STAT3 mutations underlying autosomal dominant hyper-IgE syndrome impair human CD8⁺ T cell memory formation and function

Abstract

Background

The capacity of CD8⁺ T cells to control infections and mediate anti-tumor immunity requires the development and survival of effector and memory cells. IL-21 has emerged as a potent inducer of CD8⁺ T cell effector function and memory development in mouse models of infectious disease. However, the role of IL-21 and associated signaling pathways in protective CD8⁺ T cell immunity in humans is unknown.

Objective

To determine which signaling pathways mediate the effects of IL-21 on human CD8⁺ T cells and whether defects in these pathways contribute to disease pathogenesis in primary immunodeficiencies caused by mutations in components of the IL-21 signaling cascade.

Methods

Human primary immunodeficiencies resulting from monogenic mutations provide a unique opportunity to assess the requirement for particular molecules in regulating human lymphocyte function. Lymphocytes from patients with loss-of-function mutations in STAT1, STAT3 or IL21R were used to assess the respective roles of these genes in human CD8⁺ T cell differentiation in vivo and in vitro.

Results

Mutations in STAT3 and IL21R, but not STAT1, lead to a decrease in multiple memory CD8⁺ T cell subsets in vivo, indicating that STAT3 signaling – possibly downstream of IL-21R - regulates the memory cell pool. Furthermore, STAT3 was important for inducing the lytic machinery in IL-21-stimulated naïve CD8⁺ T cells. However, this defect was overcome by TCR engagement.

Conclusion

The IL-21R/STAT3 pathway is required for many aspects of human CD8⁺ T cell behavior but in some cases can be compensated by other signals. This helps explain the relatively mild susceptibility to viral disease observed in STAT3 and IL-21R-deficient individuals.

INTRODUCTION

CD8⁺ T cell responses are essential for the control of viruses and protection against some tumors. Common γ-chain family cytokines are important regulators of CD8⁺ T cell behavior. Thus, IL-7 and IL-15 control lymphocyte homeostasis^1-3, while IL-2 regulates differentiation of naïve cells into effector or memory populations^{4, 5}. IL-21 has also been reported to control CD8⁺ T cell function. In vitro, IL-21 increases survival and proliferation of mouse^6-9 and human CD8⁺ T cells^10-12, induces effector molecules such as IFN-γ, granzyme B and perforin 6, 13-15, and transcription factors BCL6 and EOMES that control their differentiation into effector and memory populations¹⁶.

IL-21 is also implicated in controlling immune responses in vivo. Treatment of cancer patients with IL-21 resulted in upregulation of cytotoxic molecules such as granzyme B, perforin and IFN-γ in their CD8⁺ T cells and NK cells¹⁷. In mice, IL-21 enhanced memory CD8⁺ T cell responses during vaccinia infection^{9, 18} and was required for CD8⁺ T cell mediated control of chronic LCMV infection^19-21. IL-21 alone, or together with IL-15, also increased the efficacy of anti-tumor responses by CD8⁺ T cells^{8, 9, 22-25}. Thus, IL-21 is a potent inducer of CD8⁺ T cell effector function and memory with clinical relevance in both anti-viral and anti-tumor immunity.

IL-21 mediates its effects by activating JAK1 and JAK3^{15, 26, 27} leading to phosphorylation of STAT1, STAT3 and STAT5^{6, 15, 26}. IL-21 can also activate MAPK and Akt⁶. It is not clear, however, which of these pathways mediates the stimulatory effects of IL-21 on human CD8⁺ T cells. Primary immunodeficiencies (PID) resulting from mutations in single genes provide a unique opportunity to address the role of individual molecules in regulating immune responses. Autosomal dominant hyper IgE syndrome (AD-HIES) is a PID characterized by chronic eczema, elevated serum IgE levels and recurrent infections of the skin, mucosa and lungs^{28, 29}. Notably, some AD-HIES patients have impaired control of reactivation of infection with herpes viruses (HSV, VZV)^{29, 30}, and are predisposed to developing non Hodgkin’s B-cell lymphoma^29-32. The molecular lesion in AD-HIES is heterozygous mutations in STAT3, with mutant alleles working in a dominant negative manner^{33, 34}. Mutations in STAT1 also result in infectious susceptibility to particular pathogens. Thus, mono- or bi-allelic loss-of-function STAT1 mutations severely compromise responses to IFN-γ. However, responses to IFN-α/β and IFN-λ are either intact (AD STAT1 deficiency; heterozygous mutations) or impaired (autosomal recessive [AR] STAT1 deficiency; bi-allelic mutations). Consequently, these mutations result in clinical disease caused by weakly virulent mycobacteria, and occasionally non-lethal viral infection^{35, 36}. On the other hand, bi-allelic null mutations abolish STAT1-dependent cellular responses to IFN-γ, IFN-α/β and IFN-λ, thereby predisposing affected individuals to fatal infection with herpes viruses and mycobacteria^{35, 36}. The importance of IL-21 signalling in humans was recently highlighted by the identification of four patients with IL21R mutations who develop recurrent respiratory and gastrointestinal infections, particularly with cryptosporidia resulting in chronic liver disease³⁷. Two of these patients also exhibited ongoing infection with norovirus and rhinovirus, but immunity against herpes viruses and other pathogens that are commonly problematic for patients with combined immunodeficiencies (eg CMV, EBV) appeared to be intact³⁷.

Here, we used STAT3 mutant (STAT3^MUT), STAT1^MUT and IL21R^MUT patients to determine the requirement for STAT1 and STAT3 in regulating human CD8⁺ T cell responses. IL-21 in combination with IL-15 induced proliferation of, and granzyme expression in, naïve CD8⁺ T cells. Loss of STAT3 function impaired IL-21-induced granzyme B expression but did not affect its ability to induce proliferation. However, strong TCR/co-stimulatory signals could rescue granzyme expression in STAT3^MUT T cells. Loss of STAT1 function did not affect controlled the formation/maintenance of effector and memory CD8⁺ T cell subsets in vivo, as proliferation or granzyme B production. We also found that STAT3, but not STAT1, evidenced by reduced frequencies of differentiated memory populations. We also observed some memory cell deficiencies in patients with IL21R mutations, implicating IL-21 as a potential STAT3-activating cytokine required for CD8⁺ memory T cell homeostasis. These findings provide insight into some of the clinical features of AD-HIES and IL-21R deficiency, including impaired control of viral infection and susceptibility to B-cell lymphoma.

Materials and Methods

Human Samples

Buffy coats from normal donors were purchased from the Australia Red Cross Blood Service. Peripheral blood (PB) was collected from patients with mutations in STAT3, STAT1 or IL21R (Refer to Table E1 in Online Repository for Patient details). All human experiments were approved by ethics committees in Canberra, Sydney, Melbourne, Brisbane and Perth, and institutional review boards of Necker Medical School, Rockefeller University and NIH.

T cell phenotyping & isolation

PB CD8⁺ T cells were stained with mAbs to CD4, CD8, CCR7 and CD45RA. Subsets were defined as naïve (CD8⁺CD4^-CCR7⁺CD45RA⁺), central memory (T_CM; CD8⁺CD4-CCR7⁺CD45RA^-), effector memory (T_EM; CD8⁺CD4^-CCR7^-CD45RA^-) or T_EMRA (revertant CD45RA effector memory; CD8⁺CD4^-CCR7^-CD45RA⁺). For experiments with STAT3^MUT samples naïve cells were isolated from samples using a Positive Isolation Dynal Kit (Invitrogen) followed by sorting CD8⁺CCR7⁺CD45RA⁺ cells (FACSAria: BD). Due to limiting numbers of cells in STAT1^MUT and IL-21R^MUT samples, naïve CD8⁺T cells were isolated directly by sorting. For phenotyping cells were also stained for further cell surface markers (Table E2 in the Online Repository lists mAbs used)

Expression of phospho-STATs

Normal naïve CD8⁺ T cells were cultured for four days with TAE beads (Miltenyi Biotech), rested for 2 hours in OPTI-mem (LifeTech) plus Normicin (InVivogen), and then stimulated in the absence or presence of IL-2 (50 U/ml), IL-15 (50 ng/ml) and/or IL-21 (50 ng/ml) for 30 min. Cells were fixed with 2% paraformaldehyde, permeabilized with 90% methanol and stained with anti-phospho-STAT1, STAT3 and STAT5 mAbs.

In Vitro Stimulation of Naïve CD8⁺ T cells

Naïve CD8⁺ T cells were labeled with CellTrace™ Violet (CTV; Invitrogen) and then cultured (∼4 × 10⁴ cells/200μl/well) with or without TAE beads (one bead/5 cells) for 4 or 10 d respectively; alone or together with 50 U/ml IL-2 (Millipore), 50 ng/ml IL-15 or 50 ng/ml IL-21 (PeproTech). Cells were thenharvested, permeabilized and stained with anti-perforin and anti-granzyme B mAb. Cell division and phenotype were determined using FlowJo software (Tree Star, Inc.).

Quantitative PCR analysis

RNA was isolated immediately after ex vivoisolation or after 4 or 10 days of culture using the QIAGEN RNeasy kit. For quantitative PCR, total RNA was reverse transcribed using oligo-dT. Expression of genes was determined by real-time PCR using the LightCycler 480 Probe Master Mix and System (Roche). All primers (Table E3 in the Online Repository) were from Integrated DNA Technologies. All reactions were standardized to GAPDH.

RESULTS

IL-21 activates STAT1, STAT3 and STAT5 in human CD8⁺ T cells

IL-21 activates numerous intracellular signaling pathways including STAT1, STAT3, STAT5, MAPK and Akt^{6, 15, 26}. We assessed which pathways were activated by IL-21 in human naïve CD8⁺ T cells. IL-21 induced strong phosphorylation of STAT3, and a low level of STAT1 and STAT5 phosphorylation (Fig 1). We also analysed STAT activation induced by two other γc cytokines that are potent inducers of CD8⁺ T cell proliferation and differentiation, namely IL-2 and IL-15. In contrast to IL-21, IL-2 and IL-15 did not result in phosphorylation of STAT1 or STAT3, but did induce STAT5 phosphorylation (Fig 1). The combination of IL-15 and IL-21 did not alter the level of STAT phosphorylation above that observed with these cytokines alone (Fig 1A-F). Therefore, of these cytokines, IL-21 uniquely activates STAT1 and STAT3 in human CD8⁺ T cells.

Figure 1. IL-21 predominately activates STAT1, STAT3 and STAT5 in human CD8⁺T cells.

(A-C) Naïve CD8⁺T cells were activated for 4 days with TAE beads and recultured with cytokines for 30 min to determine phosphorylation of STAT1, 3 and 5. Histograms show nil or cytokine cultures and are representative of 4 experiments. (D-F) Graphs represent fold increase in MFI (mean ± SEM; n=4) of cells stimulated with cytokine over nil cultures. The dashed lines indicate a fold-change of 1 i.e. no change.

STAT1 and STAT3 mutations do not impair proliferation of naïve CD8⁺ T cells

IL-21 plays a pivotal role in inducing proliferation of CD8⁺ T cells^{10-12, 38}. However, since IL-21 activates multiple signaling pathways, it is not clear which of these underlies this proliferative effect. To address this we utilized naïve CD8⁺ T cells from patients with mutations in STAT1 (n=8), STAT3 (n=15) or IL21R (n=3).

Homeostatic cytokines support survival of CD8⁺ T cells and can induce proliferation and differentiation in the absence of extrinsic TCR stimulation. Thus, IL-15 mediates homeostatic proliferation of memory cells and combined with IL-21 drives naïve cells to effector phenotypes^{9, 12, 38, 39}. In vitro culture of naïve CD8⁺ T cells with IL-2 or IL-15 for 10 days significantly increased the recovery of viable cells (Fig 2A). In contrast, IL-21 alone did not increase survival above that seen with media alone. However, co-culture with IL-15 plus IL-21 induced significant proliferation as assessed by CTV dilution (Fig 2B). STAT3^MUT or STAT1^MUT CD8⁺ T cells stimulated with IL-15 and IL-21 showed comparable proliferation to controls (Fig 2B-D), however proliferation of IL-21R^MUT CD8⁺ T cells was strongly reduced, with the residual proliferation likely being induced by IL-15 (Fig 2C, 2D). Thus, IL-21’s involvement in the homeostatic turnover requires a functional IL-21R but is unaffected by loss-of function mutations in STAT3 or STAT1.

Figure 2. Cytokine-induced proliferation is impaired in IL-21R-deficient, but not STAT1- or STAT3-deficient, naïve CD8⁺ T cells.

Naïve CD8⁺ T cells were cultures with cytokines only and (A) the numbers of live cells were determined (mean ± SEM, n=5). (B, C) Histograms show representative CTV profiles of cells stimulated with IL-21/IL-15. Graph in (B) shows the number of divided cells in IL-21/IL-15 cultures (mean ± SEM; n=5). (D) Each bar represents an individual patient or normal donor for experiments using cells from STAT1- and IL-21R-deficient patients. * P<0.05; **, P<0.001; *** P<0.001.

STAT3 is required for IL-21 induced expression of granzyme B

The capacity of CD8⁺ T cells to produce the cytotoxic molecules granzyme B and perforin is important for their effector function⁴⁰. IL-2, IL-15 and IL-21 can all induce expression of these molecules^{7-9, 16, 41-43}. Therefore, we assessed the impact of STAT1, STAT3 and IL21R mutations on the ability of these cytokines to induce granzyme B. Co-culture with IL-21 plus IL-15 induced higher granzyme B expression in normal naïve CD8⁺ T cells than did IL-2, IL-15 or IL-21 alone (Fig 3A). However, both IL21R^MUT and STAT3^MUT CD8⁺ T cells cultured with IL-15 and IL-21 failed to upregulate granzyme B (p<0.001) to the same level as control cells (Fig 3A-C). In contrast, STAT1^MUT CD8⁺ T cells showed normal upregulation of granzyme B after stimulation with IL-15/IL-21 (Fig 3B, 3C). Thus, acquisition of the lytic machinery by naïve CD8⁺ T cells stimulated with IL-21 in combination with IL-15 was dependent on STAT3 signaling downstream of a functional IL21R.

Figure 3. Cytokine-induced expression of granzyme B is impaired in IL-21R-deficient or STAT3-deficient naïve CD8⁺ T cells.

Naïve CD8⁺ T cells were cultured with cyokines only and (A and B) graphs depict fold increase (mean ± SEM; n=5) in MFI of granzyme B over normal cells cultured with media alone (A). Each bar graph in (B) represents mean and range of duplicate cultures from an individual patient or normal donor for experiments using cells from STAT1- and IL-21R deficient patients. (A) and (C) are representative plots of cells from normal donors or the indicated patients stimulated with IL-21/IL-15 and normal donor cells from unstimulated cultures. ** P<0.001; *** P<0.001.

TCR/costimulation rescues defective IL-21 responses in STAT3^MUT CD8⁺ T cells

During an immune response CD8⁺ T cells also receive signals via the TCR and costimulatory molecules. Therefore, we examined naïve CD8⁺ T cell responses following culture with cytokines and anti-CD3/anti-CD28/anti-CD2 stimulus (provided by TAE beads). Addition of IL-15 or IL-21/IL-15 to cells from healthy donors resulted in the recovery of a significantly more CD8⁺ T cells than stimulation with TAE beads alone, or TAE beads plus IL-2 (Fig 4A). Proliferation analysis revealed that treatment of naïve CD8⁺ T cells with TAE beads plus IL-21 increased the total percentage of divided cells and the average number of divisions the cells had undergone compared to those cultured with TAE beads alone (Fig 4B,C). Interestingly, we detected no significant defect in the ability of TAE-activated naive STAT3^MUT (Fig 4A-C) or STAT1^MUT (Fig 4D) CD8⁺ T cells to respond to IL-21 or IL-21/IL-15 co-stimulation. However, mutations in IL-21R decreased recovery of viable cells and progression through division in cultures containing IL-21 (Fig 4E). IL-7, another γc cytokine, also enhanced proliferation of normal naïve CD8⁺ T cells that had been stimulated with TAE beads (not shown). Although previous studies have found that IL-7 can activate STAT3^{44, 45}, the ability of IL-7 to promote naïve CD8⁺ T cell proliferation was unaffected by mutations in STAT3 (not shown).

Figure 4. STAT3-deficient naïve CD8⁺ T cells proliferate normally in response to TCR engagement and activating cytokines.

Naive CD8⁺ T cells from healthy donors (n=6-8 [A-C], 3 [D], 5 [E]), STAT3-deficient (A-C; n=6-8), STAT1-deficient (D; n=3); or IL-21R-deficient (E; n=2) patients, were cultured with TAE beads alone or together with cytokines. Total numbers of live cells that had entered division (A, D and E) from each culture were determined (mean ± SEM). (B) Histograms show representative CTV. Numbers give % of divided cells (mean ± SEM). (C-E) Percentage of CD8⁺ T cells in each division was determined. * P<0.05; ** P<0.001; *** P<0.001

In cultures receiving cytokines alone, STAT3 mutations impaired the ability of IL-21 to upregulate granzyme B (Fig 3). We therefore determined whether signals provided by TCR/costimulation modulated this impairment. Addition of IL-2, IL-15, IL-21 or IL-21/IL-15 to TAE-stimulated cultures strongly (i.e. >20-fold) up-regulated granzyme B expression in normal naïve CD8⁺ T cells (Fig 5A,D). In contrast to these cytokines, the effect of IL-7 on granzyme B induction was modest (ie <10% granzyme B⁺ cells; ref¹⁶ and not shown).Mutations in STAT3 (Figure 5A,D) or STAT1 (Figure 5B,D) did not impair the ability of TAE-stimulated T cells to upregulate granzyme B in response to any of the cytokines tested. However, IL-21R^MUT naïve CD8⁺ T cells were unable to upregulate granzyme B following IL-21 stimulation and this was partially recovered by IL-15 (Fig 5C,D). These results demonstrate that stimulation through TCR/costimulation alters the activation of the naïve CD8⁺ T cells such that STAT3 mutations no longer prevent IL-21-induced expression of the cytotoxic mediator granzyme B.

Figure 5. IL-21 induced granzyme B production is intact in TCR-stimulated STAT3-deficient CD8⁺ T cells.

Naive CD8⁺ T cells from (A) STAT3-deficient (n=6-8), (B) STAT1-deficient (n=3), (C) IL-21R-deficient (n=2) patients or healthy donors (n=6-8), were cultured with TAE beads alone or together with cytokines. Representative histograms of granzyme B expression (filled – normal, coloured lines – patients) are depicted in (A-C). Graphs in (D) show fold-increase (mean ± SEM) in MFI of granzyme B expression by cytokine-stimulated normal and patient cells over those cultured with TAE beads alone.

Mutations in STAT3 and IL21R alter the frequencies of memory CD8⁺ T cells

Since IL-21 signals through IL-21R to activate STAT1 and STAT3 and regulates effector function we speculated that impaired IL-21R, STAT1 or STAT3 function may also affect CD8⁺ T cell differentiation in vivo. Therefore, we examined PB CD8⁺ T cell populations and phenotypes in patients with mutations in these molecules. We found that CD8⁺ T cells represented 21.0 ± 1.1% of PB lymphocytes in healthy donors. This did not differ for STAT3^MUT (23.5 ± 1.5%), STAT1^MUT (31.9 ± 3.8%) or IL-21R^MUT patients (17.4 ± 2.6%; Fig 6A).

Figure 6. Mutations in STAT3 but not STAT1 impair generation of effector/memory CD8⁺ T cells in vivo.

PB from healthy donors (n=46 or 51), STAT3-deficient (n=13), STAT1-deficient (n=5 or 6) or IL-21R-deficient (n=3) patients were assessed for the percentage of (A) total CD8⁺ T cells or (B) naïve, T_CM, T_EM and TEMRA cells. Each symbol corresponds to an individual donor or patient and lines represent means. Histograms and dot plots are from one representative donor or patient. ** P<0.001; *** P<0.001

In humans the CD8⁺ T cell population can be divided into subsets based on differential expression of CD45RA and CCR7⁴⁶. The CD8⁺ T cell compartment of normal donors thus comprises naïve (35.5 ± 2.3%), T_CM (8.9 ± 7.0%), T_EM (33.5 ± 1.8%) and T_EMRA (22.1 ± 2.1%) cells (Fig 6B). The distribution of these subsets in STAT1-deficient patients did not differ from normal controls (Fig 6B). However, the frequency of naïve CD8⁺ T cells in STAT3-deficient patients was significantly increased (62.6 ± 5.7%, p< 0.001) compared to normal controls. This was associated with substantial decreases in T_EM, T_EMRA and T_CM cells in STAT3-deficient patients (19.9%± 3.8%; 14.0 ± 2.7% and 3.6 ± 0.8%, respectively; Fig 6B). Furthermore, analysis of CD8⁺ T cell subsets from the three IL-21R-deficient patients suggested that memory in these patients may also be dysregulated, with increased naïve (57.5 ± 6.3%) and reduced T_EMRA (8.2 ± 5.0%) (Fig 6B). It was recently reported that populations of memory and effector CD8+ T cells reach adult levels by ∼5-10 years of age^{47, 48}. This is consistent with our finding that the proportions of memory and effector CD8⁺ T cells in the cohort of STAT1-deficient patients, the average age of which was 13 years (Table E1), were normal (Fig 6B). Thus, it is unlikely that the decreased frequencies of non-naïve CD8⁺ T cells in STAT3 and IL-21R-deficient individuals reflects the inclusion of some younger patients in these cohorts. These results suggest that STAT3, but not STAT1, plays an important role in generating and/or maintaining memory CD8⁺ T cells. Analysis of further IL-21R-deficient patients will be required to determine whether or not the required STAT3 activation is occurring downstream of IL-21 signaling or whether other cytokines, such as IL-10⁴⁹, are also involved.

To further understand the decrease in memory CD8⁺ T cell in STAT3-deficient patients we analyzed expression of genes that control their differentiation and survival (Fig E1). BCL2 was significantly higher in STAT3^MUT naïve compared to normal naïve CD8⁺ T cells (Fig E1). This suggests Bcl-2 may play a role in the survival, and thus increased frequency, of naïve CD8⁺ T cells in STAT3-deficient patients. However, we observed no other ifferences in expression of pro- or anti-apoptotic molecules between patients and controls (Fig E1). Similarly, although transcription factors responsible for CD8⁺ T cell function and differentiation were differentially expressed across CD8⁺ T cell populations, we observed no significant differences between normal and STAT3^MUT CD8⁺ T cells (Fig E1).

STAT3-deficient T_EM and T_EMRA CD8⁺ T cells have a phenotype suggestive of sustained activation

CD8⁺ T cell subsets were further assessed for expression of a range of molecules that change during differentiation from naïve to effector cells^{50, 51}. STAT3^MUT cells showed dysregulated expression of many of these molecules (Fig 7). For example, 2B4 was highest on normal CD8⁺ T_EM and T_EMRA cell, but was increased 2-4 fold on STAT3^MUT T_EM cells (p<0.001). In contrast, 2B4 was expressed at normal levels on STAT1^MUT CD8⁺ T cells (Fig 7, E2A, E3A). CD57 expression is associated with poorly proliferative terminally differentiated T cells^{51, 52}. Consistent with this, the greatest frequency of CD57⁺ cells was found in the TEMRA population. However, proportions of CD57⁺ cells in T_EM and T_EMRA populations of STAT3^MUT CD8⁺ T cells were significantly increased relative to normal and STAT1^MUT cells (Fig 7). CD127 (IL-7Rα) is highly expressed on naïve and T_CM cells, and downregulated on T_EM and T_EMRA cells. However, in STAT3^MUT patients all memory populations displayed significantly decreased levels of CD127 compared to healthy donors (Fig 7). Not all activation molecules displayed altered expression however as CD95 was not altered on CD8⁺ T cells from either the STAT1^MUT or STAT3^MUT patients (Fig 7, E2A, E3A).

Figure 7. CD8⁺ T cells from STAT3-deficient display a more activated phenotype.

Subsets of naïve, T_CM, T_EM and T_EMRA CD8⁺ T cells in PB of healthy donors (n=15), STAT3-deficient (n=7), STAT1-deficient (n=3) and IL21R-deficient (n=2) patients were assessed for expression of 2B4, CD57, CD95, CD127, CX3CR1, CD11a, CD11b, granzyme B or perforin. Graphs represent fold-change in MFI of the molecules relative to naïve cells, or the % of positive cells. **, P<0.001; ***, P<0.001

Chemokine receptors and adhesion molecules are important for regulating migration to secondary lymphoid organs and inflamed tissues. CX3CR1, CD11a and CD11b are highest on T_EMRA and T_EM cells from normal donors. However, their expression on STAT3^MUT, but not STAT1^MUT, T_EM and T_EMRA cells was significantly (2-3-fold) higher than on normal cells (Fig 7, E2B, E3B). Interestingly, STAT3^MUT T_EM and T_EMRA CD8⁺ T cells had significantly higher expression of granzyme B but not perforin. The altered expression of these molecules by STAT3^MUT CD8⁺ T cells suggests they have undergone aberrant differentiation in vivo and the residual effector memory populations exhibit a more senescent/exhausted phenotype.

DISCUSSION

To establish protective anti-viral and anti-tumor immunity, naïve CD8⁺ T cells must proliferate and acquire effector function, processes believed to be regulated by IL-21. Indeed, IL-21 has been found to increase in vitro proliferation and survival of mouse and human CD8⁺ T cells^{6-12, 38}, as well as induce effector molecules - IFN-γ, granzyme B, perforin - and enhance their overall cytotoxicity^{7-10, 16, 42, 43}. Furthermore, in vivo delivery of IL-21 to cancer patients upregulated granzyme B and perforin in CD8⁺ T cells¹⁷. Consistent with previous studies in multiple cell types^{6, 15, 17, 26}, IL-21 induced phosphorylation of STAT1, STAT3 and STAT5 in human CD8⁺ T cells. Yet, the contribution of individual STATs and cytokines to the development and effector function of human CD8⁺ T cells has not previously been defined. To clarify the requirements for these molecules in human CD8⁺ T cell function, we analyzed CD8⁺ T cell differentiation in vivo, and IL-21 signaling in vitro, in individuals with loss-of function mutations in IL21R, STAT1 or STAT3.

The ability of IL-21 to promote proliferation of naïve CD8⁺ T cells was unaffected by mutations in STAT1 or STAT3. In contrast, impaired STAT3 signaling abolished upregulation of granzyme B in response to IL-15/IL-21 where no exogenous TCR stimulus was provided. This paralleled our observations from cells unable to signal through the IL-21R, thereby suggesting that intact signaling through the IL-21R/STAT3 axis is required for GZMB transcription and subsequent cytotoxicity in human CD8⁺ T cells stimulated with IL-15/IL-21. These findings are physiologically relevant because they infer that the increase in expression in granzyme B observed in CD8+ T cells of cancer patients who were administered IL-21 in the absence of specific T-cell activation¹⁷ was STAT3-dependent. This is supported by work showing that IL-21-induced phosphoSTAT3 binds upstream of Gzmb in murine CD4⁺ T cells⁵³. The inability of STAT3-deficient CD8⁺ T cells to upregulate granzyme B in response to IL-15/IL-21 would be further compounded by impaired production of IL-21 by STAT3-deficient CD4⁺ T cells⁵⁴. Surprisingly, we observed that when naïve STAT3^MUT CD8⁺ T cells were provided with extrinsic TCR stimulus and IL-21 they were capable of upregulating granzyme B to similar levels as controls. IL-21-induced granzyme B upregulation was also intact in STAT1-deficient naïve CD8⁺ T cells suggesting that TCR/costimulation did not result in a switch from a STAT3 to a STAT1 pathway downstream of IL-21R for granzyme B regulation. Rather TCR/costimulatory signaling is likely to alter the sensitivity of cells to STAT3 such that residual STAT3 activity in STAT3^MUT CD8⁺ T cells is sufficient to induce a cytotoxic response.

Studies in mice have suggested that IL-21 is important for controlling some viral infections, such as vaccina¹⁸ and chronic LCMV^19-21. AD-HIES patients generally do not exhibit heightened susceptibility to primary viral infection, but some (<20%) do have an impaired ability to control reactivation of herpes viruses (HSV, EBV, VZV)^{29, 30}. On the other hand, the few IL-21R-deficient individuals documented to date appear capable of mounting protective responses to these viruses, yet experience ongoing infection with norovirus and rhinovirus³⁷. Both of these PIDs though are associated with increased susceptibility to infection with bacterial and fungal pathogens (Staphylococcus aureus, Candida albicans, Pneumocystis, cryptosporidia)^{28, 29, 37, 55, 56}. Our findings actually provide an explanation for the relatively mild susceptibility of STAT3^MUT and IL-21R^MUT individuals to primary viral infection, despite predictions from mouse models^{18-21, 49}: first, delivery of strong TCR and costimulatory signals during viral infection would facilitate normal induction of granzyme B in IL-21-stimulated STAT3^MUT cells. Second, granzyme B induction by IL-2 and IL-15 is intact in STAT3^MUT and IL-21R^MUT CD8⁺ T cells. Thus, during most viral infections, combined signals by TCR/costimulation, IL-2, IL-15 and/or IL-21 in naïve CD8⁺ T cells would be sufficient to generate a protective cytotoxic thereby circumventing the dependency on signaling through IL-21R or STAT3.

An interesting feature of AD-HIES is predisposition to B-lymphoma^{29, 31, 32, 57}. It is possible that during an anti-tumor response immune activation is not as strong as during viral infection; thus the relative contribution of IL-21R/STAT3 signaling in regulating granzyme B may be greater. Consequently, STAT3 mutations may contribute to impaired CD8⁺ T cell immune surveillance against B-cell malignancies. This is reminiscent of the susceptibility of perforin-deficient individuals to hematological neoplasms, including B-lymphoma⁵⁸.

The reported role of IL-21 in memory cell development^{12, 18, 49} prompted us to investigate the phenotype of CD8⁺ T cells from STAT1- and STAT3-deficient patients. Recent studies reported reduced frequencies of T_CM cells in AD-HIES patients³⁰ thereby implicating STAT3 in the development of „central memory’ CD8⁺ T cells^{30, 49}. We found significant decreases not only in T_CM but also in T_EM cells in STAT3-deficient, but not STAT1-deficient, individuals. These changes were not due to alterations in the total percentages of CD8⁺ T cells or age-related variations in the number of naïve versus memory cells (Fig E4A-D). T_EM and TEMRA populations from STAT3-deficient patients also displayed a phenotype of exaggerated differentiation often associated with increased/sustained exposure to antigen^{50, 51, 59}. Assessment of IL-21R-deficient patients suggested IL-21 may contribute to establishing some of the memory populations in STAT3-deficient individuals. However, to determine whether IL-21 is the STAT3-activating cytokine required for maintaining CD8⁺ T cell memory analysis of additional IL-21R-deficient patients will be required.

Several explanations can be proposed for the memory CD8⁺ T cell deficiency in AD-HIES. First, STAT3 mutations may affect CD8⁺ T cell homeostasis or differentiation. However, proliferation of STAT3^MUT naïve CD8⁺ T cells in response to IL-7 or IL-21/IL-15, which regulate homeostasis of CD8⁺ T cells^{2, 9, 38, 42, 60-64}, was normal. Thus, there is no evidence that the homeostatic proliferation and survival of naïve T cells induced by these cytokines requires STAT3, implying that the memory cell deficit is unlikely to be caused by impaired proliferation. Second, impaired STAT3 function may alter CD8⁺ memory T cell numbers through effects on differentiation. CD8⁺ T cell differentiation is regulated by multiple transcription factors that control opposing fates: Eomes and Bcl-6 favor “central memory” CD8⁺ T cell development, whereas T-bet and Blimp-1 promote differentiation to “effector” CD8⁺ T cells^{13, 65-69}. The observations that Socs3, Tbx21, Bcl6 and Prdm1 are direct targets of STAT3⁵³, together with reduced expression of these genes in Stat3^-/- murine CD8⁺ T cells post-infection with LCMV⁴⁹, and of SOCS3 and BCL6 in ex vivo isolated STAT3-deficient human naïve CD8⁺ T cells³⁰, suggest that STAT3 regulates CD8⁺ T cell differentiation by controlling expression of transcription factors. Consistent with this we observed decreased SOCS3 expression in IL-21-stimulated STAT3-deficient naïve CD8⁺ T cells in vitro. Thus, as proposed previously⁴⁹, decreased SOCS3 expression may contribute to aberrant differentiation resulting in the enhanced activated phenotype we observed in T_EM and T_EMRA cells in STAT3^MUT patients. However, we saw no significant differences in the levels of other transcription factors between normal and STAT3-deficient CD8⁺ T cells either directly ex vivo (Fig E1) or following IL-21 stimulation (Fig E5). This suggests that the IL-21/STAT3 axis proposed to drive Bcl-6 expression and CD8⁺ T_CM cell generation may be an oversimplification. Instead, STAT3 appears to have broader effects, being required for the generation or maintenance of not only T_CM cells, but also T_EM and T_EMRA populations.

Lastly, reduced memory CD8⁺ T cell frequencies in STAT3^MUT individuals may reflect reduced signaling through IL-7R (CD127). IL-7 is another homeostatic cytokine important for memory T cell maintenance^{39, 61, 63, 64}. Interestingly, CD127 was significantly decreased on all memory populations from STAT3-deficient patients compared to normal donors consistent with a more activated phenotype. Thus, reduced expression of CD127 on STAT3-deficient CD8⁺ T cells may limit their responsiveness to IL-7 signals, thereby compromising the pro-survival effects of IL-7 and compounding any decrease in memory cell numbers caused by reduced memory cell differentiation.

These findings reveal STAT3, but not STAT1, as an important downstream component of IL-21 signaling that mediates induction of effector function in CD8⁺ T cells. However, a high level of redundancy for the induction of cytotoxic function seems to exist suggesting that in most circumstances a functional level of killing would still be generated. This is consistent with mild susceptibility to viral infections in either AD-HIES or IL-21R-deficiency. In contrast, STAT3 signals, possibly initiated by IL-21, are critical for regulating the pool of all memory CD8⁺ T cell subsets in humans. Collectively, these insights significantly add to our understanding of the function of IL-21 and STAT3 in human CD8⁺ T cell development and behavior, and disease pathogenesis in individuals with mutations in STAT3 and IL21R.

Key Messages.

• loss-of function mutations in STAT3, causing autosomal dominant hyper-IgE syndrome, and IL21R, compromise differentiation of human CD8+ T cells to memory and effector cells

• mutations in STAT3, but not STAT1, abrogate the ability of human CD8⁺ T cells to differentiate into granzyme B-expressing effector cells in response to IL-21

Abbreviations

AD-HIES

autosomal dominant hyper-IgE syndrome

AR

autosomal recessive

IL-21R

IL-21 receptor

BCL-2

B cell lymphoma 2

EOMES

Eomesodermin

LCMV

Lymphocytic choriomeningitis virus

PID

primary immune deficiency

TCR

T cell receptor

T_CM

central memory T cells

T_EM

effector memory T cells

T_EMRA

effector memory T cells expressing CD45RA

TAE

T cell activation and expansion beads

EBV

Epstein Barr Virus

CMV

cytomegalovirus

VZV

varicella zoster virus

CTV

Cell Trace Violet