a critical role in protecting the body from both foreign pathogens and internal dangers such as cancer. However, little is known about how the immune system develops during human gestation. Rechavi et al. analyzed differences in B and T lymphocyte ontogeny from 12 to 26 weeks of gestational age. They found that B cell development precedes T cell development and that repertoire maturation is both temporally and spatially regulated. These data can be used as a baseline to improve immune function in developing fetuses and to assess the effects of therapeutic interventions.


Insights into the ontogeny of the human fetal adaptive immune system are of great value for understanding immunocompetence of the developing fetus. However, to date, this has remained largely uncharted territory, in large part because blood samples from healthy, early gestation fetuses have been hard to come by. In a comprehensive study, we analyzed levels of T cell receptor excision circles (TRECs), signal-joint κ receptor excision circles (sjKRECs), and intron recombination signal sequence–K-deleting element (iRSS-Kde) rearrangement, and T and B lymphocyte repertoire clonality in human fetuses from 12 to 26 weeks of gestational age. Using next-generation sequencing, we analyzed the diversity and complexity of T cell receptor β (TRB) and immunoglobulin heavy chain (IGH) repertoires in four fetuses at 12, 13, 22, and 26 weeks of gestation and in healthy full-term infants. We report the progressive increase of TREC, sjKREC, and iRSS-Kde levels over time and confirm that B cell development precedes T cell development in the human fetus. Temporally and spatially regulated maturation of B and T cell repertoire diversity and complexity during human fetal development was observed, including evidence that immunoglobulin somatic hypermutation and class switch recombination occur already during intrauterine life. Our results help define physiological levels of immunodeficiency in premature infants and may serve as a reference for future studies aimed at investigating the impact of intrauterine pathologies on fetal immune development and function.


The ability of the adaptive immune system to specifically recognize pathogens and mount protective responses depends on the expression of a largely diversified set of antigen-specific receptors, in particular the T cell receptor (TCR) and immunoglobulin (Ig) molecules. Production of TCR and Ig molecules requires V(D)J recombination, a process that involves rearrangement of variable (V), diversity (D), and joining (J) elements at the TCR and Ig loci. Generation of a functional immune repertoire is a developmental process that initiates during fetal life but matures only several years after birth. The fetal lymphocyte counts rise throughout gestation (1). In particular, B lymphocytes are detected in the fetal liver already at 8 weeks of gestation and appear in fetal blood circulation by 12 weeks of gestation (2, 3). By contrast, T cell progenitors begin migrating to the thymus by 8 to 9 weeks of gestation, and circulating mature T lymphocytes are detected only at 15 to 16 weeks of gestation (4). Consistent with this developmental program, incomplete generation of a diversified antigen receptor repertoire has been reported in neonates and especially in prematurely born babies (5, 6), and this immaturity has been correlated with an increased risk of infection early in life.

Limited accessibility to blood samples from healthy, early gestation fetuses has limited our knowledge of the ontogeny of the human adaptive immune system. Studies of liver and spleen tissue of aborted fetuses have revealed differential usage of V, D, and J elements as compared to infants and adults (79). The techniques used so far to investigate development of human fetal T and B cell repertoire could not adequately capture the diversity of the fetal immune repertoire, a problem that can now be solved by using next-generation sequencing (NGS) (10).

Signal-joint T cell receptor excision circles (sjTRECs) represent by-products of V(D)J recombination process at the TCRα/δ locus. Similarly, V(D)J rearrangement at the IGK locus in developing B cells produces signal-joint κ receptor excision circles (sjKRECs) and intron recombination signal sequence–K-deleting element (iRSS-Kde) rearrangement. sjTRECs and sjKRECs are retained within developing T and B lymphocytes and are progressively diluted with cell replication, whereas iRSS-Kde represents a genomic signature of terminal rearrangement at the κ locus (11, 12). Accordingly, T and B cell lymphopoiesis can be assessed by measuring sjTREC and sjKREC levels in peripheral blood, and B cell replication history can be analyzed by measuring levels of sjKRECs and iRSS-Kde.

By studying levels of sjTREC, sjKREC, and iRSS-Kde, TCRβ (TRB) V gene usage, and T and B lymphocyte repertoire clonality in 20 fetuses between 12 and 26 weeks of gestational age (WGA), we confirm that B cell development precedes T cell development in the human fetus. Moreover, by applying NGS to the study of blood samples from four fetuses of various gestational ages and healthy full-term infants, we demonstrate timely and spatially regulated mechanisms that govern the progressive maturation of repertoire complexity during human fetal development. Together, these data may help define the progressive attainment of immunocompetence in the developing fetus and may serve as the basis for future studies of fetal immunity in health and disease and in response to therapeutic intervention in prenatal life.


Progressive increase of sjTREC, sjKREC, and iRSS-Kde levels during fetal development

To analyze thymopoiesis and B cell development during intrauterine life, we quantified sjTREC, sjKREC, and iRSS-Kde levels in 20 and 19 (respectively) blood samples collected from fetuses at 12 to 26 WGA. As expected, sjTREC as well as sjKREC and iRSS-Kde copy numbers progressively rose with gestational age (Fig. 1 and table S1). Notably, sjKRECs and iRSS-Kde appear in peripheral blood slightly before sjTRECs. In particular, samples obtained from six fetuses at 12 WGA contained either undetectable (n = 5) or very low (n = 1) TREC levels, whereas only one of them had undetectable sjKREC and iRSS-Kde rearrangement products. Overall, these data indicate that Ig gene rearrangements occur earlier than do TCR gene rearrangements during human fetal ontogeny, consistent with the earlier appearance of B lymphocytes than of T lymphocytes in fetal peripheral blood. Because sjKRECs are progressively diluted during B cell proliferation (12), the sjKREC/iRSS-Kde ratio can be conveniently used to analyze the balance between naïve, newly generated B cells and B cells with a significant replication history. The measurement of the sjKREC/iRSS-Kde ratio of 15 fetal blood samples in which both sjKREC and iRSS-Kde were detected (table S1) was 0.9 ± 0.14 (mean ± SEM), indicating very modest replication history. Overall, these findings indicate that B cell lymphopoiesis precedes T cell lymphopoiesis during human fetal development.

Fig. 1. T and B cell receptor excision circle copy numbers throughout gestation.

(A and B) sjTREC (A) and sjKREC (B) copies in fetal blood samples at 12 to 26 WGA. Results are shown in log scale, and a trend line is presented. Five samples dated 12 weeks of gestation contained undetectable sjTRECs and are represented in the figure as a single diamond.

Diversification of T cell repertoire during human fetal development

To assess diversity of the T cell repertoire, surface membrane expression of 24 different TCR Vβ (TRBV) families was analyzed by flow cytometry on six blood samples obtained from fetuses of different gestational ages (ranging from 14 to 26 weeks), and results were compared to a normal adult control (Fig. 2, A to F; table S2). Skewed usage of TRBV families was observed early during fetal development, with progressive attainment of polyclonality at later gestational age (Fig. 2F). However, overexpression of the TRBV5.6 family was observed throughout fetal development.

Fig. 2. TCR-Vβ repertoire: Expression of individual families by gestational age.

(A to F) Relative levels of expression of 24 specific TCR-Vβ families in six fetal (F) blood samples of various gestational ages. Levels of expression in each fetus (black bars) are compared to normal controls. (G) Levels of expression (mean ± SD) of TCR-Vβ families in the six fetuses (dotted bars) as compared to normal control values (clear bars), which were obtained from the IOTest Beta Mark—Quick Reference Card controls.

Rearrangement of the TCRγ (TRG) locus occurs in all T lymphocytes, irrespective of their final TCRαβ+ or TCRγδ+ nature (13), and can be used as a genetic signature to investigate T cell clonality (14). To further analyze diversity of the TCR repertoire during fetal development, we performed polymerase chain reaction (PCR) analysis of TRG rearrangements. As shown in Fig. 3A, analysis of Vγ9/Jγ1.1/2.1 and Jγ1.3/2.3 rearrangements demonstrated a clear shift from an oligoclonal pattern in early second trimester fetal blood samples to a largely polyclonal pattern in late second and early third trimester. Similar results were demonstrated also for the rearrangements involving the other three TRGV genes analyzed (Vγ11, Vγ10, VγF1/Jγ1.1/2.1, and Jγ1.3/2.3; fig. S1). Notably, some TRG rearrangements, albeit oligoclonal in nature, were demonstrated also in samples in which no sjTREC copies were detected, possibly reflecting the notion that rearrangement of the TRG locus precedes rearrangement of TRA (13), which is required for sjTREC generation.

Fig. 3. Spectratyping of TCRγ and IGH CDR3 length.

(A) Spectratyping of TRG CDR3 region in blood samples from 10 fetuses (F) of various WGA. (B) Spectratyping of IGH CDR3 region in blood samples from 12 fetuses of various gestational ages.

Progressive increase of B cell receptor repertoire diversity during fetal development

PCR analysis of Ig heavy chain (IGH) locus gene rearrangements was performed for 12 samples obtained at 12 to 26 WGA. Similar to what was observed for TCR repertoire, analysis of B cell receptor (BCR) repertoire using VH-FR2/JH consensus demonstrated a shift from an oligoclonal pattern in early gestation to a polyclonal pattern in late gestation (Fig. 3B). In particular, all samples obtained from fetuses at 12 to 14 WGA displayed an oligoclonal or a skewed polyclonal pattern, whereas all samples obtained at 17 WGA and upward displayed a clearly polyclonal pattern. Changes in clonality paralleled the progressive increase of sjKREC and iRSS-Kde copy numbers (table S1).

Reduced diversity and uneven distribution of TRB and IGH clonotypes during early stages of fetal lymphocyte development

To investigate in greater detail the diversity and quality of T and B cell repertoire during human fetal development, we have applied NGS to four samples obtained at 12, 14, 22, and 26 WGA. The total and unique sequences of rearranged TRB and IGH products for each of the four fetal blood samples and in healthy infant controls are reported in table S3.

Tree map representation of antigen receptor repertoires, where each dot represents a unique V-J pair and the size of each dot corresponds to the frequency of that rearrangement in the total population of sequences obtained, showed that 12 to 14 WGA fetal blood samples were characterized by a restricted TRB repertoire with clonotypic expansions, whereas a more diversified repertoire was present in 22 to 26 WGA samples (Fig. 4A). By contrast, the IGH repertoire of fetal blood samples was characterized by broader diversity even at early time points, although some clonotypic expansions were present in the 12 WGA sample (Fig. 4B). To define more precisely the degree of clonal expansions, we analyzed what proportion of the total sequences was accounted for by the top 100 most abundant TRB and IGH clonotypes. For both TRB (Fig. 4C) and IGH (Fig. 4D) rearrangements, the top 100 most abundant clones accounted for a higher fraction of total sequences in 12 to 14 WGA fetal blood samples than in infant controls. By contrast, a similar degree of clonality was observed in 22 to 26 WGA fetuses and healthy infants.

Fig. 4. Diversity of fetal blood TRB and IGH repertoires.

(A and B) Tree map representation of TRB (A) and IGH (B) repertoires from four fetal blood samples ranging from 12 to 26 WGA, where each dot represents a unique V to J joining and the size of the dot represents relative frequency. (C and D) Frequency of the top 100 clonotypes for TRB (C) and IGH (D) repertoire sequences from fetal and healthy infant blood samples (error bars showing SE; n = 100 each for F4, F9, F14, and F20; n = 300 for TRB repertoire in three infant controls; n = 400 for IGH repertoire in four infant controls; unpaired t test). (E) Summary graph representing Shannon’s H and Gini-Simpson indexes of diversity for the TRB and IGH repertoires. (F and G) Graphical presentation of cumulative frequency of unique clonotypes versus descending frequency of total clonotypes for TRB (F) and IGH (G), where the x axis value at the intercept of 50% of the total clonotypes defines D50. aa, amino acids.

Ecology biodiversity parameters can be used to measure diversity and evenness of the immune repertoire. In particular, Shannon’s H entropy index measures the diversity of the repertoire, taking into account the clonal size distribution in the overall repertoire. By contrast, the Gini-Simpson index (1 − Simpson’s D) (15) measures inequality of a given repertoire so that the higher the Gini-Simpson index, the more unequal is the distribution of individual clonotypes in the sample. Finally, D50 is a parameter that defines evenness of the population and corresponds to the percentage of unique sequences that account for 50% of the total number of sequences obtained. In a population where all species (unique sequences) are equally represented, the D50 index has a value of 50. By contrast, the lower the D50, the more uneven is the distribution of clonotypes.

As shown in Fig. 4E and table S4, markedly reduced diversity and higher degree of clonality of the TRB repertoire were documented in 12 WGA fetal blood sample as compared to samples obtained from fetuses of higher gestational age and from healthy infants. On the other hand, consistent with the tree map data (Fig. 4B), the IGH repertoire appeared largely diverse throughout fetal development (Fig. 4E and table S4). Finally, uneven distribution of TRB and IGH clonotypes (indicated by marked reduction of the D50 index) was observed at 12 to 14 WGA, with normalization at 22 to 26 WGA (Fig. 4, F and G).

Characterization of the complementarity-determining region 3 region in fetal blood TRB and IGH repertoires

The complementarity-determining region 3 (CDR3) of TCR and Ig molecules plays a critical role in antigen binding (16, 17). Accordingly, modifications of CDR3 length, amino acid composition, and hydrophobicity profile have a significant impact on antigen recognition. NGS analysis of CDR3 length of the expressed TRB and IGH unique sequences demonstrated that fetal blood samples were characterized by a shorter CDR-B3 and CDR-H3 length as compared to healthy infants (Fig. 5, A and B). The average CDR-B3 lengths for fetal blood samples from 12, 14, 22, and 26 WGA were 33.42, 31.88, 31.11, and 33.22 nucleotides (nt), respectively, and were significantly shorter than the CDR-B3 length of three healthy infants (35.06 nt; P < 0.0001, multiple t test; table S5). Similarly, the average CDR-H3 lengths for fetal samples from 12, 14, 22, and 26 WGA were 36.44, 35.59, 37.01, and 39.57, respectively, and were significantly shorter than the CDR-H3 length of four healthy infants (45.51 nt; P < 0.0001, multiple t test; table S6). These data confirm and extend previously published observations of decreased average CDR-H3 length in B cells from fetal liver and cord blood (18, 19). Deconstruction analysis of the CDR3 region showed that the reduced length of both the CDR-B3 and CDR-H3 lengths in fetal blood samples was mostly due to reduced N nucleotide addition (Fig. 4, C and D, fig. S2, and tables S5 and S6), as previously reported (18, 20). Moreover, with the exception of one blood sample from a 14 WGA fetus, no increase in trimming at the 5′ and 3′ ends of the IGHD genes was observed (Fig. 5D and table S6), indicating that exonucleolytic activity does not contribute to generation of shorter CDR-H3 regions during prenatal life. Analysis of individual IGHD gene usage in fetal blood samples showed increased frequency of IGHD7-27, which is the shortest IGHD gene element (fig. S3 and table S7). Decreased N nucleotide addition and skewed usage of the IGHD7-27 gene also contributed to the different amino acid composition of the fetal CDR-H3 region as compared to the CDR-H3 of infant controls (fig. S4), with a more hydrophilic profile in the former, as indicated by a lower Kyte-Doolittle index of average hydrophobicity (fig. S5). Similar data had been previously reported for human fetal liver B cell development (18).

Fig. 5. Characterization of the CDR3 region in the fetal blood TRB and IGH repertoire.

(A and B) Distribution of the CDR-B3 (A) and CDR-H3 (B) length (mean ± SE) among unique sequences from fetal and healthy infant blood samples. (C and D) Summary of the deconstruction analysis and a representative deconstruction map of healthy infant sample, demonstrating the average length in nucleotides (nt) of various components of the CDR3 for TRB (C) and IGH (D) repertoires, with asterisks showing the values significantly different from healthy infant samples (mean ± SE, exact P and n values can be found in tables S5 and S6; multiple t test). NDN is the region of CDR3 of nucleotides of N1 nucleotide addition, D region, and N2 nucleotide addition; V, D, or J in CDR3 are the lengths in nucleotides of V, D, or J gene segments (respectively) in the CDR3; P-N1-P is the region of CDR3 including both P and N nucleotide addition between V and D gene segments; P-N2-P is the region of CDR3 including both P and N nucleotide addition between D and J gene segments; 3′V and 3′D trimming is the nucleotide loss at the 3′ of V and D gene segments, respectively; and 5′D and 5′J trimming is the nucleotide loss at the 5′ of D and J gene segments, respectively.

Timely and spatially regulated usage of IGHV, IGHD, and IGHJ gene elements during fetal development

Next, we analyzed whether spatial constrains impinge on generation of the IGH repertoire during fetal development. As shown in fig. S3 and table S7, increased usage of DH-proximal gene elements (mainly IGHV6-1 and IGHV1-2) throughout fetal development and less frequent usage of D-distal IGHV genes at early time points (12 to 14 WGA) were observed as compared to the repertoire of healthy infants. Similarly, the fetal IGH repertoire was also characterized by increased usage of the most JH-proximal IGHD7-27 gene element, especially at early WGA (fig. S3, A and C, and table S7). Moreover, the DH-proximal IGHJ2 and IGHJ3 genes were more abundantly, and the DH-distal IGHJ6 gene was less frequently, used in fetal than in healthy infants’ B lymphocytes (fig. S3, A and D, and table S7). Similar findings had been previously reported in 130-day fetal liver samples (7, 21). Finally, we have also documented skewed usage of TRB gene elements in fetal blood samples. In particular, TRBD1 was more frequently used than TRBD2 in fetal samples, but not in healthy control infants (fig. S3E).

Isotype switching, somatic hypermutation, and antigen-mediated selection in the fetal IGH repertoire

Isotype switching and somatic hypermutation (SHM) shape the secondary antibody repertoire (22); however, little information is available on these processes during human fetal development. 454 GS Junior sequencing, using RNA as template, allows to obtain longer reads (~500 base pairs) that extend from the VH to the CH region, and thus permits to define distribution of isotypes of the expressed IGH transcripts (23). Analysis of IGH heavy chain usage among total sequences showed that only IGHM- and IGHD-containing transcripts were detected in 12 to 14 WGA fetal blood samples and that class-switched Igs (mostly IgG) were expressed by late second trimester fetal blood B lymphocytes (Fig. 6A), consistent with delayed maturation of the IgG repertoire in preterm infants (24). Because of the virtual lack of class switch recombination (CSR) in early gestational age fetal blood samples, we compared the rate of SHM among unique IGHM-containing transcripts from fetal blood and infant controls. To avoid potential ambiguity and sequencing errors, reads containing a single nucleotide difference from aligned reference sequence were removed from the analysis of SHM. Furthermore, the IMGT/HighV-Quest analysis detects and corrects sequencing errors that may be introduced during 454-based sequencing because of homopolymer hybridization (25). As shown in Fig. 6B, a substantial level of SHM was detected in fetal blood samples, including 12 to 14 WGA specimens. To investigate whether the observed SHM may also reflect antigen-mediated selection, we analyzed mutations occurring in CDR1 and CDR2 regions, resulting in amino acid replacement (RCDR) versus the total number of mutations in the V genes (MV). For each unique IGHM-containing sequence, we plotted the intersect of RCDR/MV with MV (Fig. 6, C to G). In this analysis, mutations resulting from antigen selection fall outside the gray areas. A progressive increase in putative antigen-selected mutations was observed during fetal development. In particular, blood samples from 22 to 26 WGA fetuses had reached a similar rate of antigen-mediated selection of mutated transcripts as healthy infants (Fig. 6H).

Fig. 6. Characteristics of secondary antibody repertoire in fetal blood.

(A) Analysis of IGH heavy chain usage among total sequences in the fetal and healthy infant blood. (B) Summary of SHM analysis in IGHM transcripts among the total sequences (mean ± SE; n = 5689, 2722, 4322, 3640, and 11681 for F4, F9, F14, F20, and infant control samples, respectively; unpaired t test). (C to G) Graphical representation of antigen selection in IGHM transcripts, where the ratio of replacement mutations in CDR-H1 and CDR-H2 (RCDR) to the total number of mutations in the V region (MV) is plotted against MV. The dark and light gray shaded areas represent 90% and 95%, respectively, confidence limits for the probability of random mutations. Data points falling outside the light gray shaded area represent the portion for antigen selection with probability of random mutation being P = 0.1 (dark gray area) and P = 0.05 (light gray area). (H) Summary of observed frequency for antigen selection with probability of random mutation if P = 0.05.

Sharing of clonotypes in fetal blood immune repertoire

Our data have shown that fetal blood TRB (and to a lesser extent IGH) repertoire is less diverse than later in life, and that both the TRB and IGH fetal repertoires present important qualitative differences when compared to the corresponding repertoires from healthy infants. Moreover, our data of SHM suggest that antigen-mediated selection is operative during fetal lymphoid development. Antigenic pressure may determine clonotype sharing, where different subjects may show preferential usage of the same clonotype to respond to the same antigen. Alternatively, clonotype sharing may be caused by constrains that favor rearrangement of specific V, D, and J genes before the occurrence of selection. We observed that 138 unique CDR-B3 clones and 81 unique CDR-H3 clones were shared among fetal blood samples and controls. Only a striking minority of both the CDR-B3 (1 of 138) and CDR-H3 (3 of 81) clones were shared exclusively among infant controls, whereas the majority of the CDR-B3 and CDR-H3 clones were shared exclusively among fetal samples (Fig. 7A). All four fetal blood samples contained a similar proportion of shared clonotypes, and this was significantly higher than the frequency of shared clones detected in control infants (Fig. 7B). Consistent with previous observations (26), shared unique clonotypes were shorter when compared to the average length of all unique clonotypes (Fig. 7, C and D). Analysis of the top 10 clones that were shared exclusively among fetal blood samples showed that these clones accounted for a significant proportion of all total sequences, especially for CDR-B3 and for early gestational age samples (Fig. 7, E and F).

Fig. 7. Clonotypic sharing in TRB and IGH repertoires from fetal and healthy infant blood.

(A) Venn diagram showing the distribution of the shared clones among healthy infants, among fetal samples, and among both infants and fetal samples. (B) Frequency of the shared clones in TRB and IGH repertoires in each sample among the unique amino acid sequences (n = 4 for both fetal TRB and IGH repertoires, n = 3 for infant TRB repertoire, and n = 4 FOR infant IGH repertoire; unpaired t test). (C and D) Graphical representation comparing the CDR3 length of shared unique clonotypes and of all unique clonotypes for TRB (C) and IGH (D) repertoires (mean ± SE; n values are summarized in table S9; multiple t test). (E and F) Top 10 shared total clonotypes for TRB and IGH repertories, respectively.


Analysis of composition and diversity of the B and T cell repertoire during human fetal development may provide fundamental insights into maturation of immunocompetence and may help define risks of prematurely born babies to infections. Previous studies of the adaptive immune repertoire in the developing fetus have mostly focused on the analysis of fetal liver and spleen cells, especially from second and third trimester fetuses, and have been largely based on targeted amplification and sequencing of selected rearranged products (8, 9, 19, 2729). Here, we have made use of new and more powerful technologies to analyze in greater detail the shaping of T and B cell repertoire in blood samples obtained from fetuses of various gestational ages. Our results demonstrate that circulating B lymphocytes containing both sjKREC and iRSS-Kde rearrangements (indicating recent de novo lymphopoiesis) are present already at 12 WGA and that their BCR repertoire is surprisingly diverse. We have also shown that the fetal B cell repertoire is characterized by shorter CDR-H3 length with reduced hydrophobicity, decreased N diversity, and preferential usage of DH-proximal IGHV, JH-proximal IGHD, and DH-proximal IGHJ gene segments, as well as of IGHD7-27. Furthermore, we have observed a progressive increase of CDR-H3 length from an average of 30 nt at 12 to 13 WGA to an average of 36 nt by 22 WGA to 39 nt by 26 WGA. Because of the large number of reads obtained, these data strengthen previous studies where only a limited number of rearrangements had been captured and analyzed (7, 8, 27, 29, 30). Reduced N diversity has been previously observed also in preterm babies (31) and in full-term newborns (23) and recently confirmed by NGS analysis also in cord blood (32), where adult-type distribution of CDR-H3 length in full-term babies is not achieved until 2 months of life (19). Our observations on the progressive increase of CDR-H3 length and variation of its composition (from hydrophilic to neutral profile) with gestational age, along with other changes in IGHV and IGHJ gene usage, and the increase in evenness of IGH clonotype size distribution demonstrate that dynamic changes in the IGH repertoire occur during human fetal development. Repertoire restriction and emergence of expanded clonotypes were particularly present at early WGA and were likely caused also by reduced output of mature T and B cells early during human fetal development (2, 3).

As expected, we found that the fetal IGHM repertoire consists mostly of IGHM-containing transcripts. However, at 26 WGA, IGHG-containing clones were present in a similar proportion as observed in healthy infants. A few IGHA- and IGHE-expressing clones were also detected, and the possibility that other IgA-expressing B cells may have localized to mucosal tissues cannot be excluded. Interaction between antigen-specific T helper cells and B lymphocytes plays an important role in promoting CSR. We and others have shown that cord blood CD4+ T cells have reduced ability to express CD40 ligand upon in vitro activation (33, 34), and B lymphocytes from preterm infants have reduced expression of tumor necrosis factor receptor family receptors (TACI, BCMA, and BAFF-R) (35) and exhibit markedly reduced IgG and IgA production in response to CD40 ligand (CD40L) and interleukin-10 (IL-10) (34), suggesting that both T-dependent and B cell–intrinsic mechanisms contribute to low levels of CSR in human fetal B cells. Furthermore, animal data indicate that exposure to external antigens plays a critical role in inducing CSR (36, 37). A low level of antigenic exposure characterizes normal human fetal development; nonetheless, it has been proposed that antigenic peptides processed by maternal cells or complexed with maternal IgG may access the fetal circulation and prime immune responses (38, 39).

In addition to being largely restricted to the IgM isotype, antibody responses in the neonate are typically of low affinity. Whether and to what extent SHM and selection of B lymphocytes expressing Igs with high affinity for antigens are operative during prenatal life and at birth has been a matter of contention. Ridings et al. (40) have amplified and sequenced IGHV6-containing Ig transcripts expressed by cord blood B cells, and have shown a low rate of SHM without evidence of antigen-driven selection. By contrast, Scheeren et al. (41) have reported that sorted marginal zone (MZ)–like CD19+IgM+IgD+CD27+ spleen B cells from 14 to 18 WGA fetuses express VH3-Cμ transcripts in 20% of the cases that contain SHM. Although we did not characterize the phenotype of fetal B cells carrying mutated VH-Cμ transcripts, our data confirm that SHM does occur during human fetal B cell development, even at a time (12 to 13 WGA) when there is limited T lymphopoiesis. Consistent with this, expression of activation-induced cytidine deaminase (AID), a molecule critically required for CSR and SHM, has been demonstrated in fetal liver and mesenteric lymph nodes at 14 to 18 WGA (41). Finally, we observed that a significant fraction of the mutations in IGHV genes of fetal blood B cells resulted in amino acid changes and were preferentially located in CDR regions, strongly indicating a role for antigen-driven selection. Because of the limited T cell development at 12 WGA and the lack of requirement of T cell help for SHM in MZ-like fetal B cells (41), it is likely that fetal B lymphocytes harboring SHM recognize T-independent antigens. A role for self-antigens in the induction of SHM cannot be excluded. In particular, AID expression and T-independent SHM have been documented in developing mouse B cells (42, 43) and may allow immature B cells and IgM+ CD27+ B cells to control self-reactivity (43, 44).

We found that T lymphopoiesis is lagging behind B cell development at early stages of fetal development. Furthermore, the T cell repertoire of early gestational age fetuses was characterized by significantly reduced diversity, uneven representation of clonotypes, and shorter CDR-B3 because of reduced N nucleotide addition. Low levels of TdT expression have been observed in human fetal thymus until 19 WGA (45). We have observed a progressive increase of TRB repertoire diversity and evenness during fetal development so that diversity at 22 to 26 WGA was indistinguishable from what was observed in healthy control infants. Together, our findings confirm and extend previous observations of nonrandom usage of TRAV and TRBV genes (4648) and a slow increase in the extent of N nucleotide addition (45) and CDR-B3 length (28) during fetal development.

Sharing of TRB (and to a lesser extent IGH) clonotypes among fetal blood samples had not been previously reported. The emergence of shared clonotypes has been described in patients with acute and chronic infections (49, 50). Because blood samples analyzed in this study were obtained in otherwise healthy fetuses, the observed clonotype sharing cannot be attributed to selective pressure conferred by infections, although a potential role for self-antigens or pregnancy-associated antigens cannot be excluded. A more likely explanation for clonotype sharing is represented by the limited diversity of the TRB repertoire, along with nonrandom usage of V, D, and J elements, especially early during fetal human development (51).

The main limitation of this study is represented by the small sample size, which reflected the low frequency of the procedures from which the samples were obtained, the small volume of blood that could be collected (which mandated rationing of the samples among the various experiments), and the need to exclude from the study fetuses with suspected immunological disturbances or infections. In this regard, five of the samples used in the study were obtained from fetuses with various anomalies. Upon family history and literary review, we concluded that these anomalies should have no bearing upon the fetal immune system. The results obtained for these samples were consistent with the others included in this study.

In summary, we performed a comprehensive study of the ontogeny of adaptive immune system in human fetal development. We have identified differences in the kinetics of development and maturation of T and B cell repertoires. Furthermore, limited diversity and uneven distribution of antigen receptor specificities were detected in early second trimester fetuses, with progressive attainment of a diversified and more even repertoire at 22 to 26 WGA. Reduced N nucleotide addition and preferential usage of certain coding elements remained characteristic features of the immune repertoire throughout fetal development. These data provide valuable information on the degree of immunocompetence of prematurely born infants. They may also serve as a reference for future studies aimed at investigating the impact of intrauterine infections, diseases of feto-maternal interface, inborn errors of immunity, and therapeutic interventions during pregnancy (such as use of steroids to facilitate fetal lung maturation) on the immune system of the developing fetus.


Study design

This is a nonrandomized, laboratory study performed on human fetal blood samples. Approval for conducting this study was obtained from the Institutional Review Board of Sheba Medical Center, the review board of the Israeli Ministry of Health, and Children’s Hospital Boston Institutional Review Board. A written informed consent was obtained from all participants. Sample size was dictated by the rate of sample collection over a year. Fetuses with or suspected of conditions known to interfere with immune development such as maternal infection (for example, cytomegalovirus) or relevant chromosomal abnormalities (for example, Down syndrome and DiGeorge syndrome) were excluded.

Fetal samples

Blood samples were collected over a 12-month period from 20 fetuses at 12 to 26 WGA (median gestational age, 15 weeks and 6 days). In 10 cases, fetal blood was from apparently normal multifetal pregnancies requesting fetal reduction due to obstetric indications. In the remaining cases, selective fetal reduction was performed because of impending severe premature birth (n = 2), 45X0 mosaicism (n = 1), 47XXX (n = 1), severe hemophilia (n = 1), severe renal anomaly (n = 1), and other fetal anomalies (n = 1). In three cases, pregnancy was terminated because of social indications. No history suggestive of immunodeficiency was documented in any of the families who participated in this study. Fetal blood (300 to 1000 μl per sample) was collected by ultrasound-guided transabdominal cardiocentesis (22-gauge needle) and stored in EDTA-containing tubes.

Quantification of sjTRECs, sjKRECs, and iRSS-Kde rearrangements

sjTREC copy numbers were determined using quantitative real-time PCR (qRT-PCR). PCRs were performed as previously described (46) using 0.5 μg of genomic DNA extracted from the patients’ peripheral blood mononuclear cells as template. qRT-PCR was carried out using an ABI Prism 7900 Sequence Detector System (Applied Biosystems). A standard curve was constructed by using serial dilutions containing 103 to 106 copies of a known sjTREC plasmid. For each sample, serial dilutions were tested in triplicate, and the number of TRECs in a given sample was calculated by comparing the obtained cycle threshold (Ct) value of a patient’s sample to the standard curve using an absolute quantification algorithm. The amount of sjKREC and iRSS-Kde rearrangements was determined by qRT-PCR as previously described (52). Because of the scarcity of sample DNA, reactions were performed with duplicates instead of triplicates. Results were measured as described above in reference to serial dilutions of sjTREC. The threshold for Ct determination was positioned at the same level each time. Amplification of RNase P (TaqMan assay, Applied Biosystems) served as a quality control of DNA amplification for both the sjTREC and the sjKREC/iRSS-Kde assays.

Cell surface analysis of TCR-Vβ family expression

Surface expression of individual TCR-Vβ families was analyzed using flow cytometry and a set of Vβ-specific fluorochrome-labeled monoclonal antibodies (Becton-Dickinson, FACSCalibur) as described (53). Normal control values were obtained from the IOTest Beta Mark—Quick Reference Card (Beckman Coulter).

Analysis of TCR and BCR repertoire by NGS

Rearrangements at the endogenous TRB and IGH loci were amplified using as a template equal amounts of total RNA extracted from fetal blood samples (n = 4, 12 to 26 WGA) and from peripheral blood of healthy infants (n = 3 for TRB and n = 4 for IGH; age range, 9 months to 4 years). A set of nested primers specific for various V and C elements of the TRB and IGH loci were used for the first amplification cycles, followed by amplification using a communal primers according to the manufacturer’s protocol for TRB and IGH, respectively (iRepertoire Inc.) (54). Purified PCR products were sequenced using the GS Junior 454 platform (Roche Inc.).

Raw sequences were filtered for PCR errors, and tree maps were generated from the total CDR3 sequences to assess graphically V-J pairing and the relative distribution of distinct rearrangements. The filtered sequences were multiplied accordingly to reflect the number of copies observed in the sequencing using the Macro editor of Excel to generate a CSV file with the numbers of the sequences and the nucleotide sequences. Next, the sequences were converted into FASTA format using the Geneious software (Biomatters Ltd.). FASTA sequences were submitted to IMGT HighV-QUEST and analyzed for V, D, and J gene usages, composition and length of CDR3, and Kyte Doolittle index of hydrophobicity using the IgAT software (55). Shannon’s H entropy and Gini-Simpson’s D indexes were calculated using the VDJ statistics file from IgAT analysis and the PAST program (56).

Statistical analysis

Unpaired t test was used to compare the fetal blood samples to infant controls for variables with normal distribution. For nonparametric variables, Mann-Whitney test was used. The χ2 test was used for categorical values. For all multiple t tests, post hoc Bonferroni correction was used. The analyses were performed using Prism version 6 (GraphPad).


Materials and Methods

Fig. S1. TCR Vγ rearrangement analysis.

Fig. S2. N nucleotide addition in CDR-H3.

Fig. S3. Differential usage of IGHV, IGHD, and IGHJ genes in fetal blood IGH repertoire and of TRBD genes in fetal TRB repertoire.

Fig. S4. Amino acid composition of CDR-H3.

Fig. S5. CDR-H3 hydrophobicity.

Table S1. TREC, sjKREC, and iRSS-Kde copy numbers during fetal development.

Table S2. Source file for TCR-Vβ repertoire of Fig. 2

Table S3. Summary of the number of reads and sequences that were used to analyze the TRB and IGH repertoire.

Table S4. Diversity indices of Shannon’s H and Gini-Simpson of the TRB and IGH repertoires.

Table S5. Average lengths in nucleotides of CDR-B3 and components comprising the CDR-B3.

Table S6. Average lengths in nucleotides of CDR-H3 and components comprising the CDR-H3.

Table S7. Summary of P values for χ2 test (goodness of fit with post hoc Bonferroni correction for multiple comparison) for IGHV, IGHD, and IGHJ gene usages in fetal blood IGH repertoire.

Table S8. Source data for TRBD gene usages of fig. S3E.

Table S9. Percent of shared clones among the unique sequences, number of shared clones (n1), and number of unique sequences (n2) in TRB and IGH repertoires of fetal and infant samples.


Acknowledgments: We thank the families for participating in this study. The sjTREC plasmid was provided by D. Douek (Vaccine Research Center, National Institute of Allergy and Infectious Diseases, Bethesda, MD). Funding: This work was supported by the Jeffrey Modell Foundation (to R.S.) and by The Manton Foundation (to L.D.N.). Author contributions: E.R., Y.N.L., L.D.N., and R.S. designed the study, supervised analysis of the data, and wrote the manuscript; E.R., A.L., Y.N.L., A.J.S., and N.A. performed the experiments and analyzed the data; Y.N.L. performed statistical analysis; Y.Y., S.L., and B.W. provided valuable samples and gathered the data. Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article