Cutting Edge: IL-7-Dependent Homeostatic Proliferation of CD8⁺ T Cells in Neonatal Mice Allows the Generation of Long-Lived Natural Memory T Cells

Mice

C57BL/6J mice (B6; Thy 1.2⁺) and congenic B6.PL-Thy1a/Cy mice (Thy 1.1⁺) were purchased from Charles River Breeding Laboratories (Sulzfeld, Germany). RAGE-deficient (RAGE^−/−) mice (24) constitutively expressing the enhanced green fluorescent protein (eGFP), recombination-activating gene (RAG) 2-deficient (RAG2^−/−), and IL-15^−/− mice (25), all on the C57BL/6 background, were bred in our animal facility.

Adoptive T cell transfer

Single-cell suspensions of spleens and lymph nodes were prepared from male mice of the indicated strains. Total CD8⁺ T cells were purified using CD8α-specific microbeads and autoMACS (both Miltenyi Biotec, Bergisch Gladbach, Germany). The purity ranged from 85 to 95%. For the purification of naive CD8⁺ T cells expressing low levels of CD44 (CD44^low), cell suspensions were incubated with a FITC-labeled mAb specific for CD44 (IM7; BD PharMingen, Hamburg, Germany). Subsequently, anti-FITC microbeads and the CD8⁺ T cell isolation kit were used to deplete CD44^int/high and CD8⁻ cells by autoMACS (all Miltenyi Biotec). The depletion of CD44^int/high cells ranged from 98 to 99.8%. For CFSE, 1 × 10⁷/ml purified T cells were incubated with 5 μM CFSE (Molecular Probes, Eugene, OR) in PBS for 15 min at 37°C. Cells were washed twice with ice-cold PBS and were finally resuspended in PBS before injection. As indicated, 1.5 × 10⁶–4.0 × 10⁶ CD8⁺ T cells were injected i.p. into neonatal mice or i.v. into adult male mice. At the indicated time points, recipient spleens were removed and single-cell suspensions were stained with mAbs for CD8α (53-6.7), CD44 (IM7), Thy1.1 (OX-7), or CD122 (TM-β1; all BD PharMingen) and analyzed with a FACSCalibur flow cytometer (BD Biosciences, Mountain View, CA).

Intracellular detection of IFN-γ

For the polyclonal restimulation of T cells, tissue culture plates were coated with 4 μg/ml anti-CD3ε mAb (145-2C11; BD PharMingen) or were left untreated. Spleen cells were cultured in the respective plates for 6 h in RPMI 1640 plus 10% FCS, penicillin/streptomycin, MEM, and 2-ME (50 μM) containing brefeldin A. For intracellular detection of IFN-γ, the intracellular staining kit from BD PharMingen and mAbs for CD8α, Thy1.1, and IFN-γ (XMG1.2; BD PharMingen) were used.

Blockade of IL-7 function in neonates

A mixture of mAbs specific for the IL-7Rα chain (A7R34; 40 μg/g body weight) and IL-7 (M25; 20 μg/g body weight) were injected at the day of birth and subsequently every 2–3 days until the day of analysis. A control group received irrelevant rat and mouse Igs (both from Dianova, Hamburg, Germany) at identical concentrations. CD8⁺ T cells were transferred within 24 h after the first Ab injection.

CD8⁺ T cells undergo homeostatic proliferation in neonatal B6 mice

Despite any previous immunizations, the CD8⁺ T cell pool of naive B6 mice contains a population of CD44^highCD122^high memory cells (Fig, 1⇓A). Since neonatal mice are lymphopenic within the first days after birth (26) and LIP of CD8⁺ T cells allows their differentiation into memory cells (10, 11, 12), we asked whether neonatal lymphopenia was sufficient to cause the generation of the naturally occurring memory CD8⁺ T cell pool of adult B6 mice. To test this, Thy1.1⁺CD8⁺ T cells were purified from spleens and lymph nodes of B6.PL-Thy1.1 mice, labeled with CFSE, and injected into neonatal B6 mice (Thy1.2⁺) within 24 h after birth (day 1) or at days 4, 7, or 14 after birth. Age-matched neonatal lymphopenic RAG2^−/− mice were treated identically and served as positive controls for maximum LIP. Thy1.1⁺CD8⁺ T cells were also transferred to adult RAG2^−/− and B6 mice, the latter to determine the degree of background proliferation. Recipient spleens were harvested and the transferred Thy1.1⁺CD8⁺ T cells were analyzed for their CFSE content and expression levels of CD44 (Fig. 2⇓A). Nearly all Thy1.1⁺CD8⁺ T cells recovered from RAG2^−/− mice, irrespective of their age, were CFSE^lowCD44^high, showing that they had undergone proliferation and memory differentiation, as expected. In contrast, in adult B6 mice only very few Thy1.1⁺CD8⁺ T cells were CFSE^lowCD44^high. However, when Thy1.1⁺CD8⁺ T cells were recovered from neonates injected at day 1 after birth (day 1), 67% of Thy1.1⁺CD8⁺ T cells were CFSE^lowCD44^high (Fig. 2⇓B). The percentage of CFSE^lowCD44^highThy1.1⁺CD8⁺ T cells was similar after transfer into day 4 neonates (64%) and decreased to 41% in day 7 neonates. After transfer into day 14 neonates, 27% of the Thy1.1⁺CD8⁺ T cells were CFSE^low. Comparable numbers were found in adult B6 mice (26%). A contribution of exogenous Ags to the CD8⁺ T cell proliferation observed in day 1 to day 7 B6 neonates and RAG2^−/− mice is highly unlikely because the proliferative response was at background levels in day 14 B6 neonates and adult B6 mice despite identical housing conditions and thus a similar antigenic environment. As mentioned above, high numbers of T cells transferred into adult lymphopenic mice can prevent the proliferation of a second cohort of T cells transferred to the same host (16). Adult T cell numbers are reached in the spleen of neonatal mice within 16 days after birth (26), suggesting that the age-dependent decrease of LIP in B6 mice resulted from increasing numbers of host T cells. We conclude that day 1 to day 4 neonates provided a functionally lymphopenic environment that allowed LIP of Thy1.1⁺CD8⁺ T cells.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Adult untreated B6 mice contain a population of CD44^highCD122^high memory CD8⁺ T cells. Spleen cells of untreated B6 mice (A, n = 10) and IL-15^−/− mice (B, n = 10) were analyzed by flow cytometry. A and B, Shown is the log fluorescence intensity for CD44 and CD122 after gating on CD8⁺ T cells. Specificity of staining was verified using isotype-matched control Abs.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

CD8⁺ T cells proliferate homeostatically in neonatal B6 mice and acquire a memory phenotype that is maintained until adulthood. A and B, Briefly, 1.5 × 10⁶ CFSE⁺Thy1.1⁺CD8⁺ T cells were injected i.p. into B6 (upper row) or RAG2^−/− mice (lower row) within 24 h after birth (day 1: B6, n = 9; RAG2^−/− n = 8) or at day 4 (B6, n = 7; RAG2^−/−, n = 7), day 7 (B6, n = 9; RAG2^−/−, n = 6), and day 14 (B6, n = 9; RAG2^−/−, n = 7). Adult B6 (n = 9) and RAG2^−/− mice (n = 7) were injected i.v. Sixteen to 28 days after transfer, spleen cells were analyzed by flow cytometry. A, The log fluorescence intensities for CFSE and CD44 are shown for one representative mouse per group after gating on Thy1.1⁺CD8⁺ T cells. B, The results are summarized and presented as the percentages of proliferating cells (CFSE^lowThy1.1⁺CD8⁺ T cells) ± SDs in B6 (□) and RAG2^−/− neonates (▪). C and D, Neonatal B6 mice (day 1, n = 5) were injected with 4 × 10⁶ CD8⁺ T cells from eGFP-transgenic mice or were left untreated (E, n = 5). Fifty-five days after transfer, spleen cells were analyzed by flow cytometry. Shown are the log fluorescence intensities for CD44 and CD122 after gating on eGFP⁺ (C, transferred) or eGFP⁻ (D and E, endogenous) CD8⁺ T cells. The numbers indicate the percentages of cells in the respective quadrants. Specificity of staining was verified using isotype-matched control Abs.

As shown in Fig. 2⇑A, all CFSE^lowThy1.1⁺CD8⁺ T cells derived from B6 neonates were CD44^high, indicating their memory phenotype. To test whether 1) this memory phenotype was stable until adulthood and 2) whether the injection of CD8⁺ T cells at the day of birth altered the composition of the endogenous CD8⁺ T cell pool, one group of day 1 B6 neonates was injected with CD8⁺ T cells expressing the eGFP, whereas a second group was left untreated. Eight weeks after transfer, the expression patterns of CD44 and CD122 by eGFP⁺CD8⁺ T cells were compared with those of CD8⁺ T cells from neonatally treated (Fig. 2⇑D) and untreated mice (Fig. 2⇑E). Similar to the data presented in Fig. 2⇑, A and B, the majority of the neonatally transferred eGFP⁺CD8⁺ T cells showed a CD44^high memory phenotype (Fig. 2⇑C) and represented around 3% of the adult memory CD8⁺ T cell pool. Thirty percent of the eGFP⁺CD8⁺ T cells had remained naive and were CD44^low (Fig. 2⇑C). In contrast to that, 65% of the endogenous CD8⁺ T cells were CD44^low (Fig. 2⇑D). Since similar percentages of naive and memory CD8⁺ T cells were found in age-matched control mice that had not received CD8⁺ T cells (Fig. 2⇑E), this suggested that 1) the adoptive transfer of CD8⁺ T cells at the day of birth did not alter the composition of the recipients’ CD8⁺ T cell pool and 2) LIP-induced memory CD8⁺ T cells generated in neonates maintained their phenotype until adulthood and contributed to the hosts’ natural memory CD8⁺ T cell pool (Fig. 1⇑A).

LIP of naive CD8⁺ T cells in neonatal mice requires IL-7 but not IL-15

The experiments described in Fig. 2⇑ were done with total CD8⁺ T cells containing naive CD44^low and memory CD44^highCD8⁺ T cells. Since CD44^highCD8⁺ T cells have a higher proliferative capacity than CD44^lowCD8⁺ T cells (2), we asked next whether only CD44^highCD8⁺ T cells had expanded in day 1 to day 7 B6 neonates or whether the neonatal environment also promoted the proliferation of naive CD44^lowThy1.1⁺CD8⁺ T cells and their differentiation into memory CD8⁺ T cells. For this purpose, CFSE-labeled CD44^lowThy1.1⁺CD8⁺ T cells were transferred to day 1 B6 neonates. Adult B6 and RAG2^−/− mice were injected in parallel. As shown in Fig. 3⇓A, 29% of Thy1.1⁺CD8⁺ T cells were CFSE^lowCD44^high after transfer into day 1 B6 neonates. In adult B6 mice, ∼12% were CFSE^lowCD44^high (Fig. 3⇓D) and in RAG2^−/− mice 96% (Fig. 3⇓E). It is important to note that the numbers of CFSE^lowCD44^high cells obtained after transfer of naive CD8⁺ T cells (Fig. 3⇓A) were clearly lower than after transfer of total CD8⁺ T cell (Fig. 2⇑). This suggested that CD44^high as well as CD44^low cells had contributed to the generation of CFSE^lowCD44^high cells (Fig. 2⇑). Nevertheless, the data presented in Fig. 3⇓A showed that the lymphopenic environment provided by neonatal B6 mice was sufficient to induce the proliferation of naive CD44^lowCD8⁺ T cells and their differentiation into memory CD44^highCD8⁺ T cells.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Homeostatic proliferation of naive CD8⁺ T cells in B6 neonates requires IL-7 but not IL-15. 1.5–2.5 × 10⁶ CFSE⁺CD44^lowThy1.1⁺CD8⁺ T cells were transferred to B6 day 1 neonates (A, n = 19), IL-15^−/− day 1 neonates (B, n = 19), B6 day 1 neonates treated with anti-IL-7/anti-IL-7R mAbs (C, n = 10), adult B6 (D, n = 10), and adult RAG2^−/− mice (E, n = 10). A–E, Twelve to 15 days after transfer, spleen cells were analyzed by flow cytometry. Shown are the log fluorescence intensities for CFSE and CD44 after gating on Thy1.1⁺CD8⁺ cells. The numbers indicate the mean percentages of CFSE^lowCD44^high cells ± SDs.

Next, we wanted to determine the factors that were required for LIP of naive CD8⁺ T cells in neonates. Since adult IL-15^−/− mice largely lack CD44^highCD122^high natural memory CD8⁺ T cells (Fig. 1⇑B and Ref.25), we asked whether neonates had to provide IL-15 for the induction of LIP and subsequent memory differentiation of naive CD8⁺ T cells. CFSE⁺CD44^lowThy1.1⁺CD8⁺ T cells were injected into day 1 IL-15^−/− neonates and their proliferation and differentiation was measured. Twenty-eight percent of the Thy1.1⁺CD8⁺ T cells were found to be CFSE^lowCD44^high (Fig. 3⇑B), a percentage that was nearly identical to that observed in B6 neonates (Fig. 3⇑A). This demonstrated that host IL-15 was not required for the induction of LIP in neonates and suggested that the reduction of natural memory CD8⁺ T cells in adult IL-15^−/− mice (Fig. 1⇑B and Ref.25) did not result from their impaired generation in the neonatal phase but rather from the lack of IL-15-dependent survival signals (3).

IL-7 was shown to be critical for LIP of T cells in adult lymphopenic mice (4, 21, 22, 23). However, a recent study demonstrated that IL-7 was not required for LIP of CD4⁺ T cells in neonatal B6 mice (20). To test whether IL-7 was required for LIP of naive CD8⁺ T cells in neonates, CFSE⁺CD44^lowThy1.1⁺CD8⁺ T cells were transferred to day 1 B6 neonates that were treated with a mixture of mAbs against IL-7 and the IL-7Rα chain. As shown in Fig. 3⇑C, 10% of Thy1.1⁺CD8⁺ T cells were CFSE^lowCD44^high, a percentage that was similar to the background values detected in adult B6 mice (Fig. 3⇑C). LIP-induced memory CD8⁺ T cell generation in control Ig-treated neonates was similar to that in untreated B6 (Fig. 3⇑A) and IL-15^−/− neonates (Fig. 3⇑B and data not shown), suggesting that the prevention of LIP after injection of anti-IL-7/anti-IL-7Rα chain Abs (Fig. 3⇑C) was due to the specific blockade of IL-7 function rather than due to unspecific effects. Thus, the blockade of IL-7 function in neonates prevented LIP of naive Thy1.1⁺CD8⁺ T cells.

The observation that IL-7 in neonates was critical for LIP of CD8⁺ T cells (Fig. 2⇑C) but not for CD4⁺ T cells (20) demonstrates an important difference between both T cell subsets that has not yet been observed in adult lymphopenic mice (4, 21, 22, 23).

LIP-induced memory CD8⁺ T cells produce IFN-γ

As a result of LIP in adult lymphopenic mice, CD8⁺ T cells acquire effector functions such as the ability to kill target cells and to produce IFN-γ (10, 11, 12, 14, 15). To test whether this was also the case in neonates, CFSE⁺CD44^lowThy1.1⁺CD8⁺ T cells were transferred to day 1 B6 neonates, adult B6, and adult RAG2^−/− mice and splenocytes were analyzed 2 wk later. As shown in Fig. 4⇓A, ∼19% of the CFSE^lowThy1.1⁺CD8⁺ T cells from B6 neonates produced IFN-γ while CFSE^highCD8⁺ T cells did not. In RAG2^−/− mice, 40% of the CFSE^low Thy1.1⁺CD8⁺ T cells produced IFN-γ (Fig. 4⇓B) and in adult B6 mice only background levels of ∼2% could be detected (Fig. 4⇓C). Without stimulation, IFN-γ was not detectable (Fig. 4⇓, D–F).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

CD8⁺ T cells that have proliferated in B6 neonates produce IFN-γ. Spleen cells recovered from B6 neonates (A, D, G, and I), adult RAG2^−/− (B and E), and adult B6 mice (C, F, H, and J) that had been reconstituted 12–15 days before with CD44^lowThy1.1⁺CD8⁺ T cells (Fig. 3⇑) were cultured for 6 h in anti-CD3-coated (A–C, G, and H) or untreated tissue culture plates (D–F, I, and J). IFN-γ was measured by intracellular cytokine staining. A–F, Shown are the log fluorescence intensities for CFSE and IFN-γ after gating on Thy1.1⁺CD8⁺cells. The numbers indicate the mean percentages of IFN-γ⁺CFSE^low cells ± SDs. G–J, The log fluorescence intensities for IFN-γ and CD8α after gating on recipient (Thy1.1⁻) spleen cells from neonatal (G and I) and adult B6 recipients (H and J). The numbers indicate the mean percentages of IFN-γ⁺CD8⁺ cells ± SDs. Specificity of staining was verified using isotype-matched control Abs.

The results presented so far demonstrated that neonatal B6 mice provided a lymphopenic environment that allowed the proliferation and differentiation of naive CD8⁺ T cells into long-lived, IFN-γ-producing memory CD8⁺ T cells. However, it remained unclear whether the endogenous CD8⁺ T cells underwent a similar differentiation process. To answer this question, the spleen cell cultures derived from neonatal (Fig. 4⇑, A and D) and adult B6 mice (Fig. 4⇑, C and F) were reanalyzed by gating out Thy1.1⁺ cells to determine the IFN-γ production of recipient CD8⁺ T cells (Thy1.1⁻). Approximately 12% of the neonatal (Fig. 4⇑G) and 6% of the adult CD8⁺ T cells (Fig. 4⇑H) produced IFN-γ after TCR stimulation but not in its absence (Fig. 4⇑, I and J). Since IFN-γ could be produced only by CD8⁺ T cells that had previously divided (Fig. 4⇑, A–F), Fig. 4⇑G indicated that also endogenous CD8⁺ T cells acquired effector functions as a result of LIP. However, it remains to be established how many of the IFN-γ⁺CD8⁺ T cells derived from the adult were generated as a result of LIP in the neonatal phase or in response to environmental Ags encountered later in life.

It was shown that LIP can induce inflammatory bowel disease in adult lymphopenic mice (27). An important question arising from our results is how autoimmunity is prevented in neonates although CD8⁺ T cells undergoing LIP acquire effector functions (Fig. 4⇑). The fact that all adoptively transferred CD8⁺ T cells underwent LIP in T cell-deficient RAG2^−/− mice (Fig. 2⇑, A and B) but not in B6 neonates (Figs. 2⇑, A and B, and 3) suggests that neonatal recipient T cells contributed to the regulation of LIP. Whether this required particular T cell subsets or resulted from clonal competition remains to be shown.

Thus, under lymphopenic conditions, there appears to be a hierarchy that favors the generation of a memory CD8⁺ T cell pool before the constitution of the naive T cell repertoire. Due to their low activation thresholds, long life span, and stable phenotype, the natural memory CD8⁺ T cells generated in early neonatal life may contribute to immune responses against foreign Ags and the regulation of T cell homeostasis in the adult.