Blood was collected from 22 adult patients submitted to thymectomy during corrective cardiac surgery in early childhood and 20 age-matched healthy controls. All the subjects gave written informed consent for blood sampling and processing. The study was approved by the Ethical Boards of Faculdade de Medicina da Universidade de Lisboa, Centro Hospitalar Lisboa Norte, and Hospital de Santa Cruz, Portugal.

Peripheral blood mononuclear cells (PBMCs) were isolated from freshly collected heparinized blood via Ficoll-Paque PLUS (GE Healthcare, Uppsala, Sweden). Naive CD4 T-cells were subsequently purified by negative selection (purity > 96%, StemCell Technologies, Grenoble, France). Purified naive CD4 T-cells were cultured at 1 × 10⁶ cells/ml with either IL-7 (10 ng/ml; R&D Systems, Minneapolis, MN, USA) or IL-2 (20 IU/ml; NIH/AIDS Research and Reference Program, Division of AIDS, NIAID, Hoffman-La Roche), for up to 13 days (d), with media replacement at d3 and d7, as we have previously described (14, 28).

Ex vivo phenotypic analysis was performed in freshly collected whole blood using an eight-color staining protocol and a panel of monoclonal antibodies previously described (28). Purified naive CD4 T-cells were surface stained ex vivo and upon culture, followed by intracellular staining using eBioscience FoxP3 kit (eBioscience, San Diego, CA, USA), as described (28). At least 150,000 events were acquired for each sample on a BD LSRFortessa (BD Biosciences, San Jose, CA, USA). Data were analyzed using FlowJo software (TreeStar, Ashland, OR, USA) after doublet exclusion. Results are presented as proportion of a cell population or as mean fluorescence intensity (MFI) of a given marker within the specified population.

Purified untouched naive CD4 T-cells were cultured at 1 × 10⁶ cells/ml (25,000 cells/well) and stimulated with increasing concentrations of beads coated with anti-CD3 and anti-CD28 monoclonal antibodies (Dynabeads, Thermo Fischer Scientific), in order to generate a dose–response curve of TCR stimulation at the single-cell level. In agreement with a previously reported protocol (29), antibody against CD40L (clone 89-76, BD Bioscience) was added to the culture media to improve the staining. After 6 h, cells were collected and surface stained, as described (28), for the marker of cell differentiation (CD45RO) and the additional activation marker (CD69).

Signal joint (sj) and DβJβTREC analyses were conducted as described (5, 30). Briefly, multiplex PCR amplification for sjTREC, DJβ1TRECs (Dβ1-Jβ1.1 to 1.6), or DJβ2TRECs (Dβ2-Jβ2.1 to 2.7), together with the CD3γ chain was performed in triplicate on lysed PBMC. TREC and CD3γ quantifications were then performed using a LightCycler™ in independent experiments, with the same first-round serial dilution standard curve. This highly sensitive nested quantitative PCR assay allowed detection of 1 copy in 10⁵ cells for any excision circle. The sj/βTREC ratio [sjTREC/10⁵ cells/(DJβ1TRECs/10⁵ cells + DJβ2TRECs/10⁵ cells)] was calculated as described (30).

Total RNA was extracted from purified naive CD4 T-cells using Quick-RNA MicroPrep (Zymo Research Corporation, Irvine, CA, USA). cDNA was synthesized from 50 ng of RNA (SuperScript III Reverse Transcriptase, Thermo Fischer Scientific) and used to quantify the expression levels of KLF2, FOXP1, P21, BIM, DUSP4, and DUSP6 in duplicates, using TaqMan Gene Expression Assays on a ViiA7 Sequence Detection system (both from Thermo Fischer Scientific). Results are expressed as ΔCT normalized to the medium CT levels of GAPDH and HPRT.

Total RNA was extracted from 10⁵ to 10⁶ cells with RNeasy kit (Qiagen, MD, USA), and first-strand cDNA synthesized from 1 to 2 µg of RNA (SuperScript III) using an equivolume mixture of random hexamers and oligo (dT). Amplification of the TCRVβ CDR3 was performed using primers specific for each TRBV family and a common TRCB reverse primer (31), followed by a run-off reaction that extends each different PCR product with a second TRCB FAM-labeled primer; and the third step, in which each different fluorescent TRBV-TRBC PCR fragment was separated using a capillary electrophoresis-based DNA automated sequencer. Data were collected and analyzed with GeneMapper v4.0 (Thermo Fischer Scientific) for size and fluorescence intensity determination.

Statistical analysis was performed with Graph Prism Version 5.01 (GraphPad Software, San Diego, CA, USA). The following tests were used for analyzing epidemiological data and results from ex vivo studies as appropriate: Wilcoxon-Signed Rank/paired T-test for pairwise comparisons and unpaired T-test/Mann–Whitney for unpaired comparisons, for Gaussian and non-Gaussian distribution respectively. Cultures were analyzed using one-way ANOVA. Results were expressed as median (interquartile range or range when n < 4). P values <0.05 were considered significant.

We studied a cohort of 22 adults submitted to thymectomy in early childhood during corrective cardiac surgery and 20 age-matched healthy controls (Table 1; Table S1 in Supplementary Material). Of note, the thymic function is relatively stable in healthy individuals during the age-period spanned (4). The thymectomized patients were stratified into two groups according to evidence of residual thymic activity (Table 1; Table S1 in Supplementary Material). No thymic activity (∅Thy) strictly refers to cases with surgical reports of complete thymus removal and levels of sjTRECs clearly below the lower level found in controls (P < 0.0001, Figure 1A), as we previously reported (28). Individuals with some degree of thymic activity (Thy) featured sjTREC levels within the range of age-matched controls, though significantly lower (P = 0.0061, Figure 1A).

Interestingly, these two groups showed almost no overlap between age at thymectomy, which was performed during the first year of life in all Thy cases except two and later on in all ∅Thy patients (Table S1 in Supplementary Material). A previous study also found an association between preservation of thymic activity and younger age at thymectomy, which was attributed to loss of thymus regenerative capacity in children older than 18 months (24, 32). It is also noteworthy that complete thymic tissue removal is more likely to occur after the first year of life due to the surgical procedures required for the type of cardiac defects (Table S1 in Supplementary Material), as well as due to age-related anatomic specificities (21) Of note, individuals with syndromatic cardiac defects were not included (Table S1 in Supplementary Material).

In agreement with complete lack of thymic activity, the ∅Thy group featured decreases in naive CD4 T-cell frequency and absolute numbers, which were statistically significant not only in comparison with healthy individuals (P = 0.0012 and P = 0.0006, respectively, Figure 1B) but also with Thy (P = 0.0041 and P = 0.0019, respectively, Figure 1B). These cells featured a truly naive phenotype based on an extensive panel of naive markers and lack of expression of molecules associated with a memory phenotype, as we have previously reported (28).

On the other hand, Thy patients showed no reduction in lymphocyte counts (Table 1) and maintained the naive CD4 T-cell compartment (Figure 1B). In order to estimate their effective thymic output, we quantified the sj/βTREC ratio, which reflects the number of proliferation cycles undergone by precursor T-cells during their intra-thymic differentiation and directly correlates with thymic activity (30). We observed similar levels of sj/βTREC ratio in Thy and age-matched healthy individuals (Figure 1C). Together with close to normal sjTREC contents, this observation supports a major contribution of thymic recovery to the maintenance of the size of naive CD4 T-cell compartment upon partial thymectomy.

We further assessed the impact of the degree of thymectomy on the structural diversity of naive CD4 T-cells by spectratyping analysis of their TCR repertoire. The distribution of the CDR3 lengths within each different Vβ family is considered to reflect the overall sequence diversity (33). A diverse polyclonal TCR repertoire is associated with a Gaussian distribution of CDR3 lengths, whereas skewed TCR repertoires feature a reduced number of peaks. A relatively preserved TCR diversity was observed in Thy patients as compared to age-matched controls (P = 0.2620), supporting that their degree of thymic activity was sufficient to ensure the preservation of the quality of the naive CD4 T-cell compartment (Figures 2A,B). Patients with no thymic activity exhibited higher numbers of non-polyclonal Gaussian families than both Thy (P < 0.0001) and healthy individuals (P = 0.0001), as shown in Figure 2.

Of note, these differences between the two thymectomized cohorts could not be attributed to a distinct prevalence of CMV infection, since a similar proportion of individuals with IgG seropositivity against CMV was observed in ∅Thy and Thy cohorts (Table S1 in Supplementary Material). Moreover, the contraction of the naive CD4 T-cell compartment in ∅Thy individuals was not restricted to those seropositive for CMV, as previously reported (21).

Overall, peripheral homeostatic mechanisms were unable to prevent the contraction of the naive CD4 T-cell compartment upon complete thymus removal in infancy, whereas the maintenance of some degree of thymic activity allowed preservation of naive CD4 T-cells with a diverse TCR repertoire into adulthood.

The maintenance of naive T-cells is also determined by the rate of their differentiation into memory–effector cells. Of note, both central and effector memory CD4 T-cell counts in individuals lacking thymic activity were found to be similar to those in age-matched healthy controls (Figure 3A). Therefore, we assessed the expression of a panel of genes known to be involved in the regulation of cell quiescence and/or of the threshold for TCR-mediated cell activation in purified naive CD4 T-cells from thymectomized and healthy individuals (Figure 3B). No alterations were found in the expression levels of the following genes: Krüppel-like factor 2 (KLF-2) (34), the transcription factor FOXP1 (35), CDKN1A (encoding the cyclin-dependent kinase inhibitor p21^cip1/waf) (36), the proapoptotic Bcl-2 family member BIM (37), and the dual-specificity protein phosphatase DUSP4 (38). However, individuals with no thymic activity featured significantly higher DUSP6 transcript levels than controls, an increase not observed in those with some preservation of thymopoiesis (Figure 3B).

DUSP6 is highly specific for ERKs, leading to reduction of ERK activity that is critical for efficient TCR signaling (38, 39). Therefore, high DUSP6 levels might be associated with an increase in the threshold for TCR-induced activation in ∅Thy. To test this possibility, we performed a dose–response TCR stimulation of purified naive CD4 T-cells and quantified the upregulation of the early activation markers CD40L and CD69, in parallel with the downregulation of CD3 expression and induction of the memory marker CD45RO. Contrarily to our expectation, individuals completely lacking thymic activity responded to TCR stimulation as efficiently as healthy subjects (Figure 3C).

In conclusion, we found no apparent constraint in the differentiation of naive CD4 T-cells into the memory compartment in individuals completely lacking thymic activity.

The CD31 molecule has been shown to be expressed in all RTEs and to be lost upon TCR stimulation of naive CD4 T-cells (12, 18). Next, we investigated the contribution of peripheral cell survival and cell cycling to the homeostasis of the CD31⁺ and CD31⁻ naive CD4 T-cell subsets according to the degree of thymic activity.

The CD31⁺ compartment was preserved in Thy patients (Figures 4A,B). Conversely, it was significantly contracted in the ∅Thy cohort, both in frequency (Figure 4A) and absolute counts (Figure 4B), as expected in the absence of thymic activity (3, 20). Of note, the median level of CD31 expression within CD31⁺ naive CD4 T-cells was not significantly different in thymectomized individuals (CD31 MFI: healthy 4,494 [3,490–4,872]; Thy 3,573 [3,255–4,453]; ØThy 3,936 [2,658–4,170]; P > 0.05).

Importantly, despite the marked naive CD4 T-cell lymphopenia, ∅Thy featured preserved CD31⁻ naive CD4 T-cell counts (Figure 4B). This finding adds to previous data on aged individuals reporting preservation of the CD31⁻ compartment in parallel with the progressive decline of CD31⁺ cell counts (40) and argues in favor of the robustness of the homeostasis of CD31⁻ naive CD4 T-cells in individuals lacking thymic activity.

The expression levels of the survival marker Bcl-2 were upregulated in both thymectomized cohorts as compared to healthy controls (∅Thy: P = 0.0187; Thy: P = 0.0487; Figure 4C), suggesting increased naive CD4 T-cell survival irrespectively of the presence of thymic activity, which persisted for more than 20 years post-thymectomy (Table 1). This increase was observed in both CD31⁺ and CD31⁻ naive T-cells (Figure 4D). Of note, we found that the CD31⁻ compartment featured significantly higher Bcl-2 MFI than CD31⁺ cells both in healthy and Thy individuals, a difference that was attenuated in ∅Thy (Figure 4D).

Regarding proliferation, a significant increase in the frequency of cycling cells within total naive CD4 T-cells was found only in ∅Thy (P < 0.0001 to healthy, P = 0.0007 to Thy; Figure 4E). The proportion of Ki-67⁺ cells was significantly higher in the CD31⁻ than in the CD31⁺ compartment in both healthy individuals and in Thy patients (Figure 4F). Nevertheless, its relative contribution to the pool of proliferating naive CD4 T-cells is minor (<1%), given the large overrepresentation of CD31⁺ cells in these individuals (Figure 4A). On the other hand, ∅Thy individuals featured an increase in the relative representation of the CD31⁻ subset, associated with the loss of CD31⁺ cells (Figure 4A), in parallel with a significant increase in the frequency of cycling cells irrespective of CD31 expression (CD31⁻ subset: P < 0.0001 to healthy; P = 0.0048 to Thy; CD31⁺ subset: P < 0.0001 to healthy; P < 0.0001 to Thy; Figure 4F).

Thus, we showed that the CD31⁻ naive CD4 T-cell compartment was maintained in the absence of thymic output, in association with both expanded cell survival and increased proliferation.

Cytokine-driven homeostatic mechanisms are crucial for naive CD4 T-cell maintenance, and IL-7 is considered the key cytokine in these processes (13). Therefore, we hypothesized that naive CD4 T-cells adjust their intrinsic ability to respond to IL-7 in order to counteract the decline in thymic output. Thus, we investigated the impact of IL-7 on purified naive CD4 T-cells, using a 13d culture system previously optimized in our laboratory (14, 28). We were able to purify the required amount of naive CD4 T-cells from 5 ØThy and 11 Thy individuals, which were compared with samples from 14 healthy subjects. We found comparable ex vivo levels of IL-7Rα-chain (CD127) expression within naive CD4 T-cells in healthy and ∅Thy, and significantly higher levels in Thy individuals (Figure 5A). Others have shown that the proximal signaling through the IL-7 receptor is preserved in adults thymectomized in infancy, as assessed by STAT5 phosphorylation upon short-term stimulation with IL-7 (20).

As illustrated in Figure 5A, the cells preserved their naive phenotype upon culture with either IL-7 or IL-2, including those from ∅Thy individuals. The expected IL-7-mediated downregulation of CD127 expression (14, 41) was comparable in all individuals and, therefore, independent of the degree of thymic activity (Figure 5A). Of note, no changes occurred in the control culture condition with IL-2 (Figure 5A). Additionally, the upregulation of CD25 and CD95 by IL-7 (14, 42) was also similar in all cohorts, both in the CD31⁺ and CD31⁻ compartments (data not shown).

An important physiological role of IL-7 relies on Bcl-2 induction (43). We found a clear upregulation of Bcl-2 expression, in both CD31⁺ and CD31⁻ subsets in cultures with IL-7 in both thymectomized cohorts, which was not observed with IL-2 (Figure 5B).

We then investigated the proliferative response of naive CD4 T-cells using the cell-cycling marker Ki-67, which we have shown to be the best approach to reveal low-level IL-7-driven proliferation (14). Unexpectedly, we found that in contrast to healthy and Thy cohorts, ∅Thy featured no significant increase in the frequency of cycling cells in response to IL-7, as compared to both ex vivo and cultures with IL-2 (Figure 6A). As shown in Figure 6B, naive CD4 T-cell recovery per well was significantly higher upon culture with IL-7 than IL-2 in healthy (fold change 1.07 [0.86–1.54] versus 0.79 [0.51–1.63]; P = 0.0137) and Thy (fold change 1.15 [0.16–1.55] versus 0.65 [0.06–1.32]; P = 0.0259), but not in ∅Thy (fold change 0.87 [0.65–1.77] versus 0.54 [0.34–1.66]; P = 0.2234). The cell recovery upon culture with IL-7 was significantly lower in ∅Thy as compared to healthy individuals (P = 0.0236).

We have previously shown that the IL-7-induced proliferation was restricted to the CD31⁺ subset in healthy subjects (14), which was confirmed here (Figures 6C,D). Notably, a similar profile was observed in Thy individuals (Figures 6C,D). By contrast, patients lacking thymic activity featured reduced proliferation upon IL-7 stimulation (Figures 6C,D) and showed no significant increase in the frequency of cycling cells (Ki-67⁺) within the CD31⁺, as compared to the CD31⁻ subset (Figures 6C,D). As a consequence, the proportion of CD31⁺ within cycling cells was significantly lower in ∅Thy, as compared to both healthy and Thy individuals (P < 0.0001 and P = 0.007, respectively).

We have also previously shown that IL-7 upregulates the levels of expression of CD31 within the CD31⁺ naive CD4 T-cell subset in a PI3K-dependent manner (14, 28). Of note, the ∅Thy cohort reached significantly lower levels of upregulation of CD31 MFI within the CD31⁺ subset in response to IL-7, as compared to healthy (P = 0.0035), despite featuring comparable ex vivo levels (P = 0.1052) (Figure 6E). Moreover, the upregulation of CD31 MFI was also significantly lower in ∅Thy than in Thy (P = 0.0177) individuals, although the latter featured significantly higher ex vivo levels of CD31 MFI within the CD31⁺ subset (P = 0.0177) (Figure 6E). These findings suggest that naive CD4 T-cells from ∅Thy patients lost the ability to respond to IL-7, possibly through the PI3K pathway, in agreement with their impaired proliferation. Conversely, they preserved the ability to upregulate Bcl-2 in response to IL-7, which we have shown previously that is not PI3K-dependent (14, 28).

Altogether, these data showed that the ability of naive CD4 T-cells to proliferate and upregulate CD31 in response to IL-7 was impaired in individuals completely lacking thymic activity.