The CREM–IL-15 axis: decoding the persistence–exhaustion paradox in NK cell immunotherapy

Abstract

Natural killer (NK) cells are critical components of the innate immune system, renowned for their ability to recognize and eliminate malignant and infected cells without prior sensitization. NK cell immunotherapy encompasses various approaches, including adoptive transfer of ex vivo expanded NK cells, cytokine stimulation to enhance their activity, and genetic modifications to improve persistence and specificity. CREM is a cAMP-responsive transcription factor that modulates gene expression in response to receptor- and cytokine-driven signaling. Recent evidence now shows that IL-15 and CAR stimulation rapidly induce CREM in activated NK cells. Rafei (Nature 643:1076–1086, 2025) further demonstrated in CAR-NK models that CREM functions as a regulatory checkpoint limiting NK cell cytotoxicity and cytokine production, while its relevance in resting/native NK cells remains to be established. The CREM–IL-15 signaling axis has emerged as a pivotal regulator of NK cell biology, influencing their development, activation, and longevity. IL-15 is a critical cytokine for NK cell survival, proliferation, and functional maturation. Understanding this axis is vital, as it offers insights into mechanisms that sustain NK cell activity and those that lead to functional exhaustion, thereby informing strategies to enhance therapeutic efficacy. A central challenge in NK cell immunotherapy is balancing cellular persistence with functional exhaustion. Persistent NK cell activity is desirable for sustained tumor control; however, prolonged activation often results in cellular exhaustion characterized by diminished cytotoxicity and cytokine production. This paradox hampers the long-term success of NK cell-based treatments. The CREM–IL-15 axis plays a complex role in this dynamic, potentially promoting NK cell survival and persistence while also contributing to exhaustion under certain conditions. Deciphering the molecular underpinnings of this paradox is essential for developing interventions that maintain NK cell functionality over time, thereby improving therapeutic outcomes in cancer patients.

Introduction

Natural killer (NK) cells constitute a vital component of the innate immune system, recognized for their capacity to eliminate virally infected and transformed cells without prior sensitization. Their inherent cytolytic activity and rapid cytokine production position them as compelling candidates for cancer immunotherapy [1]. Adoptive transfer of activated NK cells, often augmented by cytokines, has demonstrated promise in preclinical models and clinical trials across various malignancies. For instance, recombinant human interleukin-15 (rhIL-15) has been evaluated for stimulating autologous NK cell activity in chronic lymphocytic leukemia (CLL), leading to improved leukemic cell depletion, particularly in conjunction with anti-CD20 monoclonal antibodies [2, 3]. Similarly, inhaled IL-15 showed activation of NK cells in osteosarcoma models. Clinical infusions of IL-15 have been observed to expand CD56^bright NK cell populations, enhancing their cytotoxic capabilities against target cells [4, 5]. Despite these encouraging findings, a significant hurdle for NK cell-based therapies remains their limited persistence and functional decline, commonly termed exhaustion, particularly within the suppressive tumor microenvironment [6, 7].

The efficacy of NK cell immunotherapy is directly contingent upon the sustained functionality and viability of these effector cells within the challenging tumor microenvironment. A key challenge is the phenomenon of NK cell exhaustion, where chronic activation and exposure to suppressive factors impair their cytotoxic potential and cytokine production [8]. IL-15 is a cytokine indispensable for NK cell development, survival, proliferation, and activation. While IL-15 stimulates NK cell responses, continuous exposure to high doses can paradoxically lead to functional exhaustion, characterized by reduced viability, impaired signaling, and decreased tumor control [9]. This highlights a complex interplay between IL-15 and NK cell state. Cyclic AMP-responsive element modulator (CREM) is a transcription factor family known to regulate diverse cellular processes, including immune cell differentiation and function, often in response to cyclic AMP signaling [10]. Recent work now directly links CREM to NK cell regulation: IL-15 and CAR signaling induce CREM, and CREM loss enhances NK cell cytotoxicity and cytokine production. These findings place CREM within the core transcriptional circuitry shaping NK cell persistence, activation, and exhaustion [8, 11]. Importantly, this conclusion is now directly supported in CAR-NK/IL-15 models, where CREM induction depends on PKA–CREB signaling and CREM deletion enhances cytotoxicity, 3D spheroid killing, tumor rechallenge resistance, and in vivo control [11]. Understanding how CREM might interact with IL-15 signaling to influence NK cell persistence versus exhaustion is crucial for developing durable immunotherapies.

This review investigates current knowledge concerning the critical factors that dictate NK cell persistence and functional exhaustion in the context of immunotherapy, with a particular focus on the IL-15 axis and the experimentally demonstrated role of CREM in NK cell regulation downstream of IL-15 and CAR signaling. The primary objective is to delineate the cellular and molecular mechanisms underlying the persistence–exhaustion paradox in NK cells, specifically examining the intricate signaling pathways initiated by IL-15 and exploring the potential regulatory role of CREM. We aim to integrate findings from various studies to construct a comprehensive understanding of NK cell functional dynamics within the tumor microenvironment. Furthermore, this review identifies current limitations in the understanding of the CREM–IL-15 axis in NK cells and proposes future research directions, including potential therapeutic strategies to overcome exhaustion and enhance the efficacy of NK cell-based immunotherapies.

NK cell biology: phenotypic and functional landscape

NK cells are lymphocytes of the innate immune system, originating from hematopoietic stem cells and differentiating in the bone marrow, spleen, and lymph nodes. They are characterized by the expression of CD56 and the absence of CD3, distinguishing them from T lymphocytes [12]. Two primary subsets of human NK cells are recognized: CD56^dimCD16⁺ NK cells, which represent the majority in peripheral blood (approximately 90–95%) and possess potent cytotoxic capabilities, and CD56^brightCD16^−/low NK cells, which are less cytotoxic but are prolific cytokine producers [13]. NK cells recognize and eliminate target cells through a complex balance of activating and inhibitory receptors. Activating receptors, such as natural killer group 2, member D (NKG2D), NKp30, NKp44, and NKp46, bind to stress-induced ligands on target cells, triggering cytolytic responses and cytokine release [14]. Conversely, inhibitory receptors, primarily killer cell immunoglobulin-like receptors (KIRs) and NKG2A, recognize major histocompatibility complex (MHC) class I molecules on healthy cells, preventing NK cell-mediated lysis of self-tissue [15]. The integration of these signals dictates the NK cell's decision to activate or remain quiescent. Once activated, NK cells form an immunological synapse with target cells, polarizing lytic granules containing perforin and granzymes toward the target for secretion and subsequent apoptosis induction [16]. Beyond direct cytotoxicity, NK cells produce various cytokines, notably interferon gamma (IFN-γ) and tumor necrosis factor alpha (TNF-α), which modulate adaptive immune responses and directly influence tumor growth. IL-15 signaling is critical for enhancing these effector functions, promoting NK cell proliferation, survival, and cytotoxic capacity [17, 18].

Activation state, tissue residency, and functional specialization

KIRs is a highly polymorphic family of receptors that interact with MHC class I molecules. KIRs can be either inhibitory (e.g., KIR2DL1, KIR3DL1) or activating (e.g., KIR2DS1). The specific combination of KIRs expressed by an individual NK cell, dictated by stochastic gene expression and epigenetic modifications, creates a unique "receptor repertoire" that fine-tunes its responsiveness to target cells [19]. C-type lectin receptors family includes the inhibitory CD94/NKG2A, which recognizes the non-classical MHC class I molecule HLA-E, and several activating receptors like NKG2D, which binds to stress-induced ligands (MICA/B, ULBPs) on aberrant cells [20]. NKG2C, another member, is associated with "memory-like" NK cell responses, particularly after cytomegalovirus (CMV) infection. In addition, NKp30, NKp40, and NKp46 are key activating receptors that recognize various ligands, some of which are upregulated on tumor cells or virally infected cells, contributing significantly to NK cell-mediated lysis [21]. Receptors like DNAM-1 (CD226) (Fig. 1), LFA-1 (CD11a/CD18), and NKp80 facilitate target cell recognition and conjugate formation, enhancing the efficiency of effector functions [22].

Fig. 1.

NK cell activation is governed by the balance between activating receptors (e.g., NKG2D, NKp46) and inhibitory receptors (e.g., KIRs, NKG2A), which determines whether the cell mounts cytotoxic activity or remains restrained

The phenotypic landscape is not static. NK cells undergo a complex developmental progression, maturing through various stages characterized by changes in receptor expression. Furthermore, the concept of tissue-resident NK (trNK) cells has emerged, revealing populations with unique phenotypic signatures and functions tailored to their specific microenvironments (e.g., liver, uterus, decidua) [23]. These trNK cells often exhibit distinct developmental origins, longer lifespans, and specialized roles in tissue homeostasis and local immune regulation. The functional decision-making of an NK cell is a result of a delicate molecular balance between signals received through its activating and inhibitory receptors, often referred to as a “rheostat” model [24]. This equilibrium determines whether the NK cell remains quiescent, becomes activated to kill, or enters a state of anergy. Healthy cells express high levels of MHC class I molecules, which engage inhibitory KIRs and CD94/NKG2A on NK cells, preventing their activation (inhibition dominates) [25]. Cancer cells or virally infected cells often downregulate MHC class I (the "missing self"), removing the inhibitory signal and allowing NK cell activation. Stressed, infected, or cancerous cells upregulate ligands for activating receptors (e.g., MICA/B, ULBPs for NKG2D; PVR, Nectin-2 for DNAM-1; unknown ligands for NCRs) [26]. These activating signals can overcome residual inhibitory signals, leading to NK cell activation even in the presence of MHC class I, especially if the activating signals are strong enough [27].

Intracellular signaling cascades

Upon ligand binding, activating and inhibitory receptors transmit signals through distinct intracellular signaling cascades. Many activating receptors, such as KIRs with short cytoplasmic tails, associate with ITAM (immunoreceptor tyrosine-based activation motif)-bearing adaptor molecules like DAP10 (for NKG2D) or DAP12 (for some activating KIRs and NCRs) [28]. Phosphorylation of ITAMs by Src family kinases initiates a cascade involving syk/ZAP70 kinases, leading to the activation of downstream pathways like phospholipase C-gamma (PLC-γ), mitogen-activated protein kinase (MAPK), and phosphatidylinositol 3-kinase (PI3K)/Akt [29]. These pathways ultimately result in cytoskeletal rearrangements, granule polarization, perforin/granzyme degranulation, and cytokine gene transcription [30]. CD16, signaling through the FcRγ chain, also utilizes ITAMs to trigger antibody-dependent cellular cytotoxicity (ADCC). Inhibitory KIRs (e.g., KIR2DL1) and CD94/NKG2A possess ITIM (immunoreceptor tyrosine-based inhibitory motif) sequences in their cytoplasmic tails. Upon ligand binding, ITIMs become phosphorylated and recruit SHP1 and SHP2 phosphatases [31]. These phosphatases dephosphorylate key signaling molecules in the activation pathways, effectively dampening or shutting down the activating signals and preventing NK cell cytotoxic or cytokine responses [32].

The primary mechanism involves the directed release of preformed lytic granules containing perforin and granzymes. Perforin creates pores in the target cell membrane, allowing granzymes (e.g., Granzyme B) to enter and initiate apoptotic pathways, ultimately leading to target cell demise [33]. NK cells can also induce apoptosis via FasL/Fas and TRAIL/TRAIL-R interactions. Activated NK cells secrete a range of cytokines, most notably IFN-γ and TNF-α. IFN-γ plays a critical role in anti-viral immunity, enhancing MHC class I expression on target cells, activating macrophages, and promoting T helper 1 (Th1) differentiation [34]. TNF-α has direct cytotoxic effects and contributes to inflammatory responses. NK cells also produce chemokines that recruit other immune cells to sites of infection or inflammation [35]. Through cytokine secretion and direct cell–cell contact, NK cells can influence the maturation and function of other immune cells, including dendritic cells, T cells, and B cells, acting as crucial bridge between innate and adaptive immunity [36]. Recent research has unveiled "memory-like" properties in certain NK cell subsets, particularly in response to viral infections like CMV. These cells exhibit enhanced and prolonged responses upon re-encounter with specific stimuli, challenging the traditional view of NK cells as purely innate effectors [37]. This adaptation, often linked to epigenetic modifications and receptor expression (e.g., NKG2C⁺), highlights a fascinating area of convergence between innate and adaptive immune strategies [38].

CREM: structure, isoforms, and transcriptional regulation

CREM is a member of the CREB/ATF family of basic leucine zipper (bZIP) transcription factors. These proteins bind to cyclic AMP (cAMP) response elements (CREs) in the promoters of target genes, thereby regulating gene expression in response to changes in intracellular cAMP levels [39]. The CREM gene is characterized by its complex alternative splicing patterns, leading to the generation of numerous isoforms with diverse and often opposing functions [40]. These isoforms can be broadly categorized into activators (e.g., CREMα, CREMβ, CREMγ) and repressors (e.g., ICER—inducible cAMP early repressor) of transcription [41]. CREM activators typically contain a kinase-inducible domain (KID) that allows for phosphorylation by protein kinase A (PKA), enhancing their transcriptional activity [42]. Repressor isoforms, such as inducible cAMP early repressor (ICER) (Fig. 2), often lack the KID and act as dominant-negative regulators by binding to CRE sites and preventing the recruitment of transcriptional machinery by activating CREM isoforms or other CREB family members [43]. The precise balance and expression of these isoforms are cell-type specific and highly regulated, allowing for fine-tuning of gene expression in response to diverse physiological stimuli [44]. In immune cells, CREM has been implicated in processes such as T cell differentiation, anergy, and apoptosis, although its specific role in NK cell biology, particularly concerning persistence and exhaustion, remains an area requiring dedicated investigation [45]. Given its capacity to modulate gene networks related to cell survival, proliferation, and differentiation, CREM presents a compelling candidate for influencing NK cell fate under chronic stimulation [46].

Fig. 2.

CREM contains a bZIP DNA-binding domain, a transcriptional activation region, and regulatory sequences. Repressor isoforms such as ICER lack the kinase-inducible/transactivation region and preferentially inhibit CRE-driven transcription

N-terminal regulatory domains are highly variable due to alternative splicing and contain multiple potential phosphorylation sites and distinct transactivation domains (TADs). The precise composition of these N-terminal segments largely dictates an isoform's ability to recruit co-activators and initiate transcription [47]. Kinase-inducible domains, a hallmark of activating CREM isoforms (and its close relative CREB), the KID contains critical serine residues (e.g., Ser133 in humans, Ser117 in mice) that are primary targets for phosphorylation by PKA [48]. Phosphorylation within the KID is a pivotal event, enhancing DNA-binding affinity, promoting the recruitment of co-activators like the CREB-binding protein (CBP) and p300, and ultimately boosting transcriptional activity [49]. Some CREM isoforms possess two KIDs (KIDα and KIDβ), further adding to their regulatory potential. Basic region (BR) highly conserved, positively charged region is responsible for direct sequence-specific binding to DNA, specifically the CRE consensus sequence (5'-TGACGTCA-3') [50]. The basic region contains evolutionary conserved arginine and lysine residues that make direct contact with the major groove of the DNA helix. Leucine zipper (LZ), immediately C-terminal to the basic region, is characterized by a heptad repeat of leucine residues, typically four or five [51]. This hydrophobic domain mediates dimerization, allowing CREM monomers to form stable homodimers or heterodimers with other bZIP proteins, such as CREB and ATF1 [52]. Dimerization is a prerequisite for high-affinity DNA binding. The modular nature of these domains, particularly their susceptibility to alternative splicing, is the primary driver behind CREM's extraordinary functional diversity [53].

The Crem gene, through extensive alternative splicing of its 16 exons and utilization of multiple promoters, gives rise to a remarkable array of isoforms. These isoforms can be broadly categorized based on their functional impact: activators, repressors, and more nuanced regulators [54]. The differential expression of these isoforms is highly tissue-specific, developmentally regulated, and dynamically responsive to extracellular stimuli [55]. Activating isoforms (e.g., CREMτ, CREMα) isoforms are characterized by the presence of both the DNA-binding bZIP domain and one or more functional transactivation domains, including at least one intact kinase-inducible domain [56]. Upon phosphorylation, typically by PKA, these isoforms bind to CREs in the promoters of target genes and recruit co-activators like CBP/p300, leading to enhanced gene transcription. CREMτ (tau), particularly prominent in the post-meiotic germ cells of the testis, is a potent transcriptional activator crucial for spermatogenesis [57]. It contains multiple transactivation domains and two KIDs, making it highly responsive to cAMP signals. Its absence leads to severe male infertility. CREMα is another widely expressed activating isoform found in various tissues, including the brain, thymus, and adrenal gland [58]. CREMα also contains a functional KID and acts as a positive regulator of gene expression. Repressing isoforms (e.g., ICER), in stark contrast to the activators, lack functional transactivation domains and often lack the KID, rendering them unresponsive to activating phosphorylation by PKA [59].

ICER, perhaps the most extensively studied repressing CREM isoform, is unique as its expression is driven by an alternative, cAMP-inducible promoter located within an intron of the Crem gene [60]. This allows ICER to be rapidly induced in response to cAMP signals. Crucially, ICER lacks the N-terminal transactivation domains and the KID, but contains the bZIP domain. This makes ICER a critical component of negative feedback loops, where its induction by cAMP signaling subsequently downregulates prolonged cAMP-dependent gene expression [61]. This autoregulatory mechanism is vital in contexts like the regulation of circadian rhythms and hormonal responses. The precise balance and interplay between activating and repressing CREM isoforms, dictated by tissue-specific expression and signaling cues, determine the ultimate transcriptional outcome [62].

Transcriptional regulation of CREM activity

The activity of CREM, and by extension its impact on gene expression, is subject to multiple layers of sophisticated regulation, encompassing both its intrinsic transcriptional activity and the control of its own gene expression [63]. Posttranslational modifications are the primary and most well-understood regulatory mechanism for CREM activity involves phosphorylation, particularly within its kinase-inducible domain [64]. Upon activation of G protein-coupled receptors (GPCRs) that couple to adenylyl cyclase (AC), intracellular cAMP levels rise. This activates PKA, which in turn phosphorylates activating CREM isoforms (e.g., CREMτ, CREMα) at specific serine residues within their KID (e.g., Ser133) [65]. This phosphorylation event is crucial for enhancing CREM's DNA-binding affinity and, more importantly, for creating a docking site for the co-activator CBP/p300. CBP/p300 possesses histone acetyltransferase (HAT) activity, leading to chromatin relaxation and facilitating the recruitment of the basal transcription machinery [66]. While PKA is the dominant regulator, other kinases can also phosphorylate CREM, adding layers of complexity and integrating CREM with diverse signaling pathways [67]. These include CaMKII (calcium/calmodulin-dependent protein kinase II), RSK (ribosomal S6 kinase), and GSK-3 (glycogen synthase kinase-3). This convergence of multiple signaling pathways on CREM allows it to act as an integrator of various cellular cues [68].

The transcriptional output of CREM is not solely dependent on its DNA binding but also on its ability to recruit accessory proteins. CBP/p300 are essential transcriptional co-activators that interact with phosphorylated CREM [69]. They act as molecular bridges, linking transcription factors to the basal transcription machinery and possessing HAT activity that remodels chromatin structure to promote gene expression [70]. Less extensively characterized are other cofactors that might interact with different CREM isoforms to modulate their activity, contributing to the specificity of gene regulation [71]. A defining feature of CREM regulation is its involvement in critical feedback mechanisms, particularly through the induction of its repressing isoform, ICER. As mentioned, the Icer promoter is itself cAMP-responsive [72]. Thus, an initial cAMP surge activates activators like CREMτ and CREB (Fig. 2), but also rapidly induces ICER expression. As ICER accumulates, it then competes with the activators for CRE binding or forms inactive heterodimers, thereby dampening and terminating the prolonged cAMP response [73]. This ensures that cAMP-mediated gene induction is transient and finely tuned, preventing overstimulation and maintaining cellular homeostasis. The expression levels of different Crem isoforms are themselves tightly regulated at the transcriptional level [74]. Various promoters control the expression of the Crem gene, responding to different hormonal, developmental, and environmental cues. For instance, the Icer promoter is specifically induced by cAMP, glucocorticoids, and in response to stress. This dynamic regulation ensures that the appropriate CREM isoforms are available at the right time and in the right cellular context [75].

IL-15 signaling in NK cell homeostasis and regulatory functions

IL-15 is a four α-helix bundle cytokine that shares structural and functional similarities with IL-2, yet critically diverges in its biological roles and receptor presentation. While IL-2 is primarily involved in T cell expansion, IL-15 predominantly supports NK cell and memory CD8⁺ T cell biology [76]. The biological activity of IL-15 is mediated through a heterotrimeric receptor complex comprising the IL-15 receptor alpha chain (IL-15Rα, CD215), the common beta chain (CD122, shared with the IL-2 receptor), and the common gamma chain (CD132, shared with receptors for IL-2, IL-4, IL-7, IL-9, and IL-21) [77]. IL-15 is classically delivered by trans-presentation, in which IL-15Rα-bearing accessory cells present IL-15 to CD122/γc-expressing NK cells; this mechanism is central to NK cell homeostasis and differentiation [78, 79]. This process involves IL-15 binding to its specific high-affinity receptor subunit, IL-15Rα, on the surface of accessory cells, such as dendritic cells (DCs) or macrophages [80]. This complex is then presented in “trans” to neighboring NK cells, which express the common IL-2/IL-15 receptor beta chain (IL-2/15Rβ) and the common gamma chain (γc) [81].

Upon engagement, the IL-15–IL-15Rα complex activates downstream signaling pathways within the NK cell. The primary pathway involves the Janus kinase (JAK)/Signal Transducer and Activator of Transcription (STAT) pathway, specifically JAK1 and JAK3 (Fig. 3), which phosphorylate STAT5 [82]. Phosphorylated STAT5 translocates to the nucleus, regulating the transcription of genes associated with cell survival (e.g., Bcl-2), proliferation (e.g., Myc), and functional activity. IL-15 also activates the PI3K/AKT/mammalian target of rapamycin (mTOR) pathway, which is crucial for metabolic fitness and sustained proliferation [83]. Inhibition of mTOR abrogates IL-15-induced cell function advantages, underscoring its importance. The mode of IL-15 presentation also influences the outcome: soluble IL-15Rα–IL-15 complexes primarily support NK cell survival, while membrane-associated complexes transferred via “trans-endocytosis” are more critical for proliferation. This nuanced signaling ensures NK cell maintenance and effective immune surveillance [79, 84].

Fig. 3.

IL-15 signaling through IL-15Rα activates JAK3/STAT5 and PI3K/AKT pathways, supporting NK cell survival, proliferation, and cytotoxic function

The canonical JAK–STAT5 pathway

The cytoplasmic tails of CD122 and CD132 lack intrinsic kinase activity but are constitutively associated with JAK tyrosine kinases. Specifically, CD122 is associated with JAK1, and CD132 is associated with JAK3 [85]. Upon IL-15 binding, these receptor chains undergo conformational changes, leading to the rapid trans-phosphorylation and activation of JAK1 and JAK3. Activated JAKs then phosphorylate specific tyrosine residues on the cytoplasmic tails of CD122 and CD132, creating docking sites for the Src homology 2 (SH2) domain-containing protein STAT5 [86]. STAT5, present in the cytoplasm as a latent monomer, is recruited to the phosphorylated receptor complex. JAK1 and JAK3 subsequently phosphorylate STAT5 at conserved tyrosine residues (Tyr694 on STAT5a, Tyr699 on STAT5b) [87]. Phosphorylation induces STAT5 dimerization, allowing it to translocate into the nucleus. In the nucleus, STAT5 dimers bind to specific DNA sequences (STAT5-binding elements, or SBEs) within the promoters of target genes, thereby regulating their transcription [88].

The transcriptional program orchestrated by activated STAT5 is central to NK cell homeostasis: (i) Survival: STAT5 upregulates the expression of anti-apoptotic proteins, most notably B-cell lymphoma 2 (Bcl-2) and myeloid cell leukemia 1 (Mcl-1). These proteins counteract the action of pro-apoptotic Bcl-2 family members, ensuring NK cell longevity and preventing premature cell death [89]; (ii) Proliferation: STAT5 drives the expression of cell cycle regulatory genes, including c-Myc and cyclins (e.g., cyclin D2), which are essential for NK cell expansion and the maintenance of a sufficient cell pool [90]; (iii) Differentiation and Maturation: IL-15-dependent STAT5 signaling is critical for the terminal differentiation and acquisition of full effector functions by NK cells. This includes the upregulation of cytotoxic molecules such as perforin and various granzymes, which are crucial for target cell lysis. STAT5 also contributes to the expression of activating receptors like NKG2D and the low-affinity Fcγ receptor CD16, enhancing NK cell-mediated ADCC [91, 92]; (iv) Metabolic Reprogramming: Recent evidence suggests that STAT5 also plays a role in reprogramming NK cell metabolism, promoting glycolytic flux and oxidative phosphorylation to meet the high energy demands of proliferation and effector function [93].

Beyond JAK–STAT5: the PI3K–AKT–mTOR pathway

While JAK–STAT5 is the dominant pathway, IL-15 also engages other crucial signaling cascades, most notably the PI3K–AKT–mTOR pathway. This pathway is intimately involved in cell growth, metabolism, and survival, and its activation by IL-15 significantly contributes to NK cell homeostasis [94]. Upon IL-15 stimulation, PI3K is recruited to the activated receptor complex, often via adapter molecules or direct interaction with JAKs. Activated PI3K phosphorylates phosphatidylinositol (4,5)-bisphosphate (PIP2) to generate phosphatidylinositol (3,4,5)-trisphosphate (PIP3) at the plasma membrane [95]. PIP3 serves as a docking site for proteins containing a Pleckstrin homology (PH) domain, including AKT (also known as protein kinase B). AKT directly phosphorylates and inactivates pro-apoptotic proteins like BAD and forkhead box O (FoxO) transcription factors, thereby promoting cell survival and synergizing with the anti-apoptotic effects of STAT5 [96]. AKT activates mTOR complex 1 (mTORC1), which is a master regulator of cell growth, proliferation, and metabolism. mTORC1 promotes protein synthesis, lipid synthesis, and glycolysis, providing the necessary building blocks and energy for rapidly proliferating NK cells [97]. This metabolic rewiring is crucial for sustaining the high demands of an activated NK cell. The PI3K–AKT–mTOR pathway can also influence NK cell effector functions, potentially by regulating granule polarization, cytotoxicity, and cytokine production, although these roles are still being fully elucidated [98]. The interplay between JAK–STAT5 and PI3K–AKT–mTOR is dynamic and complex. Both pathways contribute to NK cell survival and proliferation, with potential cross-talk and synergistic effects (Fig. 4). For instance, STAT5 can indirectly influence mTOR activity, and AKT can modulate STAT5 signaling [99].

Fig. 4.

Integrative model of the CREM–IL-15 axis in NK cells: persistence versus exhaustion. IL-15 trans-presentation by IL-15Rα-expressing accessory cells engages the IL-2/15Rβγ receptor complex on NK cells, leading to JAK1/3 and STAT5 activation and induction of CREM. CREM, together with CBP/p300, promotes transcriptional programs associated with NK cell survival, persistence, and function, including anti-apoptotic genes (BCL2A1, MCL1), metabolic regulators (PGC1α, CPT1A, OXPHOS-related genes), effector molecules (IFNG, PRF1, GZMB), and IL-15Rα expression, thereby reinforcing a positive feedback loop. Additional co-stimulatory inputs, including 4-1BB, CD27, CXCR6–CXCL16, IL-12/18, and type I interferons, may further support this program. In contrast, chronic antigen exposure, inhibitory receptors such as NKG2A, PD-1, and TIGIT, and suppressive cues including TGF-β, IL-10, and metabolic stress activate exhaustion-associated pathways involving TOX, NR4A2, BATF, and IRF4, with downstream inhibition of NF-κB, STAT3, and mTORC1 signaling, thereby suppressing CREM-driven IL-15 responses and favoring NK cell dysfunction and exhaustion. Solid arrows indicate established interactions; dashed arrows indicate probable or context-dependent interactions

Other associated signaling pathways

IL-15 can activate the ERK, JNK, and p38 MAPK cascades. These pathways are involved in regulating cell proliferation, differentiation, stress responses, and gene expression, including the production of various cytokines and chemokines by NK cells [100]. Besides, activation of nuclear factor-kappa B (NF-κB) by IL-15 has also been reported, contributing to NK cell survival, cytokine production (e.g., IFN-γ), and the expression of adhesion molecules. To prevent uncontrolled proliferation or immune pathology, IL-15 signaling pathways are tightly regulated by both positive and negative feedback mechanisms [78]. Suppressors of Cytokine Signaling (SOCS) Proteins family proteins, particularly SOCS1 and SOCS3, are inducible negative regulators of cytokine signaling. They are often upregulated by STAT5 activation, creating a negative feedback loop [101]. SOCS proteins inhibit JAK activity, either by direct binding or by targeting JAKs for proteasomal degradation, thereby attenuating STAT5 phosphorylation and downstream signaling. Tyrosine phosphatases such as SHP1 and SHP2 can dephosphorylate JAKs and STATs, reversing their activation and terminating signaling [102]. Signaling molecules, including receptors and kinases, can be ubiquitinated and targeted for degradation by the proteasome, providing another layer of control over signal intensity and duration [103].

rhIL-15 and modified IL-15 superagonists (e.g., N-803/Anktiva) are being investigated as potent anti-cancer agents. By expanding endogenous NK cell populations and enhancing their effector functions, these agents can augment anti-tumor immunity [103]. Clinical trials are showing promise in various solid tumors and hematological malignancies. In adoptive transfer strategies, such as CAR-NK cells, co-expression or systemic administration of IL-15 is being explored to enhance NK cell persistence, expansion, and anti-tumor efficacy in vivo [104]. Modulating IL-15 signaling could also be relevant in chronic viral infections (e.g., HIV, HBV) where NK cell function is often impaired, or in autoimmune disorders where dysregulated NK cell activity might contribute to pathology [105].

The persistence–exhaustion paradox in NK cells

NK cells are potent effectors, but their therapeutic utility is constrained by limited persistence and susceptibility to exhaustion, particularly within the suppressive tumor microenvironment. Exhaustion is characterized by a progressive loss of effector functions, including reduced cytotoxicity, impaired cytokine production (e.g., IFN-γ), and altered receptor expression [106]. This state can arise from chronic activation, sustained exposure to inhibitory signals, or metabolic stress within the microenvironment. A key example of this paradox involves IL-15, a cytokine essential for NK cell survival and proliferation [107]. While IL-15 enhances NK cell expansion and function, continuous exposure leads to decreased viability, diminished signaling, reduced mitochondrial respiration, and impaired tumor control, consistent with NK cell exhaustion [108]. This exhaustion is marked by decreased viability, diminished signaling through pathways like STAT5, reduced mitochondrial respiration, and impaired tumor control [109]. Such findings underscore that optimal NK cell stimulation requires a delicate balance to achieve sustained anti-tumor immunity without inducing functional anergy (Fig. 5). The mechanisms underlying this tipping point, particularly how chronic signaling influences transcriptional programs and metabolic fitness, remain central to overcoming this therapeutic barrier [110].

Fig. 5.

Chronic stimulation can shift NK cells from persistence to exhaustion, leading to reduced cytotoxicity, cytokine production, and overall anti-tumor function

Molecular mechanisms of the CREM–IL-15 axis in NK cells

CREM-mediated gene expression in NK cell fate decisions

Direct experimental evidence now shows that CREM is induced by IL-15 and CAR signaling in NK cells, and that CREM restrains NK cell effector function; therefore, CREM should be presented as an emerging, context-dependent regulatory checkpoint in NK cells rather than a purely speculative candidate. The same study also showed that NK cells express CREM and ICER isoforms at baseline, and that IL-15 increases CREM-specific transcripts in a dose-dependent manner [11]. Rafei et al. also showed that CREM induction converges on the PKA–CREB axis and that CREM-KO increases anti-tumor function across 2D cultures, 3D spheroids, tumor rechallenge assays, and murine models [11]. CREM, through its diverse isoforms, can act as either a transcriptional activator or repressor, influencing gene expression in response to cellular cues, including those mediated by cAMP signaling [111]. As recently demonstrated, CREM is induced by IL-15 and CAR signaling in NK cells and restrains effector function, establishing CREM as a checkpoint transcription factor in the activation–persistence balance. CREM deletion enhances NK cell cytotoxicity and cytokine production, supporting its role as a negative regulator of NK cell function [11, 112].

Sustained NK cell function, or persistence, relies on the active transcription of genes that support cell survival, metabolic fitness, and effector molecule synthesis. If CREM is involved in NK cells, activating isoforms of CREM might contribute to the expression of pro-survival genes such as Bcl-2, which is critical for NK cell maintenance and is regulated by IL-15 signaling via STAT5 [113]. They could also influence genes associated with metabolic pathways, which are crucial for maintaining energy demands during prolonged activation. Furthermore, a hypothetical role for CREM could involve the regulation of genes encoding activating receptors or cytokine receptors, thereby preserving the NK cell's capacity to receive survival and activation signals [114]. The interplay between CREM and other transcription factors, such as E-proteins like HEB or Id2, which regulate NK cell development and lineage decisions, could be critical [115].

Conversely, the induction of exhaustion in NK cells is associated with the upregulation of specific inhibitory receptors and the downregulation of effector molecules. Repressive CREM isoforms, including ICER, are detectable in NK cells, and IL-15 stimulation increases both CREM-specific and ICER-specific transcripts in CAR-NK cells; therefore, chronic IL-15 may favor a repressor-skewed CREM program that could contribute to exhaustion, although this still requires direct validation in unmanipulated human and murine NK cells [11, 116]. For example, ICER's role in attenuating gene expression might lead to reduced production of IFN-γ or granzymes, or increased expression of inhibitory checkpoints like T cell immunoglobulin and ITIM domain (TIGIT), which has been observed to be upregulated in exhausted NK cells in soft tissue sarcoma [117]. The observation that continuous IL-15 exposure leads to exhaustion with altered gene expression patterns, including cell cycle arrest, suggests a transcriptional mechanism at play. Future investigations could explore whether specific CREM isoforms are induced under conditions of chronic IL-15 stimulation and whether their expression correlates with the acquisition of exhaustion markers, thus providing a molecular link in the persistence–exhaustion paradox [118].

Integration of CREM and IL-15 signals: cellular outcomes

The integration of signals from IL-15 and CREM now has direct experimental support in NK cells, because IL-15 stimulation induces CREM, CREM occupancy increases in IL-15-exposed CAR-NK cells, and CREM knockout enhances effector function. In that study, CREM-regulated genes included exhaustion/stress-associated targets such as BTG1, DUSP2, SMAD3, NFKB2, and RGS1, whereas cytotoxic genes such as GZMB, GZMA, PRF1, and IFNG were more accessible or upregulated after CREM los [11, 119]. IL-15 robustly activates the PI3K/AKT/mTOR pathway, which directly governs metabolic processes and promotes cell survival by upregulating anti-apoptotic genes like Bcl-2. This pathway ensures the high metabolic demands of proliferating and active NK cells are met [120]. Conversely, continuous IL-15 exposure can lead to reduced mitochondrial respiration and decreased fatty acid oxidation, contributing to an exhausted phenotype. As recently demonstrated, IL-15 signaling rapidly induces CREM upregulation in NK cells, and CREM functions as a regulatory checkpoint that limits NK cell effector function downstream of IL-15 and activation signaling [11, 121]. Functionally, CREM deletion enhances NK cell effector activity, supporting the interpretation that CREM acts as a brake on persistence-associated NK cell programs rather than as a speculative metabolic enhancer [11]. These data support a model in which CREM restrains NK cell function, and its induction may contribute to dysfunction under sustained activating signals [122]. The balance between activation and CREM-mediated restraint appears to be influenced by IL-15 and CAR signaling, which together induce CREM and shape NK cell functional output [11, 123].

IL-15 is indispensable for enhancing NK cell cytotoxicity and driving their proliferation. It promotes the expression of cytolytic molecules and sustains proliferative expansion, which is essential for mounting an effective anti-tumor response [124]. However, the persistence of NK cells and their cytotoxic potential are significantly compromised under conditions of exhaustion, where continuous stimulation can lead to cell cycle arrest and reduced functional output. CREM interacts with IL-15 signaling at multiple levels, because both CAR activation and IL-15 rapidly induce CREM in NK cells [11, 125]. Functionally, CREM limits NK cell effector capacity, and CREM deletion increases anti-tumor effector function. These findings indicate that CREM influences the transcriptional program governing NK cell activation and dysfunction, thereby affecting overall functional persistence [11, 126]. Differential CREM induction in response to IL-15 and CAR signaling provides an experimentally supported mechanism linking activation to dysfunction in NK cells and helps explain the persistence–exhaustion paradox [11, 127].

Decoding the paradox: noncoding RNAs (ncRNAs) modulating the CREM–IL-15 axis

The emerging role of ncRNAs in orchestrating complex biological processes, coupled with the critical regulatory nature of the CREM–IL-15 axis, suggests a powerful intersection where ncRNAs could serve as master regulators of NK cell persistence and exhaustion [128]. Identifying microRNAs (miRNAs) or long noncoding RNAs (lncRNAs) that directly target CREM mRNA for degradation or repress its translation could alleviate its inhibitory effects [129, 130]. For instance, a specific miRNA upregulated upon NK cell activation might transiently suppress CREM expression, allowing for robust IL-15 signaling and sustained anti-tumor activity. Conversely, a lncRNA could sequester RNA-binding proteins essential for CREM stability or function, indirectly downregulating its activity [131]. Modulating such ncRNAs could represent a strategy to keep CREM expression in check during immunotherapy. Prolonged antigen exposure or chronic cytokine stimulation in the tumor microenvironment could lead to the differential expression of specific ncRNAs that drive NK cells into exhaustion [132].

miR-181a/b is a known positive regulator of NK cell function and metabolism. While its direct link to CREM in NK cells needs validation, in T cells, miR-181a targets several negative regulators of TCR signaling. A miRNA such as miR-181a could be tested as a regulator of CREM, but the present study support specifically establishing CREM as an IL-15-responsive checkpoint in NK cells rather than identifying upstream miRNA regulators [11, 133]. Perform miRNA-seq on persistent vs. exhausted NK cells. Identify miRNAs downregulated in exhaustion. Validate direct targeting of the CREM 3'UTR by luciferase reporter assays and Ago2-CLIP. Overexpression of these miRNAs should rescue NK cell function under chronic IL-15 exposure [134]. miR-15a/16–1 cluster are well-known tumor suppressors that target multiple oncogenes and also STAT3 (a node in cytokine signaling). Their overexpression could potentially dampen excessive IL-15/STAT5 signaling, preventing overactivation and exhaustion. miR-155 induced by IL-15 and generally promotes NK cell fitness and IFN-γ production [135]. However, its dysregulation could contribute to imbalance. It may indirectly influence the axis by targeting SOCS1 (a suppressor of cytokine signaling), thus potentiating IL-15 signals [136].

lncRNAs act as scaffolds, guides, or decoys to modulate gene expression in complex ways. A specific lncRNA is induced by chronic IL-15 and serves as a scaffold to recruit transcription factors (e.g., CREB/CREM) or chromatin modifiers to the CREM promoter, enhancing its expression. Lnc-Exhaust (a hypothetical name) is upregulated in exhausted NK cells [137]. It acts as a molecular sponge for miRNAs that target CREM (e.g., the putative miR-181a), sequestering them and allowing CREM expression to increase (a competing endogenous RNA, ceRNA, mechanism) and recruits histone acetyltransferases (HATs) to the CREM promoter, opening chromatin and facilitating its transcription. Use CRISPRi/CRISPRa to knock down or overexpress candidate lncRNAs and assess CREM and CD25 expression, STAT5 phosphorylation, and NK cell function [138, 139].

circRNAs are highly stable ncRNAs that often function as efficient miRNA sponges. A circRNA derived from a gene in the IL-15 signaling pathway (e.g., STAT5B or JAK3) sponges miRNAs that promote NK cell exhaustion [140, 141]. Circ-IL15Rα (a circRNA originating from the IL15RA gene) could sponge miRNAs that target positive effectors of the pathway (e.g., miRNAs that repress STAT5 or IL2RG (common γ-chain)). By absorbing these "exhaustion-promoting" miRNAs, Circ-IL15Rα would reinforce IL-15 signaling and promote persistence [142]. Use circRNA-enriched RNA-seq and bioinformatic analysis to find circRNAs correlated with a persistent phenotype. Perform FISH to confirm localization and RIP assays to identify bound miRNAs [143].

Cellular dynamics: persistence versus exhaustion in the tumor microenvironment

Adaptive responses to chronic IL-15 exposure

NK cells exhibit adaptive responses to their cytokine milieu, including prolonged exposure to IL-15. While IL-15 is fundamental for NK cell homeostasis and activation, continuous or excessive stimulation can elicit complex adaptations that ultimately lead to functional impairment. This phenomenon is particularly relevant in therapeutic contexts where sustained cytokine delivery is intended to bolster anti-tumor immunity [144]. NK cells continuously treated with IL-15 demonstrate decreased viability, diminished signaling, and reduced tumor control, alongside altered metabolic profiles [145]. This suggests that NK cells, in an attempt to cope with chronic hyper-stimulation, undergo adaptive changes that can paradoxically result in a less effective state. The precise molecular switches driving these adaptive responses, particularly how they diverge from beneficial activation to detrimental exhaustion, are areas of intense investigation [146]. Public single-cell resources also strengthen this interpretation. A large NK cell atlas integrating approximately 225,000 NK cells from 718 donors provides a standardized reference for NK cell states across healthy tissues and tumors, and the 2025 CREM study used public scRNA-seq datasets from TISCH2 and other public cohorts to show broad CREM expression in tumor-infiltrating NK cells, with higher CREM expression than in PBMC NK cells. Together, these data support the view that CREM is enriched in chronically stimulated or dysfunctional NK cell states, even though they do not by themselves prove that IL-15 is the sole upstream driver in every context [11, 147].

Epigenetic mechanisms, including DNA methylation, histone modifications, and noncoding RNAs, are crucial regulators of gene expression and cellular identity. In the context of chronic IL-15 exposure, NK cells undergo a CREM-associated regulatory program that links activation to dysfunction, indicating that prolonged signaling can reshape NK cell state [11, 148]. Such changes could lock NK cells into specific transcriptional programs, altering their responsiveness to subsequent stimuli or their inherent functional capacity [149]. For example, persistent inflammation, as observed in models of chronic cigarette smoke exposure, can result in the sustained presence of innate NK cells, suggesting epigenetic adaptations that contribute to their long-term presence [150, 151]. Epigenetic modifications could regulate the accessibility of genes involved in effector functions (e.g., perforin, granzymes, IFN-γ) or inhibitory pathways (e.g., PD-1, TIGIT), thereby defining the NK cell's long-term functional phenotype [152].

NK cell persistence is not merely the absence of exhaustion but involves the active maintenance of viability, proliferative potential, and effector functions. Molecular checkpoints that govern persistence likely involve a network of pro-survival pathways and mechanisms that mitigate cellular stress [153]. The mTOR pathway, activated by IL-15, is a significant checkpoint for sustaining NK cell metabolic fitness and overall function. Furthermore, the balance of activating and inhibitory receptor signaling, as well as the engagement of co-stimulatory molecules like CD137, can contribute to sustained NK cell activity and expansion [154]. SOCS2, a suppressor of cytokine signaling, also modulates IL-15-primed NK cell function by interacting with Pyk2, highlighting internal regulatory mechanisms. Identifying and manipulating these molecular checkpoints is crucial for designing strategies to enhance NK cell persistence in therapeutic settings [155].

Exhaustion phenotype in NK cells: defining features

NK cell exhaustion represents a state of functional anergy, characterized by distinct phenotypic and transcriptional alterations. Phenotypically, exhausted NK cells often exhibit altered expression of activating and inhibitory receptors [156]. For example, intratumoral NK cells in soft tissue sarcoma showed significantly higher expression of TIGIT, an inhibitory receptor, alongside activation markers, suggesting a co-expression pattern indicative of exhaustion [157]. Other commonly associated markers include programmed cell death protein 1 (PD-1), lymphocyte activation gene-3 (LAG-3), and T-cell immunoglobulin and mucin domain-3 (TIM-3), although their expression patterns vary depending on the context of exhaustion (e.g., chronic viral infection versus tumor microenvironment) [158]. Transcriptomically, exhausted NK cells display a distinct gene expression signature that reflects their impaired function. This can include downregulation of genes associated with cytotoxicity (e.g., perforin, granzymes) and cytokine production (e.g., IFN-γ, TNF-α) [159]. Studies on continuous IL-15 exposure leading to NK cell exhaustion observed a cell cycle arrest gene expression pattern, diminished signaling, and reduced mitochondrial respiration profiles characterized by a decrease in spare respiratory capacity dependent on fatty acid oxidation. These changes reflect a metabolic reprogramming that limits the cell's capacity for sustained effector function [160, 161].

The tumor microenvironment (TME) significantly contributes to NK cell exhaustion through suppressive mechanisms, and these signals may intersect with the CREM–IL-15 axis now shown to regulate NK cell function. Tumors release immunosuppressive cytokines, such as IL-10 and IL-27, which can directly inhibit NK cell function [11, 162, 163]. For instance, IL-27 receptor signaling has been shown to suppress NK cell activity in hepatocellular carcinoma, promoting tumor development. Similarly, IL-33/ST2 signaling in squamous cell carcinoma can impair NK cytotoxic activity [162]. Tumor cells can also upregulate inhibitory ligands, such as PD-L1, which engage PD-1 on immune cells, leading to functional suppression. Although the current study does not directly test TME-derived suppressive factors, it shows that CREM is induced by activating signals and functions as a checkpoint in NK cells, providing a mechanistic basis for future studies of TME–CREM interactions [11, 164]. This could indirectly affect the NK cell's ability to respond optimally to IL-15 stimulation. For example, if TME-induced stress signals activate pathways that favor CREM repressor expression, it could counteract the pro-survival and proliferative signals of IL-15, leading to compromised NK cell persistence [165].

Therapeutic targeting of the CREM–IL-15 axis in NK cell-based immunotherapy

Strategies to enhance NK cell persistence via CREM modulation

Modulating CREM to enhance NK cell persistence is now supported by direct experimental evidence showing that CREM deletion improves NK cell effector function. Given that CREM is a checkpoint in NK cells downstream of IL-15 and CAR signaling, targeted interventions may shift the NK cell transcriptional program toward a more persistent and functional state [11, 166]. Pharmacological approaches aimed at modulating CREM activity would involve identifying small molecules that selectively enhance the activity of activating CREM isoforms or inhibit repressive ones. Such interventions could influence the transcriptional program that governs NK cell effector function and resistance to dysfunction. For instance, if specific kinases phosphorylate activating CREM isoforms to promote persistence, agonists of these kinases could be explored [11, 167]. Conversely, if certain phosphatases or proteasomal degradation pathways specifically target activating CREM isoforms in an exhausted state, inhibitors could be developed. The challenge lies in achieving isoform-specific modulation without inducing widespread off-target effects, given the broad expression and diverse roles of CREM across different cell types [168]. Genetic engineering offers a more precise avenue for modulating CREM within NK cells. This could involve overexpression of activating CREM isoforms or knockdown/inhibition of repressive isoforms (e.g., ICER) specifically in adoptively transferred NK cells [169]. Technologies such as CRISPR/Cas9 or RNA interference could be employed to achieve this. Engineering NK cells to reduce CREM activity could enhance resistance to exhaustion signals encountered in the tumor microenvironment, consistent with the observed benefit of CREM deletion [11, 170]. This approach would necessitate a thorough understanding of the specific CREM isoforms involved in NK cell persistence and exhaustion, as well as the safety and immunogenicity of such genetically modified cells in a clinical setting [171].

Optimizing IL-15 delivery and receptor engagement

Optimizing IL-15 delivery is paramount to maximize NK cell therapeutic efficacy while mitigating the risk of exhaustion. The unique trans-presentation mechanism of IL-15 suggests that not only the dose but also the presentation format significantly impacts NK cell activation and persistence. The dose and schedule of IL-15 administration are critical [172]. Continuous IL-15 exposure has been shown to induce NK cell exhaustion, characterized by decreased viability, signaling, and tumor control. Conversely, intermittent dosing strategies or controlled-release formulations may reduce exhaustion by limiting prolonged activation, a concern consistent with the fact that IL-15 induces CREM in NK cells [173]. Various IL-15 formulations have been developed to enhance its half-life and improve therapeutic index. These include rhIL-15, IL-15 superagonists like ALT-803, and fusion proteins combining IL-15 with its receptor α-chain (IL-15Rα) or Fc domains [174]. For instance, IL-15Rα-IgG1-Fc fusion proteins mimic trans-presentation and have demonstrated enhanced NK and CD8⁺ T cell expansion with potent anti-tumor activity in preclinical models [175]. Delivery systems range from systemic intravenous infusions, which can lead to high peak concentrations and rapid clearance, to localized delivery methods or engineered cellular platforms that provide sustained, physiological presentation of IL-15 within the tumor microenvironment [176, 177].

Synthetic biology offers innovative tools for precise control over IL-15 signaling. This involves engineering cells (e.g., feeder cells, tumor cells, or even NK cells themselves) to express membrane-bound IL-15–IL-15Rα complexes, mimicking physiological trans-presentation [178]. Such engineered systems can deliver IL-15 signals in a cell contact-dependent manner and should be evaluated for their effects on CREM induction, NK cell activation, and exhaustion [179]. For example, tumor cells engineered to express IL-15/IL-15Rα led to increased tumor-infiltrating NK cells and inhibited tumor growth. Furthermore, synthetic inducible systems, such as those employing optogenetics or chemically inducible dimerization, could allow for spatiotemporal control over IL-15 production or receptor activation, enabling researchers to fine-tune the signal strength and duration to promote persistence and avoid exhaustion. This level of control offers the potential to precisely modulate NK cell responses for optimal therapeutic outcomes [180, 181].

Combinatorial approaches: overcoming exhaustion and resistance

Overcoming NK cell exhaustion and resistance in the tumor microenvironment often necessitates combinatorial therapeutic strategies that target multiple immunosuppressive pathways simultaneously. Immune checkpoint inhibitors (ICIs), such as those targeting PD-1/PD-L1, have revolutionized cancer therapy by reactivating exhausted T cells [182]. While NK cells can express PD-1, the primary impact of checkpoint blockade on NK cells may be indirect, by enhancing T cell-mediated anti-tumor responses and altering the TME. However, studies show direct upregulation of PD-L1 on tumor cells, which can protect them from NK cell cytotoxicity [183]. Combining IL-15-based NK cell activation with checkpoint blockade holds promise. For instance, in soft tissue sarcoma, combined IL-15 stimulation and TIGIT blockade significantly increased cytotoxicity against tumor targets [184]. This suggests that while IL-15 boosts NK cell numbers and activation, checkpoint blockade can simultaneously “release the brakes” on these cells, allowing them to function more effectively in the suppressive TME. Such combinatorial strategies aim to enhance both the quantity and quality of anti-tumor NK cell responses [185].

Translating these findings into clinical practice requires careful consideration of safety and efficacy. High doses of systemic IL-15 can lead to toxicities, necessitating strategies like modified IL-15 variants (e.g., RLI) that exhibit superior activity with potentially reduced systemic side effects [186]. The development of genetically modified NK cells, whether for CREM modulation or enhanced IL-15 signaling, demands rigorous safety assessments, including potential off-target effects, immunogenicity, and persistence of the engineered cells [187]. Ensuring controlled expansion and activation of NK cells while preventing overstimulation and exhaustion is a delicate balance [188]. Clinical trials are evaluating various IL-15-based therapies, alone or in combination, to define optimal dosing, scheduling, and delivery methods that maximize therapeutic benefit while minimizing adverse events. The integration of detailed immune monitoring in these trials is critical to understand the dynamic interplay of NK cell states and inform future therapeutic design [189, 190].

CRISPR-Cas9 knockout of CREM

Despite early successes, particularly in hematological malignancies, the broader application of NK cell immunotherapy has been challenged by several key hurdles. Ex vivo expanded NK cells often exhibit poor persistence in vivo, failing to engraft or expand sufficiently to maintain sustained anti-tumor responses [191]. Their trafficking to solid tumor sites can be inefficient, and once within the TME, they encounter a formidable array of immunosuppressive factors, including inhibitory cytokines (e.g., Transforming growth factor beta (TGF-β)), metabolic reprogramming, and checkpoint molecules, all of which conspire to dampen NK cell activity and promote exhaustion. Overcoming these limitations necessitates innovative approaches that enhance NK cell fitness, longevity, and resistance to suppression [192].

The advent of CRISPR-Cas9 gene editing technology provides an unprecedented level of precision and efficiency for targeted genetic modification, making it an ideal tool for implementing this strategy. CRISPR-Cas9 systems utilize a guide RNA (sgRNA) to direct the Cas9 nuclease to a specific genomic locus, where it induces a double-strand break (DSB) [193]. The subsequent cellular repair mechanisms, primarily non-homologous end joining (NHEJ), often lead to small insertions or deletions (indels) that disrupt the gene's open reading frame, resulting in a functional gene knockout [194]. For NK cell modification, CRISPR-Cas9 can be delivered using various methods, including viral vectors (e.g., adeno-associated virus (AAV), lentivirus) or non-viral approaches such as electroporation of Cas9 ribonucleoprotein (RNP) complexes (Cas9 protein + sgRNA) [195]. RNP delivery offers the advantage of transient expression, reducing the risk of off-target effects and avoiding integration into the host genome, which is particularly appealing for therapeutic applications [196]. The feasibility of using CRISPR-Cas9 to engineer NK cells has been demonstrated in several studies, primarily for the introduction of CARs or the knockout of inhibitory receptors, providing a strong precedent for its application in CREM knockout [135].

The isolated NK cells are genetically modified using CRISPR-Cas9 to disrupt the CREM gene. Optimized RNP electroporation protocols are likely preferred to ensure efficient knockout with minimal toxicity and off-target activity [197]. Successfully modified CREM-KO NK cells are validated for gene knockout efficiency and maintained ex vivo in the presence of feeder cells or stimulatory cytokines (including IL-15) to achieve sufficient numbers for therapeutic infusion. The expanded CREM-KO NK cells are then reinfused into patients undergoing immunotherapy. Preclinical validation is the crucial next step [198]. In vitro studies would need to rigorously confirm CREM knockout and meticulously characterize its functional consequences in NK cells. This would involve assessing NK cell proliferation, survival, activation marker expression, cytokine production (especially IFN-γ), and cytolytic activity against various tumor cell lines, both in basal conditions and, critically, in response to varying concentrations of IL-15. Transcriptomic and proteomic analyses would further elucidate the specific genes and pathways modulated by CREM knockout, particularly those directly or indirectly related to IL-15 signaling [199].

Subsequent in vivo studies using relevant immunocompetent mouse models of cancer would evaluate the anti-tumor efficacy, in vivo persistence, trafficking capabilities, and overall safety profile of CREM-KO NK cells. Looking ahead, CREM-KO NK cells could be synergistically combined with other therapeutic modalities [200]. For instance, combining CREM-KO with CAR-NK technology could lead to highly potent, tumor-specific NK cells with enhanced persistence and resistance to exhaustion. Similarly, pairing CREM-KO NK cells with IL-15 superagonists or checkpoint inhibitors could elicit more profound and durable anti-tumor responses [201]. Translational challenges, including establishing robust GMP-compliant manufacturing protocols for gene-edited NK cells and navigating regulatory pathways, will be paramount for bringing this innovative strategy to clinical trials [202].

Comparative assessment, limitations, contradictions, and knowledge gaps

The current body of evidence concerning NK cell immunotherapy, particularly centered on IL-15, presents a compelling yet complex picture. Preclinical studies consistently demonstrate IL-15's robust capacity to promote NK cell proliferation, survival, and enhanced cytotoxicity across various tumor models [203]. Animal models have shown that IL-15/IL-15Rα trans-presentation can inhibit tumor growth by increasing tumor-infiltrating NK cells and CD8⁺ T cells. Similarly, engineered IL-15 variants, such as RLI, have exhibited anti-metastatic properties and superior NK cell activation in mouse models [204, 205]. These preclinical data provide a strong mechanistic foundation for IL-15's therapeutic potential. Translational and early clinical data largely corroborate these findings, with IL-15 infusions leading to significant expansion and activation of NK cell populations in cancer patients [206]. For instance, rhIL-15 enhanced autologous NK cell stimulation and B leukemic cell depletion in CLL patients. IL-15 receptor stimulation also boosted the functionality of ascites-derived NK cells from ovarian carcinoma patients [207]. However, a critical divergence between preclinical optimism and clinical reality emerges with the observation of NK cell exhaustion following continuous IL-15 exposure. This highlights that while IL-15 is a potent activator, its application requires careful optimization to prevent detrimental overstimulation, a challenge less acutely observed or reported in early preclinical models designed primarily to demonstrate initial efficacy. The disparity underscores the need for more nuanced preclinical models that simulate chronic stimulation and TME-induced suppression [208].

Despite advancements, significant limitations and knowledge gaps persist in understanding the CREM–IL-15 axis and the persistence–exhaustion paradox in NK cells. A major remaining gap is not the existence of a CREM checkpoint per se, but how broadly this mechanism generalizes beyond activated CAR-NK/IL-15 cells to native NK cells, other cytokine contexts, and non-engineered therapeutic settings [11, 204]. Rafei et al. (2025) resolve several central mechanistic questions in CAR-NK/IL-15 models, but whether the same transcriptional and epigenetic program governs resting NK cells, tissue-resident NK cells, or chronic physiological IL-15 exposure remains unresolved [11, 209]. While CREM's regulatory functions are established in other immune cells, its precise molecular mechanisms and downstream targets in NK cells remain largely unexplored within the provided literature [210]. Although continuous IL-15 exposure is linked to exhaustion, the exact molecular switches and comprehensive transcriptional and epigenetic programs that drive NK cells from an activated to an exhausted state are not fully elucidated. The precise metabolic shifts and how they link to functional decline require deeper investigation [211]. The conflicting outcomes of IL-15 (activation vs. exhaustion) highlight a critical need to define optimal dosing strategies, formulations, and delivery methods that promote sustained activity without inducing exhaustion. Current clinical approaches are still refining these parameters [212]. The tumor microenvironment's multifaceted immunosuppressive mechanisms (e.g., PD-L1 expression, inhibitory cytokines like IL-10 and IL-27) profoundly affect NK cell function. The specific interactions between these suppressive pathways and the IL-15 signaling cascade, as well as their potential influence on CREM activity, are not fully mapped [213]. There is a lack of reliable biomarkers to predict which patients will respond to IL-15-based therapies or to monitor the onset of NK cell exhaustion in real time within clinical settings. This hinders personalized therapeutic approaches [214]. These limitations collectively underscore the complexity of NK cell immunobiology and the necessity for integrated, multi-omic investigations to unravel the intricate regulatory networks governing their persistence and functional state [215, 216].

Implications for personalized NK cell therapies

Understanding the CREM–IL-15 axis and the mechanisms underlying the persistence–exhaustion paradox holds significant implications for the development of personalized NK cell therapies. The observed variability in NK cell responsiveness to IL-15 and susceptibility to exhaustion suggests that a "one-size-fits-all" approach to NK cell immunotherapy may be suboptimal [217]. Identifying preexisting NK cell phenotypes or genetic predispositions (e.g., variations in IL-15 receptor expression or CREM pathway components) that predict responsiveness to IL-15 or susceptibility to exhaustion. This could involve profiling patient-derived NK cells for specific exhaustion markers like TIGIT, which correlate with outcome in soft tissue sarcoma [218]. Adjusting IL-15 dosing, frequency, and formulation based on individual patient characteristics and real-time monitoring of NK cell activation and exhaustion markers. This could involve intermittent IL-15 treatment to avoid the detrimental effects of continuous exposure [219]. Genetically modifying adoptively transferred NK cells to resist exhaustion. This might involve enhancing pro-persistence pathways (e.g., modulating mTOR signaling via IL-15) or, hypothetically, overexpressing activating CREM isoforms, or downregulating inhibitory receptors, to improve their functional longevity within the TME [220]. Rational design of combinations with other immunomodulatory agents (e.g., checkpoint inhibitors) to counteract specific suppressive factors in a patient's tumor microenvironment. Ultimately, tailoring NK cell immunotherapeutic approaches to the unique immune landscape of each patient has the potential to significantly enhance treatment efficacy and achieve more durable responses [221].

Challenges in modulating the CREM–IL-15 axis for durable responses

Modulating the CREM–IL-15 axis for durable NK cell responses faces several challenges. The existence of numerous CREM isoforms with antagonistic functions presents a challenge. Targeting specific isoforms to promote persistence without disrupting other vital cellular processes or inducing unintended side effects is complex [222]. A comprehensive understanding of the specific CREM isoforms expressed in NK cells and their precise roles in response to IL-15 and TME signals is needed. IL-15 is essential for NK cell function, yet continuous high-dose exposure leads to exhaustion [223]. Achieving the optimal balance of IL-15 signaling to drive robust activation without triggering exhaustion is a critical challenge. This requires precise control over IL-15 bioavailability and receptor engagement, which current delivery methods are still refining. NK cell populations are heterogeneous, with distinct subsets exhibiting varied responses to stimulation and different susceptibilities to exhaustion [224]. Developing strategies that uniformly enhance persistence across these subsets or specifically target desirable populations remains difficult. The TME is highly dynamic and can adapt to immune pressures, upregulating immunosuppressive factors (e.g., PD-L1, IL-10, IL-27) that counteract NK cell activity [225]. Strategies targeting the CREM–IL-15 axis must account for this plasticity and be robust enough to overcome evolving suppressive mechanisms. Systemic modulation of key signaling pathways like IL-15 and potentially CREM carries the risk of off-target effects on other immune or non-immune cells, leading to systemic toxicities [226]. Developing highly specific and localized interventions is crucial. Addressing these challenges demands an integrated research approach, combining detailed molecular biology with advanced genetic engineering and sophisticated clinical trial designs [227].

Future prospectives in molecular targeting of the CREM–IL-15 axis

Future research efforts should focus on defining how CREM and IL-15 together govern the NK cell persistence–exhaustion paradox and how this axis can be therapeutically targeted. Direct, comprehensive investigation into CREM expression and function in NK cells is warranted, because CREM has now been shown to act as a regulatory checkpoint downstream of IL-15 and CAR signaling [11, 228]. This includes identifying specific upstream signals that regulate CREM expression and its downstream transcriptional targets in NK cells, particularly in response to varying IL-15 concentrations and TME cues. Deeper multi-omic profiling (transcriptomics, epigenomics, metabolomics) of NK cells undergoing exhaustion in response to chronic IL-15 or TME suppression will reveal precise molecular signatures [229]. This will aid in identifying novel biomarkers for early detection of exhaustion and potential therapeutic targets. Developing advanced IL-15 formulations and delivery platforms that ensure sustained, physiological signaling without inducing exhaustion. This includes further optimizing IL-15 superagonists and engineered cell-based delivery systems that mimic trans-presentation, exploring controlled-release technologies, and evaluating intermittent dosing schedules [230]. Given the link between metabolic dysfunction and exhaustion, exploring therapeutic strategies that enhance NK cell metabolic fitness, such as modulating fatty acid oxidation or glucose metabolism, could be beneficial [231]. Systematically evaluating synergistic combinations of IL-15-based therapies with checkpoint inhibitors, TME-modulating agents (e.g., those targeting IL-27 or PD-L1), or metabolic pathway modulators to achieve robust and durable NK cell responses [232]. Implementing rigorous clinical trials that integrate comprehensive immune monitoring to correlate molecular signatures of NK cell state with clinical outcomes, paving the way for personalized NK cell immunotherapies [233].

Conclusion

The persistence–exhaustion paradox represents a central challenge in maximizing the therapeutic efficacy of NK cell-based immunotherapies. This review underscores that while IL-15 is an indispensable cytokine for NK cell development, survival, and robust anti-tumor function, its chronic or excessive presence paradoxically leads to functional exhaustion, characterized by reduced viability, impaired signaling, and compromised tumor control. The unique trans-presentation mechanism of IL-15 and its downstream activation of JAK/STAT5 and PI3K/AKT/mTOR pathways are critical for driving NK cell proliferation and metabolic fitness. However, continuous stimulation can induce metabolic shifts, such as reduced fatty acid oxidation, contributing to the exhausted state. Recent evidence supports a direct role for CREM as a regulatory checkpoint in activated CAR-NK cells downstream of IL-15 and CAR-associated signaling, strengthening the rationale for targeting the CREM–IL-15 axis in engineered NK cell therapies while leaving open whether the same checkpoint governs native NK cell persistence and exhaustion. The tumor microenvironment further exacerbates exhaustion through suppressive cytokines and inhibitory ligands, necessitating combinatorial strategies to sustain NK cell activity. Additionally, understanding patient-specific variations in the CREM–IL-15 axis may facilitate personalized immunotherapy regimens. Also, translating these molecular insights into safe and effective therapeutic interventions could significantly advance the field of NK cell-based cancer immunotherapy, overcoming current limitations posed by the persistence–exhaustion paradox (Table 1).

Table 1.

Therapeutic strategies for modulating the CREM–IL-15 axis

Strategy type	Specific approach	Expected outcome	Current status/challenges
CREM inhibition	Genetic Knockdown/Out (CRISPR/shRNA) of Repressive CREM Isoforms	Enhanced IL-15 responsiveness, prolonged NK cell persistence, sustained anti-tumor cytotoxicity	Preclinical (in vitro/in vivo). Challenges: efficiency in primary NK cells, off-target effects, delivery for in vivo application [175]
	Small Molecule Inhibitors of CREM	Similar to genetic modification; systemic applicability	Early discovery/development. Challenges: specificity for CREM isoforms, avoiding off-target transcription factor inhibition [181]
	Peptide-based Inhibitors	Interference with CREM activity/binding	Highly experimental. Challenges: delivery, stability, specificity [183]
IL-15 optimization	Pulsed IL-15 Dosing Regimens	Maximized NK cell activation while minimizing CREM induction and exhaustion	Clinical trials for IL-15 are ongoing; optimization of dosing schedules is a continuous effort [185]
	IL-15 Superagonists / Fusion Proteins	Enhanced and sustained IL-15 signaling	Some (e.g., N-803) in clinical trials. Challenges: systemic toxicity, potential for CREM induction [186]
Combinatorial approach	IL-15 Superagonist + CREM Inhibitor	Synergistic effect: potent IL-15 stimulation coupled with removal of the negative feedback loop	Preclinical exploration. Challenges: drug combination toxicity, pharmacokinetic compatibility [187]
	Epigenetic Modulators Targeting CREM	Indirect control of CREM expression	Early conceptual stage. Challenges: specificity, broad epigenetic effects [188, 190]

Abbreviations

NK

Natural killer

rhIL-15

Recombinant human interleukin-15

CLL

Chronic lymphocytic leukemia

CREM

Cyclic AMP-responsive element modulator

NKG2D

Natural killer group 2, member D

KIRs

Killer cell Immunoglobulin-like Receptors

MHC

Major histocompatibility complex

IFN-γ

Interferon gamma

TNF-α

Tumor necrosis factor alpha

CMV

Cytomegalovirus

trNK

Tissue-resident NK

ITAM

Immunoreceptor tyrosine-based activation motif

PLC-γ

Phospholipase C-gamma

PI3K

Phosphatidylinositol 3-kinase

MAPK

Mitogen-activated protein kinase

ADCC

Antibody-dependent cellular cytotoxicity

ITIM

Immunoreceptor tyrosine-based inhibitory motif

Th1

T helper 1

bZIP

Basic leucine zipper

cAMP

Cyclic AMP

CREs

CAMP response elements

KID

Kinase-inducible domain

PKA

Protein kinase A

ICER

Inducible cAMP early repressor

TADs

Transactivation domains

CBP

CREB-binding protein

LZ

Leucine zipper

GPCRs

G protein-coupled receptors

AC

Adenylyl cyclase

HAT

Histone acetyltransferase

CaMKII

Calcium/calmodulin-dependent protein kinase II

RSK

Ribosomal S6 kinase

GSK-3

Glycogen synthase kinase-3

DCs

Dendritic cells

γc

Gamma chain

JAK

Janus kinase

STAT

Signal transducer and activator of transcription

mTOR

Mammalian target of rapamycin

SH2

Src homology 2

Mcl-1

Myeloid cell leukemia 1

PIP2

Phosphatidylinositol (4,5)-bisphosphate

PIP3

Phosphatidylinositol (3,4,5)-trisphosphate

PH

Pleckstrin homology

FoxO

Forkhead box O

mTORC1

MTOR complex 1

SOCS

Suppressors of cytokine signaling

ncRNA

Noncoding RNA

lncRNAs

Long noncoding RNAs

miRNAs

MicroRNAs

PD-1

Programmed cell death protein 1

LAG-3

Lymphocyte activation gene-3

TIM-3

T-cell immunoglobulin and mucin domain-3

TME

Tumor microenvironment

ICIs

Immune checkpoint inhibitors

TGF-β

Transforming growth factor beta

DSB

Double-strand break

NHEJ

Non-homologous end joining

AAV

Adeno-associated virus

RNP

Ribonucleoprotein

TIGIT

T cell immunoglobulin and ITIM domain

Authors’ contributions

Amr Ali Mohamed Abdelgawwad El-Sehrawy: Conceptualization, Methodology, Investigation, Writing – Original Draft, Writing – Review & Editing, Visualization, Project Administration, Corresponding Author Mutaz Jamal Al-khreisat: Conceptualization, Validation, Writing – Review & Editing, Supervision, Resources Jalpa R. Patel: Investigation, Writing – Review & Editing, Visualization, Funding Acquisition Makhfirat Kibriyeva: Methodology, Data Curation, Writing – Review & Editing, Validation Lutfullayeva Gulnoza: Investigation, Formal Analysis, Writing – Review & Editing, Visualization Khayriddin Qosimov: Resources, Writing – Review & Editing, Validation, Project Administration Lhyb Mohammed Abdul: Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization Navin Kumar Tailor: Methodology, Software, Formal Analysis, Writing – Review & Editing, Visualization.

Funding

This research did not receive any financial support from public, commercial, or nonprofit organizations.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare that they have no competing interest.

Consent for publication

Not applicable.

Ethics approval

Not applicable.