A team from IBM Quantum, IBM Research, and the Center for Cellular Construction now reports one of the largest experimental deployments to date of quantum machine learning in the life sciences, using a 61-qubit quantum circuit to help classify the killing power of CAR T-cell designs. Writing in the journal Quantum Machine Intelligence, Filippo Utro, Meltem Tolunay, and colleagues describe how a hybrid quantum-classical technique called the projected quantum kernel, or PQK, allowed a conventional support vector machine to extract biologically meaningful patterns from a sparsely sampled library of receptor designs. The work suggests that near-term quantum processors, even without error correction, may earn their keep in data-starved biological problems.

The underlying dataset came from earlier work by Kyle Daniels and colleagues at Stanford, who built a combinatorial library of 13 intracellular signaling motifs, labeled M1 through M13, that could be placed in up to three positions within the CAR’s intracellular domain, with every construct ending in a terminal M14 motif. The team measured cytotoxicity and stemness for 246 experimentally produced constructs. Cytotoxicity, assessed by how well the engineered cells killed Nalm6 leukemia cells, showed a bimodal distribution with a natural dividing line at a survival value of 0.62, allowing the researchers to frame the problem as binary classification: high cytotoxicity versus low. Stemness, by contrast, showed a single asymmetric peak and was not amenable to clean binary labeling, so the study focused on cytotoxicity.

Encoding the designs for a quantum computer required care. Each of the 14 motifs, plus an empty placeholder for unoccupied positions, was one-hot encoded across four positions, producing a 60-dimensional binary feature vector — hence a 60-qubit circuit for the first embedding. The researchers tested two quantum embeddings drawn from the theoretical literature on quantum kernels. The first, E1, is the ZZ Feature Map, which loads each binary feature onto a qubit as a rotation angle of either pi or pi/2 and entangles qubits through controlled phase gates. The second, E2, is a Trotterized simulation of the one-dimensional Heisenberg model, which required an additional ancilla qubit for a total of 61.

The PQK trick is that the quantum computer never tries to learn anything itself. Instead, the classical data is loaded onto the circuit, where it lives in a high-dimensional Hilbert space, and is then projected back into a classical representation by measuring every qubit in the X, Y, and Z bases. These measurements yield single-qubit reduced density matrices whose expectation values become the new features. A standard classical kernel method — here a support vector classifier — then does the actual classification. This design sidesteps the training instabilities and noise sensitivity that plague variational quantum algorithms on today’s pre-fault-tolerant hardware.

The experiments ran on IBM Heron R2 superconducting processors, ibm_marrakesh and ibm_kingston, devices with fixed-frequency transmon qubits linked by tunable couplers and median two-qubit gate errors around 0.3 percent. Each expectation value was estimated with 10,000 shots distributed across 64 Pauli-twirled circuit instances, with readout error mitigated using the TREX technique and qubit layouts selected for best performance. The full run on ibm_marrakesh consumed about 100 minutes of quantum processor time, roughly 22 seconds of quantum processing per data point. Before committing to hardware, the team applied the geometric framework of Huang and colleagues to check whether the dataset even held promise for quantum advantage: the geometric separation g of 15.78 was comparable to the square root of the training set size, 13.12, and the quantum model complexity of 1.53 was well below the classical value of 6.09 — conditions consistent with a potential predictive edge.

On hardware, the ZZ embedding with a pi/2 rotation angle and eight feature-map repetitions delivered the best results, reaching median and maximum F1 scores of 0.75 and 0.81 across ten random 70/30 train-test splits, compared with 0.73 and 0.77 for an optimized support vector machine on the original features. Performance dropped at twelve repetitions, consistent with accumulated hardware noise in deeper circuits. The Heisenberg embedding never beat the classical baseline, its deeper circuits falling victim to noise. Reordering features so that correlated ones sat near each other on the chip changed nothing meaningful, suggesting the circuit depth already generated sufficient entanglement. The team also benchmarked against approximate classical simulations — operator backpropagation and matrix product state methods — which were six to ten times slower and still slightly less accurate than the quantum processor results.

The most striking result came from validation. On an independent cohort of 16 previously synthesized constructs — a severely imbalanced set with only one low-cytotoxicity sample — the PQK-based model achieved a median Matthews correlation coefficient of 0.415, versus minus 0.067 for the classical model, a statistically significant difference. The quantum-enhanced classifier caught the single minority-class sample in 8 of 10 splits, a recall of 0.8, while the classical SVM managed it in only 1 split. Precision remained low for both models, and the authors caution that the tiny, imbalanced validation set makes these findings preliminary rather than definitive proof of generalization.

Perhaps more intriguing than the headline metrics is where the quantum kernel helped most. The PQK-based classifier consistently predicted specific motif-position combinations that the classical model could not, and never the reverse: across all ten splits there was no case in which the original-data SVM significantly outperformed PQK on a given motif and position. The advantage grew as information thinned out. At the third motif position — the least sampled, most often empty or terminal — the number of source proteins benefiting from PQK rose to eight, compared with four at the first position. Motifs derived from LAT, CD40, and LAIR1 were predicted better by the quantum model at all three positions, and LAT-derived motifs are clinically relevant, having been linked to improved CAR T-cell responses in antigen-low acute lymphoblastic leukemia.

The authors frame the work as an early demonstration that hybrid quantum-classical learning can navigate underdetermined biological design spaces, and as one of the first large-scale applications of quantum machine learning to cell therapy optimization. The overall gains are modest, and the quantum hardware offered favorable runtime and predictive utility relative to the best approximate classical simulations rather than a dramatic leap. But the pattern — advantages concentrated exactly where data is sparsest — is the pattern quantum machine learning theorists have long predicted. As CAR T-cell applications expand beyond oncology and combinatorial receptor libraries keep outpacing experimental throughput, quantum-enhanced feature maps may become a practical tool for steering synthetic immune receptor design toward regions of the design space no classical model would think to explore.

Subject of Research: Hybrid quantum-classical kernel methods for predicting CAR T-cell cytotoxicity from combinatorial signaling domain designs

Article Title: Enhanced prediction of CAR T-cell cytotoxicity with quantum-kernel methods

Article References: Utro, F., Tolunay, M., Rhrissorrakrai, K., Gujarati, T. P., Shi, J., Capponi, S., Amico, M., Earnest-Noble, N., & Parida, L. (2026). Enhanced prediction of CAR T-cell cytotoxicity with quantum-kernel methods. Quantum Machine Intelligence, 8(2), Article 105. https://doi.org/10.1007/s42484-026-00446-w

Image Credits: AI Generated

DOI: 10.1007/s42484-026-00446-w

Keywords: CAR T-cell therapy, quantum machine learning, projected quantum kernel, chimeric antigen receptor, support vector machine, synthetic immunology, IBM Heron QPU, cytotoxicity prediction, combinatorial library, cell therapy, quantum computing, signaling motifs

The pipeline begins with one of the most detailed single-cell RNA sequencing atlases of triple-negative breast cancer available, dataset GSE176078, which profiles more than 100,000 cells from 26 treatment-naive patients. After rigorous quality control that removed cells with excessive mitochondrial content or too few detected genes, the researchers normalized expression data, identified 2,000 highly variable genes, and applied principal component analysis, k-nearest neighbor graph construction, and Leiden clustering to organize the cellular landscape. The resulting atlas resolved nine major cell compartments, including T cells, myeloid cells, B cells, cancer epithelial cells, and cancer-associated fibroblasts. Validation against the curated reference annotations produced adjusted rand index values between 0.288 and 0.311 and normalized mutual information values between 0.616 and 0.671, confirming that the unsupervised clustering substantially recovered known biology despite the inherent difficulty of matching 35 data-driven clusters to nine curated labels.

With the atlas established, the analysis zeroed in on the T cell compartment, extracting roughly 15,000 to 30,000 T cells and re-processing them independently to resolve functional sub-states. Two gene-set module scores quantified exhaustion and cytotoxicity across the population. The centerpiece of the study, however, is a deceptively simple quantity: the per-cell ratio of PDCD1, the gene encoding the inhibitory receptor PD-1, to CD2, the adhesion receptor that stabilizes the immune synapse between T cells and their targets. When PD-1 expression dominates and CD2 is low, inhibitory signaling suppresses activation so thoroughly that even dense CD58 ligand on an opposing surface cannot overcome it. When CD2 is moderately expressed but PD-1 only mildly elevated, calibrated CD2/CD58 ligand density can restore synapse stability and partial T cell activation. Prior work had shown that CD2 expression on tumor-infiltrating CD8 T cells correlates negatively with exhaustion markers and that CD2/CD58 co-stimulation is less sensitive to PD-1 inhibition than canonical CD28 signaling.

Aggregating these measurements per patient revealed three distinct immune phenotypes across the cohort. Roughly eight patients formed a high-exhaustion group in which more than 65 percent of CD8 T cells were classified as exhausted and PD-1 expression exceeded CD2 by three-and-a-half to five-fold. Their cells showed the lowest cytotoxic scores, consistent with a terminally exhausted state in which effector function is epigenetically silenced, making CD2-directed surface engineering alone an insufficient strategy. A second group of about ten patients retained measurable cytotoxic capacity alongside moderate exhaustion; their principal barrier appeared to be inadequate synapse stabilization rather than irreversible transcriptional shutdown, making them the most tractable candidates for calibrated CD2/CD58 ligand density optimization. A third group of roughly eight patients showed sparse CD8 infiltration altogether, representing immune-cold tumors where T cell recruitment must precede any activation engineering. Across all patients, mean PDCD1 expression spanned a seven-fold range and the resulting PDCD1/CD2 ratios ranged from approximately 0.3 to 4.1, capturing clinically meaningful variability invisible to bulk gene expression summaries.

To test whether the ratio carries clinical weight beyond the single-cell cohort, the team turned to TCGA-BRCA bulk RNA sequencing data of roughly 1,100 samples. In unadjusted Cox regression, higher bulk PDCD1/CD2 ratios were associated with lower mortality, with a hazard ratio of 0.47 and a confidence interval of 0.28 to 0.79, reaching statistical significance below 0.005. That direction appears paradoxical at first, since a high ratio reflects exhaustion at the single-cell level. The authors are careful to explain the discrepancy: in bulk tissue, PDCD1 transcripts derive from the entire immune infiltrate rather than from exhausted CD8 cells alone, so a high bulk ratio primarily signals strong immune infiltration, itself a favorable prognostic factor in breast cancer. This cell-type composition effect inverts the ratio’s meaning across measurement platforms, and the study explicitly frames the survival association as exploratory and hypothesis-generating rather than as validation of a clinical biomarker.

Cross-modal comparison of the 24 patients with matched bulk and single-cell data reinforced this interpretation with appropriate caution. Spearman correlations between single-cell-derived and bulk-derived ratios were weakly negative at approximately minus 0.30, consistent with the infiltration confound compressing the denominator, while exhaustion scores showed a weakly positive correlation of about 0.28, directionally consistent with the hypothesis that higher single-cell exhaustion burden corresponds to higher bulk exhaustion signal. Neither correlation reached conventional statistical significance given the small matched sample, but the directional agreement across platforms suggests the single-cell signals are not artifacts of normalization. Kaplan-Meier analysis stratified by the bulk ratio median showed longer median overall survival in the high-ratio group, with median survival differences of roughly 12 to 18 months and curve separation emerging after about two years of follow-up.

Beyond stratification, the pipeline delivers what previous tools have not: engineering output. A targeted ligand-receptor proxy screen across five immune axis pairs placed PD-1/PD-L1 highest in both T cell-tumor and T cell-myeloid pairings, reflecting the co-elevation of PD-1 on exhausted T cells and PD-L1 across tumor and myeloid compartments. CD2/CD58 ranked moderately at the tumor interface but low with myeloid cells, indicating that CD2-directed ligand optimization is specific to the direct tumor-T cell boundary. LAG-3/HLA-DRA showed the inverse profile, strongest with myeloid and B cell ligand sources, while CD28 co-stimulation was broadly downregulated throughout the tumor microenvironment, consistent with exhaustion-associated loss of CD28 expression. The authors emphasize these rankings are comparative heuristics from product-of-means calculations, not statistically validated communication events, and require experimental confirmation through co-culture or blocking assays.

The final translation step is the DesignPriorityScore, a rule-based metric combining normalized exhaustion burden with 50 percent weight, PDCD1/CD2 axis imbalance with 30 percent weight, and CD8 infiltration with 20 percent weight. Each of the 26 patients receives a ranked score and a corresponding recommendation: recruitment-first strategies for immune-desert tumors, combined PD-1 blockade with CD2 reinforcement for the highest-exhaustion group, CD2/CD58 axis optimization for the moderate group, and CD28 co-stimulation otherwise. Bootstrap resampling with 200 iterations, threshold sensitivity testing across exhaustion quantiles from 0.60 to 0.90, and weight sensitivity analysis confirmed ranking stability, with mean Spearman correlations above 0.85 and top-quartile patients retaining their position more than 90 percent of the time. Adjacent patients were separated by an average score gap of 0.03 normalized units, sufficient to avoid ambiguous recommendations.

The authors are candid about limitations. The pipeline was built and tested on a single cohort of 26 patients, exhaustion states were assigned by quantile-based gene-set scoring rather than experimental annotation, and the survival associations remain unadjusted for tumor purity, stage, and molecular subtype. The scoring weights are biologically motivated rather than data-trained, and no functional experiments have yet linked the recommendations to actual T cell activation outcomes. Future directions include integrating the spatial transcriptomics data that accompany the atlas to map where exhausted cells reside relative to tumor cells, training the score on labeled immunotherapy response cohorts, and connecting patient-level ratios to biophysical models of ligand density and clustering on synthetic cell membranes. As a prototype, the study demonstrates that single-cell immune phenotyping can be pushed beyond description into prescriptive design, offering synthetic immunologists a patient-specific starting point for building the engineered immune cells of tomorrow.

Subject of Research: A computational framework translating single-cell PDCD1/CD2 immune axis measurements into patient-specific synthetic immune cell engineering priorities for triple-negative breast cancer.

Article Title: Computational stratification and engineering framework of the PDCD1/CD2 immune axis in triple-negative breast cancer

Article References: Chowdhury, K. (2026). Computational stratification and engineering framework of the PDCD1/CD2 immune axis in triple-negative breast cancer. Clinical Cancer Bulletin, 5(1), Article 11. https://doi.org/10.1007/s44272-026-00063-5

Image Credits: AI Generated

DOI: 10.1007/s44272-026-00063-5

Keywords: triple-negative breast cancer, single-cell RNA sequencing, T cell exhaustion, PDCD1/CD2 axis, immune checkpoint blockade, synthetic immunology, tumor microenvironment, immune stratification, CD2/CD58 co-stimulation, DesignPriorityScore, TCGA-BRCA, ligand-receptor interactions

The research, summarized in a companion Research Briefing, combines AI-assisted de novo binder design with the assembly of complete CAR constructs and scalable testing platforms in vitro and in vivo. Rather than asking whether a computationally designed protein can bind a target antigen in a dish, the investigators pushed the designs through the full developmental gauntlet that a therapeutic candidate must survive: expression on the surface of primary T cells, signal transduction upon antigen engagement, target cell killing, and durable anti-tumor activity in living models. The resulting dataset links sequence-level and structural-level features of designed binders directly to therapeutic function, providing a design grammar that the field has previously lacked.

The work builds on a remarkable run of advances in computational protein design. In 2023, the introduction of the RFdiffusion framework dramatically raised the success rate of de novo binder campaigns by using diffusion models to generate protein backbones tailored to a desired target surface. A year later, researchers reported targeting overexpressed antigens in glioblastoma using CAR T cells armed with computationally designed high-affinity protein binders, offering early proof that AI-designed recognition domains could function as the business end of a chimeric antigen receptor. More recently, the hallucination-based design pipeline BindCraft described one-shot generation of functional protein binders, and it produced some of the highest-success design campaigns in the new study. Together with ProteinMPNN, a robust deep learning method for assigning amino acid sequences to designed backbones, these tools have made binder generation almost routine. The new work addresses the harder question: which of the many binders that pass computational filters will actually drive T cells to kill cancer?

That question matters because binding is only the beginning. A CAR binding domain operates in a demanding mechanical and biological context. It must fold correctly and traffic to the cell membrane when fused to hinge, spacer, transmembrane and signaling modules. It must bind antigen with an affinity that falls within a productive range; too little affinity produces no signal, while excessive affinity can cause antigen-independent tonic signaling, activation-induced cell death and poor persistence. It must tolerate epitope densities that vary enormously across tumor tissues. And it must be small, stable and non-immunogenic enough to be clinically deployable. Natural antibody fragments often fail several of these criteria simultaneously, which is why the pharmaceutical industry has invested heavily in screening campaigns that yield a single usable binder after months of labor.

By systematically varying binder sequences and structures within the CAR context and evaluating the resulting constructs in standardized cell models, the authors were able to infer the attributes that correlate with efficacy. The study frames a complete framework for designing efficacious CARs, in which binder attributes such as affinity, epitope choice, stability and expression behavior are treated as tunable design parameters rather than accidents of discovery. The significance of this reframing is difficult to overstate. For two decades, CAR engineering has been as much an art as a science, with laboratories borrowing fragments from existing antibodies and adjusting hinges and spacers empirically. A predictive model of what makes a binding domain efficacious turns CAR design into an engineering discipline in which candidate receptors can be specified computationally before a single experiment is run.

The experimental architecture underpinning the study is as important as its findings. The investigators paired binder design with scalable in vitro assays that measure how many of the designed constructs express on T cells, how strongly they signal, and how effectively they eliminate antigen-positive targets. In vivo models then tested whether promising designs retained activity against tumors in a physiological setting, where antigen density, immune suppression and trafficking barriers conspire to defeat otherwise potent receptors. This multi-tier funnel mirrors the path of a therapeutic candidate and ensures that the design rules extracted from the data reflect true clinical relevant properties, not merely binding measurements from immobilized proteins.

The implications extend well beyond one cancer type. Because the pipeline is generative, it is in principle antigen-agnostic: given a target surface, the same design-and-test cycle can produce panels of candidate binding domains against antigens relevant to solid tumors, autoimmune disease, fibrosis and infectious disease. Solid tumors have proven especially refractory to CAR therapy, in part because widely shared tumor-associated antigens are also expressed on essential healthy tissues and because single-antigen targeting invites escape. AI-designed binders, selected with precise affinity windows and epitope specificity, could enable new strategies such as affinity tuning to discriminate between high- and low-expressing tissues, dual-antigen logic gating, and rapidly generated panels against patient-specific neoantigens. The glioblastoma work from 2024 demonstrated that computationally designed binders could target an antigen overexpressed in one of the deadliest solid tumors; the new study supplies the general principles for making such binders reliably efficacious.

There are also cautionary notes that seasoned observers of the CAR field will appreciate. Computational design success rates, even with state-of-the-art tools, remain probabilistic, and the attributes that make a binder effective in a standardized cell line may not transfer directly to the hostile microenvironment of a human tumor. Immunogenicity of non-human-derived protein scaffolds must be assessed rigorously before clinical translation, and the regulatory pathway for wholly synthetic recognition domains is still being defined. Tumor heterogeneity, antigen loss and the immunosuppressive microenvironment remain problems that no binder, however well designed, can solve alone. What the study offers is not a finished therapy but a reproducible methodology for generating and selecting binding domains with predictable properties, which removes one of the largest sources of variability and failure in current CAR programs.

The broader scientific community has taken notice of how quickly the ingredients of this advance came together. RFdiffusion and ProteinMPNN provided the generative backbone and sequence design machinery; AlphaFold-style structure prediction supplied reliable in silico validation of designed conformations; BindCraft demonstrated that hallucination-based pipelines could deliver functional binders in single campaigns; and the earlier glioblastoma CAR study established clinical feasibility. The new research closes the loop by asking what distinguishes the binders that work in a CAR from those that bind beautifully on paper but fail on the cell surface. The answer, encoded in the sequence and structural determinants the authors report, is a practical toolkit for the next generation of synthetic immunology.

If the field can standardize on these design attributes, the consequences could be transformative. Cell therapy developers could move from years of empirical binder discovery to weeks of computational specification followed by targeted validation. Clinicians could obtain CAR constructs tuned precisely to the antigen expression profile of an individual tumor. Academic laboratories with modest resources could design receptors against orphan antigens that no commercial entity would ever fund an antibody campaign for. The convergence of generative AI and cellular immunotherapy has promised exactly this kind of acceleration for several years, and this study provides some of the clearest evidence yet that the promise is becoming an operational reality, one amino acid at a time.

Subject of Research: AI-assisted de novo design of protein binders for constructing efficacious chimeric antigen receptor T cell therapies

Article Title: Defining attributes of effective binders for AI-assisted CAR design

Article References: Defining attributes of effective binders for AI-assisted CAR design. (2026). Nature Biomedical Engineering. https://doi.org/10.1038/s41551-026-01792-7

Image Credits: AI Generated

DOI: 10.1038/s41551-026-01792-7

Keywords: CAR T cells, chimeric antigen receptor, AI protein design, de novo binders, RFdiffusion, ProteinMPNN, BindCraft, cancer immunotherapy, protein engineering, Nature Biomedical Engineering, antigen targeting, synthetic immunology