CD55, also known as decay-accelerating factor, has historically been classified as a complement regulatory protein, a guard that prevents the body’s own complement cascade from shredding healthy tissue. It has also been implicated, in T cells, as a co-stimulatory receptor that fine-tunes adaptive immune responses. The new work, led by Lingyu Li, Zhaozhi Li, Yang Liu and colleagues under the supervision of Yufeng Wang and Jianhua Yu, reveals that on natural killer cells CD55 performs a role that is far more fundamental than previously appreciated. Rather than merely assisting activation, CD55 acts as a self-sufficient primary signal initiator, the molecular ignition switch that allows NK cells to launch their cytotoxic program upon first contact with a tumor.

The researchers found that when an NK cell first encounters a tumor, the activating receptor NKG2D engages its ligands on the malignant cell and triggers a signaling cascade that activates the transcription factor p65, a component of the NF-κB family. This NKG2D–p65 axis drives a rapid upregulation of CD55 on the NK cell surface. In other words, the very act of meeting a tumor instructs the killer cell to equip itself with the molecule it needs to kill. This inducible, self-reinforcing architecture ensures that CD55 appears exactly when and where it is most needed, at the immunological synapse where killer and target cells meet.

What happens next is the mechanistic heart of the study. Unlike its role in T cells, CD55 on NK cells directly binds CD97, a receptor that is abundantly expressed on tumor cells, in what immunologists call a trans interaction, meaning the two molecules engage each other across the junction between two different cells. This CD55–CD97 engagement triggers the aggregation of lipid rafts, the cholesterol-rich, ordered microdomains that float within the cell membrane and serve as organizing platforms for signaling machinery. Within these coalescing rafts sits LCK, the Src-family tyrosine kinase that stands at the apex of the lymphocyte activation cascade. By concentrating lipid rafts, CD55 brings LCK molecules together, permitting their phosphorylation and activation, which in turn amplifies the downstream signaling that commands the release of cytotoxic granules.

The technical elegance of this discovery lies in its demonstration that NK cells do not depend on external co-stimulation to fire their killing apparatus. Where T cells typically require multiple reinforcing signals before committing to attack, the CD55–CD97 lipid raft mechanism provides a single, autonomous activation pathway. The researchers confirmed this through careful loss-of-function experiments: NK cells engineered to lack CD55 showed impaired raft aggregation, diminished LCK activation and weakened cytotoxicity, and these defects could be traced specifically to the raft–kinase axis rather than to CD55’s classical complement-regulating duties. Methyl-beta-cyclodextrin disruption of lipid rafts abrogated the activation induced by CD55 agonists, and the functional defects in CD55-deficient NK cells proved independent of the complement pathway altogether, underscoring that this is a signaling role, not a complement role.

The story darkens when the investigators followed NK cells into the tumor microenvironment. Upon prolonged exposure to tumor cells, CD55 expression on the NK cell surface progressively declines, tracking closely with the well-known downregulation of NKG2D that occurs under chronic stimulation. This erosion of CD55 is not a benign byproduct of exhaustion. Within the tumor microenvironment, the loss of CD55 causally impairs NK cell function, collapsing the raft–LCK signaling platform that the cells depend on and leaving them unable to mount effective attacks. The very chronicity of tumor exposure, which initially instructs NK cells to upregulate CD55, ultimately strips them of the molecule and with it their killer instinct.

To test whether this biology matters in human disease, the team turned to single-cell RNA sequencing data from a remarkable breadth of cancers, including nasopharyngeal carcinoma, melanoma, lung cancer, breast cancer, head and neck squamous cell carcinoma, hepatocellular carcinoma, renal cancer, colorectal cancer and pancreatic ductal adenocarcinoma, drawing on datasets from the Human Tumor Atlas Network and numerous public repositories. The analysis revealed a sobering clinical correlation: in cancer patients, low CD55 expression in tumor-infiltrating NK cells is associated with poor clinical outcomes. CD55 thus emerges not only as a mechanistic linchpin but as a potential biomarker, a molecular gauge of whether a patient’s innate immune arsenal remains armed or has gone quiet.

The most therapeutically consequential finding, however, is that this dysfunctional state is reversible. When the researchers restored CD55 expression in both conventional NK cells and chimeric antigen receptor-engineered NK cells, the effects were striking. Restored CD55 augmented LCK signaling, enhanced effector function and persistence, and improved antitumor efficacy in vivo. Engineered NK cells fortified with sustained CD55 expression proved more lethal against tumors and survived longer within the tumor microenvironment, addressing two of the most stubborn limitations of NK cell immunotherapy: inadequate persistence and functional anergy after adoptive transfer. In mouse models and in the NK92 cell line used clinically, enforced CD55 expression translated into measurably better tumor control.

For the rapidly expanding field of NK cell-based cancer immunotherapy, these findings arrive at a critical moment. CAR-NK therapies have generated enormous enthusiasm because, unlike CAR-T cells, they carry a lower risk of graft-versus-host disease and cytokine storms, and they can be derived from off-the-shelf cell sources. Yet clinical results in solid tumors have lagged behind the promise, precisely because transferred NK cells dysfunction rapidly after entering tumor tissue. The CD55 discovery provides both an explanation and a solution: by engineering CD55 expression into therapeutic NK products, or by finding pharmacological means to preserve it, cell therapists may be able to keep the raft–LCK ignition switch flipped on throughout the cell’s journey into and within the tumor.

Beyond its immediate therapeutic implications, the study reshapes fundamental understanding of how innate immunity is wired. It reveals that NK cells possess an autonomous activation circuit in which tumor recognition, through NKG2D and p65, induces a membrane organizer that then amplifies signaling through lipid raft condensation and LCK activation. It also adds CD55 and CD97 to the growing roster of ligand–receptor pairs whose trans interactions at the immune synapse determine the outcome of encounters between immune cells and cancer. And it frames NK cell dysfunction not as an irreversible slide into exhaustion but as a defined, mechanistically understood state, one whose molecular signature, CD55 loss, can in principle be measured, monitored and corrected. As immunotherapy continues its migration from blood cancers toward solid tumors, the humble complement regulator once known simply as decay-accelerating factor may prove to be one of the most important switches in the killer cell’s arsenal, and flipping it back on could mark a turning point in the effort to make natural killer cells the reliable cancer weapons they have always promised to be.

Subject of Research: The role of the membrane protein CD55 in organizing lipid raft–LCK signaling that drives natural killer cell antitumor immunity.

Article Title: CD55 organizes lipid raft-LCK signaling to potentiate NK-cell antitumor immunity

Article References: Li, L., Li, Z., Liu, Y., Fan, W., Lei, Y., Tian, L., Chen, L., Qu, Z., Shi, Y., Yu, J., & Wang, Y. (2026). CD55 organizes lipid raft-LCK signaling to potentiate NK-cell antitumor immunity. Cell Research. https://doi.org/10.1038/s41422-026-01288-8

Image Credits: AI Generated

DOI: 10.1038/s41422-026-01288-8

Keywords: CD55, natural killer cells, LCK kinase, lipid rafts, NKG2D, tumor microenvironment, CAR-NK cells, immunotherapy, CD97, antitumor immunity, organizes, lipid

Immune cells are particularly instructive subjects for this kind of analysis because their function depends on rapid, spatially organized surface events. A T cell encountering an antigen-presenting cell reorganizes its membrane within minutes, clustering receptors and adaptor proteins into a structured interface known as the immunological synapse. B cells undergo analogous rearrangements when their B cell receptor binds antigen. Natural killer cells survey target cells and form activating or inhibitory contacts whose outcomes depend on the balance of receptor signals at the contact site. In each case, the physical state of the membrane at the interface is not incidental; it determines which proteins can diffuse into or out of the contact zone, which lipid species segregate there, and how efficiently the cytoskeleton can be remodeled to stabilize or dissolve the interaction.

The biophysical basis of membrane order lies in the composition and behavior of the lipid bilayer itself. Sphingolipids and phospholipids with saturated acyl chains pack tightly and adopt extended conformations, producing regions of high order. Unsaturated phospholipids, with kinks introduced by double bonds, disrupt packing and lower local order. Cholesterol intercalates between phospholipids and has a concentration-dependent effect: at moderate levels it rigidifies fluid bilayers and promotes the coalescence of ordered domains, while at high levels it can increase order further in saturated lipid environments. These interactions underlie the long-standing hypothesis of lipid rafts, nanoscale assemblies enriched in sphingolipids, cholesterol, and certain lipid-anchored proteins that have been proposed to serve as platforms for signaling. Direct visualization of rafts in living cells proved technically elusive for decades because the domains are small, transient, and below the diffraction limit of conventional microscopy, which fueled considerable debate about their physiological relevance.

Probe-based imaging has been central to resolving this debate. Environmentally sensitive dyes such as laurdan and its derivatives report on the hydration and packing of their lipid surroundings through shifts in their emission spectra, allowing order to be quantified as a generalized polarization value. When such probes are targeted to specific leaflets of the plasma membrane or conjugated to molecules that partition preferentially into ordered or disordered phases, they provide spatially resolved readouts of membrane physics in live cells. The interpretation of these measurements requires care, because probe partitioning can itself perturb the membrane, and spectral readouts can be confounded by factors such as pH, probe concentration, and photobleaching. Advances in probe chemistry, calibration standards, and imaging modalities have progressively addressed these concerns, making it possible to compare order measurements across cell types and experimental conditions with increasing confidence.

Super-resolution microscopy techniques have further transformed the field by bringing the relevant length scales within reach. Stimulated emission depletion microscopy, photoactivated localization microscopy, and stochastic optical reconstruction microscopy each achieve effective resolutions well below the diffraction limit, revealing that proteins and lipids once thought to be uniformly distributed actually occupy discrete nanoscale clusters. Combining these structural methods with spectral imaging of order-sensitive probes allows researchers to ask whether regions of high membrane order coincide with clusters of signaling proteins, and whether such coincidence changes upon receptor activation. In immune cells, this combination has shown that ordered domains accumulate at sites of receptor engagement and that disrupting ordered lipid phases, for example by depleting cholesterol or inhibiting sphingolipid synthesis, impairs signaling outputs such as calcium flux, phosphorylation cascades, and cytokine production.

The relationship between membrane order and the actin cytoskeleton adds another layer of regulatory complexity. Cortical actin filaments exert forces on the overlying membrane, creating regions of tension and constriction that can influence lipid phase behavior. Actin-driven structures such as membrane ruffles, microvilli, and picket-and-fence arrangements compartmentalize lateral diffusion, effectively corralling proteins and lipids into transient domains. Conversely, the lipid composition of the membrane affects how actin-binding proteins attach to the cytoplasmic face, creating a bidirectional feedback loop. In migrating immune cells, leading-edge membranes enriched in unsaturated lipids and low order support the protrusive activity needed for chemotaxis, while the uropod exhibits different physical properties that promote adhesion and retraction. Mapping order across a polarized cell therefore reveals how physical heterogeneity aligns with functional polarity.

Pathogens have evolved to exploit membrane physical properties during infection, which underscores the selective pressures shaping these systems. Enveloped viruses bud from membranes whose lipid composition facilitates assembly and release, and some viruses preferentially incorporate ordered lipid domains into their envelopes. Bacterial toxins that bind cholesterol or sphingomyelin use ordered domains as points of attachment for pore formation. Intracellular pathogens manipulate host membrane traffic and lipid metabolism to create replication niches with altered physical properties. In each scenario, the immune response must contend with a membrane environment that the pathogen has actively reshaped, and measurements of membrane order in infected cells can reveal these manipulations as measurable shifts in surface biophysics.

Aging and metabolic state also leave imprints on membrane order. Dietary lipid composition influences the saturation profile of membrane phospholipids over time, and age-associated changes in lipid metabolism have been documented in immune cells from multiple organisms. Membranes from aged T cells, for example, show altered cholesterol content and modified order characteristics that correlate with diminished signaling capacity upon antigen stimulation. Metabolic diseases such as obesity and diabetes, which alter circulating lipid profiles, produce measurable changes in the membrane properties of circulating leukocytes. These observations suggest that membrane order could serve as an integrative readout of an organism’s metabolic and inflammatory history, encoded in the physical state of its immune cell surfaces.

Therapeutically, the sensitivity of membrane order to lipid metabolism opens avenues for intervention. Statins, which reduce cholesterol synthesis, have immunomodulatory effects that may partly reflect changes in membrane organization. Drugs targeting sphingolipid metabolism, such as inhibitors of sphingomyelin synthase or glucosylceramide synthase, alter ordered domain abundance and have shown effects on inflammatory signaling. Fingolimod, a sphingosine-1-phosphate receptor modulator used in multiple sclerosis, acts in part through receptor internalization but also engages with the broader biology of sphingolipid-enriched membranes. Understanding how such agents redistribute membrane order across immune cell subsets could explain some of their off-target effects and guide the design of compounds that tune immune responses through membrane biophysics rather than direct receptor antagonism.

Methodological standardization remains an important challenge for the field. Different probes report on different aspects of membrane physics, and values obtained with one dye are not directly comparable to those from another without careful cross-calibration. Sample preparation, temperature, imaging parameters, and analysis pipelines all influence measured values, and the field has not yet converged on universally accepted reference standards. Efforts to establish standardized protocols, share calibration reagents, and report measurements in ways that facilitate comparison across laboratories will be essential if membrane order is to mature from a research measurement into a reproducible biomarker. The application of machine learning approaches to extract order-related features from large imaging datasets may also accelerate progress by identifying patterns that manual analysis would miss.

The diversity of immune cell subsets presents both an opportunity and a complication. Myeloid cells, lymphocytes, and innate lymphoid cells each maintain distinct lipidomes shaped by their developmental programs and functional demands. Within a single subset, activation state, differentiation stage, and tissue microenvironment further modify membrane composition. A dendritic cell maturing in response to pathogen-associated molecular patterns remodels its membrane as part of its transition from antigen capture to antigen presentation. Tissue-resident macrophages adapt their membrane properties to the lipid milieu of their organ of residence, which differs substantially between brain, lung, liver, and adipose tissue. Comprehensive maps of membrane order across this diversity would require systematic sampling, but the resulting atlas could reveal how physical membrane states encode functional specialization in ways that transcriptomic or proteomic measurements alone do not capture.

Looking forward, the integration of membrane order measurements with other single-cell modalities promises a more complete picture of immune regulation. Combining order imaging with live-cell reporters of signaling activity, such as fluorescent biosensors for kinase activity or calcium, would allow direct testing of causal relationships between membrane physics and signal transduction at the single-cell level. Pairing order measurements with lipidomics would connect physical readouts to their molecular determinants. Spatial transcriptomics and proteomics of tissue sections could place membrane biophysical states in their anatomical and pathological contexts. As these datasets accumulate, the plasma membrane’s physical organization may come to be recognized as a fundamental layer of cellular regulation, one that immune cells exploit with particular sophistication and one that offers distinct targets for therapeutic modulation of immunity.

Subject of Research: Plasma membrane order maps functional diversity in immune cells

Article Title: Plasma membrane order maps functional diversity in immune cells

Article References: Andronico, L. A., Gurdap, C. O., Arora, A., Ragaller, F., Sandoz, P. A., Jiang, Y., Giatrellis, S., de Boer, L. L., Carannante, V., Iskrak, S., Mikes, J., Buggert, M., Österborg, A., Önfelt, B., Klymchenko, A. S., Brodin, P., & Sezgin, E. (2026). Plasma membrane order maps functional diversity in immune cells. Nature Chemical Biology. https://doi.org/10.1038/s41589-026-02322-x

Image Credits: AI Generated

DOI: 10.1038/s41589-026-02322-x

Keywords: Plasma, membrane, order, maps, functional, diversity, immune, cells, scientific research

HLA-E belongs to the family of non-classical MHC class I molecules, sometimes called class Ib, which play an essential role in immunomodulation. Unlike its celebrated cousins HLA-A, HLA-B and HLA-C, which bristle with hundreds of variants and dominate transplant matching, HLA-E has long been considered a genetic backwater. Structurally, however, it resembles the classical molecules closely: three extracellular domains pair with beta-2 microglobulin, forming a complex that presents peptides at the cell surface. What makes HLA-E unusual is what it displays. Instead of offering a wide menu of foreign peptides, it primarily binds self-peptides derived from the leader sequences of classical class I molecules, effectively monitoring whether other HLA molecules are being produced normally. When viruses or tumours suppress classical HLA expression, HLA-E levels at the surface drop, and natural killer cells, which survey the body through activating and inhibitory CD94/NKG2 receptors, interpret the silence as a warning. HLA-E also engages cytotoxic and regulatory CD8 T cells through their alpha-beta T cell receptors, giving it a dual role bridging innate and adaptive immunity.

For decades, only two HLA-E protein variants were known, differing by a single amino acid at position 107 of the alpha-2 domain: an arginine in HLA-E01:01 and a glycine in HLA-E01:03. That simplicity was partly an artefact of method. Most studies sequenced only exons 2 and 3, the segments encoding the peptide-binding groove, leaving the rest of the gene unexplored. As full-length sequencing technologies spread, researchers began finding more variability than expected. The IPD-IMGT/HLA database now lists 378 HLA-E alleles, though synonymous substitutions and one null allele mean these encode only 142 protein variants. Two large-scale efforts have since confirmed the richer picture: one analysed next-generation genotyping data from 2.5 million potential stem cell donors across 104 populations, and another used single-molecule real-time sequencing on 6,227 DNA samples, uncovering 86 novel alleles. Despite this hidden diversity, the two classic protein variants still account for roughly 99 percent of HLA-E worldwide, in nearly equal proportions, a pattern that has prompted suggestions of balancing selection maintaining functional differences between them. Laboratory work has shown that HLA-E01:03 achieves slightly higher cell surface expression than 01:01, owing to its marginally stronger peptide-binding affinity.

The Estonian population offered an intriguing test case. With around 1.3 million inhabitants, Estonia has been shaped by successive migration waves, geographic isolation and cultural influences, producing marked genetic structuring within the country, particularly between the south-east and the rest of the nation. Previous genomic studies have documented this regional differentiation in fine detail. Whether such structuring extends to HLA-E, however, was unknown, since the gene had been examined in Estonians only once before, through the large registry-based study that did not perform full-gene sequencing. The new research drew on the Estonian Biobank, selecting 143 DNA samples from a collection of more than 10,000 individuals born in Estonia and sampled in 2005. To ensure fair representation, the team chose equal numbers of participants from each of Estonia’s 15 counties, matched for age and gender, spanning ages 18 to 83, with 72 men and 71 women. The study was approved by the Ethics Review Committee on Human Research of the University of Tartu, and all participants gave written informed consent.

The methodological approach was deliberately thorough. Rather than targeting the familiar exons, the researchers amplified the complete HLA-E gene from the 5-prime untranslated region to the 3-prime untranslated region, capturing both coding sequences and the regulatory flanks in a single amplicon. Purified fragments were sequenced in both directions using the Sanger method, the gold standard for resolving single nucleotide variants in targeted regions. Allele frequencies were calculated directly as the number of observed alleles divided by twice the number of individuals, and regional comparisons employed chi-squared tests. To explore associations with the classical HLA genes, the team performed low-resolution typing of HLA-A, HLA-B and HLA-C using Luminex sequence-specific oligonucleotide probes, analysing the results with the PYPOP software package, which also confirmed that the HLA-E allelic distribution fitted Hardy-Weinberg equilibrium, with a P-value of 0.78 indicating no deviation from random mating expectations.

The sequencing effort yielded 16 distinct HLA-E alleles, four of which were entirely new to science: HLA-E01:01:01:51, 01:01:43, 01:01:01:53 and 01:01:01:54. For a population of this size, discovering four novel alleles is consistent with the yield of comparable full-length studies in other populations worldwide. Most of these variants differ only in non-coding or synonymous positions, so the 16 alleles boil down to just four protein variants: HLA-E01:01, 01:03, 01:06 and 01:09. Phenotype frequency calculations showed that 45 percent of Estonians carry both major variants in heterozygous form. The most common allele was HLA-E01:01:01:01/02 at a frequency of 0.441, followed by 01:03:02:01 at 0.357. Intriguingly, the peptide-binding grooves of the rarer variants mirror the common ones: HLA-E01:09 has a groove identical to 01:01 with its difference located in the alpha-3 domain, while 01:06 matches 01:03 in the groove but diverges in the alpha-3 domain.

The regional analysis produced perhaps the most conceptually interesting null result. Despite Estonia’s well-documented genetic split between the south-eastern counties of Põlva, Tartu, Valga and Võru and the rest of the country, chi-squared comparisons of high-resolution HLA-E allele frequencies found no significant differences, whether comparing the south-east with the north-east or with all other counties. Nor did comparison with neighbouring populations reveal major differences. Drawing on the registry data covering Estonians, Finns, Russians, Latvians, Belarusians and Ukrainians, the study found Estonian frequencies sitting comfortably within the European pattern, while showing the expected clear divergence from Asian populations such as the Japanese, Indonesians and Chinese. A finer single-nucleotide comparison with Finnish samples from the 1000 Genomes Project, justified by the deep historical and genetic ties between the two peoples, showed broadly similar SNP frequencies, with modest differences at two positions: the T nucleotide at position 424, characteristic of HLA-E01:03:02G, was somewhat more prevalent in Estonians, while the T at position 1857, which defines HLA-E01:06, was less common than in Finns.

The association analysis added a layer of evolutionary depth. Although the arginine and glycine variants appear in almost equal numbers overall, they are not distributed evenly across the classical HLA haplotypes. HLA-A01 travelled preferentially with arginine 107, appearing with R in 12 percent of cases but with G in only 1 percent. HLA-A03 showed the opposite tendency, pairing with glycine at 15 percent versus arginine at 2 percent, and HLA-C04 was found exclusively with glycine. Several other allele groups displayed near-exclusive associations, though small sample sizes caution against over-interpretation. Examining complete HLA-A to HLA-B to HLA-C haplotypes sharpened the pattern further: eight of the ten most frequent haplotypes showed a clear preference for one variant or the other. Notably, HLA-A01 and HLA-A*03 share an identical leader peptide sequence that could bind HLA-E, so the driver of these associations cannot be the leader peptide itself. Nor did the dimorphism in the HLA-B leader peptide, methionine versus threonine at position minus 21, show any coherent relationship with the HLA-E variants, ruling out one obvious mechanistic explanation.

The findings collectively reinforce the view that HLA-E performs a conserved function preserved across diverse genetic backgrounds. Whatever historical migrations and admixture events shaped the broader Estonian genome, they left their fingerprints everywhere except on this gene, whose variation remained stable across regions and indistinguishable from that of neighbouring populations. This conservation, set against the extreme polymorphism of the classical class I genes, continues to fuel speculation about HLA-E’s evolutionary pathway and the functional significance of the single amino acid difference that defines its two main variants. The preferential haplotype associations documented here suggest that the R107G dimorphism does not drift independently but is woven into the broader architecture of the HLA region, echoing earlier observations in Chinese and French populations. For transplant immunology, stem cell donor registries and cancer immunotherapy research, each full-length dataset of this kind sharpens the picture of a molecule that, though long overshadowed by its classical relatives, may hold keys to immune regulation that medicine is only beginning to turn.

The study employed sophisticated mass cytometry techniques to dissect the phenotypic landscape of CREM-expressing NK cells subjected to different stimulatory conditions. Researchers generated three distinct NK cell populations: IL-15 NK cells lacking chimeric antigen receptors (CARs), CAR70 NK cells without IL-15 secretion, and a hybrid subset termed CAR70–IL-15 NK cells that secrete IL-15 while expressing CAR70. These NK cells were co-cultured with UMRC3 cells, which express the CD70 antigen targeted by CAR70. This complex experimental framework provided a unique opportunity to parse out how CREM expression correlates with NK cell activation states.

Analyses revealed a clear hierarchy in CREM expression: CAR70–IL-15 NK cells exhibited the highest levels, followed by CAR70 NK cells, with IL-15 NK cells demonstrating the lowest expression. Such findings suggest a synergistic effect between CAR engagement and IL-15 stimulation in upregulating CREM. Mass cytometry further delineated six phenotypically distinct clusters of NK cells, each characterized by a unique constellation of surface markers and functional proteins. Of particular interest, clusters 1, 2, and 4 were predominantly represented in CAR70–IL-15, CAR70, and IL-15 NK cell populations, respectively, highlighting their differential activation signatures.

The cluster with the highest CREM expression, cluster 1, exhibited a dual expression pattern marked by robust activation alongside inhibitory markers. Notably, activation-associated molecules such as NKG2D, CD25, CD2, granzyme B, and perforin were elevated, indicating heightened cytotoxic potential. Simultaneously, this cluster upregulated inhibitory receptors including NKG2A, LAG3, TIGIT, and TIM3, markers often associated with immune exhaustion. This intriguing coexistence of activation and exhaustion phenotypes points toward a complex regulatory checkpoint, where CREM likely orchestrates a balance preventing excessive immune activation.

Examining clusters 2 and 4 revealed a tempered but similar phenotype. Cluster 2, enriched in CAR70 NK cells, and cluster 4, more abundant in IL-15 NK cells, maintained moderate expression of both activating and inhibitory markers, indicative of intermediate functional states. In stark contrast, cluster 5—prevalent in non-treated (NT) NK cells—displayed suppressed expression across both activation and inhibitory markers, consistent with a quiescent or suppressed phenotype.

Importantly, the association between elevated CREM expression and a combined activation–exhaustion immunophenotype was consistent across all experimental conditions. This widespread correlation underscores CREM’s pivotal role as a regulatory node in NK cell biology, intricately linked to the cells’ functional status. It appears that the induction of CREM is not merely a marker of activation but a key mediator of activation-induced exhaustion, potentially as a mechanism to temper immune responses and preserve cell viability.

Such discoveries have profound implications for CAR-NK cell therapy, an emergent strategy designed to harness the specificity of engineered receptors combined with the innate cytotoxicity of NK cells. The finding that CAR signalling synergizes with IL-15 to drive CREM expression suggests that while these stimuli boost NK cell activation, they may concomitantly trigger regulatory circuits that limit sustained immunity. Therefore, targeting CREM or modulating its downstream pathways could enhance the efficacy and durability of CAR-NK therapies by preventing premature exhaustion.

From a therapeutic design standpoint, these insights invite the development of next-generation CAR constructs or cytokine delivery systems tailored to fine-tune CREM activity. By decoding the molecular interplay that leads to activation-induced exhaustion, clinicians might improve persistence and potency in treated patients. Moreover, understanding CREM’s dualistic role allows for strategies that preserve beneficial activation while restraining deleterious overstimulation.

The study also presents a nuanced perspective on the phenotypic heterogeneity of NK cell populations under combined CAR and cytokine stimulation. The distinct clusters characterized by specific marker profiles emphasize the plasticity and dynamic nature of NK cell responses. Recognizing this heterogeneity is crucial for designing interventions that selectively expand optimal NK cell subsets or reprogram dysfunctional ones within immunotherapeutic contexts.

On a fundamental level, this research enriches our comprehension of NK cell biology by revealing how transcriptional checkpoints integrate signals from engineered receptors and endogenous cytokines. It prompts a reevaluation of CREM not only as a biomarker of activation exhaustion but as a potential target for enhancing immune system resilience against tumors. Additionally, it paves the way for investigations into how other transcription factors may interact with CREM to regulate NK cell fate decisions.

In conclusion, the discovery of CREM as a central regulatory checkpoint in CAR and IL-15 signalling pathways represents a significant advancement in the field of immune engineering. This study provides compelling evidence that manipulating CREM expression or function could unlock the next tier of efficacy in CAR-NK therapies, optimizing their anti-tumor capabilities. Continued exploration into the molecular circuitry governing NK cell activation and exhaustion promises to yield transformative approaches in cancer immunotherapy and beyond.

As the interface between cellular engineering and immunology deepens, the delicate balance unveiled by CREM’s regulatory role reminds us of the immune system’s complexity and precision. Harnessing such knowledge will be crucial in the design of sophisticated therapies that not only empower immune cells to attack malignancies but also safeguard against their premature attrition. With the advent of these insights, the potential of NK cells as versatile and potent effectors in the immunotherapeutic arsenal looks brighter than ever.

Subject of Research: Regulation of NK cell activation and exhaustion by CREM in the context of CAR and IL-15 signalling.

Article Title: CREM is a regulatory checkpoint of CAR and IL-15 signalling in NK cells.

Article References:
Rafei, H., Basar, R., Acharya, S. et al. CREM is a regulatory checkpoint of CAR and IL-15 signalling in NK cells.
Nature (2025). https://doi.org/10.1038/s41586-025-09087-8

Image Credits: AI Generated