Interleukin-4:STAT6 signaling delays protective CD8 T cell bystander activation by antagonizing IL-18 sensing

Abstract

Memory CD8 T cells (Tmem) are activated into innate-like killers by cytokines including IL-12, IL-15 and IL-18; but mechanisms regulating this phenomenon (termed bystander activation) are unclear. We show basal IL-4 signals antagonize IL-18 sensing and subsequent IFNγ production during Tmem bystander activation. IL-4 treatment can act directly on Tmem in a STAT6-dependent manner, to limit IFNγ–mediated control of a bystander bacterial infection. IL-4 does not simply block bystander activation but tunes effector molecule expression. Strain-specific defects in bystander activation of homeostatic Tmem partially relates to IL-4 exposure, but these differences are erased in Tmem produced by TCR activation leading to uniform IL-18 receptor expression and capacity for bystander activation/cytotoxicity. Our data demonstrate that bystander activation by inflammatory cytokines is subject to regulation by both IL-4 and by prior antigen experience. These findings underscore the importance of the cytokine milieu in dictating bystander-mediated pathogen control.

Introduction:

CD8 memory T cells (Tmem) provide rapid protection when their cognate antigen is reencountered; however, CD8 Tmem are also capable of innate-like, antigen-independent functions when activated by pro-inflammatory cytokines^1–4. The cellular consequences of this phenomenon, called bystander activation, include proliferation⁵ and cytotoxicity^6–10. The cytokines IL-12, IL-15, and/or IL-18 potently elicit bystander responses^1,11. Once activated, bystander CD8 Tmem indirectly kill by secreting interferon-γ (IFNγ)^6,9,12 and are reported to directly kill by NKG2D-dependent recognition of target cells expressing NKG2D ligands (a family of stress induced proteins)^7,8,13–15.

However, these effector functions of activated bystander Tmem are a double-edged sword^2–4: While bystander-mediated killing promotes pathogen clearance during acute infections^{7,12,13,16,17}, during sustained/dysregulated inflammation, these responses beget immunopathologies^8,9,14,15,18. Data thus far suggest all CD8 Tmem, including circulating Tmem^8,10,19,20, tissue-resident Tmem^12,16, and antigen-inexperienced “virtual memory” T cells (Tvm)^13,17,18, are capable of bystander activation and hence contribute to innate-like immune responses. Thus, there is significant interest in understanding what regulates the ability of CD8 Tmem to participate in bystander responses, and how this can be modulated therapeutically.

Conventional T cell responses, mediated through T cell receptor (TCR) engagement, are tightly regulated. Chronic TCR activation leads CD8 Tmem to express molecules like TOX and PD-1, which render the cell refractory to additional TCR stimulation in a cell-intrinsic manner^21–23. Although cytokine-mediated activation transiently increases both TOX and PD-1 expression in bystander CD8 Tmem, these cells remain functional, expressing high levels of IFNγ and cytotoxic granules like granzyme B (GzmB)²⁰. Despite indifference to conventional means of attenuation, bystander activation can be influenced by cell-extrinsic factors. Systematic analysis of cytokine combinations showed IL-4, IL-10, and/or TGF-β impair IFNγ expression during in vitro bystander activation with IL-12, IL-15, and IL-18¹¹. Indeed, IL-10 plays a role in regulating bystander activation, as IL-10–deficient animals with Lyme arthritis exhibit exacerbated activation of bystander CD8 Tmem⁹. Nevertheless, the biologic significance of signals like IL-4 during bystander activation in vivo, especially in the context of infection, remains unclear.

Here we report that mouse strains differ in their propensity for homeostatic CD8 Tmem bystander activation and that this relates, partially, to basal IL-4 signaling, which mutes IFNγ production during bystander activation by antagonizing IL-18 sensing. These strain-specific differences in bystander activation are overridden in conventional, antigen-primed CD8 Tmem, which have enforced expression of cytokine receptors necessary for bystander activation. Nevertheless, even in strains predisposed for bystander activation, exogenous IL-4 delays protective bystander responses like IFNγ secretion and attenuates control of a bacterial infection. IL-4 does not fully abrogate bystander CD8 Tmem activation but rather tunes responses by re-shaping expression of effector molecules and cytokine receptors, including downregulation of IL-18 receptor expression through a STAT6-dependent pathway. The relevance of IL-4-mediated tuning of bystander responses in the context of therapies and contemporaneous infection as well as the limitations of SPF models for understanding the full contribution of activated bystander CD8 Tmem in immune responses is discussed.

Results:

Mouse strains differentially respond to bystander cytokines

We first examined whether bystander activation of memory phenotype CD8 T cells (Tmem) varies across inbred mouse strains. To minimize T cell-extrinsic signals that could impact activation, CD8 T cells were isolated from secondary lymphoid organs of C57BL/6, BALB/c and their F1 cross (CB6F1) and stimulated with IL-12, IL-15, and IL-18 in combination (Fig. 1a). While Tmem (CD44^hi CD8 T cells) from all three strains respond to IL-12/15/18 by upregulating effector molecules, IFNγ and GzmB, and the activation marker, CD69, the magnitude of this response varied considerably, from highest in C57BL/6 to lowest in BALB/c, with cells from CB6F1 having intermediate expression of these markers (Fig 1b, c; Extended Data Fig. 1a). In contrast, TCR-induced CD69 upregulation was equivalent in all three strains arguing against generalized differences in this response (Extended Data Fig. 1a)

Figure 1. Bystander-activating cytokines differentially activate CD8 Tmem from SPF laboratory mice.

(a) Experimental layout for CD8 T cell stimulations and ex vivo receptor phenotyping, in which we isolated bulk CD8 T cells using MACS before stimulating with the bystander-activating cytokines, IL-12, IL-15, and IL-18 in combination (100ng/mL, ea). We added GolgiPlug 4h prior to collecting cells for flow analysis. (b) and (c) IFNγ and Granzyme B expression in memory phenotype (CD44^hi) CD8 T cells after stimulation with bystander-activating cytokines. (d) Basal expression of IL-18Ra in CD8 Tmem across SPF laboratory animals. Each point depicts a unique animal at a distinct timepoint (n=5 per strain) across 2 technical replicates, with bars indicating mean. Indicated statistical significance was calculated by Kruskal-Wallis tests with Dunn’s multiple comparisons tests.

To test a potential role for strain-associated microbiota, animals were co-housed for ≥3 weeks before conducting bystander stimulation assays (Extended Data Fig. 1b): strain-specific differences in CD8 Tmem responses remained intact, suggesting genetic determinants rather than the microbiome, regulates variation in bystander activation. Many Tmem in SPF animals are antigen-inexperienced “virtual memory” (Tvm), which can be identified by low CD49d expression²⁴: gating on this population showed the same strain-related hierarchy of responsiveness, indicating that these responses associate with “homeostatic” memory cells (Extended Data Fig. 1c).

Since cytokine sensing is requisite for bystander activation, we interrogated cytokine receptor expression on Tmem. IL-18Ra expression levels was highest in C57BL/6- and lowest in BALB/c-derived CD8 Tmem and Tvm, consistent with IL-18 sensitivity regulating IL-12/15/18 bystander activation (Fig. 1d, Extended Data Fig. 1d). Together, this suggested strain-specific differences in bystander activation may result from the ability to sense IL-18.

IL-4 contributes to BALB/c bystander activation deficiency

We hypothesized that impaired bystander activation in BALB/c Tmem may relate to IL-4, since BALB/c mice demonstrate elevated steady-state IL-4 expression²⁵, and exogenous IL-4 was shown to limit IFNγ expression in bystander activated BALB/c CD8 Tmem¹¹. Furthermore, IL-4R signals were reported to inhibit IL-18Ra expression during differentiation of CD4+ Th2 cells²⁶. Indeed, Tmem from Il4ra^−/− BALB/c animals exhibited elevated steady-state IL-18Ra expression (Fig. 2a) and enhanced IFNγ production following stimulation with IL-12/15/18 (Fig. 2b–c). As expected, Il4ra^−/− cells were insensitive to IL-4-mediated inhibition of bystander activation in vitro (Fig. 2c); but even in the absence of added IL-4, Il4ra^−/− BALB/c CD8 Tmem showed enhanced IL-12/15/18-induced IFNγ expression compared to littermate controls at the 24h timepoint (Fig. 2c). This advantage for Il4ra^−/− CD8 Tmem waned at later timepoints (Extended Data Fig. 1e), as did the impact of exogenous IL-4 on WT cells (Extended Data Fig. 2a–c), perhaps reflecting desensitization to IL-4 or because the inhibitory effect of IL-4 is eventually overcome by IL-12/15/18 signals. While this suggests a role for IL-4 signaling in restraining BALB/c bystander activation, other regulatory factors are also likely since CD8 Tmem from C57BL/6 mice still demonstrated greater IFNγ responses than Il4ra^−/− BALB/c (Fig. 2c).

Figure 2. Disrupting steady-state IL-4 signaling in BALB/c CD8 Tmem partially restores response to bystander-activating cytokines.

(a) Steady-state IL-18Ra expression in blood CD8 Tmem from C57BL/6, wild-type (Il4ra^+/+) BALB/c, or Il4ra^−/− BALB/c animals. (b) Experimental diagram outlining CD8 T cell isolation from Il4ra wild-type, heterozygous, and knockout BALB/c mice and subsequent stimulation with bystander-activating cytokines (IL-12/15/18, 100ng/mL, ea.) with or without IL-4 (100ng/mL). (c) IFNγ expression in CD8 Tmem from BALB/c Il4ra variants after 24h of stimulation with IL-4/12/15/18 and IL-12/15/18. Each symbol in (a) depicts and individual animal (n=9 per strain), each symbol in (c) depicts cells from an individual animal in a unique stimulation condition (n=3 BALB/c Il4ra^−/−, 5 BALB/c Il4ra^+/−, 7 BALB/c Il4ra^−/−, 8 C57BL/6 Il4ra^+/+ mice) across 2 technical replicates, with bars indicating mean. Indicated statistical significance was calculated by ordinary one-way ANOVA and (a) Tukey’s multiple comparisons tests between all samples or (c) Holm-Šídák (two-sided) multiple comparisons tests between all BALB/c samples.

Priming overwrites strain-specific bystander activation

While IL-4 regulated bystander activation of memory-phenotype CD8 T cells, it was not clear whether this would hold for antigen-driven memory cells. To address this, C57BL/6, CB6F1, and BALB/c animals underwent acute infection with LCMV and allowed to reach memory stage (Fig. 3a). LCMV-immune animals still demonstrated strain-specific differences in bystander activation when interrogating bulk, tetramer-negative CD8 Tmem (Fig. 3b), but LCMV-specific CD8 Tmem (identified using relevant peptide/MHC tetramers) uniformly expressed IFNγ in response to in vitro IL-12/15/18, regardless of strain (Fig. 3c,d).

Figure 3. CD8 T cell priming by viral infection overwrites strain-specific deficiencies in IL-18Ra expression and bystander activation.

(a) Experiment overview in which we infected C57BL/6, CB6F1, and BALB/c animals with 2 x 10⁵ PFU LCMV Armstrong to develop LCMV-specific CD8 Tmem. At the memory timepoint, we harvested CD8 Tmem for stimulations with bystander-activating cytokines (IL-12/15/18, 100ng/mL, ea.) and ex vivo phenotyping. (b) IFNγ and granzyme B expression in bulk CD8 Tmem during in vitro bystander activation. (c) Representative gating depicting identification of LCMV-specific CD8 Tmem with tetramers (e.g., K^d GP33–43, K^d NP396–404, L^d NP118–126). (d) IFNγ and granzyme B expression in LCMV-specific CD8 Tmem during in vitro bystander activation. IL-18Ra and IL-4Ra expression in (e) bulk CD8 Tmem and (f) LCMV-specific CD8 Tmem at homeostasis. (g) IL-4-mediated fold change in IFNγ expression in LCMV-specific CD8 Tmem during stimulation with IL-12/15/18. Each point depicts a unique cell population (i.e., of unique or undefined TCR specificity) from a distinct animal/stimulation (n=5 animals total) across 2 technical replicates, with bars indicating mean. Indicated statistical significance was calculated by ordinary one-way ANOVA and Tukey’s multiple comparisons test.

Corresponding with this reactivity, while tetramer-negative CD8 Tmem in BALB/c mice were IL-18Ra low (Fig. 3e), LCMV tetramer-positive CD8 Tmem expressed high levels of IL-18Ra (Fig. 3f). These data suggest that Tmem expanded by reactivity to foreign antigen gain uniformly-high expression of cytokine receptors and lose the strain-specific differences in bystander activation. Though strain-specific differences in bystander activation are not found in antigen-experienced CD8 Tmem at steady-state, strain-intrinsic sensitivity to exogenous IL-4 remained intact: LCMV-specific CD8 Tmem in BALB/c animals expressed higher levels of IL-4Ra (Fig. 3f) and demonstrated a greater fold reduction in IFNγ expression when exposed to recombinant IL-4 in tandem with IL-12/15/18 (Fig. 3g). Nevertheless, IL-4 also reduced IFNγ in LCMV-specific C57BL/6 Tmem. Thus, we next turned to models involving antigen-primed Tmem in C57BL/6 mice to investigate how IL-4 regulates bystander activation.

IL-4 limits IL-18Ra expression and IFNγ induction

To generate a defined population of conventional CD8 Tmem, naïve OT-I CD8 T cells (which express a transgenic TCR specific for OVA/K^b) were transferred into congenically-distinct C57BL/6 recipients, that were then infected with VSV-OVA (Fig. 4a). At least 60 days post VSV-OVA infection, we assessed the impact of IL-4 on IL-12/15/18-mediated activation (Fig. 4a). IL-4 reduced IL-12/15/18-induced IFNγ expression, most prominently at 24h (Fig. 4b), across a range of cytokine doses (Fig. 4c), and largely mirrored the inhibitory effects of IL-4 on memory phenotype cells from SPF animals (Extended Data Fig. 2a–c). These data confirm that antigen-driven conventional CD8 T cells retain sensitivity to IL-4 for regulation of bystander responses.

Figure 4. IL-4 limits IL-18Ra expression and IFNγ expression elicited by bystander-activating cytokines.

(a) Experimental diagram outlining OT-I Tmem generation and subsequent CD8 T cell isolation and in vitro activation with bystander-activating cytokines (IL-12/15/18, 100ng/mL, ea.) in the presence or absence of IL-4 (100ng/mL). (b) IFNγ expression in OT-I Tmem after cytokine stimulation. (c) IFNγ expression in OT-I Tmem at titrated doses of IL-12/15/18 and IL-4. (d) Experimental diagram for cytokine stimulation and IL-18Ra phenotyping, in which we stimulated bulk CD8 Tmem from OT-I memory mice with single cytokines (100ng/mL) for 24h before flow interrogation. (e) Expression of IL-18Ra in OT-I Tmem and endogenous CD8 T_naive after 24h stimulation with single cytokines. Each point in (b) and (e) depict cells from a single animal at a distinct timepoint, which are connected by animal identity, with bar indicating mean (b n=8, e n=7 across 3 technical replicates). Points in (c) depict cells from a single condition and animal, with a line indicating mean (n=5 across 3 technical replicates). Indicated statistical significance in (b) and (e) was calculated by Friedman tests with Dunn’s multiple comparisons tests comparing (b) all data groups to one another or (e) all data groups against mock-treated cells. Indicated statistical significance in c reports significance from IL-4-untreated cells and was calculated by one-way ANOVA with Geisser-Greenhouse Correction and Dunnett’s multiple comparisons tests.

Steady-state IL-4 evidently regulated expression of IL-18Ra in CD8 Tmem from BALB/c mice, hence we determined the effect of exogenous IL-4 on C57BL/6 CD8 Tmem. OT-I Tmem were stimulated with IL-4, IL-12, IL-15, IL-18 or IL-7 (which, like IL-4 and IL-15, engages a common gamma chain-dependent cytokine receptor), reavealing that IL-4 was unique in its ability to downregulate IL-18Ra (Fig. 4d, Extended Data Fig. 2d–f). In contrast, IL-12 induced increased expression of IL-18Ra, as expected²⁷. These data argue that IL-4 can alter IL-18Ra expression to reshape cellular sensitivity to bystander-activating cytokines and the subsequent expression of factors like IFNγ.

IL-4 limits protective functions of bystander CD8 Tmem

To study the impact of IL-4 on bystander activation in vivo, we utilized the ability of bystander-activated CD8 Tmem to control Listeria monocytogenes (L.m.), via rapid production of IFNγ⁶. Natural killer (NK) cells, which can also contribute to innate control of L.m., were depleted using anti-NK1.1 antibody. Treatment with IL-4 complexes (IL-4c) immediately before infection with L.m. (Fig. 5a) led to an elevated pathogen burden (Fig. 5b). Extending these findings, depletion of both NK cells and CD8 Tmem (through treatment with anti-ASGM1: Extended Data Fig. 3a–c) led to elevated L.m. burdens with or without IL-4c treatment, suggesting that the inhibitory effects of IL-4 are mediated through CD8 Tmem (Extended Data Fig. 3d).

Figure 5. IL-4 complex therapies limit protective functions of bystander CD8 Tmem.

(a) Experiment outline for establishing the effect of IL-4c on L. monocytogenes burden during the innate-like immune response, in which we depleted NK cells from C57BL/6 mice with anti-NK1.1 and subsequently infected with 1,000 CFU wild-type L. monocytogenes. We then assayed spleen and liver for pathogen at 24, 72, and 120h. (b) L. monocytogenes burden in spleen and liver in NK-depleted and/or IL-4c-treated C57BL/6 animals. c Experiment outline to test the effect of IL-4c on in vivo bystander activation. We treated OT-I memory mice with IL-4c and infected with L. monocytogenes. We injected animals with fluorophore-conjugated anti-CD8b 3 min prior to euthanasia to identify OT-I Tmem in the splenic white pulp (WP) and then harvested tissue for flow analysis. (d–f) IFNγ and IL-18Ra expression in OT-I Tmem at the splenic WP during combined IL-4c treatment and L. monocytogenes infection. (g) Outline for the generation of bystander OT-I Tmem that are the only cell population capable of IL-4 sensing and IFNg production. (h) Experimental approach to test the bystander-mediated protection against L. monocytogenes when IL-4 sensing and IFNg expression are restricted to OT-I Tmem. i Early-timepoint L. monocytogenes burdens in animals during IL-4c signaling that is restricted to bystander OT-I Tmem. Each point in (b, e, and f) represent an individual animal (b n=6 per timepoint and condition; e and f n=4 uninfected ±IL-4c, 6 L.m.-infected ±IL-4c) across 2 technical replicates. Each point in i depicts a unique animal/tissue (24h timepoints n=13 no transfer, 11 OT-I transfer −IL-4c and +IL-4c; 120h timepoints n=15 no transfer, 24 OT-I transfer −IL-4c, 23 OT-I transfer +IL-4c) across 5 technical replicates. Bars indicate mean. Indicated statistical significance was calculated by (b, i) Kruskal Wallis and Dunn’s multiple comparisons test or (e, f) ordinary one-way ANOVA and Tukey’s multiple comparisons tests.

Since OT-I TCRs cannot recognize L.m. antigens⁷, we leveraged this model to determine how IL-4c treatment affected IFNγ production by bystander CD8 Tmem in vivo. OT-I memory mice received IL-4c (or vehicle control) before low-dose L.m. infection to limit inflammation at discrete foci of the splenic white pulp (WP), wherein bystander activation occurs¹⁹. Intravenous antibody labeling was used to discriminate OT-I Tmem within the WP (Fig. 5c)¹⁹. IL-4c treatment reduced IFNγ expression in bystander OT-I Tmem in the WP 24h after infection (Fig. 5d,e). These effects were less pronounced at 72h, mirroring our in vitro stimulation data. Similarly, IL-4c treatment in vivo reduced IL-18Ra expression on OT-I Tmem from both uninfected and infected animals, although IL-18Ra levels rebounded over time (Fig. 5f), coincident with restored IFNγ production at later timepoints in IL-4c treated mice (Fig. 5d,e). Endogenous CD8 Tmem and NK cells behaved similarly (Extended Data Fig. 3e–h). IL-12 is induced during listeriosis¹⁹ and promoted IL-18Ra expression in vitro (Fig. 4e), and might restore IL-18Ra expression in opposition to IL-4 signals. Consistently, blocking IL-12 hindered IL-18Ra recovery during IL-4c therapy (Extended Data Fig. 3i). These findings suggest IL-4 delays the initiation of bystander responses, which can ultimately be overcome by opposing signals. While those additional cues help restore IL-18 sensitivity, our data suggest the initial delay in bystander activation may compromise effective control of Listeria.

Bystander activation orchestrates the microbicidal functions of antigen-presenting cells (APCs) through IFNγ²⁸, prompting examination of how IL-4c treatment affects APC function. While IL-4c treatment alone did not substantially dysregulate representation of APC subsets (Extended Data Fig. 4a), IL-4c together with L.m. infection caused decreased frequencies of pro-inflammatory Ly-6C^hi monocytes (Extended Data Fig. 4b). Ly-6C^hi monocytes are critical for host defense during listeriosis²⁸, so we tested whether IL-4c treatment impairs their antimicrobial functions. iNOS, an IFNγ-induced effector molecule, was only expressed in Ly-6C^hi monocytes from infected animals (Extended Data Fig. 4c) and the proportion of iNOS-expressing cells were reduced by IL-4c treatment (Extended Data Fig. 4c). iNOS-positive Ly-6C^hi monocytes upregulated other IFNγ-inducible markers, like CD86, PD-L1 and I-A^b (Extended Data Fig. 4d), which directly correlated with IFNγ levels in bystander OT-I Tmem and hence were reduced by IL-4c treatment (Extended Data Fig. 4e). Together, these data suggest IL-4-induced inhibition of IFNγ expression in bystander CD8 Tmem impacts other immune cells and subsequent pathogen control.

Our data indicated that CD8 Tmem were necessary to mediate the inhibitory effects of IL-4 on Listeria control. To test whether CD8 Tmem sensitivity to IL-4 was sufficient for this response, we developed a model whereby bystander CD8 Tmem were the sole population capable of producing IFNγ and also the only cells sensitive to IL-4. WT OT-I T cells were transferred into Il4ra^−/− Ifng^−/− recipient mice, which were infected with VSV-OVA to develop OT-I Tmem (Fig. 5g). Animals were treated with IL-4c (or vehicle control) then infected with L.m. and then assayed for early pathogen load (Fig. 5h). While bacterial burdens did not dramatically differ at 24h, Il4ra^−/− Ifng^−/− mice receiving no adoptive transfer had high pathogen burdens at 120hpi (Fig. 5i). Bystander OT-I Tmem restricted L.m. at 120h, likely due to their ability to make IFNγ; however, pathogen control was significantly diminished by IL-4c treatment (Fig. 5i). These data argue that IL-4, acting directly on bystander CD8 Tmem can blunt their ability to mediate protective immunity against a bacterial infection, with deleterious consequences for the host.

IL-4 and IL-18 reprogram effector properties of CD8 Tmem

While IL-4 exposure impaired production of IFNγ during in vitro and in vivo bystander activation, expression of other effector molecules, such as GzmB, showed the opposite pattern (Fig 6a,b, Extended Data Fig. 5a). Likewise, there was a trend for increased GzmB expression on CD8 Tmem (OT-I and endogenous populations) 24h after IL-4c of L.m.-infected mice, and this was significant in NK cells from the same animals (Extended Data Fig. 5b,c). These data suggest IL-4 does not simply act as a brake on CD8 Tmem stimulation but rather may tune the response to bystander-activating cytokines.

Figure 6. IL-4 and IL-18 differentially regulate T cell differentiation and fates.

(a) Experimental diagram outlining CD8 T cell isolation from OT-I memory mice and stimulation with IL-4 (100ng/mL) and/or bystander-activating cytokines (IL-12/15/18, 100ng/mL, ea.). (b) Granzyme B expression in OT-I Tmem during cytokine stimulation. (c) IL-4-mediated fold change of granzyme B expression elicited in OT-I Tmem by bystander-activating cytokines. (d) Expression of CD69, IFNγ, and granzyme B in OT-I Tmem stimulated with permutations of cytokine combinations for 24h. (e) Experimental outline for in vitro CD8 T cell stimulation, OT-I Tmem isolation, and bulk RNA sequencing. (f) Principal component analysis of RNAseq data from in vitro-stimulated OT-I Tmem. (g) Volcano plots for DEGs (in which CPM ≥ 25 in 2/3 replicates) for genes downregulated (blue) or upregulated (red) by IL-4 in the context of mock or IL-12/15/18 stimulation. (h) Experimental outline for in vitro killing assays, wherein in vitro-activated OT-I Tmem were co-cultured with fluorescently-labeled RMA target cells. (i) Specific lysis of RAE-1ε-expressing (left) or SIINFEKL-pulsed (right) RMA cells by NKG2D- or TCR-mediated killing, respectively. Each point in (b, c, and d) represents an individual timepoint/stimulation connected by animal identity (b and c n=8 mice; d n=10, 9, and 7 mice for CD69, IFNγ, and GzmB analysis, respectively) across 3 technical replicates. (f and g) depict 3 technical replicates of n=2 mice, pooled (total n = 6 mice), except for IL-12/15 (pool A), which failed cDNA library prep. Points in I depict a unique stimulation condition from one animal connected by animal identity (n=7 mice) across 5 experiments. Bars in plots indicate mean. Indicated statistical significance was calculated by (b) Friedman tests and Dunn’s multiple comparisons tests and (i) One-way ANOVA with Geisser-Greenhouse correction and Tukey’s multiple comparisons tests.

To methodically assess this, we tested the effect of IL-4 on OT-I Tmem bystander activation in response to permutations of cytokine combinations. Although individual cytokines poorly induced IFNγ or GzmB, combinations of IL-12, IL-15, and/or IL-18 induced expression of these molecules (Fig. 6d). IL-12/15 potently upregulated GzmB; but this was impaired (while IFNγ expression was simultaneously enhanced) when IL-18 was added (Fig. 6d). IL-4 appeared to partially reverse these IL-18-mediated effects, incompletely restoring GzmB expression at the expense of IFNγ (Fig. 6d). These findings are consistent with our data that showed IL-4 impairs expression of IL-18Ra (Fig. 4g). Nevertheless, IL-4 also impaired the induction of GzmB in response to the combination of IL-12 and IL-15, suggesting effects that go beyond regulation of IL-18 sensing (Fig. 6d).

To extend these studies, RNAseq on OT-I Tmem was conducted on OT-I Tmem following culture in vitro for 24h in the presence or absence of IL-4, in the context of IL-12/15, IL-12/15/18, TCR engagement or no additional stimulation (Fig. 6e) (Supplementary table 1). PCA analysis on these samples revealed that IL-4 induced transcriptional changes in all conditions, although these were relatively modest in the context of TCR stimulation (Fig. 6f). Several transcriptional changes aligned with our flow cytometry results: In the context of IL-12/15/18 stimulation, IL-4 lowered Cd69 and Ifng, while increasing Gzmb (Fig. 6g). On the other hand, we unexpectedly found that culture with IL-4 alone elevated transcription of Gzmb and, paradoxically, Ifng (Fig. 6g), despite minimal evidence for protein expression of either factor under those conditions (Fig. 6d). Nevertheless, these findings argue against IL-4 simply acting to close the Ifng locus in CD8 Tmem, a conclusion that was supported by the finding that IL-4 did not impair Ifng transcription in cells undergoing TCR stimulation, although Prf1 (perforin) transcription was decreased by IL-4 under these conditions (Extended Data Fig. 5d). Analysis of protein expression confirmed that IL-4 inhibited Perforin but not GzmB or IFNγ expression during TCR stimulation (Extended Data Fig. 5e).

These data suggested that IL-4 both reduced and increased gene expression when used alone or in combination with IL-12/15 or IL-12/15/18, consistent with IL-4 tuning CD8 Tmem differentiation rather than merely inhibiting bystander activation altogether. We interrogated whether IL-4 significantly altered the transcription of the receptor components for IL-12 (Il12rb1, Il12rb2), IL-15 (Il2rb, Il2rg), and IL-18 (Il18r1, Il18rap). While IL-4 decreased IL-18Ra expression at the protein level (Fig. 4e), it significantly decreased the corresponding Il18r1 transcript levels, albeit not within our 2-fold cutoff (Fig. 6g). Il2rb was increased by IL-4, while other cytokine receptor components necessary for bystander activation via IL-12/15/18, were otherwise unchanged (Fig. 6g, Supplementary table 1). This suggested that IL-4 affects IL-18 sensing and bystander activation at the receptor level, and likely affects other cytokine signals via other mechanisms.

Under all or most stimulation conditions, IL-4 induced Tafa3, Dusp10, Rgs1, Cis, Xdh, Slc2a6, Pdcd4, D630039A03Rik and Il4ra, many of which are known targets of IL-4 stimulation (Fig. 6f, Extended Data Fig. 5d,f). Conversely, IL-4 decreased transcription of Pdcd1 (PD-1) in the context of IL-12/15/18 stimulation, and Pdcd1 and Klrk1 (NKG2D) in the context of mock stimulation (Fig. 5g), which were validated at the protein level (Extended Data Fig. 5g). Notably, induction of NKG2D with IL-12/15 stimulation dropped to baseline with IL-4 signals (Extended Data Fig. 5g). These data echo previous studies that suggested IL-4 can impair NKG2D expression by memory CD8⁺ T cells²⁹. Together these data illustrate that, in addition to regulating IL-18 sensing, there is likely a core transcriptional program of IL-4 signaling and probable regulation of IL-12 and/or IL-15 signaling (downstream of receptors) by IL-4.

Given the complexity in IL-4-control of Ifng, Gzmb, Prf1, and Klrk1 expression, the expression of other cytotoxic factors was measured with in vitro stimulations and killing assays (Fig. 6h). Though IL-12/15/18 induced perforins and granzymes B and C, the addition of IL-4 curbed this, including the perforin/granzyme co-expression needed for effective direct cytotoxicity (Extended Data Fig. 6a,b).

Cytolysis assays assessed both TCR- and NKG2D-dependent killing in vitro (Fig. 6h, Extended Data Fig. 6c). Though NKG2D-mediated killing was not observed in these assays, IL-12/15 stimulation activated OT-I Tmem into the most effective antigen-specific killers, and that IL-4 and IL-18 had parallel roles in restricting this effect (Fig. 6h,i, Extended Data Fig. 6c). These data suggest cytokines play a complex role in regulating indirect and direct cytotoxicity when T cells are innate-like or adaptive effectors.

STAT6 controls IL-18 sensing for bystander activation

Since IL-4 acted directly on CD8 Tmem to alter their bystander activation response, the relevant signaling pathway downstream of IL-4R was dissected. Il4ra^Y500F encodes a mutated IL-4Ra that cannot signal through the PI3-K pathway³⁰. However, CD8 Tmem from Il4ra^Y500F BALB/c animals remained sensitive to IL-4-mediated downregulation of IFNγ and upregulation of GzmB expression during IL-12/15/18 stimulation (Extended Data Fig. 7a). Previous studies on in vitro-polarized Th2 CD4⁺ T cells suggested that STAT6 was involved in restraint of Il18r1 gene expression^26,31,32, but this has not been investigated in CD8⁺ T cells or in memory T cells. Initial tests using the STAT6 inhibitor, AS1517499, reduced responses to IL-12/15/18 stimulation, suggesting off-target activity (Extended Data Fig. 7b). Therefore, CRISPR/Cas9 was used to ablate Stat6 (or Thy1, as a control) (Fig. 7a). Following expansion in vitro, cells were stimulated with IL-4 and ΔSTAT6 OT-I T cells lost phospho-STAT6 induction (Fig. 7b), confirming effective targeting. ΔThy1 and ΔSTAT6 OT-I T cells were co-transferred into hosts immediately after CRISPR editing and activated through VSV-OVA infection (Fig. 7c). Importantly, peripheral ΔThy1 and ΔSTAT6 OT-I Tmem at the memory timepoint maintained their respective phenotypes and did not differ in steady-state IL-18Ra expression (Fig. 7d). The L.m. infection model was used to determine how Stat6-deficiency affects the impact of IL-4c on IL-18Ra expression and bystander activation (Fig. 7e). Both OT-I populations equivalently activated in the absence of IL-4c treatment, excluding a steady-state role for STAT6 in this C57BL/6 model. However, in contrast with control ΓThy1 OT-I Tmem, ΔSTAT6 OT-I Tmem exhibited normal activation (CD69 induction) and IFNγ expression following IL-4c treatment (Fig. 7f, Extended Data Fig. 7c). Similarly, IL-18Ra expression was high on both ΔThy1 and ΔSTAT6 OT-I Tmem at steady-state, yet IL-4c treatment caused a drop in IL-18Ra expression only on ΔThy1 OT-I Tmem (Fig. 7g, Extended Data Fig. 7c). Given the complexity of cytokine signals in vivo, parallel in vitro stimulations of ΔThy1 and ΔSTAT6 OT-I Tmem was conducted, confirming that Stat6-deficiency eradicates the ability of IL-4 to regulate IL-18 sensitivity and IFNγ expression (Extended Data Fig. 7d).

Figure 7. IL-4:STAT6 signaling controls IL-18 sensing and IFNγ expression during bystander activation.

(a) Experimental outline for Stat6 and Thy1 control CRISPR and in vitro cell expansion. (b) Phenotype and STAT6 phosphorylation in expanded CRISPR-edited OT-I T cells after IL-4 treatment. (c) Experimental outline for co-transfer and in vivo expansion of CRISPR-edited OT-I T cells. (d) Phenotypes of blood OT-I Tmem (≥60 days post VSV-OVA). (e) Experimental outline for in vivo bystander activation during IL-4c treatment. (f) CD69, IFNγ and (g) IL-18Ra expression in ΔThy1 and ΔSTAT6 OT-I Tmem during IL-4c treatment and/or bystander-activating L. monocytogenes infection. (h) Experimental outline for Il18r1 and Thy1 CRISPR-edited cell co-transfer, in vivo expansion, and in vivo bystander activation. (i) ΔThy1 and ΔIL-18Ra OT-I Tmem identification and ex vivo IL-18Ra expression. (j) CD69 and IFNγ expression in ΔThy1 and ΔIL-18Ra OT-I Tmem during bystander-activating L. monocytogenes infection. Representative flow plots depict (b) 1 experiment, (d) 10 animals, concatenated, (f, g, i, j) 2–3 animals per condition, concatenated. Graphs in (d, f, g, i, and j) connect OT-I Tmem subsets by animal. Graphs depict (b) n = 3 independent CRISPR experiments, (d) n = 10 animals, (f and g) n = 6 uninfected and 10 infected animals per condition across 3 technical replicates, (i and j) n = 4 uninfected and 6 infected animals. Points in b depict mean±SD. Bars in d, f, g, i, and j depict mean. All indicated statistical significances were calculated by paired t-tests (two-sided).

While IL-4:STAT6 signaling impairs IL-18Ra expression, gene expression analysis indicated IL-4 leads to numerous gene expression changes. To test whether defective IL-18 sensing alone would phenocopy the poor bystander activation observed in IL-4-stimulated Tmem, CRISPR/Cas9 was used to target Il18r1 in OT-I T cells, again targeting Thy1 in a control population, prior to memory induction (Fig. 7h). ΔIL-18Ra cells formed memory and maintained their responsiveness to cognate antigen, suggesting these cells are otherwise normal (Fig. 7i, Extended Data Fig. 7d). However, in response to bystander L.m. infection, ΔIL-18Ra OT-I Tmem were unable to upregulate IFNγ in an innate-like manner, while induction of CD69 was unaffected (Fig. 7j). These data demonstrate the importance of IL-18 signaling in the induction of IFNγ during bystander activation, and underscore how IL-4 could dramatically alter the bystander response through changing IL-18R expression.

Type 2 infections perturb the bystander T cell responses

IL-4 signals tuned the response to bystander-activating cytokines by altering IL-18 sensing both in vitro and in vivo in a manner most relevant to cytokine-based immunotherapies^33,34. To test if helminth infection also alters bystander activation, OT-I memory mice were infected with Heligmosomoides polygyrus, which elicits strong IL-4 expression³⁵ and 6 days later, co-infected with L.m. (Extended Data Fig. 8a). H. polygyrus infection alone did not induce CD69 or IFNγ expression in bystander OT-I Tmem, while both factors were expressed following L.m. infection alone (Extended Data Fig. 9b). H. polygyrus-infected mice showed attenuated CD69 and IFNγ induction by bystander OT-I Tmem during listeriosis (Extended Data Fig. 8b). On the other hand, H. polygyrus infection alone increased GzmB expression in bystander OT-I Tmem, an effect which was further enhanced by simultaneous L.m. infection (Extended Data Fig. 8c). Since an IL-4:IL-18 axis tuned IFNγ expression during bystander activation in reductionist models, we asked whether type 2 inflammation from H. polygyrus infection similarly impaired IL-18 sensing in bystander T cells at sites distal from the gastrointestinal tract. Indeed, IL-18Ra expression was blunted in splenic bystander OT-I T cells that did not recover during the initial stages of L.m. co-infection (Extended Data Fig. 8d). While these findings echo the effects we observed for IL-4c treatment, other data suggested additional mechanisms are at play. In contrast to the requirement for STAT6 in the downregulation of IL-18Ra following IL-4c treatment (Fig. 7G), ΔSTAT6 OT-I showed robust loss of IL-18Ra expression following H. polygyrus infection (Extended Data Fig. 8e–j). Nevertheless, Stat6-deficiency led to a small but significant restoration of IFNγ production by OT-I Tmem in mice sequentially infected by H. polygyrus and L.m. that was not observed in mice that were treated with antihelminth drugs (Extended Data Fig. 8k). H. polygyrus co-infection did not blunt the ability of OT-I Tmem to control L.m. burden, in Il4ra^−/−/Ifng^−/− hosts (Extended Data Fig. 8l,m). These data suggest that IL-4 can indeed alter bystander T cell functionality during helminth infection, but that additional pathways induced by such infections also shape bystander CD8 Tmem function. For example, H. polygyrus is known to induce robust production of IL-15³⁶ and type 1 responses³⁷, which might counterbalance some effects of IL-4. Nevertheless, several of the same phenotypic and functional changes in CD8 Tmem that we observed for IL-4c treated mice (loss of IL-18Ra, impaired induction of IFNγ, elevated GzmB expression) were replicated by H. polygyrus infection.

Inhibition of bystander protection by IL-4 requires STAT6

To this point, our data suggested that IL-4, acting on Tmem in a STAT6-dependent pathway, limited bystander activation in reductionist settings. To determine if disrupting this pathway overcomes deficient pathogen clearance during IL-4c in intact hosts, ΔSTAT6 or ΔThy1 OT-I T cells were transferred into separate wild-type mice, and induced to memory. Recipient animals were then treated with IL-4c or PBS prior to L.m. challenge. As expected, IL-4c treatment exacerbated bacterial burdens at 120h in control animals with ΔThy1 bystander OT-I Tmem, but animals carrying ΔSTAT6 bystander OT-I Tmem demonstrated significantly lower pathogen counts in both the spleen and liver (Fig 8). These data suggest IL-4 insensitivity in a fraction of bystander Tmem is sufficient to maintain innate-like competency against infections when IL-4 levels are elevated.

Figure 8. STAT6-deficient bystander CD8 Tmem improve pathogen control during IL-4c therapies.

Outline of single transfer, expansion, and bystander activation/IL-4c treatment of ΔThy1 or ΔSTAT6 OT-I T cells and subsequent L. monocytogenes colony counts in spleen and liver. Each point depicts a single animal/tissue (n=10 per condition/transfer) across 2 technical replicates. Indicated statistical significance was calculated by unpaired t-test (two-sided).

Discussion

Bystander T cells contribute to both beneficial and deleterious outcomes when activated into effectors by inflammation; however, the mechanisms that regulate this phenomenon are not fully resolved. We sought to identify regulators of bystander activation using multiple laboratory strains of mice and interrogating CD8 Tmem responses to cytokines responsible for potent bystander activation (IL-12/15/18). We found that memory phenotype CD8 T cells, including Tvm, from BALB/c mice fail to bystander activate to the same degree as those from C57BL/6 animals in a manner that was partially determined by the basal IL-4 signaling characteristic of BALB/c mice²⁵. Furthermore, augmenting IL-4 levels in C57BL/6 animals (which normally potently bystander activate) impairs IL-18 sensing that is requisite for innate-like IFNg secretion and limits bystander-mediated control of Listeria.

We identify mechanisms underlying strain-intrinsic differences in CD8 Tmem bystander activation—but these only manifest in SPF animals, as LCMV-experienced CD8 Tmem uniformly activated in response to IL-12/15/18. This extends previous studies, which found that IL-4 sensitivity affects CD8 Tmem differentiation in BALB/c mice but did not investigate bystander stimulation³⁸. While SPF conditions are not observed in microbially-experienced human adults, even antigen-experienced memory cells retained strain-specific sensitivity to exogeneous IL-4. Importantly, the SPF T cell repertoire mirrors human cord blood^39,40. Since neonates lack the antigen experience necessary for adaptive Tmem responses, these innate-like bystander responses are all-the-more critical for pathogen control in early life⁴¹. Therefore, genetic regulators of bystander activation may be important to neonatal/postnatal immunity. Conversely, since antigen experience overrides these strain-intrinsic determinants of bystander activation it suggests studies using certain SPF models (like BALB/c mice) may dramatically underestimate the contributions of bystander-activated CD8 Tmem to immunity.

While Il4ra deficiency enhances bystander activation in SPF BALB/c mice, it does not fully restore the response to those in C57BL/6 mice. This suggests additional factors also control bystander activation. Other animal models may prove useful in this regard; for instance, recombinant-inbred “collaborative cross” animals display diverging responses to bystander-activating cytokines, albeit for unknown reasons¹⁶. These tools alongside normal microbial experience will be important to fully identify how bystander activation contributes to health and disease, as well as developing new approaches for manipulation of these cells.

IL-4 sensing impairs IFNγ expression elicited by IL-12/15/18 stimulation in BALB/c CD8 Tmem¹¹; we provide a mechanistic context for this, showing that IL-4:STAT6 signaling inhibits IL-18Ra expression. This mirrors that of IL-18Ra repression by IL-4 observed during Th2 polarization^26,31,32. We build on those studies, showing that fully differentiated type-1 CD8 Tmem are also subject to regulation of IL-18Ra expression through IL-4:STAT6. We show IL-4, acting directly on CD8⁺ Tmem, is sufficient to impair IFNγ expression and impede L.m. control, while STAT6-deficient CD8⁺ Tmem escape this restraint, mediating effective bystander protective immunity in the face of elevated IL-4. These findings suggest an application for STAT6 blockade, although we found that current inhibitors likely act off-target.

IL-4 paradoxically enhances GzmB production during bystander activation, in part by limiting IL-18 signals, but hampers perforin and GzmC expression. Unexpectedly, we found no evidence of innate-like NKG2D-mediated killing by in vitro bystander-activated T cells, regardless of IL-4 exposure. Others demonstrated NKG2D-mediated bystander cytotoxicity using mouse cells activated by infection⁷ or human cells activated by longer-term IL-15 culture⁸. This could suggest additional signals and/or durable stimulation are needed to unlock direct bystander-mediated killing programs. We did, however, observe that cytokine stimulation changed TCR-mediated killing: IL-12/15-stimulated CD8 Tmem were potent killers of antigen-bearing target cells, but addition of IL-4 or IL-18 blunted this. This outcome may reflect the differential regulation of perforin and GzmB by IL-4 and IL-18, respectively, and highlight how innate-like activation can influence subsequent adaptive CD8 Tmem responses. Though IL-4 did not enhance co-expression of granzymes and perforin necessary for direct killing, there is growing evidence that granzyme expression without perforin modulates immunity^42,43. Pinpointing the relevance of IL-4-mediated GzmB upregulation remains challenging given the 10 granzyme paralogs in mice with potential functional redundancies.

Though IL-4 could curtail IL-18 sensing in a manner important for bystander activation, this axis may also be involved in other T cell fates. For instance, IL-18 promoted co-inhibitory receptor expression and T cell exhaustion during cancer^44,45, and IL-4 signaling restored functionality in dysfunctional T cells⁴⁶. We show that IL-4 reduced PD-1 expression during homeostasis and activation. Others found IL-4 impaired the formation of tissue resident T cells⁴⁷, and that PD-1 expression was critical for residency⁴⁸. It will be necessary to determine if IL-4 limits exhausted or resident states by IL-18Ra or PD-1 downregulation, respectively, but it suggests a larger-than-appreciated role of IL-4 in CD8 Tmem fate and function. This is amplified by our data on the diverse transcriptional changes induced by IL-4 exposure, which varied with accompanying stimuli.

Our data showed that IL-4, acting directly on CD8 Tmem, in a STAT6-dependent manner, blunted IFNγ production and impaired bystander control of L.m.; since helminth infections are renowned for inducing type-2 responses, we expected contemporaneous worm infection would behave similarly to IL-4c treatment. Indeed, H. polygyrus infection blunted IL-18Ra expression in bystander T cells, which also failed to express IFNγ in the spleen during L.m. co-infection. However, we found the mechanism by which H. polygyrus limited IL-18Ra and IFNγ expression in bystander Tmem was largely independent of STAT6. Unlike IL-4:STAT6 signaling, H. polygyrus infection did not impair innate-like clearance of Listeria. This probably reflects the complexity of helminth infection versus reductionist approaches like IL-4c treatment. H. polygyrus produces TGF-β mimics to alter host immunity⁴⁹, and TGF-β can reduce IFNγ expression during bystander activation^11,50. Despite being a type 2 infection, H. polygyrus can elicit type 1-like responses in the gut, which bystander activate microbiome-specific Tmem into producing IFNγ with no effect on worm burden³⁷. Other work showed that H. polygyrus infection promotes robust IL-15 production³⁶, which might counteract some effects of induced IL-4. Future work will be necessary to test if this is a mechanism by which competency to intracellular “type 1” pathogens is maintained despite type 2 signals and if localized bystander responses can lead to protective responses at distal sites. Nevertheless, our data show that the complex milieu experienced during a type 2 infection affects bystander activation. Helminth infection affects a quarter of humans and cytokine therapies are of growing clinical significance; it will be important to consider their effects on bystander T cell responses and whether these impacts are short-term or irreversible.

Overall, our data demonstrates that the protective responses mediated by CD8 Tmem bystander activation can be perturbed by IL-4 signals. We highlight that propensities for bystander activation are determined by a combination of factors, including antigen experience and strain. Together, these underscore important considerations in manipulating bystander T cell responses.

Methods:

Mice.

All experimental procedures were approved by the Institutional Animal Care and Use Committee at the University of Minnesota. All animals were maintained in specific pathogen–free facilities at the University of Minnesota and infected in modified pathogen–free facilities. Experimental groups were nonblinded; animals were randomly assigned to experimental groups; and no specific method was used to calculate sample sizes. SPF and BSL-2 housing maintained humidity between 30 and 70%, temperature between 20 and 23°C, and 14-hour light/10-hour dark cycle. Animal chow in both facilities (Teklad 2918) and water was provided ad libitum and changed at least weekly.

We purchased mice (C57BL/6J, CB6F1/J, BALB/cJ, BALB/c-Il4ra^tm1Sz/J [BALB/c Il4ra^−/−], C57BL/6J-Ptprc^em6Lutzy/J [JAXBOY], C57BL/6-Tg(TcraTcrb)1100Mjb/J [OT-I TCR transgenic], B6.129S7-Ifngtm1Ts/J [Ifng^−/−]) from The Jackson Laboratory. We maintained congenically-distinct (i.e., CD45.1⁺) OT-I TCR transgenic mice by crossing C57BL/6-Tg(TcraTcrb)1100Mjb/J and C57BL/6J-Ptprc^em6Lutzy/J mice. We obtained Il4ra^−/− mice⁵¹ from Dr. Fred Finkelman, which we then backcrossed onto C57BL/6J background. We crossed these animals with Ifng^−/− mice to maintain an Ifng^−/− Il4ra^−/− line. We euthanized mice in accordance with institutional protocols and subsequently collected spleens and LNs for experimentation. We exclusively utilized female mice for all experiments except for those in Figs. 2B–C, 5G–I, 7A–B, and Extended data Figs. 8M which used combinations of male and female mice (denoted in figures). Mice were ≥8wk of age at initiation of experimentation.

Pathogens and infections.

We used an OVA-expressing VSV construct, administering 1 x 10⁶ PFU per mouse intravenously. We used wild-type L. monocytogenes. We grew L. monocytogenes in TSB media (+50μg/mL streptomycin) to the log phase of growth (determined by OD600), administering 1–2 x 10³ CFU per mouse intravenously. We propagated and prepared infective H. polygyrus bakeri third-stage larvae (L3) as previously described⁵², administering 200 L3 per mouse via gavage. All pathogens were administered in sterile 1x PBS. To cure H. polygyrus infection, we delivered 2mg pyrantel pamoate (Columbia Laboratories) via gavage 7 days post helminth infection and validated pathogen clearance by dissecting the small intestine at harvest.

Bacterial quantification.

We dissociated spleens and livers in sterile-filtered 0.1% IGEPAL (Sigma Aldrich) in PBS using gentleMACS M tubes with a GentleMACS dissociator (Miltenyi Biotec) on the Spleen protocol twice. We plated various dilutions on TSB agar plates (+50μg/mL streptomycin) at 37°C overnight before counting discrete colonies.

IL-4 complexes.

We complexed rIL-4 (Shenandoah Biotechnologies) with anti-IL-4 (clone: 11B11; BioXCell) at a 1:5 ratio by mass. We incubated complexes for 2–10min at 20-22°C before diluting in 1x PBS for injections. We delivered 30μg IL-4c (5μg rIL-4, 25μg anti-IL-4) per mouse in a 200μL volume intraperitoneally.

Developing OT-I memory mice.

We prepared a single-cell suspension of spleen and lymph node cells that were harvested from female OT-I mice by mechanically passing tissue through a 70μm strainer. To enrich transgenic T cells, we used MACS with a CD8 Negative Selection Kit (Miltenyi Biotec). We adoptively transferred 1 x 10⁴ OT-I T cells in sterile 1x PBS i.v. per female C57BL/6 recipient and subsequently i.v. infected recipients with 1 x 10⁶ PFU VSV-OVA. We allowed OT-I memory T cells to contract to a stable memory pool (≥30d) before assaying tissues.

T cell isolation and in vitro stimulation for intracellular cytokine staining.

We harvested spleen and LN from mice and mechanically prepared single-cell suspensions. To enrich bulk CD8 T cells from single-cell suspensions, we used mouse-specific CD8 T cell negative isolation MojoSort kits (BioLegend). We plated 0.5 to 1 x 10⁶ CD8 T cells per well in 96-well V-bottom tissue culture plates. We cultured cells in RP10 media (RPMI 1640 supplemented with 10% FBS, 2mM L-glutamine, 100U/mL penicillin-streptomycin, 1mM sodium pyruvate, 0.05mM β-mercaptoethanol, and 1mM HEPES pH 7.2-7.5). We cultured T cells in RP10 with media alone, rIL-4 (Peprotech), rIL-7 (Peprotech), rIL-12p70 (Peprotech), rIL-15 (BioLegend), and/or rIL-18 (BioLegend) (each at 100ng/mL). For TCR stimulations, we coated plates with 10μg/mL, ea. of anti-CD3 (clone: 145-2C11) and -CD28 (clone: 37.51) (BioXCell) overnight at 4°C before decanting solution and plating CD8 T cells. We cultured cells at 37°C, 5% CO₂, sampling cells at 24, 48, and 72h for flow staining. For intracellular cytokine staining, we added GolgiPlug (BD Biosciences) at a 1:1000 dilution 4h prior to cell harvest.

In vivo bystander activation and ex vivo analysis.

We i.v. infected mice with 1–2 x 10³ CFU wild-type L. monocytogenes. We infected mice at day 6 of H. polygyrus infection or immediately after IL-4c administration. For assays looking at H. polygyrus clearance, we infected mice at day 25 of active infection or day 14 after cure (which was initiated at day 7 with 2mg i.g. pyrantel pamoate). At 24 and 72h timepoints, we i.v. injected mice with 3μg of allophycocyanin- or FITC-conjugated anti-CD8b.2 (clone: 53-5.8) 3min prior to euthanasia to label the splenic WP^19,53,54. We harvested blood, lymph nodes, and spleens and mechanically prepared single-cell suspensions by passing tissues through a 70μm filter. To remove contaminating red blood cells, we incubated blood and spleen suspensions in ACK buffer for 5min, and then proceeded to conduct staining for flow cytometric analysis.

Chemical inhibitors.

We obtained and reconstituted chemical inhibitors against STAT6 (AS1517499) (Selleck Chemicals) in sterile DMSO. We added DMSO (vehicle control) or chemical inhibitors to cell cultures at the same time as cytokine stimuli.

Cell lines:

We received RMA-RAE-1εcells from Dr. David Raulet (U.C. Berkeley). We validated constitutive RAE-1εexpression with cytometry. We did not conduct additional authentication as RMA are not included in the misidentified cell lines list maintained by the International Cell Line Authentication Committee. We did not test cell lines for mycoplasma.

CRISPR-Cas9 editing.

We targeted genes for CRISPR-mediated editing using two gRNAs per gene of interest: Stat6: 5’-CCUCACAGGGAGUUCCUGGU-3’, 5’-CUUGCCGCACAUCAGCACCU-3’; Il18r1: 5’-UCACCGAUCACAAAUUCAUG-3’, 5’-AUUCUGGCCAGUUGAGAUGG-3’; Thy1: 5’-CAGUCUUGCAGGUGUCCCGA-3’, 5’-CCGCCAUGAGAAUAACACCA-3’; Raet1e: 5’-ACACCAGGGAAGGUCUGCUG-3’, 5’-UUCAGGUGACCCAGGGAAGA-3’ (Synthego). Targeting Thy1 demonstrated no defects in CD8 Tmem⁵⁵, so we targeted it as a dsDNA breakage control. We separately complexed gRNAs with Cas9 (Integrated DNA Technologies) before combining the two gRNA:Cas9 complexes with P3 buffer (Lonza Biotech), enhancer (Integrated DNA technologies), and MACS-isolated, naïve OT-I T cells or RMA-RAE-1ε cells for nucleofection with a 4D Nucleofector X Unit (Lonza Biotech) using the pulse code DN-100. For OT-I T cells, we immediately transferred cells into recipient mice for in vivo expansion or we cultured cells in RP10 media on aCD3/CD28-coated plates + recombinant human IL-2 and mouse IL-7 (100U/mL, 50ng/mL, respectively; PeproTech) for 48h, before culturing an additional 5 days in the absence of TCR agonists. For RMA-RAE-1ε cells, we cultured cells in RP10 for 7 days and aseptically sorted RAE-1ε deficient cells to develop a distinct line.

VITAL in vitro killing assay.

We used the MojoSort CD8 Isolation Kit (BioLegend) to enrich the CD8 T cells from the spleens and lymph nodes of OT-I memory mice, which we cultured in cytokine-containing RP10 media (100ng/mL, each) in 15mL conical tubes for 24h. We stained bulk CD8 T cells after culture with Ghost Dye Red780 (Cytek), CD8a FITC, and CD45.1 PE, and FACS-isolated live, CD8a+, CD45.1+ OT-I Tmem using an Aurora CS (Cytek). To develop RAE-1ε-deficient target cells, we CRISPR-edited RMA RAE-1ε cells (which constitutively express the NKG2D ligand) for Raet1e (which encodes RAE-1ε) (Extended Data Fig. 6c). We sorted cells which permanently lost surface RAE-1ε expression and referred to these as RMA ΔRAE-1ε cells (Extended Data Fig. 6c). To test NKG2D-mediated killing, we labeled RMA-RAE-1ε target cells with 1μM Cell Trace Violet (Thermo Fisher) and RMA ΔRAE-1ε control cells with 1μM CFSE (Thermo Fisher). We mixed RMAs at a 1:1 ratio and incubated with OT-I Tmem in RP10 for 6 h at an effector:target ratio of 6:1 (Fig 6h). To test TCR-mediated killing, we labeled RMA ΔRAE-1ε cells with 1μM Cell Trace Violet which we then cultured with 1μM SIINFEKL in RP10 for 1-2h at 37°C. We then washed cells twice, mixed 1:1 with peptide-free CFSE-labeled RMA ΔRAE-1ε cells, and co-cultured with sorted OT-I Tmem effectors at a 5:1 effector:target ratio. After co-culture, we conducted staining with Ghost Dye Red780 and CD8b APC and determined the CFSE+ to CTV+ ratio of the remaining live cells by flow analysis. We calculated specific lysis of RMAs as previously described⁵⁶.

RNAseq.

We pooled MACS-isolated CD8 T cells from OT-I memory mice, which we stimulated with media, IL-12/15, IL-12/15/18, or SIINFEKL (all ± IL-4) (cytokines at 100ng/mL, ea; peptide at 1μM) for 24h. We flow sorted OT-I T cells directly into RLT+ buffer with 1% beta-mercaptoethanol, which were then immediately frozen at −-80°C. We isolated RNA and digested gDNA using Quick RNA MicroPrep kits (Zymo Research). The University of Minnesota Genomics Center (UMGC), prepared libraries using Watchmaker Stranded mRNA prep kits (Watchmaker Genomics), which were pooled and sequenced on an AVITI (Element Biosciences) Freestyle Medium 2x150bp run. Runs generated ≥500 million pass filter reads for the pool, with all expected barcode combinations detected and mean quality scores ≥Q35.

Flow cytometry.

For in vitro and ex vivo analyses we conducted all stains at 4°C. All reagent information is included in Supplemental table 2. We conducted LIVE/DEAD fixable blue viability dye (Thermo Fisher) or Ghost Dye Red780 (Cytek) staining in 1x PBS. For surface staining, we utilized brilliant staining buffer (BD Biosciences) as the stain diluent. Surface stains were for 25 min and were extended to 60 min for tetramer staining. For intracellular cytokine staining, we fixed cells for 20 min in 1x FOXP3/Transcription Factor Fixation/Permeabilization buffer (Cytek) and conducted intracellular/nuclear stains for 30 min using 1x Flow Cytometry Perm Buffer (Cytek) as the diluent. For panels interrogating APCs, we fixed cells in Cytofix/Cytoperm (BD Biosciences) for 20 min and conducted intracellular staining using 1x Perm/Wash (BD Biosciences) as the diluent for 30 min. For PhosFlow panels, we fixed cells in 37°C Fix Buffer I (BD Biosciences) for 15 min, permeabilized cells in ice-cold Perm Buffer III (BD Biosciences), and stained for 60 min in FACSWash (1x PBS, 2% FBS, 0.2% sodium azide). After staining, we resuspended cells in FACSWash and acquired events on a LSRFortessa (BD Biosciences), which we analyzed using FlowJo v10 (BD Biosciences).

Statistics.

We conducted statistical testing using Prism v10 (GraphPad). These tests included: Kruskal-Wallis tests with Dunn’s multiple comparisons tests; ordinary one-way ANOVA with Tukey’s multiple comparisons tests; ordinary one-way ANOVA with Holm-Šídák multiple comparisons tests; one-way ANOVA with Geisser-Greenhouse Correction and Dunnett’s multiple comparisons tests; One-way ANOVA with Geisser-Greenhouse correction and Tukey’s multiple comparisons tests; Friedman tests with Dunn’s multiple comparisons tests; paired t-tests (two-sided); unpaired t-tests (two-sided). Test types are included in the figure legends. We provided exact P values. Where no P value is provided for a relevant comparison, the result was not significant. We did not use 1-tailed tests. Data distribution was assumed to be normal but this was not formally tested. No statistical methods were used to pre-determine sample sizes, but are similar to those in other publications^18–20 The numbers of experiments and animals analyzed for each experiment are indicated in figure legends. Bars depict means. We excluded mice that rejected adoptively-transferred OT-I populations.

Extended Data

Extended data figure 1. Bystander activating cytokines differentially activate CD8 Tmem from SPF laboratory mice.

A CD69 expression in memory phenotype (CD44^hi) CD8 T cells after stimulation with bystander-activating cytokines (IL-12/15/18, 100ng/mL, ea.) or TCR agonists. B Expression of IFN-γ and CD69 in CD8 Tmem from SPF co-housed mice. C Representative gating of CD49d^lo virtual memory CD8 T cells (T_VM) and expression of IFN-γ and CD69 after exposure to IL-12/15/18. D Expression of IL-18Ra in CD8 Tmem from SPF co-housed animals. E Expression of IFN-γ in BALB/c Il4ra variant CD8 Tmem at 72h and 120h timepoints of stimulation with IL-4/12/15/18 or IL-12/15/18 (100ng/mL, ea.), as described in Figure 2B. Points in A–D depict unique animals (n=2–5) animals across (A and B) 2 and (C and D) 1 technical replicate. Each symbol in E depicts cells from an individual animal in a unique stimulation condition (n=3–8 per strain/genotype) across 2 technical replicates. Bars depict mean + SD.

Extended data figure 2. IL-4 limits IL-18Ra expression and IFN-γ expression elicited by bystander-activating cytokines.

A We stimulated cells from SPF C57BL/6, CB6F1, and BALB/c mice as outlined in Figure 2A with bystander activating cytokines (IL-12/15/18, 100ng/mL, ea.) in the presence or absence of IL-4 (100ng/mL) and interrogated cells using flow. Expression of IFN-γ in CD8 Tmem across SPF strains. B IL-4-mediated fold reduction of IFN-γ elicited by bystander-activating cytokines. C Cross-strain comparison of IL-4-mediated fold reductions of IFN-γ during in vitro bystander activation. D Expression of IL-2Rb and IL-4Ra in OT-I Tmem after in vitro cytokine stimulation as outlined in Figure 2C. E We stimulated cells from OT-I memory mice with varying concentrations of IL-4. F IL-18Ra expression across IL-4 stimulation concentrations. Points in A, B, C and D depict cells from a single animal at a distinct timepoint, where connected lines indicate shared animal identity, and bars indicating mean (A n = 13-14 across 4 technical replicates; D n=7–8 across 3 technical replicates). Points in F depict cells from a single condition and animal, with a line indicating mean. Indicated statistical significance in A and D was calculated by Friedman tests and Dunn’s multiple comparisons tests comparing A all data groups to one another or D all data groups against mock-treated cells. Indicated statistical significance in B and C was calculated by Kruskal Wallis and Funn’s multiple comparisons test. Indicated statistical significance in F reports significance from IL-4-untreated cells and was calculated by one-way ANOVA with Geisser-Greenhouse Correction and Dunnett’s multiple comparisons tests.

Extended data figure 3. IL-4 complex therapies impair protective functions of endogenous CD8 Tmem and NK cells.

A Asialo GM1 (AsGM1) expression at homeostasis in CD8 Tmem. B Experimental outline, in which we treated animals with anti-AsGM1 and monitored lymphocyte frequencies. C NK and OT-I Tmem frequencies during AsGM1 depletion. D Experimental outline, in which we treated animals with anti-AsGM1 and/or IL-4c and measured L. monocytogenes pathogen burdens. E–H We treated mice with IL-4c and infected with L. monocytogenes as described in Figure 5A. E, F IFN-γ expression correlations with (WP) OT-I Tmem in E WP endogenous CD8 Tmem and F bulk splenic NK cells. G, H IL-18Ra expression and correlations with OT-I Tmem in G WP endogenous CD8 Tmem and F splenic NK cells. I We injected mice with combinations of IL-4c, IL-12 blocking antibody (anti-IL-12p40), and L. monocytogenes and assayed splenic WP OT-I Tmem for IL-18Ra expression. Flow plots in A and C are representative of n = 4 animals across 2 replicates and n = 2 mice across 1 technical replicate, respectively. D represents n = 8–10 mice per condition across 2 technical replicates. Indicated statistical significances in D were calculated by Kruskal Wallis and Dunn’s multiple comparisons test. Each point in E–I, represents an individual animal/tissue. Plots in E–H depict n=3–6 (per timepoint and condition) across 2 technical replicates, plots in I depict n=4–5 (per condition) across 4 technical replicates. Indicated statistical significance in E–I was calculated by ordinary one-way ANOVA with Tukey’s multiple comparisons test or linear regression. Bars depict mean + SD, with each point representing a unique animal/tissue.

Extended data figure 4. IL-4c impairs IFN-γ–mediated monocyte instruction.

We treated mice with IL-4c and infected with L. monocytogenes as described in Figure 5C. A Frequencies of splenic antigen-presenting cell (APC) subsets during IL-4c treatment and/or L. monocytogenes infection. B Ly-6C expression in splenic monocytes during IL-4c treatment and/or L. monocytogenes infection. C iNOS expression in Ly-6C^hi splenic monocytes during IL-4c treatment and/or L. monocytogenes infection. D Representative plots of IFN-γ–induced gene expression in iNOS⁺ (red) and iNOS⁻ (grey) splenic Ly-6C^hi monocytes. E Correlation of IFN-γ–induced gene expression in splenic Ly-6C^hi monocytes with IFN-γ expression in splenic WP OT-I Tmem. Each point in A, B, C, and E represent an individual animal (n=3–6 per timepoint and condition) across 2 technical replicates. Indicated statistical significance was calculated by ordinary one-way ANOVA with Tukey’s multiple comparisons test or linear regression.

Extended data figure 5. IL-4 selectively impairs cytokine-mediated T cell differentiation and functions.

A–C We treated mice with IL-4c and infected with L. monocytogenes as described in Figure 5C. A–C Granzyme B expression patterns in A splenic white pulp (WP) OT-I Tmem, B WP endogenous CD8 Tmem, and C splenic NK cells during combined IL-4c treatment and L. monocytogenes infection. D–F Data generated from RNAseq approach as described in Figure 6E. D Volcano plots for DEGs (in which CPM ≥ 25 in 2/3 replicates) in OT-I Tmem during IL-4 stimulation in the context of IL-12/15 or SIINFEKL. E Cytotoxic protein expression in OT-I Tmem after in vitro stimulation with IL-4 ± SIINFEKL. F DEG Venn diagrams across genes upregulated (top) or downregulated (bottom) by IL-4 in the context of mock, IL-12/15, IL-12/15/18, or SIINFEKL stimulation. G Expression of NKG2D or PD-1 in OT-I Tmem after in vitro stimulation with permutations of IL-4, −7, −12, −15, and/or −18. Each point in A, B, and C represent an individual animal (n=3–6 per timepoint and condition) across 2 technical replicates. D and F depict 3 replicates of n = 2 animals per replicate. E Depicts n = 8–12 animals across across 5–8 technical replicates. Indicated statistical significance in A–C was calculated by ordinary one-way ANOVA and Tukey’s multiple comparisons tests and E Kruskal Wallis and Dunn’s multiple comparisons test. Bars depict mean +SD.

Extended data figure 6. Impact of IL-4 on cytotoxic effector proteins hinges on stimulation Context.

A Expression of granzyme A, granzyme C, and perforin in OT-I Tmem after 24h in vitro stimulation. B Co-expression of perforin and granzymes in OT-I Tmem after 24h in vitro stimulation. C Generation of RAE-1ε-deficient RMA target cells by CRISPR/Cas9 editing. A and B depict n = a8n imals across 5t echnical replicates, wherein plots depict individual animal/stimulation conditions and bars depict mean. C Depicts n = 1 experiment for post-CRISPR editing, and genotypes for post-sort purity depict n = 3 separate culture experiments. Indicated statistical significance was calculated by ordinary one-way ANOVA with Geisser-Greenhouse correction and Tukey’s multiple comparisons test.

Extended data figure 7. IL-4:STAT6 signaling controls IL-18 sensing and IFNγ expression during bystander activation.

A We stimulated CD8 T cells isolated from wildtype, Il4ra^−/−, and Il4ra^Y500F BALB/c mice and stimulated with IL-4 (100ng/mL) and/or IL-12/15/18 (100ng/mL) in a setup similar to Figure 2B and interrogated cells using flow. IL-4-mediated fold change of IFNγ (top) and GzmB (bottom) in CD8 Tmem stimulated with IL-12/15/18. B IFNg expression during IL-12/15/18 ± IL-4 stimulation in the presence of STAT6 inhibitor, AS1517499. C IFNg and IL-18Ra expression in ΔSTAT6 and ΔThy1 OT-I cells during in vivo bystander activation with L. monocytogenes and IL-4c treatment. D IL-18Ra, CD69, IFNg, and GzmB expression in ΔSTAT6 and ΔThy1 OT-I Tmem after in vitro cytokine stimulation. E CD69, IFNg, and granzyme B expression in ΔIL-18Ra and ΔThy1 OT-I Tmem after 24h in vitro TCR stimulation with cognate SIINFEKL antigen. Symbols in A depict a unique animal/timepoint across n = 1 technical replicate, with bars indicating mean. Symbols in B depict mean ±SD across n=4 donors in 4 technical replicates and are connected by mean. Symbols in C depict a unique OT-I T cell population from an animal (n = 6–10) across 2–3 technical replicates, with bars indicating mean+SD. Symbols in D depict a unique OT-I T cell population from an animal (n=6–10) across 2–3 technical replicates and are connected by animal identity. Indicated statistical significances were calculated by C ordinary one-way ANOVA with Tukey’s multiple comparisons test or D paired t tests.

Extended data figure 8. Type 2 infections perturb the bystander T cell response by STAT6-dependent and -independent mechanisms without exacerbating listeriosis.

A Experimental outline for H. polygyrus and L. monocytogenes infections. B–D Ex vivo phenotyping of activation/functional markers and IL-18Ra in bystander OT-I Tmem from the splenic white pulp (WP). E Experimental approach evaluating H. polygyrus infection and clearance kinetics on OT-I Tmem phenotype and L. monocytogenes-mediated bystander activation. F Frequencies of major lymphocyte populations and bulk (i.e., both ΔSTAT6 and ΔThy1) OT-I Tmem. G Early timepoint (i.e., ≤ 7d post H. polygyrus) frequencies and phenotypes of ΔSTAT6 and ΔThy1 OT-I Tmem. H Bulk OT-I Tmem frequencies post H. polygyrus clearance compared to frequencies at peak (7d) infection. I IL-18Ra expression in bulk OT-I Tmem compared to pre-infection frequencies. J IL-18Ra expression kinetics post-antihelminth therapy in ΔSTAT6 and ΔThy1 OT-I Tmem. K IFNg expression in ΔSTAT6 and ΔThy1 OT-I Tmem from H. polygyrus-infected animals during in vivo bystander activation with L. monocytogenes. L Experimental outline testing impact of H. polygyrus infection on L. monocytogenes clearance in animal model wherein bystander OT-I Tmem are the only population capable of IFNg expression and sensitivity to IL-4 signals. M L. monocytogenes burdens in mock- and H. polygyrus-infected animals wherein bystander OT-I Tmem are sole population capable of IFNg expression and IL-4 sensitivity. Each point in B–D represents an individual timepoint/stimulation connected by animal identity (n=3–6 mice) across 2 technical replicates. Each point in F, H, I represents an animal at a unique timepoint (n=3–6) across 1 technical replicate. Each point in K represents a unique cell population/animal (n=2–6 per condition) across 2 technical replicates. Indicated statistical significance was calculated by ordinary one-way ANOVA and B, C, D, M Tukey’s or H Dunnett’s multiple comparisons test, F, I Mixed effects analysis with Geisser-Greenhouse correction and Dunnett’s multiple comparisons tests, or K paired t test.

Data availability:

The data that support the findings of this study are available from the corresponding author upon reasonable request. RNAseq data is deposited under GEO accession term GSE315645.