The Magnitude of Thymic Output Is Genetically Determined through Controlled Intrathymic Precursor T Cell Proliferation¹

The ability to maintain peripheral T cell pool numbers and diversity is progressively impaired with age and can be drastically reduced by infectious diseases and therapeutic interventions (1). Thymic output is primarily determined by the number of cells reaching thymic selection, thus by the level of intrathymic precursor T cell proliferation (2). Direct quantification of the level of thymocyte proliferation could therefore allow the rapid and accurate assessment of thymic function.

To monitor thymic function, several authors have quantified TCR excision circles (TREC)⁴ (signal-joint TREC (sjTREC) molecules) in circulating blood (3, 4, 5, 6, 7); sjTRECs are indeed present in a large proportion of recent thymic emigrants (RTEs). As sjTRECs are not replicated during cell division, their frequency decreases while T cells proliferate and differentiate into naive and memory T cells. Variations in homeostatic proliferation of RTEs and other T cells thus impact on sjTREC frequency, leading to the lack of reliability of their use as a marker for the extent of thymic output (8). We have developed new tools based on the quantification of both sjTRECs and DJβTRECs (byproducts of the rearrangement between TRBD and TRBJ segments), which circumvent this inherent difficulty (9). In human, we and others have interpreted the ratio of sjTREC/DJβTREC frequencies (sj/βTREC ratio) as reflecting the magnitude of precursor T cell proliferation during their intrathymic differentiation, thereby representing thymic output (9, 10). Importantly, this marker is not influenced by peripheral T cell homeostasis given that the rates of cell proliferation and death of peripheral T lymphocytes equally affect both types of TREC. Using this method, we have demonstrated that HIV infection affects thymic output by inhibiting the intrathymic proliferation of immature thymocytes. As for the sjTREC frequency, major differences in the sj/βTREC ratio were observed among age-matched healthy individuals (9), suggesting that the extent of thymic function could also be impacted by genetically controlled individual factors (11).

The complete sequencing of the murine genome associated with the large array of identified strain-specific polymorphisms has made mouse models ideal to study the genetic control of immune functions (12, 13, 14, 15). Strain polymorphisms affecting susceptibility to pathogens (16, 17, 18) or to the development of tumors (19) as well as variability in cytokine profiles have been observed in several mouse strains. Thymic IL-7 expression levels were found to be associated to the efficacy of TRBV rearrangements in the thymus of fetal mice of the C57BL/6 (B6), BALB/c, and CBA/J strains (20). Genetic polymorphisms were also associated to the IL-7 receptor gene in humans (21, 22), to age-related variations of thymocytes subpopulations (23, 24) or thymic involution (25, 26), to thymic deletion (27), and to CD8 T cell lineage commitment (28). Most of these polymorphisms, although never demonstrated, could quantitatively alter thymic output (25, 26, 27, 28).

In this study, we have analyzed the contribution of the genetic background to the magnitude of thymic output in the murine model by measuring both sjTREC and DJβTREC frequencies, estimating intrathymic precursor T cell proliferation, and quantifying thymic output in nine laboratory mouse strains as well as in their F₁ progeny.

Age-matched females between 6 and 8 wk of age from the following strains were used: A/J (Harlan, A/JOlaHsd); BALB/c (Charles River; BALB/cAnNCrlBR); C57BL/6 (JAX, 000664); DBA/1 (JAX, 000670); C3H/He (C3H; JAX, 000659); CBA (JAX, 000656); NZB/BlN (NZB; JAX, 000684); RF (JAX, 000682); FVB/N (FVB/N; Charles River, FVB/NCrl). All mice were housed in specific pathogen-free environment according to the standards of the Canadian Committee for Animal Protection. IRB approval was obtained for all animal manipulations.

Fluorescent mAbs were purchased from BD Pharmingen: CD3ε (145-2C11); CD4 (RM4-5); CD8, CD25 (PC61); CD44 (IM7); L-selectin (CD62L; MEL-14); BrdU (3D4); and the Mouse Lineage Panel (559971). The CD127 (A7R34) was obtained from eBioscience. Thymus, spleen, and lymph node (LN) cells (cervical, axillary, mesenteric, and inguinal) were treated as previously described (29) and analyzed on a FACSCalibur or FACS LSR II using CellQuest Pro or FACSDiva software. Thymocyte subsets were sorted on a MoFlo cell sorter (DakoCytomation). Thymocyte subpopulations purity was always >95%.

For in vivo BrdU incorporation studies, mice were given sterile drinking water containing 0.8 mg/ml BrdU (Sigma-Aldrich) as described (30).

The turnover of immature thymocytes was evaluated by in vivo BrdU labeling. One milligram of BrdU was injected i.p. twice at 2-h intervals, and thymuses were sampled 1 h after the second BrdU injection. A sample of 15 × 10⁶ cells was used for staining of double-negative (DN) cells, and 10⁷ cells were used for staining of intermediate simple positive (ISP), double-positive (DP), and single-positive (SP) thymocytes. Stained cells were fixed in 1% paraformaldehyde containing 0.01% Tween 20 for 12–36 h at 4°C in the dark. Cells were washed in PBS, then in 4.2 M MgCl₂, 0.15 M NaCl (pH 5), and thereafter incubated for 1 h at 20°C in the same buffer containing 500 KU of DNase I (Sigma-Aldrich). After a new wash in PBS, thymocytes were incubated for 30 min in PBS containing anti-BrdU mAb at room temperature in the dark. For DN thymocytes studies, cells were labeled with biotinylated anti-CD4, anti-CD8, and mouse lineage markers (CD11b, CD45R/B220, Ly-6G, and Ly-6C, TER-119) revealed with SAv-PerCP, CD44, and CD25 mAbs. ISP, DP, and SP thymocytes were labeled using anti-CD3, anti-CD8, and anti-CD4 mAbs. For BrdU incorporation studies, at least 5 × 10³ cells were acquired for each DN population.

Specific primers for the sjTRECs (byproducts of the TCRδ locus excision), DJβTRECs (byproducts of TRBD1/TRBJ1.1 to TRBJ1.6 or TRBD2/TRBJ2.1 to TRBJ2.5 or TRBJ2.7 rearrangements), and CD4 gene were defined on mouse sequences (GenBank accession numbers M64239, AE000663-4-5, and AC002397). TRAV, TRBD, and TRBJ were numbered according to the nomenclature described by the international ImMunoGeneTics information system (http://imgt.cines.fr). Real-time PCR quantification of the different TRECs were performed using LightCycler technology (Roche Diagnostics) as previously described, (31) and using the CD4 gene as a housekeeping gene. Briefly, cells were lysed in Tris-HCl (pH 8.3; 10 mM), Tween 20 (0.05%), and Nonidet P-40 (0.05%) supplemented with proteinase K (100 μg/ml) for 30 min at 56°C. After proteinase K inactivation (10 min at 95°C), cell lysates were used in a first-step PCR amplification using outer primers. In this step, any TREC was coamplified together with CD4 gene used as a housekeeping gene. Such coquantification permits ignoring the exact concentration of plasmid standards as well as DNA concentration. PCR conditions for this first step amplification were: denaturation at 95°C for 10 min; amplification (95°C for 30 s; 60°C for 30 s, 72°C for 5 min) for 22 cycles; cooling at 20°C. Following this first step amplification, both the TREC molecules and the CD4 gene content were quantified by real-time quantitative PCR using inner primers and LightCycler technology. Plasmids containing any of the TREC amplicon and CD4 amplicon were used to generate standard curves. All primers and probes specific for the amplification of mouse TRECs and CD4 sequences are shown in Table I. All primer pairs lead only to the generation of a single PCR product at the expected size (DJβ1.1, 365 bp; DJβ1.2, 388 bp; DJβ1.3, 362 bp; DJβ1.4, 315 bp; DJβ1.5, 339 bp; DJβ1.6, 380 bp; DJβ2.1, 344 bp; DJβ2.2, 382 bp; DJβ2.3, 358 bp; DJβ2.4, 292 bp; DJβ2.5, 292 bp; DJβ2.7, 274 bp; sjsjTREC-58, 244 bp; sjTREC-61, 226 bp), except for Dβ2/Jβ2.4 for which the upper PCR product corresponds toTRBD2/TRBJ2.5 rearrangements (Fig. 1 A).

Quantification of the 12 DJβTREC (DJβ1.1–DJβ2.7) frequencies in various thymocyte subsets revealed that these molecules, produced in the DN3 subset, are strongly diluted during further differentiation (Fig. 3, B and C). Surprisingly a second wave of DJβ2TREC production that was not accompanied by V-DJ rearrangements, therefore indicating that it is not productive, was observed in SP thymocytes (data not shown). In the human DJβTREC, quantification was used to calculate the sj/βTREC ratio, a marker measuring intrathymic events occurring between TCRβ and TCRα chain rearrangements, this second wave of DJβ2 rearrangement occurring after TCRα chain rearrangement, the frequencies of DJβ2TRECs could not be used in the murine context to calculate the sj/βTREC ratio. Consequently, in the presented data, the sj/βTREC ratio was thus calculated as the sum of sj61 and sj58 frequencies, divided by the sum of DJβ1TREC frequencies.

The Mann-Whitney analysis and Spearman’s correlation tests were performed using StatView 4.5 software. An r value of ≥0.3 or ≤−0.3 and a p value ≤0.05 were considered significant.

To accurately quantify thymic function in mice, we adapted the methodology developed in human (see Materials and Methods) (9, 32). This method is based on the simultaneous measurement of sjTREC, generated by the deletion of the TCRδ locus before rearrangement of the TCRα sequences, and that of TRECs generated during TCRβ chain rearrangement (DJβTRECs, byproducts of the rearrangement between TRBD and TRBJ segments). In humans, the ratio of these two types of TREC was estimated as a marker for intrathymic precursor T cell proliferation and thus used as a surrogate marker for thymic output (9, 31, 32, 33, 34).

Although not essential to TCRα rearrangement and commitment to the αβ T cell lineage in mice (35, 36, 37, 38), TCRδ locus deletion is mostly driven by the rearrangement of the δREC1 element to the first TRAJ segment (TRAJ 61) (7, 39). However, other δREC elements as well as the other first TRAJ elements could also participate in TCRδ locus deletion. We thus quantified the rearrangement of the δREC1, δREC2, and δREC3 with the six first TRAJ segments, that include two pseudogenes (TRAJ61 and TRAJ60), TRAJ59 that is not preceded by a consensus recombination signal sequence, and three functional TRAJ segments (TRAJ58, TRAJ57, and TRAJ56), to evaluate their relative contribution to the generation of the murine equivalent of the human sjTREC molecule. Two rearrangements clearly dominate the deletion of the δ locus in mice: the sjTREC-61, byproduct of the rearrangement between the δREC1 and Jα61, represents ∼80% of the sjTRECs in the different strains tested; whereas δREC1/Jα58 accounts for most of the remaining rearrangements (sjTREC-58; data not shown). Accordingly, we used the sum of the frequencies of these two TRECs as an approximation of the total sjTREC frequency in our studies.

With the aim of estimating intrathymic precursor T cell proliferation through measurement of the sj/βTREC ratio (9), real-time PCR quantification of the 12 different DJβTRECs was also performed on each sample. Rearrangements between TRBD1 and the six TRBJ1 (1.1–1.6), and rearrangements of TRBD2 and 6 TRBJ2 (2.1–2.5 and 2.7) were quantified (Fig. 1,A). The reproducibility of each individual TREC quantification is shown in Fig. 1 B.

Female mice from nine different inbred strains (n = 4–11) were sacrificed at 6–8 wk of age. Both sjTREC and DJβTREC frequencies were measured in LN mononuclear cells and the sj/βTREC ratio was calculated (as described in Materials and Methods and previous paragraph). sjTREC values were quite similar when all strains were compared (except for the C57BL/6; Fig. 2,A). In contrast, DJβ1TREC values varied significantly among strains with the FVB/N showing the lowest values (p < 0.005 as compared with all other strains; Fig. 2,B). Accordingly, each strain was characterized by a distinct sj/βTREC ratio (Fig. 2 C). FVB/N mice showed a significantly higher sj/βTREC ratio than any other strain (1624 ± 476 for FVB/N as compared with 759 ± 727 (A/J), 588 ± 421 (C3H), 474 ± 262 (BALB/c), 373 ± 335 (NZB), 343 ± 279 (DBA/1); 244 ± 145 (CBA), 268 ± 180 (RF), and 73 ± 14 (C57BL/6); p ≤ 0.006). Moreover, the sj/βTREC ratio of CBA mice was significantly lower than that of A/J (p = 0.027). In contrast, the sj/βTREC ratio in the B6 strain was significantly higher than that observed in all the other strains (p ≤ 0.04). Because all animals were kept in the same pathogen-free environment and were age and sex matched, such interstrain variability in the sj/βTREC ratio suggested that this parameter could be genetically determined in mice. These results are consistent with our observation of large interindividual variability in the sj/βTREC ratio in age-matched healthy human individuals (9).

Analysis of sjTREC frequencies indicated that this parameter was relatively constant in all tested mouse strains with the exception of the C57BL/6 strain. Hence, the observed variability of the sj/βTREC ratio among analyzed strains was found to be independent from the sjTREC frequency (Fig. 2 D). This result indicated that the size of the peripheral TREC-containing T cell pool is not dependent on intrathymic precursor T cell proliferation, because a limited thymic output may be peripherally compensated by increased cell survival (2, 9). The sj/βTREC ratio in C57BL/6 was significantly lower than that of all the other strains as a consequence of its low sjTREC frequency. Such a low sjTREC frequency may be due to a different mechanism of deletion of the TCRδ locus in C57BL/6 mice. In mice, three different δREC sequences can be used, in combination with several TRAV segments to delete the TCRδ locus before TCRα chain rearrangement. Accordingly, we quantified, in the C57BL/6 strain, peripheral levels of the various TRECs generated by the rearrangement between δREC1, δREC2, or δREC3 and nine of the first TRAJ segments (Jα61 to Jα56). None of these TRECs (18 combinations) but δRec1/Jα61 and δRec1/Jα58 rearrangements demonstrated any significant frequency, suggesting that unidentified sequences are used in this particular strain to delete the δ locus or that this deletion occurs through TRAV/TRAJ rearrangements. Accordingly, we excluded the C57BL/6 strain from further analysis of the sj/βTREC ratio.

Interestingly, the sj/βTREC ratio inversely correlated with the frequency of DJβ1TRECs (r = −0.822, p = 0.006 Fig. 2,E). The sj/βTREC ratio could not be a consequence of variations in peripheral T cell homeostasis given that the latter would equally impact on the levels of both types of TRECs. Accordingly, this relationship (Fig. 2 E) suggested that the sj/βTREC ratio reflects an intrathymic event occurring before TCRδ locus excision.

We then analyzed the dynamics of TREC frequencies during thymopoiesis by quantifying each individual TREC in FACS-purified thymocyte subsets. Fig. 3,A, 3B and 3C show the results obtained with the A/J strain. As expected, the sjTREC molecules were initially detected in ISP cells, and their frequencies reached a plateau at the DP and SP stages (Fig. 3 A). In mice, δREC-TRAJ rearrangement does not necessarily dominate TCR δ locus deletion (35, 36, 37, 38). The relatively low sjTREC frequencies in DPs suggest that this event occurs only on a fraction of alleles (1/4 to 1/30). In contrast, its increase in SP CD4 also suggests that the DP subset is a heterogeneous population composed of cells that already excised TCRD locus and more immature cells that did not. Moreover, the frequency of sjTREC (61 + 58) in murine DP thymocytes was in the same range as what we observed in humans (9). Surprisingly, the SP CD8 cells demonstrated a lower sjTREC content than did SP CD4, a possible consequence of variable proliferation history.

In contrast, both DJβ1TRECs and DJβ2TRECs were generated at the DN3 stage of thymocyte differentiation (Fig. 3, B and C). However, their frequencies rapidly decreased at the following steps of maturation (Fig. 3, B and C). Because TREC molecules are deleted upon cell division, the drop in DJβTREC frequencies observed at the DN4 stage strongly suggests that these molecules are diluted as a consequence of proliferation in the DN4 compartment. One can then postulate that the number of cell divisions at the DN4 stage of maturation may be strain specific, leading to variable thymic output.

We thus quantified BrdU incorporation in the various thymocyte subpopulations, as a marker for the extent of in vivo proliferation in these subsets, in the eight studied strains of mice (Fig. 3,D and data not shown). A strong positive correlation was indeed observed between the level of proliferation of the DN4 cell subset and the peripheral sj/βTREC ratios (r = 0.888, p = 0.0013; Fig. 3 D). Although other subsets of thymocytes also proliferate, no other correlations with the peripheral sj/βTREC ratio could be established (not shown). This result is consistent with reports showing that proliferation in DN4 cells is lower in aged mice, correlating with thymic involution and lower thymic output (40).

In addition, we observed a significant correlation between BrdU incorporation in the DN4 thymocyte subset and CD127 (the IL-7R α-chain) expression level on DN4 cells (r = 0.958, p < 0.000; Fig. 3 E). This correlation was not observed in any other DN subset or in ISP cells. It is thus possible that the interaction of IL-7 with its receptor CD127 not only permitted the survival of DN4 cells but also triggered their proliferation, that in turn led to the dilution of DJβTREC molecules and to the increase of the sj/βTREC ratio. Taken together, these data demonstrate that the sj/βTREC ratio measured in periphery reflects the proliferative history of RTEs during their intrathymic differentiation, more precisely at the DN4 differentiation stage.

Extensive proliferation of the DN4 subset should result in an increased number of cells reaching the DP differentiation stage and rearranging their TCRα chain, eventually leading to increased numbers of cells that undergo positive and negative selection and are exported into the periphery (2). To confirm the influence of intrathymic proliferation on thymic output, we quantified thymic production using in vivo BrdU incorporation (30) in three mouse strains shown to exhibit different levels of intrathymic proliferation as determined by their sj/βTREC ratio (FVB/N, A/J, and BALB/c). FVB/N, A/J, and BALB/c mice were given BrdU in drinking water for 28 days and sacrificed at regular intervals. BrdU incorporation was quantified in naive (CD62L⁺CD44^low) and memory (CD62L⁻CD44⁺) T cell subsets (Fig. 4, A and B). Although the accumulation of BrdU⁺ memory T cells did not significantly differ between the three strains (data not shown), FVB/N mice did produce significantly higher numbers of BrdU^low naive T cells during the 28 days of in vivo labeling than did the A/J mice (Fig. 4,C, p = 0.032). A/J mice also showed significantly higher values of BrdU^low naive T cells when compared with BALB/c mice (Fig. 4 C; p = 0.0015). These results further confirmed the above reported heterogeneity observed in the sj/βTREC ratio as both sets of data show that thymic output is very heterogeneous in different mouse strains, thereby validating the hypothesis of the importance of the genetic background in defining thymic output values.

Our results suggested the existence of an inherited character controlling intrathymic precursor T cell proliferation given that several inbred mouse strains showed significantly different magnitude of thymic outputs whereas, within each strain, age-matched mice exhibited comparable sj/βTREC ratio levels. To analyze more precisely the transmissibility of the inherited character(s) influencing intrathymic precursor T cell proliferation, FVB/N mice, which demonstrate intensive proliferation, were crossed with either BALB/c or CBA strains, which show a low sj/βTREC ratio (Fig. 5). In both (FVB/N × BALB/c)F₁ and (FVB/N × CBA)F₁ mice, the sj/βTREC ratio of the offspring was identical with that of the BALB/c and CBA parents, respectively (498 ± 314 vs 474 ± 262 in the FVB/N × BALB/c and BALB/c, and 308 ± 84 vs 244 ± 145 in the FVB/N × CBA and CBA, respectively). In contrast, the FVB/N parents showed a significantly higher sj/βTREC ratio than the offspring (1624 ± 476; p = 0.002 and p = 0.001 for (FVB/N × BALB/c)F₁ and (FVB/N × CBA)F₁, respectively). These data demonstrate that the allele(s) defining the high thymic output in the FVB/N mouse is recessive. Indeed crosses between mouse strains showing high and low thymic output generate offspring that invariably show low thymic output. Altogether, our results demonstrated for the first time the genetic regulation of thymic output.

Age dependence of thymic function, as estimated by TREC quantification and estimation of intrathymic precursor T cell proliferation through calculation of the sj/βTREC ratio is certainly not absolute in humans. In most papers relating this age-related decline in healthy individuals, a large degree of variability is observed between age-matched subjects, suggesting that other parameters are also influencing the efficacy of the thymus in producing recent thymic emigrants. In this manuscript, through the measurement of both sjTREC and DJβTREC frequencies as well as the estimation of the sj/βTREC ratio in peripheral T cells from nine inbred mouse strains, we evidenced for the first time that the extent of thymic output is indeed genetically determined in mice and strongly depends on intrathymic precursor T cell proliferation occurring at the DN4 maturation stage.

The proliferation of the DN4 thymocyte subset, evidenced through BrdU incorporation (Fig. 3, D and E), directly influences the dilution of DJβTREC molecules between DN3 and DN4 cells. The amplitude of the DJβTREC frequency decline between DN3 and DN4 stages was different for C57BL/6 and A/J strains. The mean fold decline of the DJβTREC frequency in C57BL/6 was 12.2 (range, 1.2–18.4), whereas it was on average reduced 52.9-fold (range, 20.9–123.5) in the A/J strain (p = 0.016). Thus, the number of cell divisions at the DN4 stage of maturation is strain-specific. Assuming that TREC frequency is divided by 2 at each cell division, it should take three to four cell divisions to achieve a dilution of TRECs of 12.2 in C57BL/6, whereas five to six cell divisions would be required to generate a 52.9-fold in TRECs frequency in A/J mice. These data are perfectly consistent with observations by Hayday and colleagues (41), who demonstrated that DN4 cells from healthy C57BL/6 mice placed into reaggregated thymic organ cultures are able to sustain approximately three cell divisions (10-fold increase in numbers) during a 6-day culture period. This supports our conclusion that the sj/βTREC ratio variability between the different mouse strains is indeed a direct consequence of differences in the extent of cell proliferation occurring at the DN4 stage, which could lead to variable thymic output.

The different strains analyzed here were purchased from different facilities and thus might bring nongenetically determined factors that could influence thymic production. This is in particular the case for commensal flora that is suspected to influence various aspects of the immune system in laboratory mouse strains. However, these nongenetically inherited parameters, and in particular intestinal flora are mother transmitted. In the experiments demonstrating that the F₁ offspring have a sj/βTREC ratio similar to that of their BALB/c or CBA parent, all the mothers were from the FVB/N strain. It is thus unlikely that the observed phenotype was due to nongenetically transmitted parameter.

Several molecules have been associated with differences in thymic mass, cellularity, and thymic development implicating DN4 thymocytes. Indeed, pre-TCR, IL-7, IL-7R, Kit ligand, and other molecules are implicated in the efficacy of TCRβ chain rearrangement at the DN3 stage and in the survival capacity of TCRβ⁺ cells following β selection (for review, see Ref. 42). In particular, the IL-7-IL-7R interaction is required for survival and differentiation from the DN4 to the DP stage but does not seem to directly influence cell proliferation (41). Moreover, analysis of T cell differentiation in B7-1 and B7-2 or CD28 knockout mice has shown reduced levels of DN4 cell proliferation and survival and accelerated DN3 to DN4 transition, most likely by enhancing TCR rearrangement through increased RAG-2 expression (43). In contrast, WNT signals provide important proliferative stimuli for developing thymocytes, particularly at DN4 and ISP stages of thymic development (44). Finally, Phillips et al. (40) demonstrated the existence of an age-related decrease in DN4 proliferation, suggesting a second age-related block in thymopoiesis. It is thus possible that changes in the sj/βTREC ratio are a consequence of genetic variation in genes directly involved in this proliferation. However, proliferation in the DN4 subset is directly triggered by TCRβ chain rearrangement and β selection occurring at the DN3 stage. Genetic variations among different genes involved in the efficacy of TCRβ chain rearrangement could indirectly lead to the differences in DN4 thymocytes proliferation we have observed. The generation of congenic mice should allow the identification of the gene(s) responsible for the extent of intrathymic precursor T cell proliferation and thus of thymic output.

Overall, we have demonstrated the existence of genetically inherited character(s) controlling the extent of thymic output in mice. Such factor(s) act(s) at the level of intrathymic proliferation, more precisely between TCRβ and TCRα chain rearrangement in the DN4 immature thymocyte subset. The identification of this or these factor(s) will certainly enable a better understanding of the control of thymic output and may eventually lead to the development of new therapeutic strategies susceptible to improve thymic function, the unique way to restore naive T cell diversity in lymphopenic patients.

We appreciate the animal care by Caroline Riel and the cell sorting by Sylvain Gimmig, Éric Massicotte, and Martine Dupuis.

The authors have no financial conflict of interest.