Long-term maintenance of human naïve T cells through in situ homeostasis in lymphoid tissue sites

See allHide authors and affiliations

Science Immunology  02 Dec 2016:
Vol. 1, Issue 6, eaah6506
DOI: 10.1126/sciimmunol.aah6506

T cell life doesn’t end at 40

Naïve T cells develop in the thymus. Although thymic function declines with age, T cells are persistent throughout the human life span. Thome et al. examined human lymphoid tissues from donors ranging from 2 months to 73 years in age. They found that, although the number of double-positive thymocytes and recent thymic emigrants dropped in individuals >40 years of age, naïve T cells were functionally maintained in the lymph nodes. There was minimal overlap in clonotype between the lymph tissues, suggesting that lymph nodes may maintain a diverse set of T cell specificities. These data suggest that location really does matter—tissue compartmentalization and homeostasis are critical for maintaining naïve T cells throughout the human life span.


Naïve T cells develop in the thymus and coordinate immune responses to new antigens; however, mechanisms for their long-term persistence over the human life span remain undefined. We investigated human naïve T cell development and maintenance in primary and secondary lymphoid tissues obtained from individual organ donors aged 2 months to 73 years. In the thymus, the frequency of double-positive thymocytes declined sharply in donors >40 years of age, coincident with reduced recent thymic emigrants in lymphoid tissues, whereas naïve T cells were functionally maintained predominantly in lymph nodes (LNs). Analysis of T cell receptor clonal distribution by CDR3 sequencing of naïve CD4+ and CD8+ T cells in spleen and LNs reveals site-specific clonal expansions of naïve T cells from individuals >40 years of age, with minimal clonal overlap between lymphoid tissues. We also identified biased naïve T cell clonal distribution within specific LNs on the basis of VJ usage. Together, these results suggest prolonged maintenance of naïve T cells through in situ homeostasis and retention in lymphoid tissue.


The ability to respond to new antigens is mediated largely by naïve T cells, which are generated in the thymus and emerge into the periphery by migrating through blood and lymphatics. The production of new naïve T cells from the thymus is highest at birth and during infancy, and there is an established reduction in thymic function and volume beginning in puberty (1). It is not understood how and whether human naïve T cells are maintained in the context of decreasing thymic output throughout a lifetime. Moreover, human life span continues to increase, and the ability of individuals to maintain health and be free of infectious/chronic diseases even in advanced years (2, 3) suggests that the human immune system has specific mechanisms in place for maintaining functionality over many decades. However, identifying mechanisms for preserving immunity in humans remains difficult to assess and investigate.

The capacity of T cells to recognize diverse antigens depends on their T cell receptor (TCR) specificity. TCR gene rearrangement in developing thymocytes results in each new naïve T cell expressing a unique TCR, which, in humans, can comprise more than 100 million different specificities (4). When activated by antigen/major histocompatibility complex (MHC), clones of naïve T cells proliferate and differentiate to activated/effector T cells, of which a proportion can persist as memory T cells. Although naïve T cells predominate in peripheral blood at birth, there is a gradual accumulation of memory T cells with age, and naïve T cells comprise, on average, 20 to 40% of circulating CD4+ or CD8+ T cells in adults (57). In mice, maintenance of naïve T cells is largely dependent on thymic output, whereas in humans, naïve T cell maintenance in blood appears driven by peripheral, homeostatic expansion (8), which could occur via tonic signaling or homeostatic cytokines such as interleukin-7 (IL-7) (9). It is not known whether these apparent distinctions in naïve T cell maintenance between mice and humans are due to the sampling site [spleen and lymph node (LN) in mice compared with blood in humans], life-span differences (1 to 2 years in mice versus >80 years in humans), or other factors. In humans, blood is the major accessible sample yet only contains 2 to 3% of the total T cell complement (10), whereas naïve T cells are generated in the thymus and are seeded into and become activated in secondary lymphoid organs.

We have set up a resource to obtain multiple tissues from human organ donors through a collaboration and research protocol with the organ procurement organization for the New York metropolitan area (LiveOnNY). This access to human tissues has enabled study of human T cell subsets, function, and clonal organization in lymphoid and mucosal tissues from diverse individuals of all ages (7, 11, 12). From collective analysis of more than 70 donors, naïve T cells were found to persist in frequencies of 20 to 40%, predominantly in LNs, spleen, and blood of young adults into the seventh decade of life (7, 11, 12). We hypothesized that these lymphoid sites could serve as reservoirs for long-term maintenance of naïve T cells, and their characterization could reveal mechanisms that cannot be elucidated from studies in blood. We further considered whether specific clones of naïve T cells exhibited compartmentalization as described for subsets of memory T cells (12, 13).

Here, we present a detailed analysis of human naïve T cell development and maintenance in primary and secondary lymphoid tissues obtained from individual organ donors, aged 2 months to 73 years. We dissected mechanisms for naïve T cell maintenance through analysis of TCR clonal distribution as determined by CDR3 sequencing of naïve CD4+ and CD8+ T cells in spleen and LNs from donors spanning six decades of life. Our results reveal that each lymphoid tissue site contains a unique complement of naïve T cell clones with minimal overlap between tissues, that clonal expansions of naïve T cells are observed, particularly in individuals >40 years of age, and that these expansions are contained within specific sites. Together, these results demonstrate localization-dependent mechanisms for the maintenance of human naïve T cells through in situ homeostasis, with potential effects on priming and localization of immune responses later in life.


Structural alterations and reduced double-positive thymocytes after age 40

We obtained thymus tissues from donors of different ages and examined thymus integrity by histological analysis and thymocyte composition by flow cytometry. Thymic tissue undergoes age-related atrophy, evidenced by a decrease in thymic epithelial space with organ volume further replaced by fatty tissue (14, 15). To ensure that we were isolating T cell populations from viable thymus and not fat, we performed hematoxylin and eosin (H&E) staining of isolated tissues and looked for evidence of Hassall’s corpuscles (HCs) that are a structural hallmark of the human thymus and consist of epithelial cells within the thymic medulla (16). Thymic tissues obtained from pediatric and adult donors of different ages had visible HC structures, with pediatric thymus tissue exhibiting a significantly higher density of HC of smaller size compared with adult thymi with larger HC structures (Fig. 1, A and B), demonstrating age-associated structural changes within active thymic tissue in adults.

Fig. 1 Alterations in thymus structure and diminished thymopoiesis with age.

(A) Representative histology of thymus sections from individuals of indicated ages visualized by H&E staining, shown at 10× magnification (top), with an enlarged section shown below at 40× magnification. HC structures are circular structures stained pink in each field. (B) Average number of HC structures per viewing pane in pediatric (0 to 2 years; n = 8; white) and adult (23, 25, 30, 49, 50, and 57 years; n = 6; black) donors. **P = 0.0002. (C) Representative flow cytometry plots indicating the coordinate expression of CD4 and CD8 on T cells in thymus tissue from donors of indicated ages, delineating DP and CD4 and CD8 SP subsets with percentage of each indicated in quadrants. (D) CD69 expression by DP and SP subsets from thymus tissue obtained from donors of indicated ages. (E) Compiled percentage of DP thymocytes in thymus tissue of donors of indicated ages and gender (blue, male; red, female) from 27 donors (see table S1).

Thymocytes isolated from active thymic tissue exhibited the canonical thymocyte subpopulations delineated by CD4 and CD8 expression into subsets with characteristic frequencies including double-positive (DP) CD4+CD8+ and single-positive CD4+CD8 (20%) or CD8+CD4 (10%) cells, with DP cells comprising the majority (60 to 80%) of the total thymocytes (15). We found characteristic frequencies of these thymocyte subsets in the thymus tissue of young donors with low frequencies of DP thymocytes in the thymus of older donors (Fig. 1C, see fig. S1 for gating strategy). These subsets also exhibited characteristic CD69 staining for thymocyte subsets (17, 18), with DP exhibiting lower CD69 expression than CD4+ and CD8+ subsets (Fig. 1D, bottom). The frequency of DP populations was consistently 60 to 80% of thymocytes from infancy up until the fourth decade of life, and after age 40 to 50 years, there was a steep decline in the percentage of DP cells to <15% of thymocytes (Fig. 1E). These findings show that active thymopoiesis does not exhibit a gradual decline but may cease abruptly at some discrete point in time after 40 years of age.

Naïve T cells are differentially maintained on the basis of lineage and lymphoid tissue site

We investigated the maintenance of naïve-phenotype T cells (CD3+CD45RA+CCR7+ CD4+ or CD8+ cells; see gating strategy in fig. S1) in circulation, lymphoid, and mucosal tissues as a function of different age groups associated with high or low levels of thymopoiesis based on the results in Fig. 1. We stratified the age groups into pediatric donors (2 months to 5 years), young adult donors (15 to 39 years) with active thymic output, and middle-aged/older adult donors (>40 years, with most of the donors between 40 to 60 years of age) with diminished thymic output. In pediatric donors, appreciable proportions of naïve T cells could be found in all sites examined, with the lowest proportions seen in the small intestine and lungs, as previously described (11). In young and older adults, the proportion of naïve T cells in mucosal tissues was negligible; however, substantial proportions were found in blood, spleen, and LNs (Fig. 2A). Between young and older adult age groups, the proportion of naïve T cells decreased significantly in spleen, inguinal LN (ILN), and lung-draining LN (LLN) for both CD4+ and CD8+ T cells, and in blood for CD8+ T cells, whereas frequencies of naïve T cells in mesenteric LN (MLN) were similar in all adult age groups for both CD4+ and CD8+ T cells (Fig. 2A, compare dashed to solid black lines, and table S2), indicating differential maintenance in various sites. Even in the older adult ages, up to 20% of CD4+ T cells and 30 to 40% of CD8+ T cells were maintained as naïve-phenotype cells in LNs, suggesting long-term maintenance of naïve T cells without continuous export from the thymus.

Fig. 2 Naïve CD4+ and CD8+ T cells are differentially maintained in lymphoid sites.

(A) Mean frequencies ± SEM of naïve (CCR7+CD45RA+) CD4+ (left) and CD8+ (right) T cells in indicated tissues from donors stratified into three age groups: pediatric (0 to 2 years; n = 18; red), younger adult (15 to 40 years; n = 27; black dashed), and older adult (41+ years; n = 24; black) donors. See individual means, SEM, and P values in table S2. (B) Naïve CD4+ (top) or CD8+ (bottom) T cells within four indicated lymphoid tissue sites as a function of age, with each dot representing an individual donor tissue from a total of 69 donors (see table S1). Average percent naïve CD4/naïve CD8 T cells in each site: SP, 31/35 (P = 0.16); ILN, 57/42 (P < 0.001); LLN, 49/38 (P < 0.001); MLN, 48:41 (P = 0.01). (C) Naïve T cell frequency in multiple tissues of individual donors (total, 40), where all four tissue sites were obtained, shown as a heat map showing relative percent naïve T cells in each site ranked by age in donors.

We further investigated the dynamics of naïve T cell loss and maintenance in specific lymphoid sites (spleen, ILN, LLN, and MLN) for both CD4+ and CD8+ T cells and their differential persistence in tissues within individuals (Fig. 2, B and C). Overall, there was a higher proportion of naïve CD8+ T cells maintained in LNs examined at all ages compared with naïve CD4+ T cells (Fig. 2B), with steep reductions in naïve T cell frequency in ILN, LLN, and MLN by 30 to 40 years of age for CD4+ T cells, whereas frequencies for CD8 T cells were maintained in these sites in certain individuals until >50 years of age (Fig. 2C). In contrast to the differential maintenance of naïve CD8+ compared with CD4+ T cells in LN, the spleen exhibited a similar steep decline of naïve T cells beginning in the young adult years (early 20s) for both CD4+ and CD8+ T cells (Fig. 2, A and B), which could be attributed to the higher turnover of naïve T cells in spleen versus LN based on an analysis of Ki67 in a limited number of donors (fig. S2). Together, this analysis reveals previously unknown disparities in maintenance of naïve T cells over life in different lymphoid compartments, with certain LNs serving as potential niches for long-term persistence of naïve-phenotype cells, and with spleen representing a more transient compartment for naïve T cells during youth.

Loss of recent thymic emigrant content and compartmentalization in lymphoid tissues over 40 years of age

Recent thymic emigrants (RTEs) can be detected in the periphery by the presence of extrachromosomal DNA resulting from the TCR-δ gene rearrangement event, which occurred during thymic development (1921). These TCR excision circles (TRECs) are present at measurable levels in RTE but gradually become diluted out with each cell division because of homeostasis or antigen-driven activation (21). We sorted naïve (CD45RA+CCR7+) CD4+ and CD8+ T cells from lymphoid sites (spleen, ILN, LLN, and MLN) and SP CD4+ or CD8+ thymocytes and assessed TREC content using a well-established polymerase chain reaction (PCR) assay (Fig. 3A; see Materials and Methods) (21, 22).

Fig. 3 Naïve T cells retain functionality independent on the extent of RTEs.

(A) RTE content among naïve T cells was measured by quantitation of TRECs using a PCR-based approach (see Materials and Methods). Graphs show TREC content per 100,000 sorted CD4+ (top) and CD8+ (bottom) naïve T cells from indicated lymphoid sites (circle, spleen; upside-down triangle, ILN; square, LLN; triangle, MLN), CD4+CD8 or CD8+CD4 cells from each thymus indicated by “X”, and total T cells from tissues obtained from the 17- and 73-year-old donors. Each color represents an individual donor (individual values in table S3). (B) Naïve and TEM CD4+ and CD8+ T cells were sorted from spleen, ILN, and MLN, stimulated with anti-CD3/CD28/CD2 beads, and the cytokine content in supernatants was assessed using the BD Cytokine Bead Array Kit (see Materials and Methods). IL-2 and IFN-γ production (in picograms per milliliter; means ± SEM) by naïve CD4 and CD8 T cells isolated from tissues of donors <35 years of age (two to four donors; white bars) and >50 years of age (two to four donors except for spleen CD4 T cells, which are from one donor; black bars) is shown. (C) IL-2 and IFN-γ production by naïve and memory (TEM) CD4 and CD8 T cells from the LLN of individual donors <35 years of age (four donors: aged 29, 25, 26, and 34) and >50 years of age (four donors: aged 54, 56, 52, and 59). Cytokine levels are normalized by donor and indicated by Z score [(cytokine level − mean cytokine level)/SD]. Values for each cytokine measured are shown in table S4.

On average TREC levels at a given age were similar between naïve CD4+ and CD8+ T cells. However, two types of age-associated changes in TREC levels were observed. First, there was an overall decrease in TREC content of naïve CD4+ and CD8+ T cells and thymic SP cells with age, with the most notable reduction in TREC levels to <1000 after 40 years of age (Fig. 3A; see table S3 for individualized TREC values), providing additional evidence for an abrupt reduction in thymic output in humans after age 40.

The second age-associated change in TREC levels was observed between tissues within an individual donor. In pediatric and young adult lymphoid tissues, the TREC content of naïve T cells was not equivalent within a single individual [Fig. 3A (different tissue values for each individual share the same color) and table S3]. For example, in a 3-year-old donor (donor 82; table S3), the highest TREC content of naïve CD4+ T cells was found in the thymus followed by ILN and then spleen, whereas for a 12-year-old donor, higher TREC levels were detected in splenic naïve T cells, followed by LLN and ILN (Fig. 3A and table S3). Dissimilar TREC levels between naïve T cells isolated from different sites were observed mostly in individuals younger than 40 years (Fig. 3 and fig. S3), with individuals older than 40 years having comparable low TREC levels in distinct sites. Variations in TREC content between lymphoid sites of an individual could reflect differential thymic seeding and/or differences in activation or maintenance of naïve T cells in situ.

We also compared CD31 expression by naïve CD4+ T cells in donors of different ages (fig. S4), as a marker of RTE (23). Although the overall frequency of CD31 on total CD4+ T cells was indicative of reduced thymic output and seeding of naïve T cells in mucosal sites in adults (fig. S4, A and B), its expression by naïve-phenotype cells in adults did not decrease, coincident with the decreased TREC levels found in donors (fig. S4C). These results suggest that CD31 is not a good marker to gauge thymic output in adults.

Naïve T cell subsets maintain functionality with age

To assess whether naïve T cells in lymphoid tissue maintained their functionality and naïveté at different ages associated with the presence or absence of thymic output, we measured the cytokine profile of naïve CD4+ and CD8+ T cells isolated from lymphoid tissues after anti-CD3/anti-CD28–mediated activation ex vivo. We assessed the production of multiple cytokines in culture supernatants to determine whether naïve T cells predominantly produced IL-2 or had acquired the capacity to produce effector cytokines more associated with memory T cell responses including interferon-γ (IFN-γ). However, naïve T cells from all tissue sites (spleen, ILN, and LLN) and from donors of diverse ages predominantly produced IL-2 with low to negligible levels of IFN-γ (Fig. 3B and table S4), IL-4, and IL-10 (fig. S5). Memory T cells, by contrast, exhibited an enhanced capacity to produce substantial levels of IFN-γ, IL-4, and IL-10 compared with naïve T cells within the same site from the same donor (Fig. 3C, fig. S5, and table S4). Together, these results indicate that naïve T cells defined by phenotypic expression of CD45RA and CCR7 remain functionally naïve even in the presence of waning thymic output.

TCR sequencing of T cell populations shows differences in sequence diversity both with age and with subset

The above analyses revealed that LNs served as potential reservoirs for functional maintenance of naïve T cells and that naïve CD4+ and CD8+ T cells were differentially maintained as a function of age. To gain new insights into the mechanisms for human naïve T cell maintenance in these key sites during periods of active (<40 years of age) and low (>40 years) thymic output, we analyzed the clonal distribution of naïve T cells within and between tissue sites using the immunoSEQ platform to sequence all possible human TCR CDR3β sequences (24, 25). Use of immunoSEQ has been applied to dissect the clonal origin of memory subsets (26) and to detect antigen-specific T cell clones in clinical samples (27, 28). Here, we applied immunoSEQ to assess how human naïve T cells were clonally distributed in spleen (SP), ILN, and LLN. We extracted DNA samples from CD4+ and CD8+ T cells from a total of 19 donors, including naïve subsets from 13 donors and effector memory (TEM) populations from 10 donors, of which naïve and TEM subsets were analyzed together in tissues from 4 of these donors (table S5). From these data, we assessed naïve TCR diversity in different tissues and donors as a function of age, and the tissue overlap of individual naïve T cell clones regarding the extent to which a clonal population in one tissue is found in other sites from the same individual. We compared these aspects of TCR clonal diversity and distribution in the corresponding TEM subset because our previous studies had revealed differences in diversity and tissue overlap in CD4+ and CD8+ TEM cells, although age was not a contributing factor (7).

For all T cell subsets used for TCR analysis, we obtained read numbers of adequate size for calculating diversity and tissue overlap (105 to 106 reads per sample; see table S5). We defined a distinct clonotype by nucleotide sequence and assessed the clonal diversity of naïve T cells in tissues as a function of age using Simpson’s index, a diversity measure that gives the average probability that two clonotypes randomly selected from a population are identical (see Materials and Methods). Analysis of naïve TCR repertoires from spleen, ILN, and LLN from 13 donors aged 1 to 60 years allowed us to assess the overall influence on diversity due to decreasing thymic output. We identified an overall drop in diversity (increased Simpson’s index) after 40 years of age, though with high variability across donors (Fig. 4A). For naïve CD8+ T cells, there was a broader range of values between samples of a given age range, yet higher diversity was still observed at ages >40 years (Fig. 4A, right). We did not find significant differences in naïve T cell diversity and/or clonal expansion between lymphoid sites when comparing spleen, ILN, and LLN from all donors examined (Fig. 4A and fig. S6).

Fig. 4 TCR repertoire diversity of naïve T cells decreases with age and is distinct from TEM subsets.

Naïve (CD45RA+CCR7+) CD4+ and CD8+ T cells were sorted from spleen, ILN, and MLN, and CDR3β sequences were amplified and sequenced (see Materials and Methods). (A) Repertoire diversity within spleen (red), ILN (blue), and LLN (green) tissues is quantified in bulk by Simpson’s index (see Materials and Methods) and plotted against donor age, as shown for CD4+ (left) and CD8+ (right) lineages compiled from 11 and 13 donors, respectively (see table S5), with each dot representing a single tissue from an individual donor. (B) Maximum clone frequency is higher for memory T cells than for naïve T cells. The frequency of the largest sequenced clone in the sample is plotted for CD4+ and CD8+ T cells for both naïve and TEM subsets. Significant P values based on Student’s t test are indicated between CD4+ and CD8+ TEM cells and between naïve and TEM populations within the CD4+ or CD8+ lineage. (C) Clonal diversity of every observed VJ combination as computed by Shannon entropy for naïve and TEM subsets for CD4+ (top) and CD8+ (bottom) lineages. Naïve populations are depicted by triangles, and TEM populations are depicted by circles. Clones were pooled from every donor and separated by their VJ cassette. The curve log2N depicts the maximum possible diversity for a fixed number of clones.

We further assessed how expansion of naïve clones compared to antigen-experienced TEM clones. Because Simpson’s index assigns greater importance to clones present at a higher frequency, we posited that the frequency of the largest clone in each T cell population would be sufficient to distinguish differences in the diversity of these two subtypes (Fig. 4B). Among the TEM repertoire, the highest-frequency clones were significantly greater than those in the naïve repertoire, with the most expanded TEM clones present in frequencies 50- to 200-fold greater than the maximally expanded naïve T cell clones (Fig. 4B). Moreover, naïve and TEM repertoires for CD8+ T cells were more expanded than the corresponding CD4+ T cell subsets (Fig. 4B).

TCR diversity of naïve and TEM cells was compared on the level of VJ cassette recombination. Clone frequencies were calculated for all samples, and samples of the same lineage (CD4+ or CD8+) and subset (naïve or TEM) were pooled together. For all clones generated by the same VJ cassette, pair diversity was computed by Shannon entropy (Fig. 4C and Materials and Methods). Across different VJ pairs, the Shannon entropy of naïve CD4+ and CD8+ T cell repertoires was close to the maximum possible diversity, log2N, where N is the number of clonotypes generated by a particular VJ pair. By contrast, the VJ entropy of TEM samples was much lower than that of naïve T cells and was significantly reduced compared with the maximum. Together, the diversity and VJ entropy of naïve and TEM cells in lymphoid sites indicate that naïve T cell clones exhibit considerably less clonal expansion than TEM clones, with naïve T cell clonal expansions observed more frequently in individuals >40 years of age.

Lymphoid tissues share very few naïve clones regardless of donor age

A great advantage of our samples from organ donors is the ability to compare clonal distribution of naïve (or TEM) subsets between distinct sites, enabling an assessment of whether clonal expansions observed with naïve T cells resulted in sharing of specific clones between tissue sites. We assessed sequence overlap among the top 1000 clones for naïve and TEM cells from each of three tissues within individuals of different ages. (Actual numbers were slightly greater than 1000 to account for multiple clones observed with the same read count.) As previously reported (7), and consistent with additional donors from all ages whom we examined (Fig. 5A and fig. S7), there was an appreciable overlap in TEM clones between sites, with 20 to 30% of CD4 TEM clones and >40% of CD8 TEM clones found in more than one tissue site. By contrast, there were remarkably few naïve T cell clones found in more than one site, with the vast majority of naïve T cell clones unique to either spleen, ILN, or LLN from the same donor (Fig. 5A). This minimal sharing of naïve T cell clones between tissue sites was observed in all donors examined independent of donor age (Fig. 5A and fig. S7).

Fig. 5 Naïve repertoire exhibits minimal sharing between tissues.

The nucleotide sequence of each clone in multiple tissues from individual donors was analyzed for clonal overlap. (A) Venn diagrams displaying nucleotide sequence overlap of the top 1000 clones for naïve and TEM CD4+ (top) and CD8+ (bottom) cells between three tissue sites (red, spleen; blue, ILN; green, LLN) of a 21-year-old donor (donor 125, left) and a 51-year-old donor (donor 201, right). Actual numbers are slightly greater than 1000 to account for additional clones present in identical read numbers to the 1000th clone. (B) Overlap of naïve (gold) or TEM (blue) clones between tissue pairs as a function of clone frequency (quantified as read count) for the representative donors shown in (A). For clones with a given read count in the first tissue, the fraction of overlap with all clones in the second tissue is plotted. Counts greater than 20 are logarithmically binned into 25 bins. (C) Overlap versus clone frequency plot calculated as in (B), where, for every tissue, clones have been pooled from multiple donors analyzed. The number of donors is indicated in each individual graph.

Given that naïve T cell clones were present in reduced overall frequencies compared with TEM cells, it was important to establish that the lack of overlap of naïve T cell clones between sites was not due to sampling or other quantitative differences in clone frequency. We therefore investigated how the fraction of clonal overlap between two tissues was related to the clonal read count or frequency for naïve and TEM cells for each donor (Fig. 5B). In the two representative donors, the overlap frequency for CD4+ and CD8+ TEM cells is low but measurable for clones with lower read counts, following a steep linear increase after a certain read count, with the most expanded clones having a high probability of being detected in both tissues (Fig. 5B, blue curve). By contrast, the clonal overlap versus read count curve for naïve T cells shows a greatly reduced association compared with that for TEM cells and differs as a function of age and for CD4+ T cells versus CD8+ T cells (Fig. 5, B and C). For naïve CD4+ T cells, clonal overlap is negligible for all read counts in the younger and older donor, showing a flat line (Fig. 5B), indicating minimal or negligible clonal overlap. For naïve CD8+ T cells, higher clone frequencies are associated with clonal overlap for the 21-year-old donor, but not for the 51-year-old donor (Fig. 5B). In the 51-year-old donor, the expanded naïve T cell clones did not exhibit overlap between sites even when compared to a similar quantitative expansion of memory T cell clones (Fig. 5B, bottom). When all data were compiled from 14 donors, minimal tissue sharing between expanded populations of naïve T cell clones is even more apparent, particularly for CD4+ T cells (Fig. 5C). Together, these quantitative analyses reveal an unexpected compartmentalization of expanded populations of naïve CD4+ and CD8+ T cells in lymphoid sites, suggesting in situ expansion during their maintenance in vivo.

We also sequenced naïve and TEM cells from replicate samples to calculate the detection power at a given frequency (see Materials and Methods and table S6). Clonal overlap was calculated for each set of replicates and for corresponding nonreplicate tissue samples from the same donors. For every read count, the ratio between the fraction of clones shared among replicates was compared to the fraction shared between different tissues to obtain an average overlap rate. This analysis yielded rates between 0.1 and 0.3 for naïve CD4+ T cells, between 0.3 and 0.5 for naïve CD8+ T cells, between 0.6 and 0.75 for TEM CD4+ T cells, and between 0.6 and 0.9 for TEM CD8+ T cells (table S6). Compared to baseline overlap frequencies from the replicate analysis, the negligible frequency of intertissue overlap of naïve T cell clones is still consistent with in situ maintenance.

Distribution of VJ cassette use among naïve T cells becomes more dissimilar with age

We hypothesized that the distribution of VJ combinations would be similar between T cells in different sites when seeded by the thymus but may diverge when maintained independent of thymic output. We tested this using Jensen-Shannon distance (JSD; see Materials and Methods) (29). JSD is a measure of the distance between two probability distributions, with two identical distributions having a distance of 0 and two maximally different distributions having a distance of 1. The VJ distribution for the 21-year-old donor is similar for both SP and LLN, and hence, the value of JSD is low (Fig. 6A, left). However, VJ usage for these tissues in the 51-year-old donor exhibits many differences, and consequently, the intertissue distance is significantly larger (Fig. 6A, right). Computing JSD for naïve T cells in tissue pairs for each donor reveals a clear increase in VJ distance among older donors for CD4+ T cells, with a similar trend for CD8+ T cells, consistent with the biased expansion and maintenance of naïve T cells in specific sites (Fig. 6B).

Fig. 6 VJ usage between tissues shows divergence of the naïve T cell repertoire among older donors.

(A) The top 50 VJ frequencies for CD4+ T cells from spleen and LLN are plotted side by side for two representative donors, aged 21 and 51. Corresponding VJ distance for the sample is given in the top right corner. (B) The VJ distance for eight donors (seven CD4+ and seven CD8+). Distance is computed for every pair of tissues from the same donors and plotted against donor age. Distances between spleen and LLN (white), between ILN and spleen (gray), and between LLN and ILN (black).


The maintenance of naïve T cells over a lifetime ensures that the immune system can respond to new antigens not previously encountered, reducing the likelihood of succumbing to infectious pathogens at different life stages. The extent to which naïve T cells are maintained in humans and the mechanisms for their long-term persistence have proved difficult to address on the basis of the limited sampling of peripheral blood. Here, we took an approach not previously reported to analyze human naïve T cell development, maintenance, and repertoire in primary and secondary lymphoid sites from a total of 128 organ donors spanning over seven decades of life. Our findings identify LNs as major reservoirs for the maintenance of naïve T cells and a diverse TCR repertoire in the presence of waning thymic output, which is markedly diminished after age 40. We further reveal tissue compartmentalization and in situ homeostasis as previously unknown mechanisms for preserving naïveté in the human T cell compartment.

Age-associated changes in the thymus are well documented and include reduction of thymic volume, loss of thymic epithelial cells, increase in the perivascular space, and predominance of adipose tissue (15, 30, 31). From our analysis of thymic tissue of different ages, DP CD4+CD8+ thymocytes were 60 to 80% of the total thymocytes up until the fifth decade of life, after which there is a significant reduction of 5 to 15% in DP cells. Although this overall conclusion that precipitous changes in thymic output are occurring in middle age has not previously been emphasized, low and variable DP frequencies and reductions in thymocyte number were previously reported in adult compared with pediatric thymi (32, 33). The precise mechanisms for this decline in thymopoiesis are unclear but could be due to alterations in thymic epithelial cells, consistent with the finding that expression of the FoxN1 transcription factor required for thymic function in mice (34, 35) was recently found reduced in adult thymi (36).

This steep reduction in thymic output after 40 years of age also paralleled the reduction in TREC levels (as indicators of RTE) in naïve T cells from spleen and two LN sites. Previous studies showed reduced TREC levels with age in peripheral blood naïve T cells, with 10-fold reductions in TREC levels between the third and fifth decades of life (8, 37). Here, we reveal differential TREC content in lymphoid tissue naïve T cells in younger individuals (<40 years of age) but equivalent TREC levels between tissues in older individuals (>40 years of age). In mouse models, RTEs become mature naïve T cells upon entry into LNs (38), suggesting that unequal TREC content in tissues may indicate active thymic output to certain sites. Together, our results suggest an age-related program involved in cessation of thymic output during the middle years of life.

Despite these reductions in thymic export, 20 to 40% of T cells in LNs (and lower percentages in spleen) are maintained as phenotypically and functionally naïve. To dissect mechanisms for this long-term persistence of naïve T cells, we used deep sequencing of TCRβ chains in naïve T cells in spleen, ILN, and LLN of donors aged 1 to 60 years. Integrating naïve TCR results from all donors and all sites reveals that the TCR diversity of naïve T cells is largely maintained in tissues, with clonal expansions of specific naïve T cell clones observed mostly in donors >40 years of age but still much reduced compared with those observed with memory T cells. This modest loss of diversity of naïve T cells with age has also been reported by deep sequencing of TCR of naïve T cells in blood (4). Thus, human naïve T cells in tissues preserve both their diversity and functionality independent of thymic output.

Unexpectedly, we found minimal overlap of expanded naïve T cell clones between individual lymphoid sites in donors of all ages, especially for CD4+ cells. Overlap of highly expanded clones of naïve-phenotype CD8+ T cells was detected; however, subsets of human primed CD8+ T cells can exhibit naïve phenotypes (39, 40). By comparing clonal overlap as a function of clone frequency, we demonstrate that for a given clone frequency, most of the naïve T cell clones are detected in a single site, whereas memory T cell clones are detected in multiple sites. These results suggest two possible mechanisms for naïve T cell maintenance: first, that naïve T cells remain resident in specific sites and do not readily migrate between tissues and, second, that clonal expansions are occurring within specific sites through in situ signals, particularly for older donors with negligible thymic output.

In contrast to our finding of specific clones of naïve T cells being specific to a tissue site, it is generally understood from mouse studies that naïve T cells continuously recirculate between lymphoid sites, lymph and blood (41). In mice, adoptive transfer of naïve T cells results in dissemination to multiple secondary lymphoid organs (42) via CCR7 expression (43). Naïve T cells in mice require cognate interactions of the TCR with MHC molecules for survival and/or functional maintenance (44) and homeostatic cytokines such as IL-7 (45, 46)—interactions that are more likely to occur in LNs and not during transit through circulation.

It is now recognized that a substantial proportion of mouse (and human) memory T cells can be retained in tissues as tissue-resident memory T cells distinguished by CD69 expression, which is also a marker of early T cell activation (7, 12, 13, 47). However, tissue residence has not been previously associated with naïve T cells. We found that, in human tissues, ~20% of naïve T cells up-regulate CD69, indicating signaling or potentially transient retention in LN tissue sites. Because mouse naïve T cells do not exhibit CD69 up-regulation and are largely maintained by thymic output throughout life (8), it is possible that their migration behaviors may be distinct from those of long-lived human naïve T cells exhibiting peripheral homeostatic expansion. We propose that, in humans, retention of naïve T cells in LNs may be required for their long-term preservation.

The site-specific clonal expansion of naïve T cells identified here suggests a nonrandom nature to naïve T cell persistence in a particular site. Elegant studies in mice showed that naïve CD4+ T cell survival was linked to their clonal abundance and specificity (48, 49). Because human naïve T cells need to persist for decades, we propose that cognate interactions may determine which naïve T cell clones are retained or survive in a particular site. This bias in naïve T cell specificity in different sites is supported by our VJ usage difference results, revealing a distinct repertoire skewing in one lymphoid site compared with another that was most notable after cessation of thymic output. These results indicate a biased expansion of specific clones in tissue sites.

Our study provides a “snapshot” of human T cell subset composition within lymphoid and mucosal sites in individuals over a broad age range that spans seven decades of life. There are however, caveats in our study that need to be considered. We did not specifically examine antigen- or pathogen-specific responses. The maintenance of antigen-specific naïve T cells may follow different kinetics or dynamics from the phenotypic naïve subsets examined here. There are caveats to the TCR clonal analysis, which are based on sequences obtained from a fraction of the total naïve T (or TEM) cells in a particular tissue—it was not possible to sequence every single T cell in human spleen, for example, because of the impracticality of this endeavor. We used bioinformatics analysis and calculations of frequency to determine clonal overlap and cannot rule out that there are clones that were not captured by our analysis that may exhibit different behavior with regard to tissue overlap.

In conclusion, our investigation of T cells in human primary and secondary lymphoid organs has revealed new insights into naïve T cell maintenance that cannot be extrapolated from sampling of peripheral blood. Our findings provide alternative mechanisms for the conservation of the naïve immune response by in situ homeostasis and maintenance in lymphoid tissues that may be specific to humans. These results that show site-specific repertoires for naïve T cells have implications for the design of vaccines and immunotherapies for promoting and regulating immune response, particularly in the middle years and beyond.


Study design

Research objectives

This study aimed to examine naïve T cell longevity and mechanisms for maintenance and to correlate thymic output in human lymphoid organs.

Research samples

Tissue samples from organ donors aged 2 months to 73 years were used.

Experimental design

Lymphocytes were isolated from tissues, and CD4+ and CD8+ T cell subsets were analyzed by flow cytometry for phenotypic parameters; by histology, cytokine production, and PCR for detection of TRECs; and by DNA sequencing of TCR genes.


Tissues were obtained from brain-dead organ donors in the New York metropolitan area and were obtained as available. All donor data obtained were included in this study. Results were not blinded.

Sample size

Data are compiled from tissues from 128 organ donors aged 2 months to 73 years. The number of donors analyzed for each different experimental approach is indicated in the figure legends and text.

Acquisition of human tissues

Human tissues were obtained from deceased (brain-dead) organ donors at the time of organ acquisition for clinical transplantation through an approved research protocol and material transfer agreement with LiveOnNY. All donors were free of chronic disease and cancer and were hepatitis B–negative, hepatitis C–negative, and HIV-negative (table S1). Tissues were collected after the donor organs were flushed with cold preservation solution and after clinical procurement was completed. Acquisition of these samples does not qualify as “human subjects” research, as confirmed by the Columbia University Institutional Review Board (IRB), because tissues were obtained from deceased individuals. In some cases, thymus tissue was also obtained as discarded tissue from patients undergoing pediatric cardiac surgery through the Human Studies Core of the Columbia Center for Translational Immunology (CCTI), with IRB approvals maintained by this core. Thymus tissue was removed by trained cardiothoracic surgeons during organ donor acquisition and was further confirmed by H&E staining.

Lymphocyte isolation from tissue sites

Tissue samples were maintained in cold saline and brought to the laboratory within 2 to 4 hours of organ procurement. Samples were rapidly processed using enzymatic and mechanical digestion as previously described (11, 12, 50), resulting in high yields of viable lymphocytes.

Thymus histology

Thymus tissue samples obtained from donors as described above were cryopreserved and fixed in optimum cutting temperature matrix (Tissue-Tek) and maintained at −80°C before sectioning. Sections were stained with H&E by the Histology Core service within the Department of Pathology, Columbia University Medical Center. The presence and number of HCs were counted on the basis of structures per viewing frame at 10× magnification in three separate tissue sections.

Flow cytometry analysis and sorting

For analysis of cell surface markers via flow cytometry, single-cell suspensions were stained with fluorochrome-conjugated antibodies in flow cytometry staining buffer (1% fetal bovine serum/0.1% sodium azide in phosphate-buffered saline) presented in table S5. Control samples included unstained and single fluorochrome–stained compensation beads (OneComp eBeads, eBioscience). Stained cells were analyzed using LSR II or FACSCanto Flow Cytometer (BD Biosciences) with FACSDiva (BD Biosciences) and were analyzed using FlowJo software (Tree Star). For isolation of subsets, T cells stained as described above were sorted using BD Influx (BD Biosciences), with single-cell compensation controls acquired as described above. Representative gating strategies used for the thymocytes and T cell subsets are shown in fig. S1.

T cell stimulation and cytokine analysis

T cells were cultured in 96-well plates (100,000 cells per well) with or without anti-CD2/CD3/CD28–coated beads (1 bead:1 cell) for 2 days in Clicks medium at 37°C. Supernatants from T cell cultures were analyzed for cytokine content by cytometric bead array (CBA) using the BD Biosciences Human Th1/Th2 Cytokine Kit II. Standard analytes were acquired on the BD LSR II Flow Cytometer using provided templates, and a standard curve was generated to calculate sample concentration.

Quantification of human TRECs

Naïve (CCR7+/CD45RA+) CD4+ and CD8+ T cells were sorted from spleen, ILN, LLN, and MLN of individual donors. Single positive CD4+ and CD8+ thymocytes were sorted from thymus tissues, and cell pellets were stored at −80°C before analysis. TREC content was quantified using an established real-time PCR approach with a standard curve of known molecules of human TREC (51). The following primer and probe sequences were used: 5′ primer, 5′-CACATCCCTTTCAACCATGCT-3′; 3′ primer, 3′-GCCAGCTGCAGGGTTTAGG-3′; probe, 5′-6-FAM-CAGGGCAGGTTTTTGTAAAGGTGCTCACTT-3′BHQ1 (Black Hole Quencher).

Statistical analysis and data visualization for cellular data

Descriptive statistics (means, SDs, and counts) were calculated for each T cell subset and tissue in Microsoft Excel. Frequency variance was determined for each subset and tissue by Holm-Sidak post hoc multiple comparison following two-way analysis of variance (ANOVA) to exclude subset-dependent effects in GraphPad PRISM (GraphPad Software Inc.). The resulting two-tailed P values and r values were graphed in Microsoft Excel and GraphPad Prism.

TCR sequencing

Naïve (CD45RA+CCR7+)–phenotype and TEM (CD45RACCR7)–phenotype CD4+ and CD8+ T cells were sorted from two to three whole LNs for ILN and LLN and human spleen (7-cm2 pieces) from individual donors (table S1). In some cases, DNA was isolated from cell pellets using the AllPrep DNA/RNA Mini Kit (QIAGEN) in conjunction with QIAshredder columns (QIAGEN), and DNA concentration was assessed using NanoDrop (Thermo Scientific). Either cell pellets or DNA was sent to Adaptive Biotechnologies for TCRβ deep sequencing using the immunoSEQ platform (24). Cell number and productive reads for each population from each donor are presented in table S5.

TCR data acquisition and quality control

The immunoSEQ platform used for TCR sequencing has a standardized protocol that uniquely identifies and amplifies the CDR3, using spiked-in synthetic templates and clustering techniques to correct for PCR and sequencing errors (24, 27). The resultant sequencing data were downloaded from Adaptive servers (nucleotide, amino acid, V and J genes, and read counts). Data sets were filtered to select for productive sequences (in-frame, absent any premature stop codons) using the “frame type” designation (Adaptive Biotechnologies). We verified data quality by subsampling our data and identifying plateaus in the number of unique clones observed, indicating saturation of sequenced chromosome fragments.

To identify contaminating clones within a donor due to sorting (which is 99% accurate), we applied a filter to remove low-level contamination between CD4+ and CD8+ samples. For each clone observed in any of the samples in a donor, we assigned CD4+ and CD8+ identity on the basis of maximum frequency (p) among all samples from the same donor. For a clone assigned as CD4+, if it was present in any of the CD8+ samples from the same donor with frequency <0.5p, then this clone was removed from the CD8+ samples as a contaminant from CD4+ (and vice versa); otherwise, this clone was determined to be ambiguous and was removed from all samples. On average, the fraction of reads that were filtered was <0.4% from naïve samples and <1.5% from TEM samples (fig. S8).

Statistical methods for analyzing TCR sequencing data

Clonal diversity was measured using Simpson’s index and entropy. Simpson’s index was used to quantitate diversity for naïve T cells, which exhibited low clonal expansion, and is defined as the sum of the squared clonal frequencies:Embedded ImageShannon entropy was used to compare diversity of clones from each type of VJ gene combination. Entropy is defined as the negative expected value of the log of observed clonal frequenciesEmbedded Imageand provides a balance between species richness (number of unique sequences) and evenness (their frequency in the population); it is closely related to the JSD used in divergence calculations (29), which is given by the following equation:Embedded ImageJSD was computed for each donor between every pair of tissues P and Q, as previously described (29). Distance was computed for VJ gene pairs. Calculations were performed in R.

Measuring clonal overlap between tissue sites

The top 1000 clones were selected by nucleotide sequence, with additional clones included if they were present at the same read count as the thousandth clone in the sample after sorting by read count. Clonal overlap was measured as the number of nucleotide sequences found in all samples being compared. Overlap between pairs of tissues across frequency was computed in R, with read count of clones in tissue 1 present on the x axis and frequency of overlap with all clones of tissue 2 on the y axis. To resolve cases with only a few clones observed at larger counts resulting in an overlap frequency of 0 or 1, clones present at read counts greater than 20 in tissue 1 were grouped into 25 bins of equal size on the log10 scale. Identical sequences from different donors were kept separate to avoid changing read counts corresponding to a sample. The same approach was then applied to the aggregated data across the two tissues.

To correct for variability in detection power of shared clones due to differences in the average clonal frequency, we used replicate sharing as a baseline for intertissue sharing (see table S6). For every read count, we compared the fraction (f1) of overlapping clones observed between two sites with the fraction (f0) of overlapping clones from replicates, defining overlap rate as r = f1/f0 for naïve and TEM samples. An average ratio over all read counts was then computed, weighted by the number of clones present in each bin to adjust for detection power.


Fig. S1. Flow cytometry gating strategy.

Fig. S2. Naïve T cell proliferative turnover in tissues.

Fig. S3. Compartmentalization of RTEs in tissues.

Fig. S4. CD31 expression by CD4+ T cells in donor tissues with age.

Fig. S5. IL-4 and IL-10 production by naïve and memory T cells in lymphoid sites.

Fig. S6. TCR diversity of naïve T cells in lymphoid sites.

Fig. S7. Clonal overlap of naïve and memory T cells between tissues of individual donors.

Fig. S8. Selection of productive naïve TCR sequences by filtering.

Table S1. Donor information and figure usage for this study.

Table S2. Descriptive statistics for naïve T cell frequencies in different tissue stratified by age groups.

Table S3. TREC values for naïve T cells in thymus and lymphoid tissue of individual donors.

Table S4. Source data for all cytokines measured in this study.

Table S5. Summary TCR sequencing data for all tissue naïve and TEM cells analyzed.

Table S6. Calculation of overlap detection power using replicate samples.

Table S7. Antibody panels used in this study.


Acknowledgments: We thank M. Samo for technical help with the signal-joint TREC analysis, performed in the Immunology Unit of the Regional Biocontainment Laboratory at Duke Medical Center, which received partial support for construction from the National Institute of Allergy and Infectious Diseases, NIH (UC6-AI058607). We acknowledge B. Levin as a statistical consultant. We acknowledge the generosity of the organ donor families and the efforts of the LiveOnNY transplant coordinators and staff for making this study possible. We also thank P. Sims and M. Miron for critical reading of the manuscript. Funding: This work was supported by NIH grants AI106697 and AI100119 awarded to D.L.F. J.J.C.T. was supported by NIH grant F31AG047003, a BD Bioscience Research Grant, and an Adaptive Biosciences Young Investigator Award. These studies were performed in the CCTI Flow Cytometry Core funded in part through S10 Shared Instrumentation Grant, S10RR027050, with the excellent technical assistance of S.-H. Ho. Author contributions: J.J.C.T. designed the experiments, carried out data acquisition and analysis, created the figures, and wrote and edited the manuscript; B.G. analyzed the TCR sequence data, created the figures, and wrote and edited the manuscript; B.V.K. acquired the CBA data and created the figures; M.K. and Y.O. acquired donor tissues; H.L. coordinated tissue acquisition; G.D.S. carried out TREC assay procedures; Y.S. analyzed the TCR sequence data, created the figures, and wrote and edited the manuscript; and D.L.F. designed the experiments, analyzed the data, created the figures, and wrote and edited the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The flow cytometry data for this study will be deposited in the ImmPort database (, and the TCR sequencing data are available through the Adaptive Biotech website (
View Abstract

Stay Connected to Science Immunology

Navigate This Article