B cell development in chromosome 22q11.2 deletion syndrome

Abstract

Chromosome 22q11.2 deletion syndrome is a common immune deficiency associated with thymic hypoplasia. Most patients did not survive until the mid-1980s and now there is a growing adult population. B cell and immunoglobulin defects have been described and appear to be increased in the adult population. We used flow cytometry, B cell stimulation and repertoire analysis to understand B cell function. B cell production at early stages appeared to be normal in patients but adult patients exhibited a deficit of switched memory B cells. Follicular helper T cells were present at higher percentages in patients and they exhibited a more activated phenotype in patients compared to controls. In spite of that, somatic hypermutation was decreased in patients compared to controls at all ages. Fewer mutations per clone were seen, strongly implicating aberrant T cell help. Therefore, patients with chromosome 22q11.2 deletion syndrome have a progressive decrease in switched memory B cells and evidence of compromised T cell help. In children, evidence of compromised T cell help is limited to decreased somatic hypermutation. With age, greater manifestations become apparent even though a minority of patients have hypogammaglobulinemia. As this population ages, this has important implications for management.

1. Introduction

Chromosome 22q11.2 deletion syndrome (22q11.2DS) is a common primary immune deficiency with an estimated frequency of one in 4000 live births [1,2]. Due to advances in cardiac care in the 1980s, the population frequency is increasing because larger numbers of patients are reaching adulthood and reproducing. The diagnosis is typically established after the identification of a clinical constellation that can include conotruncal cardiac defects, hypocalcemia, palatal abnormalities, developmental delay, dysmorphic facial features, and diminished T cell numbers [3]. The immune deficiency in 22q11.2DS can be extremely diverse, with some patients having no evidence of an immune deficiency and some patients having a phenotype consistent with severe combined immune deficiency [4-6]. Most patients have a mild to moderate decrement in their T cell numbers and at least initially have preserved T cell function [7]. Over time, with homeostatic pressure on the T cell compartment, T cell function can decline [8,9]. Evidence of T cell exhaustion occurs in adulthood [10,11]. In addition, there can be repertoire imbalances that also arise as a consequence of the homeostatic pressure on a limited number of founder cells [7,9,12,13]. Clinical manifestations of the immune deficiency include recurrent infections and autoimmune disease [11,14,15].

While the defect in T cell production has been well characterized and is known to be secondary to thymic hypoplasia, there is less information about 22q11.2DS and B cell function. Several studies have identified low IgA or low IgG levels [14,16,17]. While vaccine responses and antibody production are largely preserved, there are now emerging data suggesting that the B cell compartment can be compromised, particularly in adulthood [8,14,16,18]. Unlike the T cell compartment, the B cells do not appear exhausted [19]. Antibody defects have been associated with a significant infection history and autoimmunity [14,16]. A greater understanding of the etiology of these B cell defects is critical as this population increases in size and grows older.

Our data have previously demonstrated diminished numbers of switched memory B cells in adults with 22q11.2DS and a large registry study has identified hypogammaglobulinemia in approximately 6% of patients [8,17]. This current study was undertaken to understand whether the abnormalities in the B cell compartment and antibody production were intrinsic to the B cells themselves or secondary to diminished T cell help. The gene felt to be responsible for the majority of the phenotypic features in 22q11.2DS is TBX1 [20-23]. TBX1 is expressed in early embryologic development and elicits its major effect through promoting growth of mesodermal cells and neural crest cells in the pharyngeal arches. When the pharyngeal arches fail to develop properly, there can be abnormalities in cell mass or migration of the parathyroid glands, the thymus, and the heart. It has not been thought that TBX1 is expressed in adult bone marrow. However, it is possible that early in development, haplosufficiency for TBX1 could alter the differentiation program for B cell precursors. This study was undertaken in an effort to define adequacy of T cell help for B cell production of antibodies.

2. Methods

2.1. Subjects

This study was approved by the Institutional Review Board at The Children's Hospital of Philadelphia. Subjects were recruited from clinic patients at The Children's Hospital of Philadelphia and Dalhousie University/IWK Health Center. A total of 27 adults and 44 children with 22q11.2DS were examined. The diagnosis for patients was established by fluorescent in situ hybridization, MLPA, or single nucleotide polymorphism array confirmation of a chromosome 22q11.2 deletion. Adult and pediatric controls were from The Children's Hospital of Philadelphia and the University of Pennsylvania. The demographic data of patients and controls are given in Table 1. Not all patients and controls were included in each study due to limitations imposed by blood volumes. The N for each experiment is given in the figure legend.

Table 1.

Subject demographics.

	Number	Age
Flow cytometry of fresh samples
Adult 22q11.2DS	27	25 ± 7
Adult Control	35	29 ± 7
Child 22q11.2DS	44	9 ± 6
Child Control	32	8 ± 5
B cell stimulation
Adult 22q11.2DS	3	22 ± 8
Adult Control	7	29 ± 7
Child 22q11.2DS	7	10 ± 7
Child Control	14	6 ± 4
TREC and KREC
Adult 22q11.2DS	11	26 ± 4
Adult Control	14	28 ± 7
Child 22q11.2DS	21	8 ± 6
Child Control	15	9 ± 6

Table 1. Demographic characteristics of the study population. The numbers of patients and their ages are given for each subset of data.

2.2. Immunologic assays

Flow cytometric analysis of T cell subsets and B cell subsets was performed with fixation with 1% paraformaldehyde. Samples were run on an LSR Fortessa (BD Biosciences, San Jose, CA) and analyzed using FlowJo software version 8.8.7 (TreeStar, Ashland, OR). B cells were defined by physical parameters and CD19 expression and CD4 T cells were defined by physical parameters and CD4 expression. We defined follicular helper T cells (TFH) as CD4 +, CXCR5 +, and ICOS +. Active TFH were defined as CD4 +, ICOS +, CXCR5 +, CCR7lo, PD1hi or CD4 +, CXCR5 +, CCR7lo, PD1hi. We defined transitional B cells as CD27-CD38 ++, non-switched memory B cells as CD27 + IgM +, switched memory B cells as CD27 + IgM- and naïve B cells as CD27-IgM +. Plasmablasts were defined using two different gating schemes, as either CD27 ++ and CD38 ++ or CD38 ++ and IgM-.

2.3. B cell stimulation

B cells were isolated from PBMCs with the Dynal Human Untouched B cell Kit and cultured for 7 days. Cells were stimulated with anti-CD40 at 2 μg/ml, anti-IgM at 7.5 μg/ml, and IL-21 at 50 ng/ml. Anti-IgG at 7.5 μg/ml was the negative control. The percent of CD80 positive cells is reported after flow cytometric analysis gating on CD19 positive cells.

2.4. RT-PCR, TREC and KREC analysis

RT-PCR was performed using ABI primers for IL-21 on a TaqMan 9600. TRECs and KRECs were measured as previously described [24]. The assay was performed as described by Sottini with additional calculations as described by van Zelm [25,26]. PBMC DNA was isolated using the Gentra Puregene kit (Qiagen). 50 ng of DNA was used in each PCR reaction. Duplex PCR reactions were performed on a SDS 7500 Fact Realtime PCR system. Primer sequences are as follows:

• KREC forward- 5′ TCAGCGCCATTACGTTTC 3′.

• KREC reverse- 5′ GTGAGGGACACGCAGCC 3′.

• KREC probe- 5′ 56-JOEN/CCAGCTCTTACCCTAGAGTTTCTGCACGG 3′ BHQ-1.

• TREC forward- 5′ CACATCCCTTTCAACCATGCT 3′.

• TREC reverse- 5′ GCCAGCTGCAGGGTTTAGG 3′.

• TREC probe- 5′ 56-FAM/ACACCTCTGGTTTTTGTAAAGGTGCCCACT 3′ Tamra.

• TRAC forward- 5′ TGGCCTAACCCTGATCCTCTT 3′.

• TRAC reverse- 5′ GGATTTAGAGTCTCTCAGCTGGTACAC 3′.

• TRAC probe- 5′ 56-JOEN/TCCCACAGATATCCAGAACCCTGACCC 3′ BHQ-1.

• ACTIN forward- 5′ CCGGCGCTGTTTGAACC3’.

• ACTIN reverse- 5′ CGGCCGCGTTATTACCATAAA 3′.

• ACTIN probe- 5′ Cy5/ACGCCTCCGACCAGTGTTTG 3′ BHQ-2.

Numbers of TRECs, TRACs, and KRECs were calculated using a standard curve from known copy numbers. The plasmid with the insets was supplied by Imberti [27]. Actin was used to assess the overall quality of the DNA. Samples with poor actin amplification were excluded. Calculations were:

(mean of TRECs or KRECS number)/(mean of TRACs number/2) * 10⁶.

2.5. IgH sequencing

Immunoglobulin heavy chain family-specific PCRs were performed on genomic DNA samples from peripheral blood CD19 + B cells of healthy controls and subjects with 22q11.2DS. Five children and five adults from the control and patient sets were sequenced (N = 20 subjects total). Three multiplexed mixes were employed with forward primers matching VH family leader sequences and reverse primers matching JH gene segment sequences, using a similar strategy to that described in Meng et al. [28]. Heavy chain gene rearrangements (VDJ) were amplified with AmpliTaq Gold (Life Technologies, Carlsbad, CA) and 10× buffer at a final concentration of 1.5 mM MgCl₂, 0.2 mM dNTPs and 0.4 mM of each primer cocktail. The primers each contained an appropriate adaptor sequence for subsequent Illumina sequencing. Amplicons were purified by QIAquick PCR purification kit (Qiagen) followed by size selection with Pippin prep (Sage Science, Beverly MA). Library quality was evaluated using Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA) and loaded onto an Illumina MiSeq using a 2 × 300 bp paired end kit (Illumina MiSeq Reagent Kit v3, 600 cycle).

2.6. IgH data analysis

Fastq files were obtained after sequencing and the PRESTO suite of tools (website: http://clip.med.yale.edu/presto/, [29]) was used to perform Quality score trimming using Q30 10-nucleotide sliding window and single nucleotide <Q30, to replace low quality nucleotide calls with N-nucleotides. (N′s are not counted as somatic hypermutations). All sequences were length trimmed (any sequence with fewer than 60 nt was discarded). Clonally related sequences were collapsed, allowing for up to 3 nucleotides to be N values when matching sequences. Quality filtered sequences were then assigned to germline IGHV gene segment alleles, based on alignments to a database of known IGHV alleles, using the IMGT/HighV-QUEST server [30]. Mutation levels were analyzed on the basis of “100 — % of VH identity to the nearest germline allele in the IMGT database” obtained from IMGT/HighV-QUEST analysis. A conservative cut-off of >5% SHM was chosen to distinguish bona fide somatic mutations from mutations that may have been due to sequencing errors based in part upon published reports of sequencing error frequencies using the Illumina platform [31,32]. CDR3 lengths were obtained from IMGT/HighV-QUEST output and density graphs of their distributions were generated using Kernel Density Plots in R. VH usage was computed using VH gene assignments on collapsed clones. VH genes that were not used by all of the study subjects and/or were present at a frequency of < 1% were discarded from the analysis. No VH rearrangements were uniquely found in only patients or only controls. Resulting VH usage data (represented as the percentage of clones using a particular VH gene) were visualized as a heat map using GENE-E software [33]. Mutation frequencies (mutations compared to the nearest germline VH per 1000 base pair) were calculated for all unique sequences in each study subject separately in the CDR2 and FR3 using immunoglobulin analysis tool (IgAT) software [34].

2.7. Statistical analysis

A Mann Whitney test was used for the comparisons of individual variables between groups for all data. Significance for all statistical tests was set at p ≤ 0.05.

3. Results

3.1. B cell maturation

We had previously described diminished switched memory B cell counts in adult patients with 22q11.2DS [7,8]. We started by confirming this finding in a new cohort. Subjects and controls were divided into children (age 1–17 years) and adults (ages 18 years and above). CD19 + B cell percentages in total lymphocytes were higher in patients than controls in both age groups, a reflection of the lower T cell frequency in the lymphocyte population. We again found that switched memory B cells were diminished compared to controls in adults, with a compensatory increase in naïve B cells (Fig. 1). To best define changes in the distribution of related subsets, populations were defined as percentages of the CD19 population. To ensure that the lack of switched memory B cells was not due to altered metabolism of surface IgM, we also utilized IgD as a surface marker and defined switched memory B cells as CD19 + CD27 + IgD-. When examined in this way, we again found that switched memory B cells were diminished in patients compared to controls. Non-switched memory B cells, defined as CD19 + CD27 + IgM +, were lower in adult patients than in controls as well. The frequencies of transitional B cells were similar between patients and controls (Fig. 1). Therefore, we did not observe a block in development with accumulation of precursors. We reasoned that if global fitness of B cells was compromised, there would be attrition of cells at subsequent developmental stages. The major circulating antibody secreting cells are plasmablasts. Plasmablasts were analyzed two different ways. The first was on the basis of CD27 ++ and CD38 ++ expression [35]. Because B1 B cells (which are IgM +) can also express CD27 and CD38 [36], we also analyzed a subset of plasmablasts that were IgM-, CD38 ++. Plasmablasts using either gating scheme were comparable between patients and controls in both age groups (Fig. 1). Collectively, these data suggest an increasing deficit of switched memory B cells with age that is specific to that subset and not progressive attrition of all later developmental stages.

Fig. 1.

B cell subsets in 22q11.2DS. Flow cytometry was performed on peripheral blood samples from 27 adult patients with 22q11.2DS, 35 adult controls, 44 children with 22q11.2DS and 33 child controls for B cell maturation markers. Plotted are the percentages of different B cell subsets within the B cell gate (CD19 + lymphocytes) in patients vs. controls. Each symbol represents a single subject. The phenotype of 22q11.2DS was more pronounced in adults with fewer switched memory B cells defined by either absence of IgM or IgD. Non-switched memory B cells, which express IgM and are thought to represent a distinct lineage that is T cell independent were also decreased. The horizontal bar represents the mean and the error bars are SEM. The P values compare the bracketed populations. NS = non-significant.

Diminished switched memory B cells is a frequent finding in common variable immune deficiency [37]. A corollary finding is the expansion of a B cell subset characterized by CD21lo expression, also seen in HIV [38]. These B cells are thought to represent an accumulation of exhausted B cells [39]. We therefore examined the CD19 +, CD38lo, CD21lo B cell subset in patients and found no difference between patients and controls (Fig. 1). This suggests that exhaustion, at least as associated with expansion of the CD21lo subset, was not the mechanism of loss of switch memory B cells.

To further evaluate the possibility that exhaustion was occurring in the B cell compartment, we performed a kappa-deleting recombination excision circle (KREC) analysis, reasoning that proliferative pressure would deplete KRECs, as is seen for the T cell recombination excision circles (TRECs) in T cells [40]. TRECs were lower in children with the deletion compared to controls but the effect did not reach significance in the adult group. Patient KREC content was slightly higher than that seen in controls for the adult 22q11.2DS population without reaching significance (Fig. 2A). For confirmation, we expressed the results using nanograms of DNA as the denominator and found comparable results (data not shown). The trend towards higher KRECs may reflect the higher frequency of naïve B cells in the patients. We therefore concluded that B cells in 22q11.2DS patients did not exhibit signs of proliferative pressure or exhaustion.

Fig. 2.

Lymphocyte status. A) PCR for KREC and TREC targets was performed on peripheral blood samples from 11 adult patients with 22q11.2DS, 14 adult controls, 21 children with 22q11.2DS and 15 child controls. Child patients had lower TRECs than control children. KRECs were higher in adult patients than controls. The horizontal bar represents the mean and the error bars are SEM. The P value reflects a t test comparing the bracketed populations. NS = non-significant. B) B cell activation was measured using expression of CD80 after stimulation of peripheral blood mononuclear cells with anti-IgM, anti-CD40, and IL-21. IgG was the mock stimulation. Stimulation induced the expression of CD80 in all groups with all but the adult patients reaching significance. The horizontal bar represents the mean and the error bars are SEM. The P values compare the bracketed populations.

3.2. B cell function

With data supporting aberrant switched memory B cell development or survival in 22q11.2DS, it was important to examine effector functions. We had previously noted diminished specific antibody production after influenza vaccination in adult patients [8]. We had further identified low levels of IgG, IgA and IgM in some patients [17, 41]. To understand whether B cells were globally compromised in antibody production, we assessed the function of B cells by stimulating with anti-IgM, anti-CD40, and IL-21 to mimic T cell help. To measure the effects of stimulation, we examined CD80 expression, a marker of activation (Fig. 2B). Stimulation led to increased expression of CD80. These data suggest that the B cells are intrinsically able to respond to stimulation.

3.3. T cell help

We hypothesized that switched memory B cells could be reduced due to impaired production of precursor cells, impaired transition from naïve B cells, increased cell death at the switched memory stage, or increased differentiation into later differentiation stages such as plasmablasts or plasma cells. Our B cell flow data did not support the concept that B cells failed to develop globally or that naïve B cells were limiting. Furthermore, normal plasmablast frequencies suggested that increased conversion from switched memory B cells into plasmablasts was not occurring. We therefore examined the T cell compartment to define defects that might be associated with compromised T cell help, known to be critical for the development of switched memory B cells. As anticipated, the CD4 T cell percentages within the lymphocyte gate were overall diminished in both children and adults with 22q11.2DS. We defined TFH cells with CD4 +, CXCR5 +, and ICOS + and found that both adult and child patients had an increased frequency of TFH cells (Fig. 3). We had hypothesized that lack of TFH cells or lack of TFH function was responsible for the switched memory deficit in adults with 22q11.2DS. To further evaluate the TFH function, we examined the active subset of TFH cells using CD4 + CXCR5 + ICOS + CCR7loPD1hi. To our surprise, the active TFH cells were expanded in 22q11.2DS in both children and adults. We measured active TFH cells using an alternative strategy (CD4 + CXCR5 + CCR7loPD1hi) and again found that the active TFH cells were expanded in patients. We noted that in control subjects, the TFH fraction was inversely related to the overall CD4 fraction of lymphocytes. In the patients, this relationship was lost (Fig. 4). We further considered whether expanded TFH could be a compensatory mechanism in response to the low switched memory B cells. We performed logistic regression analyzing each patient's switched memory B cell and TFH percentages. We found no relationship (data not shown). These analyses suggested that lack of TFH cells did not account for the deficit in switched memory B cell in 22q11.2DS patients.

Fig. 3.

Follicular helper T cells in 22q11.2DS. Flow cytometry was performed for TFH on peripheral blood samples from 27 adult patients with 22q11.2DS, 35 adult controls, 44 children with 22q11.2DS and 33 child controls for follicular helper T cell markers. TFH percentages were generally higher in patients than controls. Activated TFH (analyzed on a subset of subjects) were significantly higher in both child and adult patients compared to corresponding controls. Quantitative RT-PCR was performed for IL-21, normalizing to the 18S signal. The horizontal bar represents the mean and the error bars are SEM. The P values comparing the bracketed populations. NS = non-significant.

Fig. 4.

TFH and CD4 association. To determine if there was a relationship between the TFH percent of CD4 and the CD4 percent of lymphocytes in the peripheral blood, we used linear regression. There was a negative association in controls that was not seen in 22q11.2DS patients in either age group.

3.4. TFH function

TFH cells are thought to act predominantly by providing IL-21 and CD40L to naïve B cells [42,43]. We therefore performed qRT-PCR for IL-21 on B cell depleted PBMCs from patients and controls. The IL-21 levels were higher in adult patients, suggesting that the active TFH fraction was as capable of IL-21 production in patients as controls (Fig. 3). Therefore, TFH cells were present and producing the critical cytokine responsible for class switching. It remained a possibility that aberrant TFH engagement with naïve B cells compromised class switching, however, global production and cytokine production appeared to be intact.

3.5. B cell repertoire

At this point, B cell intrinsic function appeared to be normal and TFH production and function appeared to be largely intact. TFH function in secondary lymphoid organs could not be probed in vivo, however. As a strategy to gain insights into the influence of TFHs on B cell development, we performed next generation sequencing of the immunoglobulin heavy chain gene rearrangements from CD19-purified B cells on five adult patients, five adult controls, five child patients and five child controls. Because we could not rule out aberrant localization of TFH cells or impaired synapse formation with B cells, we sought indirect evidence of defective T cell help by comparing somatic hypermutation (SHM) in 22q11.2DS and controls. In this analysis, the frequency of SHM in the patients was significantly lower than that seen in controls (Fig. 5A). Control subjects exhibited an increase in somatically mutated IgH sequences with age, but 22q11.2DS patients did not. To determine if the lower frequency of SHM in 22q11.DS subjects was due to a lower proportion of switched memory B cells, we plotted the fraction of B cells with SHMs vs. the fraction of switch memory cells (Fig. 5B). While there was a clear association between switched memory B cell percent and mutational burden in controls, this was not seen in 22q11.2DS subjects. These data were strongly suggestive that the rate of somatic hypermutation was not a direct result of limited numbers of switched memory B cells.

Fig. 5.

IgH repertoire features in 22q11.2DS patients compared to controls. IgH sequencing was performed on B cell DNA and the VH usage, CDR3 length and number of somatic hypermutations (SHMs) were computed. A) Aggregate SHM frequencies in IgH sequences were compared between 22q11.2DS patients and controls. Each symbol represents the average percentage of unique sequences that have more than 5% SHM for a given subject. B) The data from 7A were plotted as a percentage of switched memory B cells. When the fraction of sequences with >5% SHMs was related to the percentage of switched memory B cells in the sample, a positive association was noted in controls but not in patients. Black squares represent adult values and gray circles represent child values. C) VH usage was compared in unique in-frame rearrangements of 22q11.2DS vs. control subjects. Only VH genes found in 1% or more of unique sequences of all subjects are shown. No strong associations were observed between the usage of particular VH genes and whether subjects were patients or controls. D) CDR3 length distributions of unique sequences are compared for in-frame (and fully functional—lacking stop codons) vs. out-of-frame rearrangements for controls vs. 22q11.2DS. No significant differences were noted in CDR3 length distributions between patients and controls. E) The mutation count per sequence is graphed comparing patients and controls for both children and adults. Patients have fewer mutations per sequence. The patients (red bars) are skewed towards the left, with fewer mutations per clone. At the higher end of the mutation scale, the blue bars (controls) are higher.

To evaluate selection of somatically mutated sequences, mutation frequencies in CDR2 and framework 3 (FR3) regions were counted in all unique sequences and replacement/silent (R/S) ratios were computed. Under conditions of positive selection by antigen, an elevated R/S ratio is expected in the CDRs, whereas more silent mutations are expected in the structurally constrained framework regions [44]. Mutations in both CDR2 and Framework 3 were lower in both adult and pediatric patients compared to controls (Table 2). These data implicate a subtle defect in access to T cell help for 22q11.2DS B cells.

Table 2.

Somatic mutations in CDR2 and FR3 regions of patients vs. controls.

Subject ID	Subject	Subject	CDR2	CDR2	FR3	FR3
	Age	Disease	R/S	Mutation rate	R/S	Mutation rate
DS-1	Child	22q11.2DS	4.534	50.19	2.778	23.71
DS-2	Child	22q11.2DS	4.401	46.43	2.745	22.29
DS-3	Child	22q11.2DS	4.805	61.43	2.964	28.29
DS-4	Child	22q11.2DS	4.443	57.45	2.932	26.21
DS-5	Child	22q11.2DS	5.057	56.57	3.214	25.25
DS-6	Adult	22q11.2DS	4.458	63.89	2.838	27.36
DS-7	Adult	22q11.2DS	4.389	55.69	3.16	24.46
DS-8	Adult	22q11.2DS	4.488	48.48	2.978	24.3
DS-9	Adult	22q11.2DS	4.412	53.92	3.03	24.18
DS-10	Adult	22q11.2DS	4.25	48.38	3.141	23.45
C-1	Child	Control	4.83	72.78	2.774	35.22
C-2	Child	Control	3.936	84.27	2.766	40.72
C-3	Child	Control	4.931	77.71	2.687	36.37
C-4	Child	Control	4.728	58.1	2.859	27.84
C-5	Child	Control	4.06	62.21	3.021	28.73
C-6	Adult	Control	4.462	65.17	2.764	31.12
C-7	Adult	Control	4.916	85.69	2.654	38.07
C-8	Adult	Control	4.485	78.69	2.543	39.03
C-9	Adult	Control	4.79	72.85	2.744	33.39
C-10	Adult	Control	4.602	89.19	2.61	44.8

Table 2: Somatic mutation analysis in CDR2 and FR3 of 22q11.2DS patients vs. controls. The ratio of somatic hypermutations that change the amino acid sequence (replacement) to those that do not (silent) are shown for all clonotypes sequenced within each subject within the second complementarity determining region (CDR2) and the third framework region (FR3). Also shown are the global somatic hypermutation frequencies (mutations per 1000 bp) in these regions. Calculations were performed on all clonotypes using immunoglobulin analysis tool (see Methods). Individual subjects are labeled DS for 22q11.DS and C for control. The overall SHM frequency and R/S ratios are higher in CDR2 than in FR3, consistent with grossly similar selection in patients and controls. The R/S ratios and mutation frequencies are slightly higher in CDR2 and FR3 in adult controls than in adult patients, consistent with a subtle mature B cell selection defect in 22q11.2DS patients (p < 0.05, Mann Whitney).

We further examined the repertoire data for evidence of intrinsic B cell developmental defects in immunoglobulin gene rearrangements. Since most of the circulating B cell repertoire is naïve, VH gene usage among clonotypes provides a window into early B cell selection processes. Overall VH usage was very similar between patients and controls, suggesting that early selection processes were intact in 22q11.2DS subjects (Fig. 5C). To further evaluate early selection, we analyzed the distribution of CDR3 lengths in in-frame vs. out of frame rearrangements. Here, in-frame rearrangements were defined as being in the correct reading frame and lacking stop codons. Out of frame rearrangements were not in the correct reading frame and represent sequences that are not selected, but have undergone V(D)J recombination. Under conditions of normal IgH selection, the median CDR3 length of functional rearrangements is shorter than the median CDR3 length of non-functional rearrangements in the (non-fetal) human B cell repertoire [45]. Consistent with intact IgH gene rearrangement, the CDR3 length distribution of out of frame rearrangements was comparable in 22q11.2DS patients and control subjects. Also consistent with intact early selection, CDR3 length distributions of in frame rearrangements were comparable between patients and controls, with in-frame rearrangements harboring a shorter average and median CDR3 length than out of frame rearrangements (Fig. 5D). Similar results were obtained when productive vs. non-productive rearrangements were compared (not shown). These data suggest that early developmental processes, including V(D)J recombination and initial selection of circulating naïve B cells, proceeds in a grossly normal fashion in patients with 22q11.2DS.

Therefore, there was diminished SHM in patients at all ages and B cell intrinsic development appeared intact. To solidify our thesis that T cell help was compromised, we examined the percentage of mutated bases per clone. This variable is unaffected by the percentage of switched memory B cells. As seen in Fig. 5E, patients with 22q11.2DS had fewer mutated bases per clone. Comparing all patients vs controls, the P value is 0.01, demonstrating that the mutation frequency is significantly different. This establishes compromised T cell help as a likely mechanism for the B cell dysfunction.

4. Discussion

Immune deficiency in 22q11.2DS occurs primarily because of thymic hypoplasia, with a consequence of mild to moderate deficits in T cell counts resulting in a variable spectrum of immune dysfunction [6]. Both infection and autoimmunity are seen with increased frequency in this population [13,16,46]. Increased risk of organ-specific autoimmunity reflects the complex immune dysregulation that derives from the limited T cell production and the homeostatic proliferation invoked as a result [9,47]. Effects on regulatory T cells may also occur thereby contributing to autoimmunity [48]. New data have extended the traditional view of this immune deficiency by highlighting humoral defects [16,41]. This may represent an additional pathogenic mechanism underlying the increased susceptibility to infection and autoimmunity. Humoral defects also represent a treatable source of infection susceptibility. Previous data suggested that IgM levels declined with age [17] and the growing population of adults with 22q11.2DS represents a substantial cohort at risk for humoral deficiencies. Therefore, a better understanding has critical implications for patient care and health policy because of the high and climbing frequency of 22q11.2DS in the adult population.

B cells mature and differentiate and the final immunoglobulin repertoire is shaped in the germinal center (GC). The GC reaction supports immunoglobulin class-switch and affinity-maturation and is heavily reliant on T cell help [49]. We had hypothesized that diminished humoral function would be due to compromise in the delivery of T cell help because the thymus is known to be affected in 22q11.2DS and the critical transcription factor, TBX1, is not known to be expressed in B cells. The help signals provided by TFH cells consist of both cytokines (IL-4, IL-21) and cell-surface receptors (CD40L, ICOS). In humans, CXCR5 + TFH cells are a major component of T cell memory, and when expressing low amounts of programmed cell death-1 (PD-1) they constitute the most polarized and functional TFH cells in blood [50-53].

PD-1, also a member of CD28 co-stimulatory family regulating selection and survival events, is highly expressed on activated TFH cells, and its ligands PD-L1 and PD-L2 are expressed on germinal centers B cells [54]. Since there are different activation levels of TFH cells, we focused on activated, ICOS +, CCR7lo and PD1hi TFH cells. Activated TFH cells were present in increased percentages, and the activation level and function, as seen by their ability to produce cytokine IL-21, appeared normal. Therefore, there was no clear defect in TFH production or function among circulating cells. Increased circulation of TFH cells in peripheral blood has been reported in patients with SLE, Sjogren's syndrome and juvenile dermatomyositis and they are hypothesized to contribute autoimmune diseases by facilitating the aberrant generation of autoantibodies [55,56]. Recently, a set of monogenic immune deficiencies were described with diminished TFH, suggesting that this finding of increased TFH in 22q112DS is relatively specific [57]. Autoimmunity occurs with high frequency in 22q11.2DS and elevated TFH cells may represent a mechanism that contributes to the high rate of autoimmunity.

Humoral abnormalities, low levels of immunoglobulins, and diminished specific antibody production to influenza and polysaccharide antigens as a consequence of decreased memory B cells subsets have been described in 22q11.2DS patients [7,8,16,46,58]. The mechanism has not been clear. Many if not all T cell defects are associated with B cell dysfunction, reflecting the important roles of IL-21 and CD40L in class switching and maturation of B cells. Indeed, early descriptions of “complete DiGeorge syndrome” with athymia referenced combined immune deficiency with poor immunoglobulin production [59]. Nevertheless, broader recognition of the humoral defect in 22q11.2DS is relatively recent.

Our results confirmed that the proportion of class-switched memory B cells is significantly decreased in adult patients with 22q11.2DS compared with controls. B cells were able to maintain high KRECs showing no signs of proliferative pressure or exhaustion. The normal VH usage, normal out of frame CDR3 length distribution, normal early B cell production and an intact capacity to respond to stimulation support a view where B cells are intrinsically normal. This is also consistent with the current view that B cells do not express TBX1. Our data cannot exclude defects in stromal cells affecting B cell differentiation but at this point, there are no data implicating an intrinsic B cell defect. Instead, we favor a subtle lack of T cell regulation of B cell differentiation. This conclusion is supported by the low SHM rate in patients and the known T cell compromise in this syndrome. We found that the total mutation rate was decreased and the number of mutations per clone was decreased in patients. This latter finding eliminates skewing due to limited numbers of class switched B cells for sequencing. Other immune deficiencies associated with impaired T cell help such as class switch deficiencies, have more obvious blocks in B cell differentiation. In 22q11.2DS, there appears to be a mild compromise in T cell help or B cell access to T cell help.

Unusual aspects of the B cell dysfunction in 22q11.2DS are that it is highly variable, increases with age, and that the B cell phenotype is more pronounced than the effect on immunoglobulin production. The germinal center reaction is self-reinforcing with T–B communication in both directions. In settings where T cell help is suboptimal, the consequences may accrue gradually. Existing long-lived plasma cells may support serum immunoglobulin levels for prolonged periods of time, potentially maintaining relatively normal secreted immunoglobulin levels in 22q11.2DS patients. Thus it is possible that while the switched memory B cell defect may be significant, the downstream secreted immunoglobulin phenotype remains mild until memory cell precursors to plasma cells are depleted and/or plasma cell homeostasis is deranged.

The immunological features of 22q11.2DS are heterogeneous, and for the most part the T cell impairment is well described. We report here that dysfunction in the humoral immune system involving the memory B cell compartment may represent an additional pathogenic mechanism underlying the increased susceptibility of 22q11.2DS patients to infections and autoimmunity. Our data have confirmed diminished switched memory B cells in 22q11.2DS. The new sequencing data support a defect in T cell help, which is consistent with our understanding of the etiology of this syndrome. Weaknesses in our study include a small sample size for B cell activation and sequencing and lack of longitudinal data. There is a critical need to understand biomarkers of impending antibody failure. The high and growing frequency of this syndrome, the aging of patients and the impact of humoral dysfunction all mandate a greater understanding.