Engineered T-BET Pushes Lab-Grown NK Cells Toward Inflammation, Not Full Maturity

Scientists at Karolinska Institutet have engineered stem cell-derived natural killer cells to overproduce a key transcription factor, and the result is a striking split in the cells’ behavior: the engineered cells pumped out far more inflammatory signaling molecules, yet they failed to acquire the full repertoire of markers that define terminally mature NK cells, and their killing capacity did not measurably improve. The study, published in Frontiers in Pharmacology, suggests that inflammatory specialization and cytotoxic maturation in these lab-grown immune cells can be pulled apart, a finding with implications for how researchers design off-the-shelf cell therapies.

Natural killer cells are innate immune effectors prized by immunotherapists because they can recognize and destroy malignant cells without prior sensitization and without requiring human leukocyte antigen matching. Expanded peripheral blood-derived NK cells currently dominate clinical applications, but they carry practical drawbacks: they depend on donors, vary between batches, and are difficult to scale. Induced pluripotent stem cell-derived NK cells, or iNK cells, offer an alternative that could in principle provide an unlimited, standardized supply. The problem is that current differentiation protocols yield cells that look immature, with low levels of terminal maturation markers such as KIRs, CD16 and CD57, and reduced cytotoxic function compared with their peripheral blood counterparts. Transcriptomic analyses have even suggested that iNK cells resemble fetal rather than fully mature NK cells.

T-BET, encoded by the gene TBX21, is a T-box transcription factor long regarded as a central regulator of NK cell differentiation. Its expression rises along the NK maturation trajectory and is associated with the acquisition of effector functions, including interferon gamma production and expression of cytotoxic molecules such as perforin and granzyme B. T-BET also cooperates with other transcriptional partners, including eomesodermin and ZEB2, to regulate terminal maturation programs. Given this track record, the Swedish team asked a direct question: could forcing supraphysiological levels of T-BET during feeder-free iNK differentiation push the cells all the way to terminal maturation?

To find out, the researchers built a doxycycline-inducible overexpression system in iPSCs using a piggyBac-based genomic integration strategy. One construct expressed the reverse tetracycline transactivator constitutively; a second, tetracycline-responsive construct carried either mCherry alone as a control or T-BET fused via a T2A sequence to mCherry, so that the fluorescent reporter marked cells with active transgene expression. Validation in iPSCs showed tight control: doxycycline induced mCherry in nearly all cells in both conditions, T-BET appeared specifically in the overexpression condition, and background expression was minimal without the drug. The team then differentiated the engineered lines into NK cells using a feeder-free, spin-embryoid-body-based protocol, adding doxycycline from the onset of NK lineage commitment. For most downstream analyses, mCherry-positive cells were selectively sorted or analyzed to ensure that only cells with active transgene expression were assessed.

The first characterization confirmed what earlier work had suggested about baseline iNK cells. Using three independent iPSC lines, the researchers compared iNK cells with peripheral blood NK cells cultured for one week under matching conditions. The frequency of CD45-positive, CD56-positive NK cells was comparable between the two cultures, but the iNK cells expressed lower levels of CD16, CD94, KIRs and CD57, along with differences in activating receptors including NKp44, NKG2D and DNAM-1. Single-cell RNA sequencing showed that iNK and pNK cells occupied distinct transcriptional spaces. Notably, the iNK-enriched clusters expressed several cytotoxic effector genes, including GNLY, PRF1 and NKG7, at levels comparable to or higher than those in pNK-enriched clusters, while maturation-associated receptors and transcription factors such as PRDM1, EOMES and ZEB2 were reduced. The iNK cells had much of the cytotoxic machinery but lacked the receptor and transcription-factor networks of mature cells.

When T-BET was overexpressed during differentiation, the effects were selective rather than sweeping. The percentage of CD45-positive, CD56-positive cells rose over time in both conditions with no significant difference between them, indicating that T-BET did not alter early NK lineage commitment. NKp46 expression was unaffected. But T-BET overexpression significantly increased the percentage of CD94-positive cells over time, so that by the late NK stage, CD94 expression in the engineered cells reached a mean of 86 percent, close to the 99 percent seen in primary NK cells and far above the 42 percent in controls. T-BET also strongly induced multiple KIRs, including KIR2DL1/S1, KIR3DL1/S1 and KIR2DL2/3, as early as week three of lineage commitment, while these receptors remained minimally expressed in controls. CD16, however, stayed low in both conditions across all time points, and CD57 remained low as well. The authors concluded that T-BET alone was insufficient to drive full terminal maturation.

Transcriptomics told a similar and more nuanced story. In the single-cell comparison at the late NK stage, iNK control and T-BET-overexpressing cells largely overlapped on the UMAP, and neither shifted toward the primary NK cell state. Differential expression analysis revealed a selective upregulation of inflammatory genes in the engineered cells: the chemokines CCL3, CCL4 and CCL5, along with granzyme A, were increased with moderate to strong log2 fold changes, and the increases were distributed across a broad fraction of cells rather than confined to a small subset. In contrast, the terminal cytotoxic genes GZMB, PRF1 and GNLY showed minimal changes, with log2 fold changes below roughly 0.5 and largely overlapping expression distributions between conditions. Interestingly, IFNG transcript levels were moderately decreased by T-BET overexpression at steady state, a point the authors return to when discussing post-transcriptional regulation. Both iNK conditions also expressed lower levels of GZMB and GNLY than primary NK cells, consistent with an incompletely mature state.

Functional assays sharpened the dissociation. Degranulation, measured by CD107a surface expression at a one-to-one effector-to-target ratio, was comparable between control and T-BET-overexpressing iNK cells across all target conditions, and both showed reduced responses against rituximab-coated Raji targets, consistent with their low frequency of CD16-positive cells, since CD16 mediates the Fc receptor-dependent recognition required for antibody-dependent cellular cytotoxicity. In chromium-51 release assays against K562 cells, cytotoxicity increased with higher effector-to-target ratios in all conditions; control iNK cells were significantly less cytotoxic than primary NK cells at the lowest ratio tested, but no significant difference separated the engineered cells from either controls or primary cells, and a two-way repeated-measures ANOVA found a significant effect of ratio but no effect of condition. The clearest functional change was in cytokine output: T-BET overexpression increased the frequency of IFN gamma-positive iNK cells after stimulation with Raji, rituximab-coated Raji and Daudi cells, and the engineered cells even exceeded primary NK cells in IFN gamma positivity after those same stimulations. A numerical increase after K562 stimulation did not reach statistical significance.

The authors are candid about the limits of their study. The transcriptomic analysis was based on a single iNK differentiation and one primary NK cell donor, so the cell-level differences should be read as descriptive rather than as statistically replicated population-level effects, although phenotypic and functional data across independent differentiations supported the transcriptomic observations. The in vitro system does not fully recapitulate the microenvironmental cues of human NK ontogeny, the signaling pathways downstream of T-BET were not assessed, effects on cytokines other than IFN gamma, including TNF alpha, remain unexplored, and variability in the chromium-release assay means small effects on cytotoxicity cannot be completely excluded. The engineered populations were generated from a single iPSC line, limiting generalizability, and transposon-mediated integration leaves open possible differences in integration sites or copy numbers. The team also notes that IFN gamma transcript levels were not increased at steady state despite T-BET’s known role in directly activating IFN gamma transcription, which is consistent with T-BET enhancing cytokine production through mechanisms beyond transcriptional activation, possibly involving post-transcriptional regulation.

From a translational standpoint, the authors argue that T-BET-driven iNK cells represent a functionally specialized inflammatory NK state rather than a purely cytotoxic product. The strong increase in IFN gamma and inflammatory chemokine production could, they suggest, promote immune-cell recruitment, antigen presentation and T-cell activation within a tumor microenvironment, potentially complementing immune checkpoint blockade or T cell-based therapies, though these therapeutic scenarios remain speculative and were not tested experimentally. More broadly, the study supports the idea that NK-cell engineering can be tailored toward distinct functional objectives, whether direct cytotoxicity or inflammatory immune modulation, rather than a single mature endpoint. Whether the inflammatory and cytotoxic programs are mechanistically independent, the authors caution, remains to be established and would require independent experimental manipulation of each program. Future work, they propose, should examine whether T-BET directly regulates chromatin accessibility at inflammatory genes, how it modifies responsiveness to IL-15 and IL-2 signaling, and how these inflammatory NK cells influence surrounding immune populations in co-culture systems.

Subject of Research: The role of the transcription factor T-BET in driving inflammatory specialization versus terminal maturation of iPSC-derived natural killer cells

Article Title: T-BET promotes inflammatory specialization without driving terminal maturation of iPSC-derived NK cells

Article References: Marsal, C., Kashif, M., Borate, S., Hållstrand, C., Haazen, L., Gilljam, M., Dölalan, Y., Burduli, N., Meinke, S., Norman, A., Hussain, A., Lanner, F., Sitnicka, E., & Alici, E. (2026). T-BET promotes inflammatory specialization without driving terminal maturation of iPSC-derived NK cells. Frontiers in Pharmacology, 17, 1952429. https://doi.org/10.3389/fphar.2026.1952429

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1952429

Keywords: natural killer cells, induced pluripotent stem cells, T-BET, TBX21, NK cell maturation, inflammatory programming, IFN gamma, single-cell RNA sequencing, immunotherapy, cell therapy, degranulation, chemokines