A subset of patients with common variable immunodeficiency (CVID), group 1a of the Freiburg classification, is characterized by increased B cells expressing low levels of CD21 (CD21low), lymphoproliferation and autoimmunity. The CD21low B cells have been shown to be profoundly anergic, and defects of BCR-mediated calcium signaling and of T cells have been described in CVID 1a. We found that also the classical naïve B cells from CVID 1a patients, but not from CVID non-1a patients, proliferated poorly. The B cells of CVID 1a patients had a reduced capacity to divide reminiscent of the proliferative arrest associated with replicative senescence. Thus, we investigated whether lymphocyte dysfunction in CVID 1a was related to telomere-dependent replicative senescence, and found that both the B and the T cells from CVID 1a patients had significantly shorter telomeres compared with B and T cells from CVID non-1a patients. Telomere lengths in B and T cells were significantly correlated, indicating that the rate of telomere attrition in lymphocytes is an individual characteristic of CVID patients. Our findings suggest that telomere-dependent replicative senescence contributes to the immune dysfunction of CVID 1a patients, and may provide an important clue for a better understanding of the pathogenesis of CVID.
Common variable immunodeficiency (CVID) is a heterogeneous disorder characterized by reduced antibody levels with low to normal numbers of circulating B cells 1. Only a small proportion of CVID patients can be classified by the underlying genetic defect 2, and current classifications are mainly based on the B-cell phenotype 3, 4. A subset of CVID is hallmarked by the expansion of an unusual population of B cells with reduced expression of CD21 (CD21low); these patients are classified as CVID group 1a according to the Freiburg study 3. Clinically, CVID 1a is characterized by a late onset, autoimmunity and benign polyclonal lymphoproliferation 3, 4. In addition, several T-cell abnormalities have been described in CVID 1a patients 5.
The CD21low B cells of CVID patients express an array of inhibitory receptor and fail to proliferate in response to BCR-dependent and -independent stimuli 6, 7. However, anergy is not limited to CD21low B cells, since all mature B cells from CVID 1a patients fail to flux calcium upon BCR triggering 8. CD21low B cells are also increased in patients with HIV infection 9, systemic lupus erythematosus 10, or mixed cryoglobulinemia secondary to HCV infection 11, 12, all conditions characterized by B-cell hyperactivation. Although the CD21low B cells of patients with HIV infection or HCV-associated mixed cryoglobulinemia have mutated immunoglobulin genes 10–12, the CD21low B cells of CVID 1a patients are unmutated and mostly autoreactive 6, 7. Similarly to those of patients with CVID, the CD21low B cells of patients with HIV infection are profoundly anergic to B-cell stimuli 9.
In this study, we observed that the B cells from CVID 1a patients proliferated poorly in response to B-cell stimuli compared with the B cells from CVID non-1a patients, and that also the classical naïve B cells of CVID 1a patients failed to proliferate in response to BCR-dependent stimuli. In addition, we observed that the B cells of CVID 1a patients had a reduced capacity to divide, reminiscent of the proliferative arrest associated with replicative senescence. Thus, we investigated telomere-dependent replicative senescence in lymphocytes from CVID 1a patients. We found that the mean length of telomeres was significantly reduced in the B and T cells from CVID 1a patients compared with those from CVID non-1a patients and from healthy donors. Our findings indicate that accelerated telomere-dependent replicative senescence is associated with the functional defects of B and T lymphocytes in CVID 1a.
B-cell and T-cell abnormalities in patients with CVID 1a
Both T and B cells have been reported to be dysfunctional in patients with CVID 1a 5–8. We could confirm in our series the previous observation 5 of a significant reduction of naïve CD4+CD45RA+ T cells in CVID 1a patients compared with CVID non-1a patients (CVID 1a: n=22, mean±SD 24.9±16.7%; CVID non-1a: n=16, 36.2±19.2%; p=0.03 by Mann–Whitney U-test), and a parallel increase of memory CD4+CD45R0+ T cells (p=0.04).
The function of CD21low B cells expanded in CVID 1a patients has been previously investigated in depth 6, 7. These cells are profoundly anergic to B-cell stimuli, and display an array of inhibitory molecules that are likely to induce their unresponsiveness 6, 7. However, defects of calcium influx upon BCR triggering have been described in all B-cell subsets of CVID 1a patients except in their transitional B cells 8. We observed that the percentage of B cells entering proliferation (precursor cohort) after stimulation of TLR7 or TLR9 was reduced in CVID 1a patients compared with CVID non-1a patients and healthy donors (Fig. 1A). However, both CVID 1a non-1a CVID patients appear to have a differentiation defect, reflected by a reduced generation of plasmablasts compared with healthy donors (Fig. 1A). To discern whether the proliferative defect of CVID 1a B cells was restricted to the CD21low population, we investigated the proliferative responses of purified B-cell subsets. As previously reported 9, 10, the naïve-like CD21lowCD27− B cells were profoundly anergic to the stimulation of TLR9 with CpG and to a combination of BCR-dependent and T cell-dependent stimuli (anti-Ig, CD40L and IL4) (Fig. 1B). However, we observed that also the classical naïve (CD21+CD27−CD38+) B cells of CVID 1a patients proliferated poorly in response to a combination of BCR- and T-cell-dependent stimuli (Fig. 1B). Anergy of classical naïve B cells from CVID 1a patients is possibly related to their reduced Ca2+ influx upon BCR stimulation 8, as well as to their low-level expression of inhibitory and transcription factors [e.g. Fc-receptor-like 4 (FCRL4) and SOX5] characteristically overexpressed by CD21low B cells 6.
The CD21low B cells expanded in CVID 1a and in other conditions, such as HIV 6 or HCV 11, 12 infection, are either naïve-like CD27− or memory-like CD27+. However, only the phenotype and function of CD21lowCD27− B cells has been investigated in detail 7–9, and the CD21lowCD27+ B cells have been proposed as precursors of short-lived plasmablasts rather than exhausted cells as their CD27− counterparts 13. We further investigated the phenotype of CD21lowCD27+ B cells, and found that they share with CD21lowCD27− B cells the expression of a peculiar array of receptors (Fig. 2) associated with anergy and homing 6, 9, thus suggesting similar functional properties.
Collectively, our findings are consistent with the concept 5, 8 that multiple subsets of B and T cells are dysfunctional in CVID 1a patients.
Reduced replicative potential of B cells from CVID 1a patients
We observed that most of the B cells from CVID 1a patients activated in vitro by CpG underwent only few rounds of division compared with B cells from healthy donors (Fig. 3). Since differentiation to the plasmablast stage requires a critical number of cell divisions 14, it is likely that the low proliferative capacity of CVID 1a B cells prevented their differentiation. Indeed, the few plasmablasts generated by CVID 1a B cells derived from the few B cells that reached the threshold number of four divisions (Fig. 3). The limited capacity of B cells from CVID 1a patients to divide is reminiscent of the cell-cycle arrest called replicative senescence 15, which results from an excessive telomere shortening 16, 17. This prompted us to investigate telomere length in lymphocytes from CVID 1a patients.
Telomeres are shortened in B and T lymphocytes from CVID 1a patients
In agreement with the previous studies 18, 19, we observed that telomere length in B and T cells from healthy donors decreased with age (Fig. 4A); thus, telomere lengths in CVID patients were compared with those in healthy donors of the same or older age. The B cells from CVID 1a patients had, on average, significantly shorter telomeres than those from healthy donors or from CVID non-1a patients (Fig. 4B). No correlation was observed between the mean telomere length in whole B cells from CVID 1a patients and the frequency of CD21low B cells, suggesting that telomere attrition was independent on B-cell phenotype. Indeed, very short telomeres (8.5 kb) were observed in a patient with a large predominance of CD21+CD27− classical naïve B cells (72%) over CD21low B cells. To directly address this issue, we FACS purified CD21+CD27− classical naïve, CD21lowCD27− naïve-like, and CD21lowCD27+ memory-like B cells from a CVID 1a patient, and found that the mean telomere length was similarly reduced in these subpopulations (9.4, 9.2 and 9.3 kb, respectively) compared with whole normal B cells (range, 9.5–11.6 kb). Collectively, our findings point to increased telomere attrition in CD21low as well as in classical naïve B cells of CVID 1a patients, paralleling our observation in these patients of a proliferative defect to B-cell stimuli involving both B-cell types.
In addition to B cells, the T cells from CVID 1a patients also had significantly shorter telomeres than those from healthy donors or CVID non-1a patients of comparable age (Fig. 4B). Since telomeres are longer in human naïve rather than memory CD4 or CD8 T cells 16, 20, reduced telomere length in whole T cells from CVID 1a patients could result from the depletion of naïve CD4 T cells in these patients 5. However, we found no significant correlation between the average telomere length in whole T cells and the relative proportions of naïve and memory CD4 T cells in CVID patients. Thus, we investigated FACS-purified T-cell subpopulations from three CVID 1a patients. The strategy for cell sorting and telomere length analysis of T-cell subpopulations from a representative CVID 1a patient is shown in Supporting Information Fig. 1. We found (Fig. 5) that in CVID patients, similarly to normal individuals 16, 20, telomeres were longer in CD4 naïve compared with CD4 memory or CD8 T cells. However, telomere length in CD4 naïve T cells from two out of the three CVID 1a patients was significantly lower than that in T cells from healthy donors. Thus, the overall telomere shortening in T cells from CVID 1a patients results both from a reduced telomere length in CD4 naïve T cells and from a relative increase of CD4 memory T cells.
Finally, we observed that in CVID patients the mean telomere length in B cells was correlated with the mean telomere length in T cells (Fig. 4C). Thus, our findings suggest that the rate of telomere shortening in T and B cells is an individual characteristic of CVID patients, not correlated with the relative predominance of different lymphocyte subsets or with age.
Our study shows that the B and T cells from CVID patients with increased proportions of CD21low B cells, group CVID 1a in the Freiburg classification 3, display accelerated telomere-dependent replicative senescence. A previous report described overall normal telomeres in T and B cells from CVID patients 21; however, the mean telomere lengths varied greatly from patient to patient. Since patients were not classified according to the immunophenotype, the data reported in that study appear consistent, rather than contradictory, with distinct subgroups of CVID patients with different rates of telomere attrition.
Shortening of telomeres in B cells as well as in T cells of CVID 1a patients is consistent with the notion that both cell types are dysfunctional in this subgroup of CVID 5–8. The mean telomere length in whole B- and T-cell populations was independent on the relative predominance of specific subsets such as CD21low B cells or memory T cells. Rather, the linear relationship between the mean telomere lengths in T and B cells of CVID patients indicates that the rate of telomere attrition is a distinctive patient characteristic.
Accelerated telomere shortening in immune cells has been observed in some pathological conditions. In ataxia-telangiectasia, a progeric immunodeficiency disorder, telomere attrition is caused by the defects in genetic mechanisms that ensure telomere maintenance 22. A cause of accelerated telomere shortening in lymphocytes, independent on genetic background, is chronic stimulation by infectious agents, as for example cytomegalovirus 20, HCV 23 or HIV 24 infection. Patients with CVID are continuously stimulated by an array of microorganisms that chronically infect the respiratory tract and the intestine. Thus, it is possible that more intense infectious stimuli drive, in certain CVID patients, accelerated telomere shortening and the development of the clinical and immunological profile characteristic of CVID 1a. In this regard, it is of interest that patients chronically stimulated by HCV or HIV have an expansion of CD21low B cells 9, 11, 12. However, this possibility is made unlikely by the fact that CVID 1a patients are characterized by idiopathic lymphoproliferation and autoimmunity, but not by a higher rate of infections than CVID non-1a patients 25. Nevertheless, it is possible that still unrecognized defect(s) in the telomere maintenance machinery make CVID 1a patients more susceptible to the accelerated shortening of telomeres driven by chronic infection.
An appealing clue to explain telomere attrition in CVID 1a may arise from the recent observation of defective BCR-induced Ca2+ influx in all mature B cells of these patients 8. Ca2+ is a ubiquitous second messenger controlling several cellular functions including growth, differentiation and, as highlighted by recent studies, cell senescence 26. Mitochondrial dysfunction gives rise to a Ca2+-mediated retrograde response that may be, in part, involved in telomere-dependent senescence 27. Moreover, Ca2+ influx from extracellular space positively regulates NFAT, which in turn controls cell growth and proliferation 26. Interestingly, NFAT positively regulates the transcription of the telomerase reverse transcriptase (TERT) gene, and NFAT silencing down-regulates TERT mRNA expression 28. A defect of extracellular Ca2+ influx can, therefore, result in NFAT silencing, reduced TERT expression and defective maintenance of telomeres 29. Thus, the upstream defect of BCR-induced Ca2+ influx in CVID 1a patients 8 could cause pathologic telomere shortening in all mature B-cell populations. Concerning T cells, a defect of Ca2+ influx in response to TCR triggering has been described in a subset of CVID patients with impaired TCR-mediated proliferative responses 30, 31. This subset may constitute up to 60% of CVIDs 30, but its identity with CVID 1a has not been determined. Investigating TCR-induced calcium signaling in CVID 1a patients will help to clarify whether a defect of Ca2+ influx may also play a role in the dysfunction and, possibly, telomere attrition of T cells in these patients.
Our findings suggest that telomere-dependent replicative senescence is involved in the pathogenesis of CVID 1a. The analysis of telomere length in lymphocytes may provide a clue for a better understanding of the pathogenesis in distinct subgroups of CVID patients, and an additional biomarker for characterizing primary immunodeficiency disorders.
Materials and methods
A total of 44 CVID patients (25 female/19 male, age 15–84, median 52 years), and 25 healthy subjects were studied. All patients fulfilled the criteria for CVID based on the International Union of Immunological Societies definition 1; 23 patients were classified as group 1a (>20% CD21low B cells among B cells) and 21 as group non-1a (<20% CD21low B cells) according to the Freiburg classification 3. All subjects provided informed consent, in accordance with the Institutional Review Boards of the Sapienza University of Rome and of the University Hospital Freiburg, and with the Declaration of Helsinki.
Flow cytometry and cell sorting
PBMC were obtained by density-gradient centrifugation. Immunophenotyping was performed with the combinations of fluorochrome-labeled monoclonal antibodies, all obtained from BD Biosciences except for anti-human FCRL4 32 that was provided by G.R.A. Ehrhardt, Atlanta, GA. Unlabeled anti-FCRL4 was counterstained with FITC-or PE-conjugated goat anti-mouse IgG, (BD Biosciences), using mouse IgG as control. FACS analyses were performed on a FACSCalibur instrument (BD Biosciences) using the CellQuest (BD) and FlowJo (Tree Star) software.
Whole B- and T-cell populations were negatively purified (>95%) from PBMC by immunomagnetic sorting (Dynal). For obtaining B cells, CD2+, CD14+ CD16+, CD36+, CD43+ and CD235a+, non-B cells were depleted, whereas for obtaining T cells HLADR+, CD56+ and CD16+ non-T cells and activated T cells were depleted. B- or T-cell subpopulations were purified (>95%) from the PBMC of healthy donors and CVID 1a patients by FACS sorting, using a MoFlow (DakoCytomation) or a FACSAria (Becton-Dickinson) instrument. The sorted B-cell populations were classical naïve (CD19+CD27−CD21+CD38+), classical memory (CD19+CD27+CD21+), naïve-like CD21low (CD19+CD27−CD21−/lowCD10−CD38−) and memory-like (CD19+CD27+CD21−/lowCD10−CD38−) B cells. Classical memory B cells could be obtained only from healthy donors since patients had only minimal percentages of these cells, whereas naïve-like CD21lowCD27− and memory-like CD21lowCD27− B cells could be obtained only from CVID 1a patients. The sorted T-cell populations were CD4 naïve (CD45RA+), CD4 memory (CD45R0+) and CD8+ T cells.
Cell proliferation and differentiation
Cell proliferation was measured by the CFSE dilution method 33. PBMCs were labeled with CFSE (Invitrogen) and cultured at 2×105 cells/well in 96-well U-bottom plates in the absence or presence of the TLR9 ligand CpG (Sigma Genosys; 2.5 μg/mL) or of the TLR7 ligand R848 (Invivogen; 0.25 μg/mL). Stimulation with R848 was routinely done in the presence of IFN-α (1000 U/mL; Schering), in order to investigate TLR7 function independently on possible defects of IFN-α production by CVID plasmacytoid B cells 34. Cell proliferation was measured at day 5 of culture by flow cytometry, and cell division rates were estimated as described by Lyons 33. Briefly, the precursor population in each CFSE peak was calculated by dividing the number of cells in each peak by 2i, where i represents the division number of the CFSE peak. The sum of the precursor population from the first division represents the number of original, undivided precursor cells entering proliferation (precursor cohort).
Proliferation of FACS sorted B-cell populations was measured by 3H-thymidine incorporation. Briefly, sorted naïve and CD21low B cells were seeded in duplicates at 50 000 cells/well. Cells were cultured for 4 days in the presence or absence of 12.5 μg/mL anti-IgM (Jackson ImmunoResearch Laboratories), 100 U/mL IL-4 (BD Pharmingen), 5 μg/mL anti-CD40 (Alexis Biochemicals, San Diego, CA). One μCi of tritiated (3H)-thymidine (Amersham) was added to each well on day 3. Incorporated 3H-thymidine was measured by a 96-well scintillation counter (β-Counter Matrix, Packard).
Differentiation of IgM B cells to plasmablasts was assessed by flow cytometric analysis of cells permeabilized with Permeabilizing-Solution 2 (BD Biosciences) and stained with anti-IgM, anti-CD20 and anti-CD38 antibodies; plasmablasts were identified as CD20low/negCD38+ cells with high cytoplasmic IgM content (cIgMhigh).
Measurement of telomere length
Telomere length was measured by a modification 35 of the flow-FISH assay originally described by Rufer et al. 36. A telomere-specific Cy5-conjugated protein nucleic acid (PNA) probe Cy5-OO-(CCCTAA)3 (20 nM; Panagene) was used for detecting telomere sequences, and ethidium bromide (EB, μg/mL; Sigma-Aldrich) for determining total DNA content. Briefly, 3×105 purified B or T cells were mixed with 3×105 cells of the human B-cell line Ramos (ECACC cat. no. 85030802). The known mean telomere length of Ramos cells at G1/G0, approximately 17 kilobase (kb), was used as internal standard to calculate the mean telomere length in lymphocytes cells by comparing the respective MFIs.
Statistics were done using the StatView 5.0.1 software (SAS Institute). A p-value <0.05 was considered significant.
The authors are grateful to the patients for their willingness to participate in our study, and to the Associazione Italiana Immunodeficienze Primitive (AIP) for long-lasting support. The authors thank Goetz Ehrhardt (Emory University, Atlanta, GA) for providing anti-FCRL4 and for helpful discussion. The study was supported by grant HEALTH-F2-2008-201549 from EURO-PADnet (European Primary Antibody Deficiency network).
Conflict of interest: The authors declare no financial or commercial conflict of interest.