Suppressor of Cytokine Signaling 1 Regulates IL-15 Receptor Signaling in CD8⁺CD44^high Memory T Lymphocytes ¹

The development, maturation, and function of hemopoietic cells are critically regulated by cytokines (1, 2). Most of the type I and type II cytokines signal via the Janus kinase (JAK) ³ family protein tyrosine kinases and activate the JAK-STAT pathway, leading to the expression of cytokine-responsive genes (3). The IL-2 family of cytokines is essential for homeostatic control of T lymphocytes in the periphery (4). IL-2 augments the proliferation of activated T cells at the initiation of the immune response, but causes activation-induced cell death (AICD) at later stages (5, 6). IL-15 is essential for the homeostatic expansion of CD8⁺ memory T cells and can block IL-2-induced AICD (7, 8). IL-7 is required not only for the survival and homeostasis of naive CD4⁺ T cells, but also for the generation of CD8⁺ memory T cells (9, 10). IL-7 can influence the survival and expansion of CD8⁺ memory T cells (11, 12, 13, 14). IL-2, IL-15, and IL-7 transduce signals via the common γ-chain (γ_c) using the JAK-STAT pathway involving JAK3 and STAT5 (15, 16, 17). IL-2 and IL-15 also signal via the IL-2Rβ-chain using JAK1 and STAT3 (16, 17).

The magnitude and duration of cytokine receptor signals are tightly controlled by a number of regulatory mechanisms to prevent excessive signaling and abnormal cellular activation (18). These signal attenuation mechanisms include competition for cytokines by soluble receptor subunits, receptor internalization and degradation, inactivation of JAKs by phosphatases, and inhibition of STATs by the protein inhibitors of activated STATs family of proteins. In addition, many cytokines induce the expression of the suppressor of cytokine signaling (SOCS) family adaptor proteins, which function as negative feedback regulators by directly binding and inhibiting JAKs (19). Genetic ablation of SOCS1, SOCS2, and SOCS3 genes have shown that SOCS molecules play a nonredundant role in attenuating cytokine receptor signaling (20, 21, 22, 23, 24, 25).

SOCS1 mRNA is rapidly induced by both type I and type II cytokines (19). The expression of SOCS1 protein is regulated by cis-acting translational control elements (26) and by rapid protein turnover (27). In overexpression studies, SOCS1 inhibits most cytokines that induce its expression by virtue of its ability to inhibit all JAK kinases (19). SOCS family proteins have a characteristic modular organization with a central SH2 domain, a highly variable N-terminal sequence and a conserved C-terminal SOCS box motif. The SOCS box is present in >40 proteins distributed in nine different families (28). Through its SH2 domain, SOCS1 specifically binds to the positive regulatory tyrosine in the JAK activation loop, thereby occluding the accessibility of substrates to the enzyme active site (29, 30, 31). In addition to its activity as a pseudosubstrate inhibitor of JAKs, SOCS1 functions as a specificity factor for a protein-ubiquitin ligase complex. As such, SOCS1 targets JAK2 and Vav1 for ubiquitin-dependent protein degradation (32, 33, 34, 35, 36).

Forced expression of SOCS1 has been shown to inhibit IL-2R and IL-7R signaling (37, 38, 39). However, it remains unclear whether SOCS1 is essential to regulate these cytokines and other γ_c cytokines in vivo, because SOCS1^−/− mice die within 2–3 wk after birth due to uncontrolled IFN-γ signaling. Since SOCS1^−/− mice survive in an IFN-γ-deficient background, we analyzed T cells from SOCS1^−/−IFN-γ^−/− mice. Our results show that SOCS1 is an indispensable regulator of IL-15 and IL-2, but not IL-7, signaling selectively in CD8⁺ T cells and thus plays a critical role in maintaining their homeostasis in the periphery.

SOCS1^+/−IFN-γ^−/− mice were gifts from Dr. J. Ihle, and these animals were in a mixed genetic background of BALB/c and Sv129. SOCS1^−/−IFN-γ^−/− mice were bred in our animal care facility from SOCS1 heterozygous parents because SOCS1^−/−IFN-γ^−/− mice breed very poorly. SOCS1^+/− mice have been backcrossed to C57BL/6 for more than six generations, and SOCS1^+/−IFN-γ^−/− have been selectively bred to maintain the H-2^b/b haplotype. Recombinase-activating gene 1-deficient (Rag1^−/−) mice in the C57BL/6 background were bred in our facility. All experiments using mice were performed following institutional guidelines.

Abs against mouse CD4, CD8, CD25 (IL-2Rα), CD16/CD32 (Fc block), CD44, CD62L, CD69, CD122 (IL-2Rβ), CD132 (γ_c), Ly6C, TCRαβ, Bcl-2, TNF-α, and TCR Vβ-chains 3, 5.1/5.2, 6, pan 8, 8.2, 11, 13, and 17 conjugated to FITC, PE, or biotin were purchased from BD PharMingen (Palo Alto, CA). Streptavidin-Spectral Red and anti-Bcl-x_L-FITC were obtained from Southern Biotechnology Associates (Birmingham, AL). The J558/mIL-7 cell line secreting mouse IL-7 was a gift from Dr. A. Cumano (Pasteur Institute, Paris, France). Rabbit polyclonal Ab against phospho-STAT5 (pSTAT5) was obtained from Cell Signaling Technology (Beverly, MA). Goat anti-rabbit IgG conjugated to Oregon Green 488 and CFSE were purchased from Molecular Probes (Eugene, OR). Culture supernatants (CS) from X630/mIL-2 and J558/mIL-7 cells were used as sources of IL-2 and IL-7. Recombinant human IL-15 was purchased from R&D Systems (Minneapolis, MN), and swine IL-15 was obtained from BioSource International (Nivelles, Belgium). Opti-MEM cell culture medium and FBS were obtained from Invitrogen/Life Technologies (Burlington, Canada). Polyinosinic-polycytidylic acid (Poly I:C) was purchased from Amersham Pharmacia Biotech (Piscataway, NJ). Brefeldin A, PMA, ionomycin, and 5-bromo-2′-deoxyuridine (BrdU) were purchased from Sigma-Aldrich (Oakville, Canada).

Single-cell suspensions in PBS containing 5% FBS and 0.05% sodium azide were preincubated with Fc block for 10 min. The expression of various cell surface markers was estimated by standard three-color staining using FITC-, PE-, and biotin-conjugated primary Abs, followed by streptavidin-Spectral Red. Data acquisition and analysis were performed on a FACSCalibur using CellQuest software (BD Biosciences, Mountain View, CA).

Total lymph node (LN) cells from SOCS1^−/−IFN-γ^−/− or SOCS1^+/+IFN-γ^−/− mice were stimulated with IL-15 (40 ng/ml), IL-2, or IL-7 (both at 1/20 dilution of the CS) for 36–72 h at 2 × 10⁶ cells/ml. For TNF-α detection, cultures were incubated during the last 3 h with 20 ng/ml PMA, 1 μg/ml ionomycin, and 5 μg/ml brefeldin A. The cells were stained for CD4 and CD8 and were fixed in 1% paraformaldehyde (PFA). Following permeabilization with 0.5% saponin in PBS/5% FBS, cells were incubated with anti-Bcl-2-PE, anti-Bcl-x_L-FITC, or anti-TNF-α-PE; washed; and analyzed.

Total LN cells (1 × 10⁵) or 5 × 10⁴ purified T cell subsets were stimulated with the indicated concentrations of cytokines or mitogens for 48 h in 96-well culture plates. One microcurie of [methyl-³H]thymidine (NEN, Boston, MA) was added per well during the last 8–10 h. The cells were harvested onto glass-fiber filter mats, and the incorporated radioactivity was measured in a Top Count (Canberra/Packard, Meriden, CT). The CFSE dye dilution assay was used to estimate the number of cell division cycles within each T cell subset following cytokine stimulation. Total LN cells were labeled with CFSE by incubating cells at 2 × 10⁷ cells/ml in PBS containing 5 μM CFSE for 10 min at room temperature. The reaction was quenched with an equal volume of FBS. The cells were washed twice and stimulated with IL-15 (40 ng/ml), IL-2, or IL-7 (both at 1/20 dilution of the CS) for 72 h and then labeled for surface markers. Sequential reduction in dye content reflecting successive cell division was followed within each gated T cell subset.

Poly I:C treatment and BrdU incorporation were performed as previously described (40). Briefly, 8- to 12-wk-old SOCS1^−/−IFN-γ^−/− or SOCS1^+/+IFN-γ^−/− mice were given 150 μg of Poly I:C dissolved in PBS by the i.p. route. Littermate controls received PBS. The mice were provided with 0.8 mg/ml BrdU in the drinking water, which was replenished daily. After 3 days LN cells were stained for surface CD4 and CD8, or CD8 and CD44. The cells were fixed in ethanol and PFA, then treated with DNase before staining with anti-BrdU-FITC.

Total LN cells from SOCS1^−/−IFN-γ^−/− or SOCS1^+/+IFN-γ^−/− mice were labeled with CFSE as described above. Cells (20 × 10⁶) in 200 μl of PBS were injected i.v. into 8- to 12-wk-old Rag1^−/− mice. After 3 days, LN cells from the recipients were stained for CD4 and CD8, and the extent of homeostatic proliferation was examined by dilution of CFSE within each subset.

Cytokine induced STAT5 phosphorylation in different T cell subsets was examined by intracellular staining and flow cytometric analysis. Briefly, freshly isolated cells in Opti-MEM containing 1 mg/ml BSA and 0.5% FBS were suspended at 2 × 10⁶ cells/ml and stimulated with IL-15 (40 ng/ml) or IL-2 (at a 1/20 dilution of the CS) for 20 or 30 min to achieve maximal STAT5 phosphorylation. Immediately after stimulation, an aliquot of cells was fixed in 2% PFA at room temperature for 10 min. The rest of the cells were washed twice and returned to culture without cytokines. Aliquots were drawn at 30-min intervals and fixed. Fixed cells were washed once in PBS and incubated in cold methanol/acetone (1/1, v/v) on ice for 30 min to achieve cell permeabilization and additional protein denaturation. The cells were rehydrated by washing twice in PBS/5% FBS at room temperature. After incubation with anti-phospho-STAT5 Ab diluted in PBS/FBS for 30 min at room temperature, anti-rabbit IgG conjugated to Oregon Green 488 was added for another 30 min. After blocking with normal rat IgG, the cells were stained for surface markers. Unstimulated control cells were subjected to the same fixation and staining procedures.

The perinatal lethality of SOCS1^−/− mice precluded a detailed analysis of the effects of SOCS1 deficiency on peripheral T cells. Prevention of this lethality by crossing SOCS1^−/− mice into an IFN-γ-deficient background facilitated the analysis of peripheral lymphoid organs of adult SOCS1^−/−IFN-γ^−/− mice. IFN-γ^−/− mice do not differ from IFN-γ^+/+ mice in total T cell numbers or the frequencies of CD4⁺ and CD8⁺ subsets in the thymus or spleen (41). Even though IFN-γ^−/− mice showed a delayed contraction of CD8⁺ T cells following infection with Listeria monocytogenes, total splenocyte numbers returned to near basal levels after the bacterial load was cleared (42). LN and spleen from SOCS1^+/+IFN-γ^−/− and SOCS1^+/+IFN-γ^+/+(C57BL/6) mice, maintained under specific pathogen-free conditions, were comparable in size, cellularity, and the CD4/CD8 ratio (Fig. 1 A and data not shown). Therefore, we compared SOCS1^−/−IFN-γ^−/− mice to SOCS1^+/+IFN-γ^−/− littermate controls to investigate the effects of SOCS1 deficiency on the peripheral T cell compartment.

LN of 8- to16-wk-old SOCS1^−/−IFN-γ^−/− mice were markedly enlarged compared with those of SOCS1^+/+IFN-γ^−/− littermate controls or C57BL/6 mice (Fig. 1,A). The total cellularity of pooled inguinal, auxillary, brachial, cervical, and mesenteric LN of SOCS1^−/−IFN-γ^−/− mice was 2- to 3-fold higher than that of SOCS1^+/+IFN-γ^−/− controls (95 ± 27 × 10⁶ vs 36 ± 5 × 10⁶; see Fig. 1,E). The proportions of TCRαβ⁺ T cells in LN from SOCS1^−/−IFN-γ^−/− and SOCS1^+/+IFN-γ^−/− mice were comparable (Fig. 1,B; 75 and 78%, respectively), indicating that the total T cell numbers are increased in SOCS1^−/−IFN-γ^−/− mice. Subset analysis of the LN T cells revealed that SOCS1^−/−IFN-γ^−/− mice harbored an increased proportion of CD8⁺ T cells resulting in a decrease in the CD4/CD8 ratio of T cells from 2/1 in the SOCS1^+/+IFN-γ^−/− or SOCS1^+/+IFN-γ^+/+ (C57BL6) mice to almost 1/1 in SOCS1^−/−IFN-γ^−/− mice (Fig. 1 B and data not shown). Splenic T cells exhibited a similarly skewed CD4/CD8 ratio (data not shown). These observations revealed that SOCS1 deficiency causes a lymphoproliferative disorder with a preferential expansion of CD8⁺ T cells.

To investigate whether the increase in CD8⁺ T cells in SOCS1^−/−IFN-γ^−/− mice resulted from acute cell proliferation, we evaluated the expression of activation markers such as CD25, CD69, and CD62L on CD4⁺ and CD8⁺ T cells from these mice. The level of expression of these molecules and the proportion of positive cells were similar on CD8⁺ T cells from SOCS1^−/−IFN-γ^−/− and SOCS1^+/+IFN-γ^−/− mice (Fig. 1,C), indicating that CD8⁺ T cells were not acutely activated in SOCS1^−/−IFN-γ^−/− mice. Examination of other cell surface markers revealed that a large proportion of CD8⁺ T cells from SOCS1^−/−IFN-γ^−/− mice showed a marked increase in the expression of the memory T cell markers such as CD44, CD122, and Ly6C (Fig. 1,D). The expression of these memory markers on CD4⁺ cells was comparable between SOCS1^−/−IFN-γ^−/− and SOCS1^+/+IFN-γ^−/− mice. The absolute number of CD8⁺CD44^high cells in SOCS1^−/−IFN-γ^−/− mice was 10 times greater than that in SOCS1^+/+IFN-γ^−/− mice (Fig. 1 E; 22 ± 10 × 10⁶ vs 2 ± 1.6 × 10⁶). In comparison, there was no significant increase in the number of CD8⁺CD44^low cells in SOCS1^−/−IFN-γ^−/− mice (10 ± 5 × 10⁶ vs 8 ± 2 × 10⁶). These findings showed that a majority of the peripheral CD8⁺ T cells in SOCS1^−/−IFN-γ^−/− mice exhibit a memory phenotype.

To investigate whether the CD8⁺CD44^high memory phenotype T cells in SOCS1^−/−IFN-γ^−/− mice originated from a restricted set of T cell clones, we examined the TCR Vβ repertoire of LN T cells from SOCS1^−/−IFN-γ^−/− mice using representative Vβ Abs. Fig. 1 F shows that the diversity of the TCR Vβ repertoire of SOCS1^−/−IFN-γ^−/− CD8⁺ T cells was comparable to that of SOCS1^+/+IFN-γ^−/− cells. A similar distribution of Vβ utilizationwas observed within the CD8⁺CD44^high cells of SOCS1^−/−IFN-γ^−/− mice (data not shown). These results indicate that the enlarged CD8⁺ T cell pool in SOCS1^−/−IFN-γ^−/− mice was polyclonal in origin.

Several laboratories have established a role for IL-15 and IL-7 in the generation, maintenance, and proliferative renewal of CD8⁺CD44^high memory T cells (7, 10, 43, 44, 45, 46). To address whether SOCS1 regulates IL-15- and IL-7-induced proliferation of CD8⁺CD44^high memory T cells, first we tested the proliferative response of total LN cells from SOCS1^−/−IFN-γ^−/− mice and their littermate controls to IL-15 and other γ_c cytokines, IL-2 and IL-7. DNA synthesis induced by IL-15 was markedly elevated in SOCS1^−/−IFN-γ^−/− lymphocytes compared with that in cells from control mice (Fig. 2,A). This increased proliferative response was primarily due to the expansion of CD8⁺ T cells. as measured by the dilution of CFSE fluorescence intensity in T cell subsets (Fig. 2,B). Within the first 3 days of IL-15 stimulation, 66% of SOCS1-deficient CD8⁺ T cells had undergone at least three cell divisions compared with only 17% of SOCS1^+/+IFN-γ^−/− CD8⁺ T cells. IL-2, which uses the same signal transducing receptor subunits as IL-15, stimulated marked proliferation of SOCS1-deficient CD8⁺ T cells, albeit less strongly than IL-15 (Fig. 2, A and B). It should be noted that SOCS1-sufficient CD8⁺ T cells express functional signaling receptors for IL-15 and IL-2, as these cells strongly up-regulate Bcl-2 following IL-2 or IL-15 stimulation to levels comparable to those of SOCS1-deficient CD8⁺ T cells (see Fig. 4, upper panel, middle row). Compared with IL-15 or IL-2, IL-7 induced only a minimal proliferation of peripheral T cells even in the absence of SOCS1 (Fig. 2, A and B). These results showed that while SOCS1 is a critical regulator of IL-15R and IL-2R signaling in CD8⁺ T cells, IL-7R signaling appears to be additionally controlled by other negative regulatory mechanisms.

One of the properties of CD8⁺CD44^high memory phenotype T cells is their ability to undergo Ag nonspecific bystander proliferation mediated by IFN-induced IL-15 during viral infections (7, 40, 47). This process can be mimicked by Poly I:C, a powerful inducer of type I IFNs (40). To examine whether the polyclonal expansion of CD8⁺ T cells in SOCS1-deficient mice might have resulted from increased bystander proliferation, we injected Poly I:C into SOCS1^−/−IFN-γ^−/− and SOCS1^+/+IFN-γ^−/− mice and assessed the proliferation of T cell subsets through BrdU incorporation. Three days after Poly I:C administration, 22% of the CD8⁺ T cells from SOCS1^−/−IFN-γ^−/− mice had proliferated compared with only 6% in SOCS1^+/+IFN-γ^−/− mice (Fig. 2 C). Poly I:C-induced BrdU incorporation in CD4⁺ T cells was low in both SOCS1-deficient and control mice. Nearly all the BrdU⁺ cells within the CD8⁺ T cell pool from both Poly I:C-treated mice and control mice showed high levels of CD44 (data not shown). The elevated frequency of CD8⁺CD44^high T cells in SOCS1^−/−IFN-γ^−/− mice with markedly increased bystander proliferation of CD8⁺ T cells suggested that the lack of SOCS1 may increase the sensitivity of CD8⁺CD44^high T cells to IL-15, leading to their accumulation in SOCS1^−/−IFN-γ^−/− mice.

IL-15 is also essential for acute homeostatic proliferation of CD8⁺CD44^high memory T cells adoptively transferred into T cell-depleted or lymphopenic hosts (13, 48, 49, 50, 51, 52). To investigate whether SOCS1-deficient CD8⁺ T cells also showed enhanced homeostatic proliferation, LN cells from SOCS1^−/−IFN-γ^−/− mice or littermate controls were labeled with CFSE and injected into Rag1^−/− hosts. Examination of the dilution of CFSE fluorescence intensity 3 days after the adoptive transfer showed that CD8⁺ T cells showed a greater capacity to undergo homeostatic expansion than CD4⁺ T cells (Fig. 2,D). SOCS1 deficiency greatly augmented this tendency of CD8⁺ T cells to expand, as >90% of CD8⁺ T cells from SOCS1^−/−IFN-γ^−/− donors had divided at least once compared with 56% of SOCS1^+/+IFN-γ^−/− CD8⁺ T cells (Fig. 2,D). Moreover, 53% of SOCS1-deficient CD8⁺ T cells had undergone more than three cell division cycles within 3 days compared with only 16% of SOCS1-sufficient CD8⁺ T cells. All the proliferating CD8⁺ T cells expressed high levels of CD44, while most of the nondividing cells showed low levels of CD44 expression (Fig. 2 D).

The increased proliferative response of SOCS1-deficient CD8⁺ T cells to IL-15 in vitro and to bystander and homeostatic stimulation in vivo could result from hypersensitivity of CD8⁺CD44^high memory cells lacking SOCS1 to IL-15 or could simply reflect the elevated frequency of this subset in SOCS1-deficient mice. To address this issue, we sorted CD44^high and CD44^low cells from CD8⁺ and CD4⁺ T cell subsets and assessed the responses of the sorted cells to IL-15, IL-2, and mitogens (Fig. 3). As previously reported for SOCS1^−/− splenic T cells (22), we observed that unfractionated LN cells from SOCS1^−/−IFN-γ^−/− mice showed elevated proliferative responses to soluble anti-CD3 and anti-CD28 stimulation. Purified T cells did not proliferate in response to anti-CD3 and anti-CD28 stimulation, because these Abs were added in solution and were not plate bound. The CD8⁺CD44^high, but not the CD8⁺CD44^low, subset proliferated strongly in response to IL-15 or IL-2. Importantly, CD8⁺CD44^high cells from SOCS1^−/−IFN-γ^−/− mice showed a 5-fold stronger proliferation to IL-15 than the same number of cells from SOCS1^+/+IFN-γ^−/− mice. All other subsets responded poorly to both cytokines. All sorted T cell populations from both strains of mice showed comparable response to polyclonal activation by PMA and ionomycin, indicating that the sorted cells were viable and responded to other stimuli. These results demonstrated that the CD8⁺CD44^high T cell subset derived from SOCS1^−/−IFN-γ^−/− mice is distinctly hyper-responsive to IL-15 and IL-2.

Cytokine-induced expression of the antiapoptotic protein Bcl-2 plays an important role in the homeostatic control of lymphocyte survival (4, 53). CD8⁺CD44^high memory T cells show elevated Bcl-2 expression mediated by IL-15 (54, 55). We examined Bcl-2 up-regulation in T cells derived from SOCS1^−/−IFN-γ^−/− mice after cytokine stimulation. Strikingly, IL-2, IL-7, and IL-15 caused a similar up-regulation of Bcl-2 in CD8⁺ T cells from both SOCS1^−/−IFN-γ^−/− and SOCS1^+/+IFN-γ^−/− mice (Fig. 4, upper panel, middle row), even though SOCS1-sufficient CD8⁺ T cells showed a markedly reduced proliferation in response to these cytokines compared with SOCS1-deficient cells (Fig. 4, upper panel, top row). Neither the magnitude of induced Bcl-2 expression in CD8⁺ T cells nor the frequency of responding cells was affected by SOCS1 deficiency (Fig. 4, upper panel, middle row). Induction of Bcl-x_L in CD8⁺ T cells by IL-15 was also comparable between SOCS1-deficient and control cells (Fig. 4,A, upper panel, bottom row). IL-7, but not IL-2 or IL-15, induced significant Bcl-2 up-regulation in CD4⁺ T cells from both strains of mice (Fig. 4, lower panel, middle row). These results suggest that SOCS1 is not required to regulate Bcl-2 or Bcl-x_L expression induced by the γ_c cytokines, and cytokine receptor signals required for inducing Bcl-2 up-regulation and cell proliferation are uncoupled.

Recently, IL-15 has been shown to induce IFN-γ and TNF-β independently of TCR signaling (56). Since the synthesis of IFN-γ or TNF-α can be used interchangeably to assess the functional capacity of CD8⁺ memory T cells (57, 58), we evaluated TNF-α expression following IL-15 stimulation to assess whether the CD8⁺CD44^high T cells that accumulate in SOCS1^−/−IFN-γ^−/− mice are functional. Both IL-15 and IL-2, but not IL-7, strongly induced TNF-α synthesis in SOCS1^−/−IFN-γ^−/− CD8⁺ T cells compared with that in SOCS1^+/+IFN-γ^−/− cells (Fig. 5). SOCS1^−/−IFN-γ^−/−CD4⁺ T cells did not produce TNF-α in response to IL-15, IL-2, or IL-7. These results show that the SOCS1-deficient CD8⁺ T cells exhibit dysregulated cytokine-induced TNF-α production.

To investigate whether the increased responsiveness of SOCS1^−/−IFN-γ^−/−CD8⁺ T cells to IL-2 and IL-15 is due to increased sensitivity of SOCS1-deficient CD8⁺ T cells, merely reflects an increased frequency of CD44^high cells within the CD8⁺ population in SOCS1-deficient mice, or both, we sorted CD8⁺CD44^high and CD8⁺CD44^low cells and evaluated their TNF-α production in response to γ_c cytokines. The data presented in Fig. 5,B show that only CD8⁺CD44^high cells from SOCS1-sufficient mice responded to IL-15. In comparison, both CD8⁺CD44^high and CD8⁺CD44^low cells from SOCS1-deficient mice responded strongly to IL-15. IL-2 was as efficient as IL-15 in stimulating CD8⁺CD44^high cells from only SOCS1^−/−IFN-γ^−/−, but not SOCS1^+/+IFN-γ^−/− mice. IL-2 also stimulated CD8⁺CD44^low cells from SOCS1-deficient mice, albeit less strongly than IL-15. Direct comparison of IL-15-induced TNF-α levels in CD8⁺CD44^high cells from SOCS1^−/−IFN-γ^−/− and SOCS1^+/+IFN-γ^−/− mice showed that the maximal levels of response were comparable; however, the proportion of cells producing higher levels of TNF-α was markedly elevated in the absence of SOCS1 (Fig. 5 C). These results show that the increased responsiveness of SOCS1^−/−IFN-γ^−/− CD8⁺ T cells to IL-2 and IL-15 arises from 1) an increase in frequency of CD44^high cells, 2) a higher proportion of responder cells within the CD8⁺CD44^high population, and 3) sensitivity of CD8⁺CD44^low cells to IL-2 and IL-15 in the absence of SOCS1.

To understand the biochemical basis for the increased sensitivity of SOCS1-deficient CD8⁺ T cells to IL-15 and IL-2, we examined whether SOCS1 deficiency modulates the expression of IL-2R chains following cytokine stimulation. CD8⁺ T cells from SOCS1^−/−IFN-γ^−/− mice expressed higher levels of IL-2Rβ (CD122), but not IL-2Rα (CD25), γ_c (CD132), or IL-7Rα (CD127; Fig. 6,A). Examination of the CD44^high and CD44^low cells within the CD8⁺ T cell subset revealed that not only the SOCS1-deficient CD8⁺CD44^high cells, but also the CD8⁺CD44^low cells, showed markedly elevated IL-2Rβ expression even though the level of expression in CD8⁺CD44^high cells was higher (Fig. 6,A). Analysis of CD8⁺ T cells undergoing proliferation following stimulation with IL-15 or IL-2 showed that IL-2Rβ and IL-2Rα were similarly up-regulated in both SOCS1-deficient and SOCS1-sufficient cells following cell division while γ_c was down-modulated in both strains (Fig. 6 B). These results show that the increased sensitivity of SOCS1-deficient CD8⁺ T cells to IL-15 and IL-2 is at least partially due to increased receptor expression.

Since SOCS1 attenuates cytokine receptor signaling by inhibiting JAKs, we examined whether SOCS1 deficiency potentiates signaling via the IL-2R complex in CD8⁺ T cells, which would explain their increased sensitivity to IL-15 and IL-2. As an indicator of the γ_c-associated JAK3 activity, we examined the intensity and duration of STAT5 phosphorylation in CD8⁺ T cells by a flow cytometric assay, which we have established to obviate the need for sorted cell populations for biochemical assays (75). Total LN cells were stimulated with IL-15 or IL-2 for 20–30 min, washed, and then cultured in the absence of cytokines. Aliquots of cells before and after stimulation and at various time points after cytokine withdrawal were fixed, stained for intracellular pSTAT5 and various cell surface markers, and analyzed by flow cytometry. Even though SOCS1-deficient CD8⁺ T cells showed elevated IL-2Rβ expression, the magnitude of IL-15- or IL-2-induced pSTAT5 signal in CD8⁺ T cells at the peak response was comparable between SOCS1^−/−IFN-γ^−/− and control mice (Fig. 7). However, the IL-15-stimulated pSTAT5 signal persisted in SOCS1-deficient cells for >180 min, while the signal was completely extinguished in control cells by that time. Similarly, IL-2 stimulated a sustained pSTAT5 signal in SOCS1-deficient CD8⁺ T cells, albeit with shorter kinetics. Evaluation of IL-2- or IL-15-induced phospho-STAT5 levels on gated CD8⁺CD44^high T cells showed similar differences in decay kinetics between SOCS1-deficient and SOCS1-sufficient cells (data not shown). These results show that the increased proliferative responsive of SOCS1^−/−IFN-γ^−/− CD8⁺ T cells to IL-15 or IL-2 (Fig. 2,B) and their functional activation (Fig. 5) can be attributed not only to increased IL-2Rβ expression (Fig. 6 A), but also to sustained STAT5 phosphorylation resulting from SOCS1 deficiency.

Interestingly, IL-15 stimulated STAT5 phosphorylation in CD4⁺ T cells comparable in magnitude to that in CD8⁺ T cells (Fig. 7), yet CD4⁺ T cells showed only a marginal proliferation in response to IL-15 even in the absence of SOCS1 (Fig. 2,B), suggesting that the threshold level of cytokine receptor signaling required for stimulating cell proliferation is lower in CD8⁺ T cells than in CD4⁺ T cells. Moreover, SOCS1 deficiency did not significantly affect the kinetics of the pSTAT5 signal induced by IL-15 or IL-2 in CD4⁺ T cells (Fig. 7), indicating that SOCS1 is not a critical regulator of these cytokines in CD4⁺ T cells. Similarly, SOCS1 did not appear to be essential to regulate IL-7 signaling in either CD4⁺ or CD8⁺ T cells, as SOCS1 deficiency did not perturb the decay kinetics of the pSTAT5 signal induced by IL-7 in these cells (Fig. 7). Collectively, these results highlight the specificity of SOCS1 in regulating IL-15 and IL-2 signaling, particularly in CD8⁺ T cells.

In normal, healthy mice the pool of recirculating T lymphocytes and the ratio of CD4⁺ and CD8⁺ T cells in the periphery are maintained at a constant level by homeostatic mechanisms (59). SOCS1^−/−IFN-γ^−/− mice suffer from a lymphoproliferative disorder characterized by lymphadenopathy and a decrease in the CD4/CD8 ratio, indicating that SOCS1 deficiency perturbs the T cell homeostatic mechanisms. We provide evidence that the dysregulated T cell homeostasis in SOCS1^−/−IFN-γ^−/− mice results from the increased sensitivity of SOCS1-deficient CD8⁺ T cells to IL-15. Further, our data show that SOCS1-mediated attenuation of signaling stimulated by γ_c cytokines is cell type specific and is selective to particular cytokines.

Basal numbers of CD8⁺ T cells in SOCS1^−/−IFN-γ^−/− mice are increased, but these cells do not show evidence of acute activation or restricted clonal expansion. The increased numbers of SOCS1^−/−IFN-γ^−/− CD8⁺ T cells is accounted for by an increase in the CD8⁺CD44^high memory phenotype T cell subset. CD8⁺CD44^high memory T cells have a high potential to undergo Ag-independent bystander and homeostatic proliferation compared with CD8⁺CD44^low naive T cells (48, 49, 50, 51, 52). While a low level, basal homeostatic proliferation occurs in healthy animals with full T cell compartments to maintain the memory CD8⁺CD44^high T cell numbers, acute homeostatic expansion occurs in lymphopenic hosts toward establishing a full T cell compartment (50). Since homeostatic proliferation of CD8⁺CD44^high memory T cells is not accompanied by the up-regulation of acute activation markers such as CD69 or CD25 (51), the CD44^highCD25^lowCD69^low phenotype of SOCS1-deficient CD8⁺ T cells indicated that these cells most likely accumulated due to dysregulated homeostatic proliferation. Consistent with this, CD8⁺CD44^high T cells derived from SOCS1^−/−IFN-γ^−/− mice undergo markedly elevated bystander and homeostatic proliferation compared with the same subset of T cells from SOCS1^+/+IFN-γ^−/− mice. The selective expansion of CD8⁺CD44^high T cells in SOCS1^−/−IFN-γ^−/− mice does not perturb the size of the CD8⁺CD44^low naive T cell compartment, in keeping with the fact that the sizes of the naive and memory T cell homeostatic niches are independently regulated (59).

CD8⁺CD44^high memory T cells survive and cycle in response to IL-15 (7, 43, 44). IL-15 appears to be both necessary and sufficient to maintain the CD8⁺CD44^high memory T cell pool, as CD8⁺CD44^high memory T cells are markedly reduced in IL-15- or IL-15Rα-deficient mice (45, 46) and are increased in IL-15 transgenic mice (60). SOCS1^−/−IFN-γ^−/− CD8⁺CD44^high T cells proliferate with increased sensitivity to IL-15 and IL-2 compared with control cells. This increased sensitivity of SOCS1-deficient CD8⁺ T cells to IL-15 and IL-2 results from at least two mechanisms. First, the elevated expression of IL-2Rβ (CD122) on CD8⁺CD44^high cells from SOCS1^−/−IFN-γ^−/− mice is likely to deliver a higher magnitude of signals. Second, the absence of SOCS1 induces sustained STAT5 phosphorylation in SOCS1-deficient CD8⁺CD44^high cells, presumably allowing these cells to achieve the temporal threshold level of signal required for inducing proliferation or effector functions. Both appear to play a cooperative role in SOCS1-deficient CD8⁺ T cells. In SOCS1-deficient CD8⁺CD44^high T cells, the high level of IL-2Rβ expression and sustained signaling up-regulates Bcl-2 expression and induces proliferation following IL-15 or IL-2 stimulation. In CD8⁺CD44^low T cells, SOCS1 deficiency increases Bcl-2 expression, but is not sufficient to induce proliferation, presumably because of the lower level of IL-2Rβ expression, suggesting that the signaling threshold required for Bcl-2 up-regulation is substantially lower than that required for inducing proliferation. Therefore, SOCS1 appears to regulate CD8⁺ T cell homeostasis 1) by preventing uncontrolled basal or homeostatic proliferation of CD8⁺CD44^high memory T cells, and 2) by promoting the survival of CD8⁺CD44^low naive T cells.

SOCS1-deficient CD8⁺CD44^high T cells also showed increased sensitivity to IL-2 in vitro, with sustained STAT5 phosphorylation, TNF-α production, and augmented proliferation. However, IL-2 is less likely to play a role in the expansion of CD8⁺CD44^high T cells in SOCS1^−/−IFN-γ^−/− mice, because, unlike IL-15^−/− or IL-15Rα^−/− mice (45, 46), the lack of IL-2 or the components of the IL-2Rα or IL-2Rβ do not produce specific diminution of the CD8⁺CD44^high memory T cell compartment (61, 62, 63). Moreover, secretion of IL-2 is restricted to activated T cells, whereas IL-15 is widely expressed in multiple cell types of nonlymphoid origin (64). Lastly, IL-2 signaling on activated and memory T cells can cause AICD (5, 6). If SOCS1 attenuated IL-2-induced AICD, the lack of SOCS1 would lead to more AICD of CD8⁺CD44^high memory T cells rather than causing their accumulation. Since IL-15 signaling can block IL-2-induced AICD (65), it is conceivable that increased IL-15 signaling not only induces the proliferation of CD8⁺CD44^high T cells in SOCS1^−/−IFN-γ^−/− mice, but also inhibits IL-2-induced AICD in these cells.

Recent studies have shown that IL-7 influences the generation and homeostasis of CD8⁺ T cells. Bcl-2 induction by IL-7 appears to be required for efficient generation of memory CD8⁺ T cells (10). IL-7 can substitute for IL-15 to support acute homeostatic proliferation in lymphopenic mice (11, 12, 14). However, the markedly diminished CD8⁺CD44^high T cell population in IL-15- or IL-15Rα-deficient mice (45, 46) indicates that endogenous IL-7 is insufficient to maintain the basal homeostatic proliferation of memory CD8⁺ T cells in mice with a full T cell compartment. This is probably because IL-7, being avidly used by CD4⁺ T cells, is limiting to memory CD8⁺ T cells (13). Despite the fact that IL-7 can mediate acute homeostatic proliferation of memory CD8⁺ T cells in vivo, our results show that IL-7 does not stimulate these cells in vitro. However, IL-7-stimulation caused strong STAT5 phosphorylation and Bcl-2 expression in CD8⁺ T cells, suggesting that IL-7 signaling alone is insufficient to induce the proliferation of memory CD8⁺ T cells unless it synergizes with certain other signaling events that occurs in vivo, but not in in vitro cultures.

IFN-γ has been implicated in homeostatic control of activated T cells. Activation-induced death of CD4⁺ T cells following stimulation by viral and bacterial Ags in vivo or by anti-CD3 plus anti-CD28 Abs in vitro is severely compromised in mice lacking the IFN-γ gene (66, 67), increasing their susceptibility to induction of autoimmune encephalomyelitis (68, 69, 70). A similar requirement for IFN-γ signaling in regulating the numbers of CD8⁺ T cells following viral or bacterial infection has been shown to significantly delay the kinetics of the immune response (42, 71). However, mice deficient in IFN-γ, IFN-γR1, or IFN-γR2 do not show any abnormality in T cell development or homeostasis (41, 72, 73), and their prolonged immune responses following infections eventually return to homeostasis (42, 71). It is possible that the increased total T cell number and the elevated frequency of CD8⁺CD44^high cells in SOCS1-deficient mice on the IFN-γ^−/− background could arise not only from increased responsiveness of CD8⁺CD44^high cells to γ_c cytokines, but also from their inability to undergo apoptosis following activation. However, the fact that SOCS1^−/−IFN-γ^+/+ mice also show an increased frequency of CD8⁺CD44^high cells (data not shown) suggests that IFN-γ deficiency is not a prerequisite for their accumulation. On the other hand, the failure of CD8⁺CD44^high cells to accumulate in SOCS1^+/+IFN-γ^−/− mice suggests that the primary activation signals for these cells in SOCS1-deficient mice could be cytokines, dysregulated by the lack of SOCS1. Clearly, further investigation is needed to resolve this issue.

SOCS1 has been shown to inhibit most cytokines that induce its expression (19). However, the analysis of SOCS1^−/−IFN-γ^−/− mice has revealed remarkable specificity in the role of SOCS1 in attenuating signaling stimulated via γ_c cytokines. For example, IL-7 signaling is attenuated by forced expression of SOCS1 (37, 38); however, endogenous SOCS1 appears not to be a critical regulator of IL-7 signaling. IL-7 regulates the homeostasis of CD4⁺ T cells (9), and SOCS1 deficiency did not cause either dysregulated proliferation of CD4⁺ T cells or superinduction of Bcl-2 in these cells. Similarly, IL-7-induced STAT5 phosphorylation and Bcl-2 induction are comparable between SOCS1-deficient and control CD8⁺ T cells. These observations suggest that IL-7 signaling may be regulated by alternative signal attenuation mechanisms, probably involving other SOCS family members.

We have shown that SOCS1-deficient CD8⁺ T cells produce excessive amounts of TNF-α in response to IL-15 and IL-2, even in the absence of specific antigenic stimulation. While secretion of proinflammatory cytokines is beneficial in secondary Ag-specific immune responses, excessive production of these cytokines in an Ag-nonspecific manner may lead to tissue damage and trigger autoimmune reactions. Therefore, SOCS1, in addition to regulating the proliferation of the CD8⁺ memory T cells, prevents inappropriate activation of CD8⁺CD44^high cells and subsequent release of TNF-α into sites of ongoing inflammation. Consistent with this idea, aged SOCS1^−/−IFN-γ^−/− mice accumulate chronic inflammatory lesions in various organs (74).

Numerous mechanisms exist to negatively regulate cytokine receptor signals, such as receptor down-modulation, activation of tyrosine phosphatases, and sequestration and destruction of STAT molecules by the protein inhibitors of the activated STAT family of proteins (18). Clearly, none of these mechanisms is sufficient to regulate the duration of IL-15 signaling in SOCS1^−/− CD8⁺ T cells. By attenuating the duration of IL-15-induced signaling events, SOCS1 not only regulates the homeostatic proliferation of CD8⁺CD44^high T cells, but also restrains them from producing proinflammatory cytokines in the absence of specific antigenic stimulation. Thus, this study of SOCS1^−/−IFN-γ^−/− mice has revealed an essential and nonredundant role of SOCS1 in regulating IL-15 signaling and T lymphocyte homeostasis.

We gratefully acknowledge Dr. J. Ihle for the SOCS1^+/− and SOCS1^+/−IFN-γ^−/− mice, and Dr. Ana Cumano for the J558/mIL-7 cell line. We thank Claude Cantin for expert technical assistance with flow cytometry, Trista Steel for Ab purification, and Dina Finan and Brandon Reinhart for critical reading of the manuscript.