Cutting Edge: CD4-Independent Development of Functional FoxP3⁺ Regulatory T Cells

Mice

CD4^−/−, CD3^−/−, I-Aβ^−/−, and MHC-II^Δ/Δ mice (all of the B6 background) were described previously (10, 11, 12).

Flow cytometry

Fluorescently conjugated mAbs directed against CD3 (145-2C11), CD4 (L3T4), CD8 (Ly2 53-6.7), CD25 (PC61), CD44 (IM7), I-A/I-E (AF6–120.1), and Ki-67 (B56) from BD Biosciences or to CD62L (MEL-14), CD103 (M290), CD152 (UC10-4B9), GITR (DTA-1), TCRδ (UC7–13D5), TCRβ (H57-597), and MHC-II (ME/114.15.2) from eBioscience were used. V_β usage was determined using biotinylated anti-Vβ2 (B20.6), anti-Vβ3 (KJ25), anti-Vβ4 (KT4), anti-Vβ6 (RR4-7), anti-Vβ8 (KJ16), anti-Vβ10 (KT10b), and anti-Vβ11 (KT11) staining followed by PE-conjugated streptavidin (eBioscience). Intracellular FoxP3 (FJK-16s) staining was performed following the manufacturer’s protocol (eBioscience). Analyses were performed on a FACSCalibur or FACSCanto flow cytometer (BD Biosciences) using the FlowJo software.

Immunoscope analysis

Lymph node (LN) cells were enriched using a CD4 negative isolation kit (Invitrogen) and stained with anti-CD8, anti-CD25, and anti-CD3 Abs. CD3⁺CD8⁻CD25⁺ or CD3⁺CD8⁻CD25⁻ cells were sorted (purity ≥ 97%) using a FACSAria flow cytometer (BD Biosciences). Total RNA was extracted from 2–10 × 10⁶ sorted cells using TRIzol (Sigma-Aldrich). Protocols for Vβ-Cβ RT-PCR and gel analysis were described elsewhere (13). Immunoscope software was used to obtain CDR3 length profiles.

Functional Treg assays in vitro

LN cells were enriched using a CD4 negative isolation kit, yielding a CD3⁺CD8⁻ population (purity ≥ 98%). Then, CD25⁻ (purity ≥90%) and CD25⁺ (purity ≥89%) cells were isolated using PE-labeled anti-CD25 staining followed by immunomagnetic sorting with anti-PE beads (Miltenyi Biotec).

CD3⁺CD8⁻CD25⁺ cells were added at different ratios to CD4⁺CD25⁻ responder cells (5 × 10⁴) from B6 mice. Cells were activated for 72 h by 2.5 μg/well Con A in the presence of 5 × 10⁴ splenocytes from syngeneic CD3^−/− mice and pulsed with 1 μCi/well [³H]TdR for the last 24 h of culture.

Alternatively, CD3⁺CD8⁻CD25⁺ or CD3⁺CD8⁻CD25⁻ cells (1 × 10⁵) were cultured separately for 4 days with anti-CD28/CD3 Dynal beads (Invitrogen) before IL-2 was assayed in supernatants on a Bio-Plex/Luminex apparatus (Bio-Rad).

Functional Treg assay in vivo

FACS-purified CD3⁺CD8⁻CD25⁻ T cells (1 × 10⁶) from CD4^+/+ mice supplemented or not with 0.2 × 10⁶ FACS-purified CD3⁺CD8⁻CD25⁺ Tregs (purity ≥ 97%) from CD4^+/+ or CD4^−/− mice were injected i.v. into B6 CD3^−/− recipients. The weight curve was monitored for 7 wk after cell transfer. Then, colons were removed and 5-μm paraffin-embedded sections stained with hematoxylin/eosin/saffron.

RNA preparation and RT-PCR

Total cellular RNA was extracted from sorted CD3⁺CD8⁻CD25⁺ or CD3⁺CD8⁻CD25⁻ cells using TRIzol (Invitrogen). The Moloney murine leukemia virus enzyme was used to synthesize cDNA from 500 μg of RNA. Primers for PCR were murine cKrox-forward (5′-ACATGAGGACCCACACTGGTG-3′) and murine cKrox-reverse (5′-CTTCCTCTTCCTCC TCCTCAG-3′).

FoxP3⁺ T cells develop despite the absence of CD4

In mice that lack expression of the CD4 glycoprotein, T cells of the CD4 lineage are identified according to their CD3⁺CD8⁻ “CD4-like” phenotype (14). As expected from previous reports (9 , 10), the frequency of CD4-like T cells recovered from peripheral lymphoid organs was dramatically reduced in CD4^−/− mice; CD3⁺CD8⁻ T cells represented <10% of LN mononuclear cells as compared with >30% in CD4^+/+ mice (Fig. 1⇓A), corresponding to a ∼70% reduction in absolute numbers (Fig. 1⇓B). To further study the CD3⁺CD8⁻ T cell subset found in CD4^−/− mice, we investigated the expression of the CD4-specific transcription factor cKrox (also called ThPOK) (15). Similarly as MHC class-II restricted CD4⁺ OT-II but not MHC class-I restricted CD8⁺ OT-I TCR-transgenic T cells, cKrox was expressed by CD3⁺CD8⁻ T cells from CD4^+/+ as well as CD4^−/− mice (Fig. 1⇓C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Increased frequency of FoxP3⁺ cells within the CD3⁺CD8⁻ population in the LNs of CD4^−/− mice. A, Thymocytes and LN cells from 4- to 5-wk-old CD4^−/− (n = 14) or CD4^+/+ (n = 12) mice were stained with FITC anti-CD8, PE anti-FoxP3, PerCP anti-CD25, and allophycocyanin anti-CD3. Dot plots display the results of flow cytometry analysis after gating on lymphocytes based on forward and side scatter (left) or on CD3⁺CD8⁻ cells (right). Numbers indicate the frequency (mean ± SEM) of CD3⁺CD8⁻ cells within the lymphocyte gate (left) or the frequency of FoxP3⁺ cells within the CD3⁺CD8⁻ gate (right). B, Histograms indicate normalized mean cell numbers in CD4^−/− mice relative to CD4^+/+ mice taken as 100% (error bars represent SEM). The values indicated above the histograms are mean absolute cell numbers (×10⁶) of CD3⁺CD8⁻FoxP3⁺ or CD3⁺CD8⁻FoxP3⁻ cells within the thymus (left) or 12 pooled LNs (right). ∗∗, p < 0.01; ∗∗∗, p < 0.001 (Mann-Whitney U test). C, RT-PCR analysis of cKrox mRNA expression in immunomagnetic bead-purified LN CD4⁺ cells from OT-II or CD8⁺ cells from OT-I mice and in FACS-sorted LN CD3⁺CD8⁻CD25⁺ or CD3⁺CD8⁻CD25⁻ cells from CD4^−/− or CD4^+/+ mice.

We next investigated whether the Treg compartment would be similarly affected by the lack of CD4. In agreement with the healthy status of CD4^−/− mice that do not exhibit obvious autoimmune manifestation, FoxP3⁺ cells were readily found within the thymi and LNs of mice lacking CD4 and mainly corresponded to the cells expressing CD25 (Fig. 1⇑A). Analysis of thymocytes from animals of different ages shows that the percentages of FoxP3⁺ within the CD3⁺CD8⁻ subset represent ∼3–4% throughout the 40-wk observation period, whereas their total numbers tend to decline in aged mice (supplemental Fig. 1).⁶ In 4- to 5-wk-old mice, the numbers of CD3⁺CD8⁻FoxP3⁺ thymocytes were reduced 4-fold in CD4^−/− mice as compared with CD4^+/+ mice, whereas the conventional FoxP3⁻ thymic compartment was 3-fold diminished (Fig. 1⇑B). Thus, the genetic deletion of the CD4 coreceptor affects the development of both conventional and regulatory CD4-like T cells within the thymus, indicating that the proposed higher affinity of Treg TCRs for self-MHC/peptide complexes may not be able to compensate for the lack of CD4 contribution to TCR avidity. Alternatively, this result suggests that Tregs have a degree of autoreactivity to self similar to that of their Tconv counterparts when CD4 is absent. Such an interpretation would be in agreement with the significant overlap found between the TCR repertoires of CD4⁺ Tconv cells and Tregs, which are more closely related than the TCR repertoires of CD4⁺ and CD8⁺ cells (16).

Surprisingly, despite lower absolute numbers of FoxP3⁺ cells in the thymi of CD4^−/− mice, the percentages of CD3⁺CD8⁻FoxP3⁺ T cells within the peripheral compartments appeared to be dramatically increased in CD4^−/− mice (23% ± 0.9) as compared with their wild-type counterpart (12% ± 0.4; p < 0.001) (Fig. 1⇑A). Hence, whereas CD4^−/− mice had reduced total numbers of CD3⁺CD8⁻FoxP3⁻ Tconv peripheral cells (3-fold reduction), the total numbers of cells in the LN CD3⁺CD8⁻FoxP3⁺ compartment were less affected (2-fold reduction) (Fig. 1⇑B). These cell numbers appeared stable over the lifetimes of the mice, because 8- to 10-wk, 15- to 20-wk, and 20- to 40-wk animals had 0.5 ± 0.14 × 10⁶ (n = 6), 0.4 ± 0.04 × 10⁶ (n = 8), and 0.7 ± 0.15 × 10⁶ (n = 6) LN CD3⁺CD8⁻FoxP3⁺ cells, respectively.

CD3⁺CD8⁻FoxP3⁺ T cells from CD4^−/− mice are enriched in memory-type Tregs

We further determined whether the CD4-like FoxP3⁺ cells present in CD4^−/− mice display the classical phenotypic features of Tregs. CD3⁺CD8⁻FoxP3⁺ cells were almost exclusively TCRαβ⁺ and virtually no TCRγδ⁺ cells were detected in this subset (supplemental Fig. 2A). They did not express the NK and invariant NK T cell marker CD49b (supplemental Fig. 2A). As in normal mice, the vast majority (77% ± 3) of CD3⁺CD8⁻FoxP3⁺ cells from CD4^−/− animals were CD25⁺ (Fig. 1⇑A). Also, expression of other classical Treg markers such as surface GITR (glucocorticoid-induced TNF receptor family-related protein) and intracellular CD152 was comparable in CD4^−/− and CD4^+/+ mice (supplemental Fig. 2B).

One hypothesis for explaining the high frequency of FoxP3⁺ cells among the peripheral CD4-like compartment could be a preferential survival of FoxP3⁺ cells and/or a higher level of homeostatic proliferation as compared with Tconv cells. Although we did not find any difference in the frequencies of proliferating Ki-67⁺ cells in adult mice (∼30% of CD3⁺CD8⁻FoxP3⁺ in both CD4^−/− and CD4^+/+ mice; data not shown), the proportion of memory-type CD44⁺CD62L^low cells within the CD3⁺CD8⁻FoxP3⁺ subset appeared increased in CD4^−/− mice, as was the frequency of CD103⁺ cells (supplemental Fig. 3A). The proportion of these CD44⁺CD62^low and CD103⁺ cells increased in aged animals (supplemental Fig. 3B), presumably because either activated/memory Tregs gradually expand and/or adaptive Tregs are generated to compensate for the lower number of Tregs. These results suggest that homeostatic mechanisms contribute to the maintenance of a significant Treg pool in the periphery of CD4^−/− mice.

CD4-like Tregs from CD4^−/− mice display a diversified TCR repertoire

We next evaluated whether the absence of CD4 would impair the development of a normally diversified repertoire. TCR-Vβ usage of CD3⁺CD8⁻FoxP3⁺ cells appeared undistinguishable in CD4^−/− and CD4^+/+ mice (supplemental Fig. 4A), as was also the case for the CD3⁺CD8⁺FoxP3⁻ subset (data not shown). To further assay the level of TCR diversity, we also analyzed the TCR-Vβ CDR3 length distribution using the Immunoscope method. This confirmed that CD4-like Tregs from CD4^−/− mice had a fully diversified TCR repertoire (supplemental Fig. 4B). This was also the case when comparing the Immunoscope profiles of CD3⁺CD8⁺CD25⁻ cells in both groups (not shown). Hence, generation of the Treg compartment in CD4^−/− mice is not due to the peripheral expansion of a pauciclonal cell population. Rather, our data indicate that the lack of CD4 does not impair the production of a highly diversified Treg repertoire.

FoxP3⁺ T cells can develop in CD4/MHC class II double-deficient mice

The observation that FoxP3⁺ Tregs can develop in the absence of CD4 raises the possibility that their generation may be, at least partly, independent of interactions with MHC class II molecules. To investigate this point, we bred CD4^−/− mice under an MHC-II-deficient I-Aβ^−/− background (11). Surprisingly, the frequencies and absolute numbers of peripheral CD3⁺CD8⁻FoxP3⁺ cells in CD4^−/−I-Aβ^−/− mice were not further reduced by the absence of MHC-II molecules as compared with that of CD4^−/− mice (supplemental Table I). Because I-Aβ^−/− mice could still harbor some cross-paired I-Aα I-Eβ MHC-II chains, we additionally used another model of mice that lack all conventional MHC-II chains due to a large deletion of the entire class II region, i.e., MHC-II^Δ/^Δ mice (12). Even in this more stringent system, CD3⁺CD8⁻FoxP3⁺ cells from CD4^−/−MHC-II^Δ/^Δ were still found at similar frequencies and absolute numbers than in CD4^−/− mice (supplemental Table I). Hence, CD4-like Tregs can develop in the absence of both CD4 and MHC-II molecules, demonstrating that the generation of Tregs can be, at least partly, independent of MHC class II recognition. This conclusion is consistent with the observation that not only CD8⁺ Tregs (17) but also CD4-like Tregs may be found in MHC class II-deficient mice (18), and even in double-MHC class I- and class II-deficient mice (19).

CD3⁺CD8⁻CD25⁺ T cells of CD4^−/− mice are functionally suppressive in vitro and in vivo

To evaluate the functional characteristics of CD4-like Tregs, we investigated their proliferative response to Con A activation and their capacity to suppress the proliferation of responder T cells. As expected for Tregs, CD3⁺CD8⁻CD25⁺ cells were profoundly hyporeactive in vitro in response to mitogenic activation (Fig. 2⇓A). This in vitro hyporesponsiveness was further attested to by the fact that upon stimulation by anti-CD3/anti-CD28 beads for 4 days, CD3⁺CD8⁻CD25⁺ cells from both CD4^−/− and CD4^+/+ mice similarly produced ∼100-fold less IL-2 than CD3⁺CD8⁻CD25⁻ cells (Fig. 2⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Suppressive function of CD3⁺CD8⁻CD25⁺ Tregs from CD4^−/− mice in vitro and in vivo. A, Proliferative response of Tregs from CD4^−/− or CD4^+/+ mice upon activation with Con A. One representative result of three independent experiments is shown. B, IL-2 secretion by Tregs from CD4^−/− or CD4^+/+ mice upon activation by anti-CD3/anti-CD28 beads. One representative result of three independent experiments is shown. C, Inhibition of Con A-activated effector T cell proliferation by CD3⁺CD8⁻CD25⁺ Tregs from CD4^+/+ or CD4^−/− mice. One representative result of four independent experiments is shown. D, Inhibition of CD3⁺CD8⁻CD25⁻ (Tconv)-induced IBD by CD3⁺CD8⁻CD25⁺ Tregs from CD4^+/+ or CD4^−/− mice. The graph in the left displays variations of body weight. The pictures in the right correspond to colon histological analysis of H & E/saffron-stained paraffin-embedded sections (7 wk after transfer). One representative result of two independent experiments (n = 6 for each group) is shown.

To evaluate the suppressive function of CD4-like Tregs, different ratios of CD3⁺CD8⁻CD25⁺ cells were assayed for their capacity to inhibit the proliferative response of syngeneic CD4⁺CD25⁻ responder cells. CD3⁺CD8⁻CD25⁺ cells from CD4^−/− mice were found as efficient as those from CD4^+/+ animals in suppressing the proliferation of effector cells in a dose-dependent manner (Fig. 2⇑C).

To further evaluate the suppressive capacity of Tregs in the absence of CD4, we also analyzed their potency to prevent inflammatory bowel disease (IBD) in vivo. For this, purified Tconv cells were transferred with or without purified Tregs from CD4^+/+ or CD4^−/− mice to T cell-deficient CD3^−/− animals. Mice injected with Tconv cells alone lost 14% of their maximal weight 7 wk after transfer (Fig. 2⇑D). This weight loss was associated with diarrhea, rectal prolapse, and local inflammation as indicated by histological signs of microvilli destruction and extensive mononuclear cell infiltrates. The coinjection of Tregs from CD4^+/+ or CD4^−/− mice completely prevented the wasting syndrome and also histological signs of inflammation (Fig. 2⇑D). Together, these data demonstrate that CD4-like Tregs from CD4^−/− mice are functionally suppressive in vitro and, remarkably, also in vivo.

It follows from the present findings that CD4-like Tregs can develop in the absence of both CD4 and MHC-II molecules, indicating that the generation of Tregs can be, at least partly, independent of TCR/MHC class II recognition. This is in marked contrast with the current view that Treg development is highly dependent on high-avidity interactions with MHC class II molecules (3, 4, 5). This prevailing paradigm relies on experiments performed in TCR/Ag transgenic mice (i.e., with a repertoire biased toward the recognition of given specificities). In fact, it was shown that the total numbers of developing Tregs do not increase upon an encounter with the cognate Ag on thymic epithelial cells even if their proportion appears enriched relative to conventional CD4⁺ thymocytes (6). This has suggested that selective survival, rather than induced differentiation, is the reason that explains the apparent Treg enrichment observed in previous studies. This hypothesis is compatible with the view raised by the present data that high affinity for self-Ags may not constitute the only mechanism promoting differentiation toward the Treg pathway. Indeed, if high affinity to self was to play a major role, we would expect that deletion of the CD4 coreceptor, by limiting the possibility to recognize MHC with high overall avidity, should more severely impact the differentiation of Tregs than their conventional counterparts. Our results rather show that the absence of CD4 similarly impairs the thymic differentiation toward the Treg lineage and toward the Tconv pathway (Fig. 1⇑).

Yet, this interpretation relies on the underlying assumption that CD4-like lymphocytes generated in CD4^−/− mice are restricted to MHC class II molecules. Our finding that FoxP3⁺ CD4-like lymphocytes are also found in CD4/MHC class II double-deficient mice raises the possibility that Treg differentiation may not strictly depend on MHC-class II. In line with this view, it has recently been suggested that lymphocytes that differentiate in mice deficient for all coreceptors (CD4/CD8) are autoreactive and, surprisingly, are not restricted to MHC molecules (20). Indeed, by linking p56^lck recruitment to MHC recognition, coreceptors restrict the specificity of nascent thymocytes to MHC molecules while inhibiting MHC-independent signaling by sequestering p56^lck. Because the CD4 coreceptor binds more abundantly to p56^lck than CD8, it is plausible that the thymocytes developing in CD4^−/− mice are partly independent of MHC molecules. This is consistent with our findings, because such thymocytes would be independent of the presence of both CD4 and MHC class II for their development and could be reactive to yet unidentified autoantigens. In such a situation, affinity for MHC-independent Ags would not be dependent on the presence of CD4 and could equally affect the selection of both Tconv cells and Tregs, as found in this study.

In conclusion, the present findings demonstrate that functional FoxP3⁺ CD4-like cells displaying a fully diversified TCR repertoire can develop in the absence of a CD4 coreceptor. Although their differentiation in the thymus is impaired, as it is for Tconv cells, the situation is different in the periphery where Tregs are found in higher proportions than conventional CD4-like cells. Consistently, their total numbers are less affected by lack of CD4 than conventional T cells when compared with wild-type mice. Conceivably, this differential behavior may be linked to peripheral homeostatic mechanisms—as suggested by the augmented representation of memory-type CD44⁺CD62^low and CD103⁺ Tregs—that contribute to the prevention of autoimmune manifestations in CD4^−/− animals. Consistent with this notion, FoxP3⁺ CD4-like cells from these animals are fully capable of preventing IBD mediated by the transfer of CD4⁺ T cells in CD3^−/− hosts. Thus, caution should be taken in studies that use CD4^−/− or even MHC class II-deficient animals as models of Treg-free mice. Finally, it would be interesting in future studies to define the fine antigenic specificities of Tregs that can develop in CD4^−/− as well as in CD4/MHC class II double-deficient mice. Their characterization would provide a proof of principle that non-MHC-restricted Tregs may function independently of individual MHC backgrounds, paving the way to a potential therapeutic use upon activation by non-MHC-dependent ligands.