Correlation between recent thymic emigrants and CD31+ (PECAM-1) CD4+ T cells in normal individuals during aging and in lymphopenic children



CD31+CD45RA+RO lymphocytes contain high numbers of T cell receptor circle (TREC)-bearing T cells; however, the correlation between CD31+CD4+ lymphocytes and TREC during aging and under lymphopenic conditions has not yet been sufficiently investigated. We analyzed TREC, telomere length and telomerase activity within sorted CD31+ and CD31 CD4+ lymphocytes in healthy individuals from birth to old age. Sorted CD31+CD45RA+RO naive CD4+ lymphocytes contained high TREC numbers, whereas CD31+CD45RARO+ cells (comprising ⩽5% of CD4+ cells during aging) did not contain TREC. CD31+ overall CD4+ cells remained TREC rich despite an age-related tenfold reduction from neonatal (100 : 1000) to old age (10 : 1000). Besides a high TREC content, CD31+CD45RA+ROCD4+ cells exhibited significantly longer telomeres and higher telomerase activity than CD31CD45RA+ROCD4+ cells, suggesting that CD31+CD45RA+ROCD4+ cells represent a distinct population of naive T cells with particularly low replicative history. To analyze the value of CD31 in lymphopenic conditions, we investigated six children after allogeneic hematopoietic stem cell transplantation (HSCT). Reemerging overall CD4+ as well as naive CD45RA+ROCD4+ cells predominantly expressed CD31 and correlated well with the recurrence of TREC 5–12 months after HSCT. Irrespective of limitations in the elderly, CD31 is an appropriate marker to monitor TREC-rich lymphocytes essentially in lymphopenic children after HSCT.


anchored return primer


acute lymphoblastic leukemia


anti-T cell globulin


cord blood


hematopoietic stem cell transplantation


recent thymic emigrant


telomeric repeat amplification protocol


T cell receptor excision circle


terminal restriction fragment


telomerase primer


As the productive capacity of the thymus decreases during progressive physiological involution, the numbers of episomal T cell receptor excision circle (TREC)-bearing T cells and recent thymic emigrants (RTE) gradually decline 13. Overall T cell levels are maintained despite thymic involution by a variety of peripheral thymus-independent homeostatic mechanisms 47. One of the best ways to assess RTE is the direct analysis of TREC within mononuclear or CD4+ T cells. TREC are stable DNA episomes formed during TCR rearrangement, which are not replicated and become diluted by successive cell divisions 1, 8, 9. TREC numbers among CD4+ T cells decrease 50–100 times during aging 1. Especially under lymphopenic conditions, absolute numbers of TREC can be influenced by dilution factors, e.g. by peripheral expansion of naive and memory T cells, e.g. after hematopoietic stem cell transplantation (HSCT) 8, 9, or by unoccupied niches, e.g. in HIV-infected patients after highly active antiretroviral therapy (HAART) 7, 10, 11. The turnover of TREC and RTE is much faster than that of established naive cells, indicating that their incorporation into the general naive population is variable and regulated 1113. Recently, CD31 (PECAM-1), a 130-kDa transmembrane glycoprotein expressed by endothelial cells, platelets, monocytes, neutrophils, and certain T cell subsets, has been proposed as a cell surface marker expressed preferentially by naive TREC-rich T cells that have undergone a low number of cell divisions 14, 15. CD31 is down-regulated on about 50% of the CD8+ and the majority of CD4+ lymphocytes upon their transition to the memory phenotype 16, 17. A progressive decrease of percentages and absolute numbers of naive CD31+CD45RA+ ROCD4+ T cells has been found associated with aging, bearing analogy with the decline of TREC, while absolute numbers of naive CD31CD45RA+ROCD4+ T cells remain fairly stable throughout life 14, 15. In addition, CD31 is likely to be down-regulated during homeostatic expansion of naive T cells after ligation of the TCR with specific MHC/self peptide ligands within secondary lymphoid organs 14. Hence, naive CD31CD45RA+CD4+ T cells are characterized by a drastically reduced TREC content and a restricted oligoclonal TCR repertoire 15. Notwithstanding that the age-related dynamics of CD31 expression on CD4+ T cells has as yet not been satisfactorily investigated, CD31 was rapidly adopted in the clinical practice 1820. Our study was initiated to analyze the correlation between TREC-rich RTE and CD31-expressing CD4+ T cells during aging and in children after T cell-depleting therapies.


TREC in CD4+ T cells are almost exclusively confined to CD31+ cells

We compared the TREC content of sorted CD31+ and CD31 naive CD45RA+ROCD4+ T cells in four individuals (age: 30–42 years) (Fig. 1). Purity controls revealed between 3.7 and 16.8% “contaminating” CD31+ cells within the CD31 sorted subset. Despite these CD31+ contaminants, the TREC content in CD31+CD45RA+ROCD4+ T cells was on average 18 times higher than in CD31CD45RA+ROCD4+ lymphocytes (range 9.8–26.8).

Figure 1.

TREC in CD31-sorted T cell subsets. FACS-sorted CD31+CD45RA+ROCD4+ and CD31CD45RA+ROCD4+ T cells of four adult donors (age: 30–42 years). The purity of cell fractions (in %) is given below the graph. Fractions enriched for CD31 cells were “contaminated” with CD31+ cells (3.7–16.8%). The differences in TREC content are expressed as the ratio of the TREC copy number in CD31+ cells divided by the copy number in CD31 cells.

CD31 expression in subsets of CD4+ T cells

We analyzed CD31 expression on three subpopulations: naive CD45RA+RO cells (gate R2), memory CD45RARO+ cells (gate R7) and intermediate CD45RAlowROlow cells (gates R3 to R6; Fig. 2A) of ten healthy donors (age: 27–57 years). CD45RARO+CD4+ T cells expressing CD31 (gate R7) remained exceedingly rare (⩽5%) in all analyzed individuals. Intermediate CD45RAlowROlow cells represented about 27% of CD4+ T cells in the analyzed samples, irrespective of the donor's age (mean 26.85 ± 5.84%). The mean fluorescence intensity reflecting the density of CD31 at the cell surface was similar in all gates.

Figure 2.

Percentages of CD31+ cells in T cell subsets and in individuals with the CD45 C77G polymorphism. (A) CD4+ T cells (gate R1) were analyzed for CD31 expression according to their expression level of CD45RA and CD45RO. Percentages of CD31+ and CD31 cells are shown in the quadrants of the dot plots corresponding to the gates R2 to R7. The figure shows the results for one donor representative of all ten subjects analyzed. (B) On the left, the normal pattern of CD45RA/RO isoform expression of CD4+ T cells. On the right, the expression pattern in a donor heterozygous for the CD45 C77G polymorphism.

Two donors with the C77G polymorphism on the CD45 gene

Using PCR and restriction analysis of the amplification product, two individuals were identified as heterozygous for the C77G polymorphism on the CD45 gene (data not shown). Individuals with the C77G polymorphism are unable to splice out exon A, and memory/effector lymphocytes continue to express both CD45RA and CD45RO isoforms (Fig. 2B) 21. Both donors were excluded from analysis.

CD31+CD4+ cells and their TREC content during aging

Eleven cord blood (CB), ten pediatric and 26 adult volunteers were analyzed for percentages of CD31-expressing CD4+ T cells. In CB, CD31 was expressed on more than 80% of CD4+ T cells. In donors aged 50 or older, CD31 was present in only about 30–40% of the CD4+ T cells. A markedly decreased frequency of CD31 expression was already apparent at a young age. Infants below 1 year of age expressed only 60% CD31+CD4+ T cells, and children of 2–3 years of age already presented CD31 percentage values similar to those found in adults. During aging, CD31+ expression of CD45RARO+ memory cells never exceeded 5%, except for two CB samples (7 and 9%, respectively) (n = 31; Fig. 3A).

Figure 3.

Evolution of TREC content and CD31 expression during aging. (A) Percentages of CD31 expression on overall CD4+ T cells are depicted as black rhombi (n = 47). Percentages of CD31+CD45RO+RACD4+ cells (white rhombi; n = 31) did not exceed 5%, except for two CB samples. Time point 0 refers to CB samples. The inset in the upper right corner of the graph highlights the sharp decline of CD31+CD4+ T cells in children from birth to 2 years of age (n = 21). (B) The TREC copy number/1000 CD31+CD4+ T cells from birth to old age (n = 22). (C) Correlation between percentages of CD31 expression on overall CD4+ T cells (R2 = 0.6881; white quadrangles) and CD45RA+ROCD4+ T cells (R2 = 0.5037; black quadrangles) with TREC/1000 PBMC (n = 16). (D) Absolute numbers of CD31+ on overall CD4+ T cells (white quadrangles; R2 = 0.7557) and CD31+CD45RA+ROCD4+ T cells (black quadrangles; R2 = 0.7587) from healthy infants and children (n = 16; age: 0–15 years).

TREC within CD31+CD4+ T cells (n = 22) decreased as the donors’ age increased. We found an on average tenfold decrease from 100 copies per 1000 cells in CB to 10 copies per 1000 cells at the age of 69 (R2 = 0.5636) (Fig. 3B). TREC values varied between subjects, but TREC values in individuals aged 50 or older differed significantly from the values in middle-aged individuals or newborns (p <0.05).

The correlation of CD31 percentages expressed by overall CD4+ T cells and by naive CD45RA+ROCD4+ T cells, respectively, with TREC/1000 PBMC was slightly superior if measured on overall CD4+ T cells (n = 16; R2 = 0.6882 vs. 0.5047; Fig. 3C).

Absolute numbers of CD31-expressing CD4+ T cells were measured in healthy children (n = 16; age: 0–15 years) and reached adult levels (400–600 cells/µL) 17 after the first 3 years of life. CD31+CD45RA+ ROCD4+ T cells comprised the majority of CD31+ overall CD4+ T cells (Fig. 3D). The correlation between absolute numbers of CD31+ overall CD4+ T cells and naive CD31+ CD45RA+ROCD4+ T cells in children was excellent (R2 = 0.9881; data not shown).

Telomere length in CD31-divergent T cells

We sorted CD31+CD4+ and CD31CD4+ T cells from three healthy human volunteers (age: 29, 36, 37 years), analyzed the terminal restriction fragment (TRF) size by Southern blot and determined the relative length by PCR. In all three donors, the TRF size was significantly longer in CD31+ cells than in CD31 cells. The mean TRF size of CD31+ cells was 7.45 ± 1.5 kb while that of CD31 samples was 6.29 ± 0.69 kb. Corresponding results were obtained when Southern blot and the PCR-based technique were compared (data not shown).

Since the PCR method required substantially less DNA input (∼10–3-fold less), we were able to compare the telomere length and TREC content within purely (>95%) sorted CD31+ and CD31 naive CD45RA+ ROCD4+ T cells of three blood donors (age: 25, 30, 60 years). TREC within CD31 naive CD45RA+ ROCD4+ T cells were shown to be below detection limits. TREC were only demonstrated in the CD31+ population, and their content differed significantly between the youngest (25 years: 25 TREC/1000 cells) and the older individuals (30 years: 1.14 TREC/1000 cells; 60 years: 1.08 TREC/1000 cells). Telomere lengths were significantly longer within the CD31+ population (Fig. 4) while telomere lengths decreased with age in both CD31+ and CD31 subsets (data not shown). A similar age-related decline of telomere length was observed in unseparated whole blood samples of the same individuals (data not shown). Furthermore, we obtained enough cells from the 60-year-old donor to sort for CD31+CD45RA+RO, CD31CD45RA+RO, CD31+CD45RARO+ and CD31CD45RARO+ cells with >95% purity. TREC within sorted CD31 and CD31+ memory CD45RARO+CD4+ T cells were shown to be below detection limits. CD31 naive CD45RA+ROCD4+ T cells exhibited longer telomeres than CD31+ memory CD45RO+RACD4+ T cells. The shortest telomeres were found in the CD31 memory CD45RO+RACD4+ T cell subset (Fig. 4).

Figure 4.

TREC content and telomere length of CD31-sorted CD4+ T cells according to their CD45RA/RO isoform expression. TREC content and telomere length of sorted CD31+CD45RA+ ROCD4+ (filled bars) and CD31CD45RA+ROCD4+ cells (striped bars) from three adult individuals (age: 25, 30, 60 years). In addition, TREC content and telomere length of sorted CD31+CD45RARO+CD4+ (dotted bars) and CD31CD45RA RO+CD4+ (brick-filled bars) cells from a 60-year-old volunteer. TREC numbers are given as TREC/1000 sorted cells. Experimental n was between 3 and 5. The confidence interval was calculated for n – 1 degrees of freedom and alpha (95%); bd: below detection limit with Ct values >40. Relative telomere length calculations were based on the individual's CD31+CD45RA+ROCD4+ cell population (= 1).

Telomerase activity in CD31-divergent T cells

Telomerase activity within sorted CD31+ and CD31CD4+ T cells was analyzed in 12 healthy subjects from birth to old age. Telomerase activity was variable between different individuals. The difference of mean telomerase activity between CD31+ and CD31CD4+ T cells within each individual was significant and remained comparably different during aging. Telomerase activity was consistently (on average sixfold) higher in CD31+ compared to CD31CD4+ T cells, irrespective of the donor's age (positive sign test) (Fig. 5A). After further sorting of the naive CD45RA+ROCD4+ subset for CD31 (two volunteers: age 25 and 44 years), CD31+ cells showed a significantly higher telomerase activity than the CD31 population (80 vs. 38% and 87 vs. 47%, respectively), irrespective of age (Fig. 5B).

Figure 5.

Telomerase activity in CD31-sorted CD4+ T cells. (A) Relative telomerase activity in CD31+CD4+ (striped bars) and CD31CD4+ T cells (dotted bars). Cells were isolated from four CB samples (age: 0) and eight samples of peripheral blood from healthy adult volunteers (age: 30–69 years). Enzyme activity was normalized to equal amounts of nuclear protein input and expressed as arbitrary units (AU) relative to the telomerase activity of 1000 HL60 cells. Mean activities (± SD) for the newborn and the adult age groups are shown on the right. (B) Relative telomerase activity of sorted naive CD31+CD45RA+ROCD4+ (grey bars) and naive CD31CD45RA+ RO-CD4+ T cells (striped bars) from two adult volunteers (25 and 44 years) (positive control = human kidney/293 cells = 100%).

Correlation between CD31+CD4+ T cells and TREC after HSCT

Six children were prospectively analyzed after allogeneic HSCT. Four children received matched unrelated (patients A, B, C, E) and one patient matched sibling bone marrow (patient F) whereas one patient had haploidentical peripheral stem cell transplantation after CD34 positive selection (CliniMACS; Miltenyi) (patient D). All patients except one (patient F) received in vivo T cell depletion with high-dose anti-T cell globulins (ATG): ATG-G (Thymoglobuline; Genzyme; 10 mg/kg; patient A), ATG-F (ATG-Fresenius; 60 mg/kg; patients B, C, D), and Alemtuzumab (MabCampath; 1 mg/kg; patient E). No infectious complications or graft-versus-host disease (GVHD) occurred. Donor chimerism remained 100% donor in all patients. Patient D died due to ALL relapse at month +6. Absolute numbers of TREC/µL were detectable for the first time at very low numbers at month +5/6 after HSCT (Fig. 6A) and increased steadily until month +12. Absolute numbers of reemerging CD31+CD45RA+RO CD4+ (Fig. 6B) and CD31+CD4+ T cells (Fig. 6C) followed a similar pattern. Until month +12, CD31+CD45RARO+CD4+ memory T cells showed a slow but modest increase after HSCT (<75 cells/µL; Fig. 6D). Fairly low numbers of CD31+CD45RAlowROlow CD4+ intermediate T cells (<50 cells/µL) were observed (data not shown). Absolute numbers of TREC/µL correlated with CD31+CD4+ (R2 = 0.8447) and C31+CD45RA+RO CD4+ T cells (R2 = 0.9248) between months +5 and 12 after HSCT (Fig. 6E).

Figure 6.

CD31+ naive and overall CD4+ T cells correlate with the recurrence of TREC in six children after HSCT. Course of reemerging TREC, CD31+ naive and overall CD4+ T cells in patient A (white squares), patient B (white triangles), patient C (black squares), patient D (black circles), patient E (black triangles) and patient F (white circles) during 12 months after myeloablative HSCT. (A) Absolute counts of TREC/µL; (B) CD31+CD45RA+ROCD4+ cells/µL; (C) CD31+CD4+ cells/µL; (D) CD31+CD45RO+RACD4+ cells/µL; (E) correlation between absolute cell counts/µL (white squares = CD31+CD4+ T cells, R2 = 0.8447; black squares = CD31+CD45RA+ROCD4+ T cells, R2 = 0.9248) and TREC/µL.


In our initial experiments, the average TREC content in sorted CD31+CD45RA+ROCD4+ T cells was 18 times higher than in CD31CD45RA+ROCD4+ T cells (Fig. 1), confirming the strong correlation between CD31-expressing naive CD4+ cells and the presence of TREC that has been reported previously in healthy middle-aged adults 14, 15, 22. More precise sorting with very high purity (>95%) reveals that CD31CD45RA+ ROCD4+ T cells are actually TREC negative (Fig. 4).

Here, we show that CD31+ memory CD45RO+ RACD4+ T cells are rarely found and contribute to only up to 5% of CD4+ T cells in peripheral blood of normal individuals throughout life (Fig. 2A, 3A). Very low numbers of CD31+CD45RARO+ T cells in peripheral blood of adult volunteers have already been described previously, suggesting that signals provided in vivo during transition from the naive to the memory phenotype accelerate CD31 loss 16. It has also been shown that the emerging population of CD31 effectors after T cell receptor stimulation resulted from irreversible CD31 down-regulation and not from overgrowth of a pre-existing CD31 population 18. The observed down-regulation of CD31 has been reported to occur after the CD45RA to RO switch, but earlier than the loss of CD27 and CD62L 16, 23.

Independent of age, CD31 is also expressed on CD45RAlowROlow intermediate CD4+ T cells (Fig. 2A). Here, we demonstrate that CD31+CD45RAlowROlow intermediate CD4+ T cells are consistently detectable in up to 27% of CD4+ lymphocytes, irrespective of age. Essentially, we did not detect cells with an intermediate level of CD31 expression in any of the gates defined by the CD45 isoform expression pattern. Intermediate CD45RAlowROlowCD4+ T cells are suggested to contain primed T cells, some of which may be capable to reacquire an “unprimed” phenotype in the absence of antigenic stimulation 23. Therefore, CD31+CD45 RAlowROlow intermediate CD4+ T cells may still contain TREC.

Analyzing the age-related TREC content within CD31-expressing CD4+ lymphocytes of healthy individuals, we show a tenfold decrease of TREC content (R2 = 0.5636; Fig. 3B) from birth to old age. Poulin et al.24 who have analyzed sorted naive CD62L+CD45RA+ROCD4+ T cells have similarly discovered a tenfold decline (R2 = 0.5026) of TREC content in healthy adults during aging. Harris et al.9 have measured TREC within sorted naive CD62L+CD45RA+CD4+ T cells of healthy middle-aged individuals (n = 6; 30–45 years of age) and detected on average 27 TREC/1000 CD62L+CD45RA+CD4+ T cells (range: 3–100), while we found on average 40 TREC/1000 CD31+CD4+ T cells (range: 8–90) in a cohort of comparable age (n = 10; Fig. 3B). Hence, the number of TREC within sorted CD31+CD4+ T cells is comparable or even higher than in sorted naive CD62L+CD45RA+CD4+ T cells 7.

The age-related decline of TREC within sorted CD31+CD4+ T cells is novel and has not been reported before. Limited peripheral division of CD31+ cells or altered dynamics or stability of TREC within the CD31+CD4+ T cell pool may explain this finding 9, 10. We have not further investigated the replicative history of sorted CD31+CD45RA+ROCD4+ T cells by comparing the absolute telomere lengths of these cells between different individuals since the overall relationship between chronological age and individual telomere length in peripheral blood cells is fairly weak 2527.

However, we show for the first time that naive CD31+CD45RA+ROCD4+ T cells exhibit significantly longer telomeres than naive CD31CD45RA+ROCD4+ lymphocytes (Fig. 4), confirming the postulated higher replicative history of the latter 14, 15.

In addition, CD31+ memory CD45RARO+CD4+ T cells – despite being TREC negative, as expected – exhibit shorter telomeres than CD31 naive CD45RA+ ROCD4+ cells, indicating a higher replicative history of the former. As expected, CD31 memory CD45RA RO+CD4+ T cells exhibit the shortest telomeres of all four sorted subpopulations (Fig. 4). Our results confirm that the absence of TREC in CD31 as well as in CD31+CD45RO+ CD4+ cells must have been caused by TREC dilution after considerable cell proliferation 16, 17.

Although CD31+ memory CD45RARO+CD4+ T cells are shown to have shorter telomeres and a higher replicative history than naive CD31CD45RA+ROCD4+ cells, their scarcity in peripheral blood probably explains why “contaminating” CD31+ memory CD45RO+ RACD4+ T cells do not significantly alter the considerable replicative difference between CD31+ and CD31 overall CD4+ T cells (mean difference of telomere length: 1.1 kb; data not shown). It is noteworthy that a difference of 1 kb corresponds to an average loss of telomere length in human lymphocytes during 20–50 years of aging 28.

Endogenous telomerase activity is crucial to sustain telomere length and replicative capacity 29, 30; therefore, we have analyzed the telomerase activity of CD31+ and CD31CD4+ T cell subpopulations. Studies of unstimulated T cells have shown that only cells of very early stages of development, e.g. human thymocytes and neonatal T cells, have a high spontaneous expression of telomerase activity 3033. According to previous reports (and own experiments; data not shown), CD31+CD4+ T cells do not express activation markers, and only 0.3% of these cells are in the G2M phase of the cell cycle 23, 34. Since the majority of CD31+ T cells are obviously quiescent lymphocytes, absent or very low telomerase activity is expected in these cells 2835. We demonstrate, age independently, on average six times higher telomerase activity in CD31+ T cells compared to their CD31 counterparts (Fig. 5A), corresponding to the expected finding that the majority of CD31+CD4+ T cells represent naive cells.

After sorting of naive CD45RA+ROCD4+ T cells for CD31, we further demonstrate that naive CD31+CD45RA+RO cells exhibit significantly higher telomerase activity compared to naive CD31CD45RA+ROCD4+ T cells, even at a different age (Fig. 5B), corroborating the notion that CD31 discriminates between two distinct populations of naive CD4+ T cells with different replicative history. Very similar results have been obtained recently in naive and early-stage (CCR7+CD27+) memory CD4+ T cells showing a higher telomerase activity and telomere length than mature end-stage memory (CCR7CD27+ and CCR7CD27) CD4+ T cells 33. Whether CD31 signaling is functionally involved in the tight – but not yet unraveled – regulation of telomerase activity and telomeric reserve of T cells is currently unknown. It is noteworthy that in humans there is recent evidence that CD31 signaling stimulates telomerase activity and suppresses Bax-induced apoptosis on hematopoietic cells 3639. In contrast, CD31 knockout mice have revealed a regulatory function of CD31 in B lymphocytes, but no obvious phenotype related to its absence from T lymphocytes has been described 40.

Our findings in lymphopenic children during 12 months after myeloablative HSCT and profound T cell depletion are novel and demonstrate that the majority of reemerging CD4+CD45RA+RO T cells do express CD31 (Fig. 6B), mimicking the CD31 expression pattern in early childhood (Fig. 3D). At the same time, CD31+ memory CD45RO+RACD4+ T cell numbers increase only slowly and remain fairly low expressed (<75 cells/µL) until month +12 (Fig. 6D), as observed in normal individuals during aging (Fig. 3A). At 5–6 months after HSCT, reemerging CD31+CD4+ T cells and CD4+CD45RA+RO CD4+ T cells coincide and correlate with reappearing absolute numbers of TREC (Fig. 6E).

Our results in lymphopenic children are in concordance with a report by Muraro et al.20 who has analyzed adult multiple sclerosis patients after autologous HSCT and used CD31 in combination with CD45RA and CD45RO in order to identify RTE in the CD4 subset. Examining the correlation between mean percentages of CD31+CD45RA+CD45ROCD4+ cells and TREC levels in purified CD4+ T cells during a 2-year follow-up, they found no correlation in the early (month +6) post-therapy phase, but a strong correlation at the 1-year and 2-year follow-up. Our analysis demonstrates for the first time a satisfactory correlation between absolute numbers of reemerging TREC and absolute numbers of either CD31+ overall CD4+ or CD45RA+ROCD4+ T cells in a pediatric allogeneic HSCT setting, thereby complementing Muraro's findings. Our results also confirm that thymic reactivation in children occurs at least 6 months earlier than in adult HSCT patients [5]5.

In summary, we demonstrate that the majority of CD31+CD4+ cells are essentially CD31+ naive CD45RA+ROCD4+ T cells containing high numbers of TREC with high telomerase activity and preserved telomeres, which is typical of CD4+ lymphocytes at early stages of development and of low replicative history 32. However, the finding that TREC decrease about tenfold within sorted CD31-expressing CD4+ T cells indicates that limited peripheral expansion of CD31+CD4+ T cells without CD31 down-regulation or altered regulation of life spans of RTE within the CD31+ T cell pool may occur during aging. Despite certain limitations in the elderly, CD31 expression on CD4+ T cells is a suitable marker for thymic reactivation essentially in lymphopenic children and adolescents after allogeneic HSCT. Using CD31 as a marker for TREC-rich CD4+ T cells in lymphopenic settings or diseases would be advantageous over using the CD45RA/RO isoforms since T cells carrying the C77G polymorphism of the CD45 gene (2–3% of the Caucasian population) 21 (Fig. 2B), potential revertants from the CD45RO+ memory to the naive CD45RA+ phenotype 41 and emerging non-naive CD45RA+ T cells with effector function and phenotype 42 would be excluded. Further studies of CD31-expressing CD4+ lymphocytes are needed to analyze the correlation with TREC in children with GVHD, lymphopenic adults and seniors after allogeneic HSCT and in other lymphopenic or autoimmune diseases, as well as to unravel the function and dynamics of CD31+CD45RO+RA memory CD4+ T cells.

Materials and methods

Blood sample collection and cell culture

After scientific board review and informed (individual or parental) consent, 10–20 mL of heparinized blood from a total of 26 healthy human adult volunteers (age: 25–69 years) and 13 CB samples (healthy newborns: weeks 38–42 of gestation) was analyzed for age-related percentages of CD31+CD4+ T cells, TREC content and telomerase activity. Ten healthy adult donors (age: 25–60 years) donated 40 or 80 mL blood and were chosen for sorting experiments, consecutive telomere, telomerase and TREC analysis. Sixteen healthy infants and children (age: 0–15 years) donated 2 mL EDTA blood for flow cytometry of absolute CD31+CD45RA+ROCD4+, CD31+CD4+ T cell counts and percentages of CD31 expression on CD4+ T cells. Six children (age: 3–13 years) with chronic hereditary anemias (patients A and E) and acute lymphoblastic leukemias (ALL) (patients B, C, D, F) were prospectively analyzed for 12 months after myeloablative conditioning and HSCT on a monthly basis. Analysis included absolute numbers of TREC/µL (months +1, +3, +5/6, +9/10 and +11/12), flow cytometry (FACS) of CD31-expressing CD4+ T cells, and chimerism using the variant number of tandem repeat (VNTR) technique.

Cell separation and fluorescent staining

PBMC were separated from erythrocytes and granulocytes using a Ficoll gradient. CD31+CD4+ and CD31CD4+ T cells were purified from PBMC by manual magnetic separation in two steps, using first the CD4+ T cell isolation kit (MACS) and then CD31 microbeads according to the manufacturer's instructions (Miltenyi Biotech, Bergisch Gladbach, Germany). For some FACS sorting experiments, the CD4+ sorted cell fraction was stained for CD31- and CD45RA/RO-defined cell populations. The purity of the fractions in the initial experiments using 20–40 mL samples was at least 85% (Fig. 1). Later, for the sorting of 80-mL samples, CD4+ T cells were presorted from mononuclear cells by automated MACS negative selection (depletion of cells positive for CD8, 14, 16, 19, 36, 56, 123, TCRγδ and/or glycophorin A) followed by FACS sorting of CD31+ and CD31 strictly CD45RA+RO- or CD45RARO+-expressing CD4+ T cells, leading to >95% purity (Fig. 4, 5B). For immunostainings, the following antibodies were used: anti-CD4-APC mAb (Becton Dickinson, Heidelberg, Germany), anti-CD45RA-FITC mAb (Becton Dickinson), anti-CD45RO-PerCP mAb (eBioscience, San Diego, CA), and anti-CD31-PE mAb (eBioscience). At least 3 × 105 cells were washed in CellWASH and incubated with the relevant antibodies. After further washing, cells were fixed in CellWASH supplemented with 0.5% paraformaldehyde and analyzed on a FACSCalibur (Becton Dickinson) using the CellQuest software. The remaining cells were frozen in liquid nitrogen.

DNA isolation

For DNA purification, frozen batches of PBMC, or CD31+CD4+, or CD31CD4+ T cells were thawed and resuspended in a solution of 2 × SSC, 10% SDS and 0.1 mg/mL proteinase K, and incubated at 55°C for 2–3 h. DNA was precipitated with salt and ethanol, centrifuged, washed, and the dried pellet was resuspended in distilled water.

Real-time PCR for TREC measurement

To detect signal-joint TREC (sjTREC), genomic DNA was assayed by real-time quantitative PCR as described 8. Each PCR reaction was performed in a final reaction volume of 30 µL containing 50 ng genomic DNA (corresponding to about 10 000 cell nuclei), 15 µL 2 × TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, USA), forward and reverse primers at 900 nM each, and a fluorescent probe targeting either the sjTREC or the control sequence at 200 nM. The sequences of the primers and probes used for sjTREC measurement and quantification of the internal reference were as reported 8. TREC numbers were given as TREC/1000 PBMC (normal individuals) or TREC/µL (after HSCT). The detection threshold of our assay was two TREC copies in 10 000 cells.

Telomere length analysis

TRF length analysis was performed using the TeloTAGGG Telomere Length Assay kit from Roche Applied Sciences (Roche, Rotkreuz, Switzerland). Between 1 and 1.5 µg DNA from sorted lymphocyte subpopulations was assayed. The procedure was performed according to the supplier's instructions. The signal from the hybridized, fluorescent probe was quantified by computer-assisted scanning densitometry using the Scion Image software ( Individual samples were assayed between two and four times. The relative length of telomeres was determined using a PCR-based method developed by Cawthon 43 and modified by Kurz 44. With this technique, between 0.1 and 12 ng genomic DNA was assayed in five replicates each, using the ABI prism sequence detection 7900 system.

Nuclear protein extraction for telomerase activity assays

Cell samples were processed according to published protocols 45. Briefly, 1 × 106 cells were lysed in 100 µL 0.5% CHAPS lysis buffer (10 mM Tris-HCl pH 7.5, 1 mM MgCl2, 1 mM EGTA, 0.1 mM PMSF, 5 mM β-mercaptoethanol, 0.5% CHAPS, 10% glycerol) and incubated for 30 min on ice. After centrifugation at 16 000 × g for 30 min at 4°C, aliquots of the supernatant were flash frozen and stored at –80°C. Telomerase activity in flash frozen CHAPS samples was stable even after one freeze-thaw cycle.

Analysis of telomerase activity

For stored samples (Fig. 5A), the SYBR Green RQ-telomeric repeat amplification protocol (TRAP) assay was utilized. Primer sequences were as described by Kim et al.46 [telomerase primer (TS): 5′-FAM-AAT CCG TCG AGC AGA GTT-3′; anchored return primer (ACX): 5′-FAM-GCG-CGG-[CTTACC]3 CTA ACC-3′]. The assay was done in a reaction volume of 25 µL using 5 µL of CHAPS protein extracts (equivalent 5 × 104 cells), 12.5 µL 2 × SYBR Green PCR Master Mix (Applied Biosystems), TS at 700 nM, and ACX at 200 nM. The sequences of the TS and ACX primers were as described 24, but without the 5′ fluorescent label. For the primer extension by telomerase, samples were incubated for 20 min at 25°C, followed by 50 two-step PCR cycles (30 s at 95°C, 90 s at 60°C) using the ABI Prism 7700 thermal cycler (Applied Biosystems). HL60 cells or T lymphocytes were cultured in RPMI 1640 medium supplemented with 10% fetal calf serum and 1% antibiotics (penicillin/streptomycin). Cells were incubated at 37°C in an atmosphere containing 5% CO2. For fresh samples (Fig. 5B), the TeloTAGGG Telomerase PCR enzyme-linked immunosorbent assay (ELISA) kit (Roche Diagnostics, Basel, Switzerland), also based on the TRAP method, was used. The relative telomerase activity was expressed in % by comparing the signal of the sample to the signal of the highly positive control template (human kidney/293 cells = 100 %).

Statistical analysis

Significance was determined by comparing TREC values between age groups using Student's t-test. The sign test was used for the differences in telomerase activity. R2 correlation analysis was performed by using Microsoft Excel software.


The authors are grateful to Eva Niederer for FACS sorting, to Sandra Balli, Anette Schütz, and Maya Rutishauser for their expert technical assistance, and to Dr. Johann Peter Hossle for stimulating discussions. We greatly appreciate Prof. Berger's support for the telomere length analysis. This work was supported by grants from ZLB Behring AG (Zürich, Switzerland), Genzyme GmbH (Neu-Isenburg, Germany) and Fresenius Biotech GmbH (Graefelfing, Germany) to T.G.Conflict of interest: These authors declare no financial or commercial conflict of interest.


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5


  6. 6