Regulation of T Cell Responses in the Developing Human Fetus¹

Phenotypic and functional differences are known to exist between the fetal and the adult immune systems (1, 2, 3, 4, 5, 6). Billingham et al. (2) initially advanced the concept that Ag presentation in utero, as opposed to Ag presentation in the adult, normally leads to tolerance instead of adaptive immunity. Thereafter, the notion that fetal immune responses are functionally distinct from adult immune responses has been widely accepted (1, 6). This view has since been modified by observations that the fetus can generate vigorous B and T cell responses to foreign Ags under certain circumstances, e.g., after transplacental spread of infectious agents (7, 8, 9). These observations, by extension, support the hypothesis that an active form of immune suppression, rather than a passive functional deficit, exists during fetal development.

CD4⁺CD25^high T_Reg cells suppress a broad spectrum of immune responses (10, 11, 12, 13, 14) and recently have been described to exist in the human fetus, both in the cord blood and in fetal lymphoid organs (15, 16, 17, 18). In cord blood, the frequency of CD4⁺CD25^high T regulatory (T_Reg)⁶ cells is higher (∼12% of CD4⁺ T cells) at gestational week (g.w.) 25 than at birth, when it approximates that found in healthy adult peripheral blood (18). In the thymus, the frequency of CD4⁺CD25^high T_Reg cells is maintained at an intermediate level (mean 8% of CD4 single-positive thymocytes) throughout gestation (16, 17). We hypothesized that fetal CD4⁺CD25^high T_Reg cells play an important role in maintaining a general state of immune suppression during fetal development. Our findings reveal that the frequency of CD4⁺CD25^high T_Reg cells is higher in fetal lymph nodes (LN) than that observed in adult LN, and suggest that these CD4⁺CD25^high T_Reg cells play an important role in the suppression of fetal T cell responses during development.

Fetal tissue (mesentery, spleen, and thymus) and matched maternal blood were obtained from Advanced Bioscience Resources. Adult blood from healthy individuals was collected in heparinized Vacutainer tubes (BD Biosciences) after informed consent. Fresh LN from adults were obtained from the National Disease Research Interchange. Paraffin-embedded adult LN were verified (by W. Finkbeiner, Department of Pathology, University of California, San Francisco General Hospital, San Francisco, CA) to be noninflamed and noncancerous by H&E staining. Cord blood was obtained from the Department of Obstetrics, Gynecology, and Reproductive Sciences at San Francisco General Hospital, after informed consent from women giving birth to full-term babies. Mesenteric LN (MLN) were dissected from the fetal mesentery. LN, thymus, and spleen were incubated with 0.2 mg/ml collagenase B (Roche Diagnostics) in R15 medium (RPMI 1640 (Mediatech) supplemented with 15% FCS (Gemini Bio-Products), 2 mM l-glutamine, 10 mM HEPES, and 100 U/ml penicillin/streptomycin (Invitrogen Life Technologies) for 1 h at 37°C and processed into single cell suspensions by passing the cells through a 40-μm cell strainer (BD Falcon). PBMC from adult and full-term cord blood were isolated using Ficoll-Hypaque^Plus (Amersham Biosciences) density gradient centrifugation and washed in R15 medium before phenotypic analysis or functional assays. When necessary, RBC were lysed by incubation with ammonium-chloride-potassium lysis buffer (Quality Biologicals). In some cases, fetal MLN and thymus were frozen or Formalin fixed (10% neutral buffered Formalin) and embedded in paraffin for tissue sections.

For phenotypic analysis, blocking, and cell sorting by flow cytometry, the following Abs were used: anti-CD3 FITC, anti-CD4 PerCP, anti-CD4 Pacific Blue, anti-CD4 PE, anti-CD25 PE, anti-CD25 allophycocyanin (clone M-A251), anti-CD8 allophycocyanin, anti-CD8 Pacific Blue, anti-CTLA-4 PE, anti-CD45RA FITC, anti-CCR7 PE-Cy7, and anti-IFN-γ allophycocyanin (all from BD Biosciences); purified anti-IL-2 (clone 5334.21), purified anti-IL-7 (clone 7417.111), and anti-gluticocorticoid-induced TNFR (GITR) FITC (R&D Systems); anti-CD69 allophycocyanin and anti-CD4 PE-Cy7 (Caltag Laboratories); and anti-CD127 PE (Beckman Coulter). For phenotyping, the cells were incubated with the relevant Abs diluted in PBS-1% BSA for 30 min on ice, followed by two washes with PBS-1% BSA, and then fixed in 1% paraformaldehyde. The panel of Abs used for phenotyping in Fig. 1 a was anti-CD3 FITC, anti-CD25 allophycocyanin, and anti-CD4 PerCP. Anti-TCR Vβ Abs were from Beckman Coulter. Intracellular detection of scurfin was performed using anti-FoxP3 allophycocyanin (clone PCH101), and the accompanying staining kit was provided by eBioscience, in accordance with the manufacturer’s protocol. In all assays with more than four colors, anti-mouse Ig compensation particles (BD CompBeads; BD Biosciences) single stained with each of the Abs were used as compensation controls for software-based compensation using FlowJo software (Tree Star). For four-color flow cytometry, the samples were analyzed on FACSCalibur (BD Biosciences). For five- to eight-color flow cytometry, we analyzed samples on a LSR II flow cytometer (BD Biosciences) modified from the standard configuration by the addition of a 150 mW green (532 nm) diode laser, and the upgrade of the blue and red lasers to 100 and 25 mW, respectively. The green diode was used for the excitation of all the PE tandem conjugates. All data were analyzed using FlowJo software (Tree Star).

Fetal thymus and MLN were frozen in Tissue Tek OCT (Sakura Finetek) and sectioned into 6- to 8-μm-thick sections for immunofluorescent staining. For all immunohistochemistry, fetal and adult MLN were first fixed in 10% buffered Formalin and embedded in paraffin. For immunofluorescence, the Abs used were mouse anti-human CD4 (DakoCytomation), mouse anti-human CD8 (DakoCytomation), mouse anti-human CD25 (Novus Biologicals), and goat anti-human FoxP3 (scurfin) (AbCam). Immunohistochemical detection of scurfin was performed using a rabbit anti-human scurfin polyclonal Ab (AbCam). For immunofluorescence microscopy, sections were fixed in cold acetone and blocked with TBS (pH 8.5) containing 5% human AB serum. The sections were then incubated for 1 h with combinations of mouse anti-human CD4 (1/200 dilution), mouse anti-human CD8 (1/200 dilution), or mouse anti-human CD25 (1/200 dilution), and goat anti-human scurfin (1/50 dilution) diluted in TBS. The sections were washed in TBS/Tween 20 for 5 min and incubated with anti-mouse Ig Cy3 (1/500 dilution; Jackson ImmunoResearch Laboratories) and anti-goat Alexa488 (1/200 dilution; Molecular Probes) for an additional 45 min. The slides were mounted with Gel/Mount (Biomedia) and analyzed using a Zeiss LSM5100 confocal microscope. Data were processed with the Zeiss LSM software package. Formalin-fixed, paraffin-embedded tissues were cut into 5-mm-thick sections and deparaffinized with Histosol, followed by rehydration in graded ethanol. The sections were incubated with hydrogen peroxide (3% in PBS), followed by a protein block (DakoCytomation) for 30 min at room temperature. The sections were incubated overnight at 4°C with a rabbit anti-scurfin polyclonal Ab (1/2000 dilution). After primary incubation, the sections were washed in TBS/Tween 20 (0.5% Tween 20) and incubated for 45 min with a mouse anti-rabbit secondary Ab (1/200 dilution; DakoCytomation), followed by washing with TBS/Tween 20 and a 45-min incubation with streptavidin-conjugated HRP (1/200 dilution; DakoCytomation). The sections were developed with diaminobenzidine (DakoCytomation), counterstained with Gil’s hematoxylin (Sigma-Aldrich), followed by a final dehydration step, and mounted for analysis. Scurfin⁺ cell frequencies were assessed by counting ×40 magnification images onto which a grid was applied and comparing cells staining positive for scurfin as a fraction of total cells per image.

Magnetic beads were used for separation of CD25⁺ and CD25⁻ cells in the experiments shown in Figs. 1–3. CD25⁺ T cells were separated from CD25⁻ cells by MACS CD25 microbeads (Miltenyi Biotec), according to the manufacturer’s instructions. Typically, the CD25⁺ fraction contained >90% CD4⁺ T cells, of which >85% were CD25^high. Staining by intracellular flow cytometry for scurfin demonstrated that <1% of the remaining CD4⁺ T cells were CD25⁺scurfin⁺.

Sorting by FACS was used for experiments shown in Figs. 4 and 5. MLN from g.w. 20 were processed, as described above, to obtain a single cell suspension and stained with anti-CD25 PE, anti-CD69 allophycocyanin, and anti-CD4 PE-Cy7. Sorting was performed on a BD DIVA flow cytometer. A lymphocyte gate was set on the basis of forward and side scatter. To control for potential nonspecific effects of sorting, 10⁶ cells were sorted based on the lymphocyte gate alone (unseparated MLN cells). Total MLN cells were sorted to deplete CD4⁺CD25^high cells from total MLN cells (termed CD25⁻ cells). Similarly, total MLN cells were sorted to isolate CD4⁺CD25^highCD69⁺ cells from the remaining MLN cells (termed CD25⁻ with CD25⁺CD69⁻ cells). Finally, CD4⁺CD25^high and CD25⁻CD69⁺ cells were separated from total MLN cells, and the remaining cells (termed CD25⁻CD69⁻ cells) were collected.

Cells from MLN depleted of CD25⁺ cells (CD25⁻ cells) by MACS beads (see above) were labeled with 0.5 μM CFSE (Molecular Probes). The CFSE-labeled CD25⁻ cells, alone or admixed with nonlabeled CD4⁺CD25⁺ T cells or nonlabeled CD25⁻ cells, were stimulated with 5 μg/ml staphylococcal enterotoxin B (SEB) (Sigma-Aldrich) or with R15 medium alone, and then incubated for 24 h. Brefeldin A (Sigma-Aldrich) was added at a final concentration of 5 μg/ml for the last 6 h of incubation. After stimulation, the cells were harvested, incubated with anti-CD4 PE (BD Biosciences), washed twice, fixed in 1% paraformaldehyde, and permeabilized with FACS permeabilizing solution (BD Biosciences) for 20 min before being stained with anti-CD3 PerCP and anti-IFN-γ allophycocyanin (BD Biosciences), and then analyzed on a FACSCalibur flow cytometer (BD Biosciences). In the analysis of IFN-γ-producing cells, we gated on the CFSE⁺ cells, to exclude cells added back to the assay from the analysis.

In the analysis of proliferation in the absence of exogenous stimulation (see Figs. 2–5), the responder cells (unseparated MLN, CD25⁻ cells, CD25⁻CD69⁻ cells, and CD25⁻CD69⁺ cells) were labeled with 1 μM CFSE. For unseparated MLN cells, CD25⁻ cells, and CD25⁻CD69⁻ cells, 3 × 10⁵ responder cells were cultured in 96-well U-bottom plates in R15 medium. To analyze the proliferation of CD25⁻CD69⁺ cells (see Fig. 4,e), 2 × 10⁵ CFSE-labeled CD25⁻CD69⁺ cells were cultured with 2 × 10⁵ nonlabeled CD25⁻CD69⁻ cells. After 4 days of culture, cells and supernatants were harvested (the supernatants were analyzed for IFN-γ, as described in the next paragraph). The harvested cells were stained with anti-CD3 energy-coupled dye, anti-CD4 PE-Cy7, anti-CD8 Pacific Blue, and ethidium monoazide (EMA) (Molecular Probes). EMA was cross-linked to DNA in dead cells by 8-min light exposure, before washing in MACS buffer, fixation in 1% paraformaldehyde, and analysis on a LSR II flow cytometer (BD Biosciences). BD CompBeads single stained with the respective Abs and cells single stained with EMA and CFSE were used for software-based compensation with FlowJo software (Tree Star). All analyses were performed using FlowJo software (Tree Star). Nonlabeled cells added back to the assays were excluded from the analysis on the basis of lack of CFSE fluorescence. The frequency of CFSE^low cells was used as a measurement of total T cell proliferation. In proliferation assays using anti-IL-7 and anti-IL-2 Abs (see Fig. 3), CFSE-labeled MLN cells depleted of CD4⁺CD25⁺ T cells by magnetic beads (CD25⁻ cells) were cultured for 4 days in R15 medium alone or together with neutralizing Abs recognizing IL-2 and IL-7. The Abs were added to the cultures at the start of the assay both separately and in combination at a concentration of 5 μg/ml. The concentration used for each neutralizing Ab in these assays was ∼5–50 times higher than that required for complete neutralization of rIL-7 and rIL-2, as described in the manufacturer’s protocols. IL-2 was undetectable in the presence of the neutralizing anti-IL-2 Ab after 4 days of culture of CD25⁻ cells, whereas it could be detected in parallel cultures in the absence of anti-IL-2 Ab (data not shown). Proliferation was analyzed, as described above.

Cell culture supernatants from the proliferation assays in Fig. 4 were collected after 4 days of culture. The IFN-γ concentration was measured using cytokine bead arrays (BD Biosciences), according to the instructions from the manufacturer. Data were collected on a FACSCalibur and analyzed using FlowJo software (Tree Star).

As the thymus becomes a functional organ of T cell production between the 7th and 16th g.w. of human development, CD4⁺ and CD8⁺ T cells move into the periphery (19). The frequency of CD4⁺CD25^high T cells was assessed in MLN and in spleen at varying gestational ages. Extending the findings of recent reports (16, 17), a large fraction of CD3⁺CD4⁺ T cells in both organs expressed high levels of CD25 by g.w. 16 and 20 (Fig. 1, a and b, and data not shown). Similar frequencies of CD4⁺CD25^high T cells could be detected in MLN at g.w. 12 (Fig. 1,a), whereas no T cells could be detected at that time in the fetal spleen. Analysis of fetal thymus at time points from g.w. 12 through 20 revealed that high levels of CD25 were expressed by ∼8% of CD3^highCD4⁺ single-positive thymocytes, similar to that previously reported in both fetal and neonatal thymus samples (16, 17). To assess the frequency of CD4⁺CD25^high T cells in fetal LN in comparison with adult LN, we performed a direct analysis of adult and fetal lymphoid tissues. The frequency of CD4⁺CD25^high T cells was lower in the adult MLN and inguinal LN (range 0.7–6.8%, n = 4 donors, 12 total LN) compared with fetal MLN, and comparable to frequencies found in adult PBMC (Fig. 1, a and b).

Scurfin, a transcription factor encoded by the FOXP3 gene, is considered to be a specific marker for CD4⁺CD25^high T_Reg cells (20). As examined by immunohistochemistry and immunofluorescent microscopy, scurfin⁺ cells were found to be scattered throughout the parenchyma of fetal MLN (Fig. 1,c) as well as in the medulla of the fetal thymus (data not shown). Scurfin⁺ cells could also be found in adult LN, albeit at a significantly lower frequency than in fetal MLN (Fig. 1, d and e). In the fetal MLN, scurfin⁺ cells were CD4⁺ (Fig. 1,f), but not CD8⁺ (Fig. 1,g). An average of 87% of the fetal CD25⁺ cells were found to be scurfin⁺ (Fig. 1,h), and occasional CD25⁺ scurfin⁻ cells (Fig. 1,i, arrow) were also observed. Conversely, ∼90% of the scurfin⁺ cells in fetal MLN were CD25⁺, with relatively few CD25⁻ scurfin⁺ cells identified in each section analyzed (Fig. 1,j, arrow). The presence of a small fraction of CD25⁻ scurfin⁺ cells in the fetal MLN is consistent with findings in mice, in which a small fraction of scurfin⁺ cells in the lymph nodes was found to not express CD25 (21). Flow cytometric detection of scurfin expression was performed on fetal MLN cells to confirm the frequencies of CD4⁺CD25^high and CD4⁺CD25⁻ T cells expressing scurfin. In accordance with the immunofluorescent microscopy results, the majority of scurfin⁺ cells were found to express CD25 (∼80%), with a small fraction of CD25⁻ cells expressing low to moderate levels of scurfin (Fig. 1 k). Additionally, intracellular staining by flow cytometry confirmed that scurfin expression was predominantly restricted to the CD4⁺ T cell compartment of the fetal MLN (data not shown).

To confirm that the fetal CD4⁺CD25^high T cells were able to suppress T cell responses to a known Ag, functional studies were conducted in vitro. After stimulation of CD25⁻ cells (MLN-derived cells depleted of CD4⁺CD25⁺ T cells) with SEB, a large proportion of fetal CD4⁻CD25⁻ and CD4⁺CD25⁻ T cells produced IFN-γ (Fig. 1,l, left and right panels, respectively). Cytokine production was suppressed upon admixture of autologous CD4⁺CD25⁺ T cells at a ratio of 1:3 (1 × 10⁵ CD4⁺CD25⁺ T cells and 3 × 10⁵ CD25⁻ cells) (Fig. 1 l), whereas addition of the same number of autologous CD25⁻ cells had no effect (data not shown). The data above demonstrate that fetal MLN contain a large population of T cells that are phenotypically and functionally similar to naturally occurring CD4⁺CD25^high T_Reg cells. Interestingly, the frequency of such cells in the developing human fetus is much higher than that observed in the adult human.

Several studies have proposed that changes in the frequency of T_Reg cells might play a determinant role in regulating the balance between activation and tolerance within a specific lymphoid environment (14, 22, 23). In tumor-draining LN of adult humans, an increase in the frequency of CD4⁺CD25^high T_Reg cells correlates with a decrease in the capacity to mount a productive immune response (23). We reasoned that the high frequency of CD4⁺CD25^high T_Reg cells in fetal LN might establish an environment favoring suppression over activation and that removal of the CD4⁺CD25^high T_Reg cells in vitro might reveal potentially autoreactive T cells. To test this possibility, the proliferation of fetal MLN cells was measured using a CFSE dilution assay. In the presence of CD4⁺CD25^high T_Reg cells, unstimulated CD4⁺ and CD8⁺ MLN T cells proliferated only to a small extent (Fig. 2,a, upper and lower panels, respectively). After depletion of CD4⁺CD25^high T_Reg cells, however, CD4⁺ and CD8⁺ T cells in cultures of MLN cells proliferated vigorously (Fig. 2,b, upper and lower panels, respectively). By contrast, no CD4⁺ or CD8⁺ T cell proliferation could be detected in unstimulated cultures of adult LN, either before or after depletion of CD4⁺CD25^high T cells (Fig. 2, c and d, upper and lower panels, respectively). As expected, CD4⁺ and CD8⁺ T cells in cultures of adult LN proliferated vigorously in response to SEB (Fig. 2,e, upper and lower panels, respectively), demonstrating that the adult T cells were capable of proliferating in response to a known Ag. A summary of the frequency of dividing fetal CD4⁺ and CD8⁺ T cells (CFSE^low cells) in the presence or absence of CD4⁺CD25^high T_Reg cells from eight different donors is depicted in Fig. 2, f and g, respectively. These data indicate that fetal CD4⁺CD25^high T_Reg cells suppress the proliferation of both CD4⁺ and CD8⁺ T cells in cultures of unstimulated fetal MLN cells. It should be noted that the observed proliferation in the absence of CD4⁺CD25^high T_Reg cells could be due to homeostatic as well as autoreactive mechanisms and does not per se prove that the proliferating cells are autoreactive.

The proliferation of fetal MLN T cells observed in the absence of CD4⁺CD25^high T_Reg cells could be driven by cytokine-mediated homeostatic mechanisms. T cells adoptively transferred into neonatal lymphopenic mice undergo multiple rounds of division and gain the capacity to become effector cells upon Ag stimulation in an IL-7-dependent manner (24), and IL-7 and IL-2 are known to be regulators of neonatal T cell homeostasis and proliferation (25). Most fetal T cells in MLN, with the exception of the CD4⁺CD25^high T_Reg cells (data not shown), were found to express high levels of the high affinity IL-7Rα (CD127). To determine whether IL-2 and/or IL-7 were responsible for the observed proliferation in cultures of unstimulated fetal MLN T cells depleted of CD4⁺CD25^high T_Reg cells, neutralizing Abs to each cytokine were added to the assay. Compared with the levels of CD4⁺ and CD8⁺ T cell proliferation observed after depletion of CD4⁺CD25^high T cells (CD25⁻ control) (Fig. 3,b, upper and lower panels, respectively), the addition of neutralizing Abs to IL-2 had only a minor effect in reducing the proliferation of both CD4⁺ and CD8⁺ T cells (Fig. 3,c, upper and lower panels, respectively), and the addition of neutralizing Abs to IL-7 had no effect (Fig. 3,d, upper and lower panels, respectively). CD4⁺ and CD8⁺ T cell proliferation in cultures with both anti-IL-2 and anti-IL-7 Abs was similar to that observed in cultures with anti-IL-2 alone (Fig. 3,e, upper and lower panels, respectively). The effect of adding anti-IL-2- and anti-IL-7-neutralizing Abs on CD4⁺ and CD8⁺ T cell proliferation in three independent experiments is summarized in Fig. 3, f and g, respectively. Proliferation is depicted as a percentage of the proliferation seen in cultures of CD25⁻ fetal MLN cells without addition of Abs (control). Our data suggest that IL-2- and IL-7-induced homeostatic mechanisms are not likely to be responsible for the proliferation of fetal CD4⁺ and CD8⁺ T cells following depletion of CD4⁺CD25^high T_Reg cells from fetal MLN cultures.

If fetal MLN T cells are not proliferating as a result of homeostatic cytokine-driven mechanisms, it is possible that they are being activated by Ags present in the periphery of the developing fetus. In mice, for instance, there are reports of increased frequencies of T cells bearing autoreactive TCRs during the initial period of T cell colonization of the peripheral lymphoid tissues (26). This time frame is comparable to the second trimester of human fetal development, when T cell colonization of the periphery occurs (5, 19). To determine whether a population of autoreactive human T cells might exist in utero, we first examined MLN and splenic T cell populations for the expression of CD69, a cell surface protein transiently up-regulated within hours of TCR stimulation. Expression of CD69 was consistently detected on a subset of CD25⁻CD4⁺ and CD25⁻CD8⁺ T cells examined directly ex vivo (Fig. 4,a), both in MLN (left panels) and in spleen (right panels). To test the possibility that the CD69⁺ T cell population represented T cells subject to regulation by fetal CD4⁺CD25^high T_Reg cells, the proliferation of unstimulated fetal CD4⁺ and CD8⁺ T cells was measured in cultures of unseparated MLN cells (Fig. 4,b, upper and lower panels, respectively), in cultures of MLN cells depleted of CD4⁺CD25^high T_Reg cells (CD25⁻ cells; Fig. 4,c, upper and lower panels, respectively), and in cultures of MLN cells depleted of both CD4⁺CD25^high T_Reg cells and CD25⁻CD69⁺ cells (CD25⁻CD69⁻ cells; Fig. 4,d, upper and lower panels, respectively). As shown in previous figures (Figs. 2 and 3), depletion of CD4⁺CD25^high T_Reg cells from cultures of fetal MLN cells resulted in proliferation of a substantial fraction of both CD4⁺ and CD8⁺ T cells (Fig. 4,c). When CD25⁻CD69⁺ T cells were removed from the cultures of CD25⁻ fetal MLN cells, the levels of T cell proliferation (Fig. 4,d) decreased to those seen in whole MLN cultures containing CD4⁺CD25^high T_Reg cells (Fig. 4,b). This observation suggests that it is the CD25⁻CD69⁺ T cell population that is normally suppressed by CD4⁺CD25^high T_Reg cells. This notion is further supported by the vigorous proliferation of CD25⁻CD69⁺ T cells observed in experiments in which sorted, CFSE-labeled CD25⁻CD69⁺ cells were admixed with unlabeled CD25⁻CD69⁻ cells at a 1:1 ratio (Fig. 4 e).

These data indicate that the CD4⁺CD25⁻CD69⁺ T cells represent a population of T cells that has been activated in response to Ags present in the fetal tissues in utero, and that this population is actively suppressed from proliferating by the large fraction of CD4⁺CD25^high T_Reg cells present in fetal lymphoid tissues. Importantly, we could not detect similar proliferation in the absence of exogenous stimulation in cultures of adult LN depleted of CD4⁺CD25^high T cells, despite the presence of a large fraction of CD69⁺ T cells in adult LN (Fig. 2, c and d, and data not shown). The frequency of CD4⁺ and CD8⁺ T cells expressing CD69, like the frequency of fetal CD4⁺CD25^high T_Reg cells, was fairly consistent from donor to donor, implying that the underlying mechanism(s) accounting for the apparent activation of these cells is not likely attributable to aberrant Ag exposure within a given fetus.

Although CD69 is generally considered to be associated with TCR stimulation and early events leading to T cell activation, it is possible that IL-2- and IL-7-independent events not associated with the acquisition of T cell effector functions may trigger proliferation of the CD25⁻CD69⁺ T cell subset. To test this possibility, we measured secreted IFN-γ concentrations in the supernatants of 4-day cultures of unseparated fetal MLN, CD25⁻ cells, CD25⁻CD69⁻ cells, and CD25⁻CD69⁺ T cells mixed with CD25⁻CD69⁻ cells at a ratio of 1:1 (Fig. 4,f). In three separate donors, the concentration of IFN-γ was low in cultures of unseparated fetal MLN cells and substantially increased upon removal of CD4⁺CD25^high T_Reg cells. IL-10 was measured simultaneously and was consistently detected only at very low levels in cultures of MLN, with only small increases upon depletion of CD4⁺CD25^high T_Reg cells (data not shown). Depletion of CD69⁺ cells from the CD25⁻ cells resulted in a decrease in IFN-γ secretion, while readdition of CD25⁻CD69⁺ cells with CD25⁻CD69⁻ cells at a 1:1 ratio led to an increase in IFN-γ secretion to levels greater than those seen in cultures of CD25⁻ cells (Fig. 4 f). These findings demonstrate that the large population of fetal CD4⁺ and CD8⁺ T cells that is activated in utero acquires effector functions (e.g., IFN-γ secretion) after removal of fetal CD4⁺CD25^high T_Reg cells in vitro.

To investigate differences in the activation status of CD4⁺CD25^high T_Reg cells in relation to their suppressive function, we measured expression of CD69 and GITR (a candidate phenotypic marker of T_Reg cells) by flow cytometry. Approximately half of the CD4⁺CD25^high T_Reg cells were found to express CD69, and expression of GITR was found almost exclusively within the CD69⁺ fraction of fetal CD4⁺CD25^high T cells (Fig. 5,a, left panel). As indicated by the lower expression levels of CD45RA and CCR7, these CD4⁺CD25^highCD69⁺ T cells (Fig. 5,a, right panel) have a memory/effector phenotype, consistent with the possibility that they have been activated in utero. It was suggested recently that CD4⁺CD25^high T_Reg cells acquire suppressive function against autoantigens after activation in the periphery, as measured by CD69 and GITR expression (16). This is an attractive hypothesis and is consistent with the above data (Fig. 4), showing that CD69 expression on CD25⁻ T cells reflects their ability to proliferate and to secrete cytokines such as IFN-γ. To determine whether both the CD69⁺ and the CD69⁻ CD4⁺CD25^high T_Reg cells could suppress T cell responses, the ability of each population to suppress spontaneous proliferation of fetal MLN T cells in vitro was tested. In cultures of MLN cells depleted of the CD69⁺CD4⁺CD25^high T_Reg cells only (CD25⁻ with CD25⁺CD69⁻; Fig. 5,d), proliferation was similar to that observed in cultures of unseparated MLN cells (Fig. 5,b). Conversely, addition of the sorted CD69⁺CD4⁺CD25^high T_Reg cells suppressed proliferation of CD25⁻ T cells to a level similar to that observed in cultures of unseparated MLN cells (Fig. 5,e). Thus, the expression of CD69, and hence GITR, did not correlate with enhanced suppressive activity, as both CD69⁻ and CD69⁺ CD4⁺CD25^high T_Reg cells suppressed fetal T cell proliferation to a similar extent (Fig. 5 f).

We show in this study that human fetal T cells exist in a dynamic balance between activation and quiescence, and not in a passive state of inactivity. Furthermore, our data indicate that this balance is regulated by the presence of a large population of CD4⁺CD25^high T_Reg cells, as removal of this population resulted in substantial T cell proliferation and IFN-γ production. A majority of the T cells that spontaneously underwent proliferation and produced IFN-γ upon removal of the CD4⁺CD25^high T_Reg cells could be identified directly ex vivo by the expression of the early activation marker, CD69. These findings indicate that fetal T cells are, in the absence of CD4⁺CD25^high T_Reg cells, highly responsive to stimulation and provide evidence for an important role for CD4⁺CD25^high T_Reg cells in maintaining peripheral T cell tolerance in utero.

The conclusions of this study are corroborated by clinical findings from patients with the inherited disease, immunodysregulation, polyendocrinopathy, enteropathy, X-linked syndrome (IPEX), who have mutations in the FOXP3 gene and, therefore, are thought to have deficiencies in CD4⁺CD25^high T_Reg cell development and/or function (27). Severe cases of IPEX have been described in which the disease manifests itself in utero, causing the affected newborn to present with insulin-dependent diabetes and an inability to tolerate oral feeding due to severe chronic inflammation of the gastrointestinal tract at the time of birth (28). In conjunction with these clinical findings, our data suggest that the elevated frequencies of CD4⁺CD25^high T_Reg cells observed in the fetal periphery may play an essential role in the regulation of potentially devastating activation of self-reactive T cells during fetal development. Because only a subset of patients diagnosed with IPEX show severe signs of autoimmunity at birth, it seems likely that both genetic differences and environmental differences might play an important role in determining the temporal progression and severity of autoimmune disease seen in these patients (29).

Our finding that depletion of CD4⁺CD25^high T_Reg cells from adult MLN cultures did not lead to T cell proliferation, as seen after depletion of CD4⁺CD25^high T_Reg cells from fetal MLN cultures, suggests that the fetal and adult adaptive immune systems exhibit significant differences with respect to the requirements for maintenance of peripheral tolerance. The high levels of proliferation and IFN-γ production observed in the absence of CD4⁺CD25^high T_Reg cells in fetal MLN cultures reveal an increased frequency of potentially autoreactive T cells during fetal development, as has been reported in mice (26). Such an increase could be explained by several different mechanisms, including: incomplete central tolerance during early thymic development (26), a competitive advantage for self-reactive T cells in the fetal periphery due to lymphopenia (30), or intrinsic differences in activation requirements/thresholds of fetal and adult T cells. Another important consideration is that the human fetus may be exposed to maternal alloantigens during development via cross-placental trafficking of maternal cells or proteins. In mice and humans, there is now evidence to suggest that maternal CD4⁺CD25^high T_Reg cells play an important role in controlling maternal alloreactivity to the developing fetus (22). A better appreciation of the immunological environment during fetal ontogeny should benefit future efforts to understand the development of autoimmune diseases and the establishment of peripheral tolerance in humans.

We thank Dr. Walter E. Finkbeiner for provision of paraffin-embedded adult MLN and histological evaluation of tissues, the Gladstone Flow Cytometry Core for help with flow cytometry, Jane Gordon for help with confocal microscopy, Drs. Abner Korn and Juan Vargas for provision of full-term cord blood, Drs. Akiko Kobayashi and Karen Smith-McCune for help staining paraffin-embedded tissues, and Mark Weinstein for help preparing paraffin-embedded tissues. We also thank Drs. Susan Fisher, Einar Martin Aandahl, and Robert Wildin for thoughtful discussions and for review of the manuscript.

The authors have no financial conflict of interest.