Latent Regulatory Programs Generate Synthetic T Cell States with Enhanced Therapeutic Potential

Abstract

Transcription factors (TFs) govern cell fate through coordinated gene-regulatory networks, yet the full potential of these networks to generate non-native, therapeutically advantageous cell states in vivo remains largely unexplored. We hypothesized that systematic gain-of-function (GOF) overexpression of TFs in CD8⁺ T cells, central mediators of immune protection, could reveal latent, or "hidden," regulatory programs capable of generating synthetic T cell states with therapeutic utility. To test this, we developed single-cell GOF sequencing (scGOF-seq), a multiplexed platform for unbiased, in vivo mapping of GOF effects on T cell fate in immunocompetent mouse models of infection and cancer. scGOF-seq uncovered unexpected regulators of T cell differentiation and accumulation, including SOX2, OCT4, and GATA2, which are normally silenced during T cell differentiation. Notably, outside its native regulatory context, supraphysiologic cMyc GOF reprogrammed CD8⁺ T cells into a synthetic stem-effector hybrid state, enabling >5,000-fold antigen-dependent expansion and antitumor activity, contrasting sharply with its native function in driving terminal differentiation. scGOF-seq further identified TF modules that cooperate with cMyc GOF to promote robust CD8⁺ T cell responses in solid tumors. Together, these findings establish GOF perturbation as a powerful strategy for revealing latent immune regulatory programs and engineering synthetic immune states with therapeutic potential.

Fig. 1.. scGOF-seq screens uncover hidden regulators of CD8⁺ T cell differentiation in vivo.

(A) Expression dynamics of differentially expressed (D.E.) genes, invariant (Inv.) genes, and lowly expressed genes in canonical CD8⁺ T cell states in LCMV Arm based on GSE290082 with representative TFs highlighted (left). Representative processes enriched based on gene ontology analyses of D.E. genes and lowly expressed genes (right). (B) Expression dynamics D.E., Inv., and lowly expressed genes in canonical CD8⁺ T cell states in LCMV CL13 from reanalysis of GSE290082 with representative TFs highlighted (left). Representative processes enriched based on gene ontology analyses of DE genes and lowly expressed genes (right). (C) Gene expression and motif deviation scores based on paired scRNA-seq and scATAC-seq profiling (GSE290082) of most TFs included in the initial screening library. TFs lacking known motifs are indicated as “NA.” (D) P14 CD8⁺ T cells were transduced with the RV-ORF library and transferred into mice infected with LCMV Arm or CL13. EGFP⁺ donor cells were sort-purified 2wks post-infection for 5’ scRNA-seq analysis. A paired ORF library was generated based on amplification of EGFP-containing transcripts to assign ORF identities. (E) scGOF-seq UMAP colored by condition and T cell states. (F) Representation of each ORF in LCMV Arm and LCMV CL13 normalized to input. (G) Composition of cell states for select ORFs (ORFs with sufficient recovery included). (H) Representative abundance of ORFs projected into the UMAP space from E. (I) P14 CD8⁺ T cells transduced with an expanded TF-GOF library (74 vectors) were transferred into mice bearing B-ALL-GP (top). EGFP⁺ donor cells were sorted 10 days post-transfer for scGOF-seq and projected in the UMAP space (bottom). (J) Representation of each ORF in the B-ALL-GP screen normalized to input. (K) Composition of cell states for each ORF from I (L) Representative abundance of ORFs projected into the UMAP space from I. (M) Gene module enrichment for exhaustion, anergy, stemness, cell cycle, and terminal differentiation signatures. (N) Gene module enrichment for stemness and exhaustion signatures for each ORF in B-ALL-GP scGOF-seq. (O) Mixed transfer analysis of cMyc GOF, Oct4 GOF, KLF4 GOF, SOX2 GOF, and Ctrl cells in LCMV Arm (D14), LCMV CL13 (D14), and B-ALL-GP (D12). Graphs show mean ± SEM of n=7 mice pooled from two independent experiments (Fig. 1O). *p<0.05, ***p < 0.001, n.s. not significant, paired Student’s t-test. For scORF-seq screens cells were pooled from n=5-11 mice (D) or n=2 mice (I).

Fig. 2.. Artificially sustained cMyc expression restrains exhaustion differentiation.

(A-B) RV-cMyc and RV-Ctrl P14 cells were mixed and transferred into congenic mice infected with LCMV CL13 or LCMV Arm. (A) Donor P14 cells enrichment were evaluated on days 5, 8, 15, and >42 post-infection by spectral flow cytometry. (B) UMAP representation and clustering of donor T cells from A (top) and relative expression of key proteins (bottom). Green dashed regions mark clusters predominantly composed of day 42 cMyc GOF cells. (C) WT and Myc^GOF-Tg mice were infected with LCMV CL13. Tetramer⁺ cells were analyzed via spectral flow cytometry >D42 post-infection. (D) The terminal exhaustion phenotype of H2-Db GP33–41 tetramer⁺ and H2-Db GP276–284 tetramer⁺ cells from panel C was analyzed based on co-expression of exhaustion markers (TIM-3, PD-1, TIGIT) and a TIM-3⁺ Ly108⁻ or CD62L⁻ profile, indicating loss of memory/progenitor markers. (E) Tex-prog frequency via measuring CD62L⁺Ly108⁺ cells from C. (F) scRNA-seq analysis of cMyc GOF phenotype. CD8⁺ T cells were sort purified on days 7 and 15 post-infection from WT or Myc^GOF-Tg mice. UMAP representation and clustering of CD8⁺ T cells by condition. (G) UMAP representation and clustering of CD8⁺ T cells by cell state annotation according to key cell state-defining genes expression (fig. S6A-B, Fig. 2J). (H-I) Feature plots reflecting enrichment of exhaustion and dysfunction gene modules (H, top) and relative expression of exhaustion-associated genes, Tox, Havcr2 (TIM3), and Cd101 (I, bottom). (J) Gene expression analysis of key genes. (K) Trajectory and pseudotime analysis generated using Monocle3. Dashed regions indicate distinct terminal differentiation endpoints for WT and Myc^GOF-Tg cells. (L) Relative abundance of each cell state. (M) Summary of cell state composition of day 15 WT and cMyc^GOF-Tg mice based on scRNA-seq profiling. (N) Gene set comparison of WT control cells versus the Tex-myc^UNIQUE population. (O) UMAP representation and clustering of RV-Ctrl and RV-cMyc P14 cells donor cells harvested from LCMV CL13-infected mice on day 14-16 of infection (right). Relative expression of GzmA, KLRG1, CX3CR1, and Ki-67 in donor cells from O (left). (P) Enumeration of GzmA⁺Ki67⁺KLRG1⁺ donor cells. Graphs show mean ± SEM of n=5-16 mice from one representative experiment or pooled from two or more independent experiments (Fig. 2A-G). *p < 0.05, **p < 0.005, **p < 0.001, paired Student’s t-test.

Fig. 3.. Multimodal reprogramming of exhausted T cells by artificially sustained cMyc activity.

(A) cMyc reporter (cMyc-EGFP fusion) P14 cells were transferred into congenic distinct mice infected with LCMV CL13. Donor cells were analyzed on days 4, 10, and >42 post-infection by spectral flow cytometry. (B) Pathway enrichment analysis comparing canonical cMyc-associated programs across T cell states in WT cells and day15 MYC^GOF-Tg cells from the scRNAseq dataset in Fig. 2F. (C–H) Histograms and quantitative comparisons of cMyc expression, glucose uptake, GLUT1 expression, MitoTracker-ROS, MitoTracker Deep Red (MTDR), regulator of ribosome synthesis 1 (RRS1), puromycin uptake, Ki-67 expression, and IL-2 production in cMyc-RV versus control-RV cells in a mixed transfer setting in LCMV Cl13 on d12-16 of infection. (I) RV-Ctrl, RV-cMyc, and RV-cMyc^T58A P14 cells were mixed and transferred into mice infected with LCMV CL13. (J) Log2 accumulation of cMyc-RV and cMyc^T58A-RV cells relative to RV-Ctrl cells in mLN, siIEL, and spleen during LCMV Arm or CL13 infection. Tissues were harvested on days 14–16 of infection. (K) Representative histograms comparing mitochondrial function, glucose uptake, and protein translation in donor cells from J. (L) TIM3 expression and IL-2 production on donor cells from J. (M) Glucose uptake and KLRG1 expression on donor cells from J (N) KLRG1 expression based on graded cMyc expression from J. (O) Taiji analysis of TF activity in cMyc-GOF cells. Color of circles indicates the TF activity difference between cMyc GOF vs. Ctrl. Line thickness represents the functional interaction or shared target genes between each TF and cMyc in cMyc GOF condition (green) and WT condition (gray). (P) RV-Ctrl, RV-BLIMP-1, RV-cMyc^T58A, and RV-BLIMP-1+cMyc^T58A P14 cells were mixed and transferred to mice infected with LCMV CL13. Spleens were harvested on days 14–16 post-infection. (Q) Flow cytometry profiling of donor cells from P. (R) Representative flow cytometry plots and quantification of Tex-term differentiation, gated as TIM3⁺CX3CR1⁻ and TIM3⁺Ly108⁻ populations. Graphs show mean ± SEM of n=3-20 from one representative experiment or pooled from two or more independent experiments (Fig. 3A, C-N, P-Q). *p < 0.05, **p < 0.005, **p < 0.001, n.s. =non-significant, paired Student’s t-test, ***p < 0.001, **p < 0.01, *p < 0.05.

Fig. 4.. cMyc GOF enhances cell therapy efficacy in leukemia.

(A) Gene set comparison of WT control cells versus the cMyc GOF population in B-ALL. (B) Pathway enrichment analysis comparing canonical cMyc-associated programs in WT control cells versus the cMyc GOF population in B-ALL. (C) Expression of cMyc, PD-1, and TIM3 in RV-Ctrl, RV-cMyc, and RV- cMyc^T58A transduced P14 donor cells transferred into B-ALL-GP bearing mice and analyzed on day 10 post T cell transfer. (D) cMyc GOF induced synthetic state identification of donor cells from C and enumeration of GzmA⁺Ki67⁺KLRG1⁺PD-1⁻ cells. (E) RV-Ctrl, RV-cMyc, or RV-cMyc^T58A transduced P14 were transferred individually into B-ALL-GP bearing mice, and survival was measured. (F) cMyc expression dynamics in human CD8⁺ T cells under standard CAR-T cell expansion conditions. (G) In vitro killing assay comparing human CD8⁺ T cells transduced with CD19-CAR (ctrl) and CD19-Myc^T58A-CAR co-cultured with CD19⁺ Daudi Leukemia cells. (H) Expression levels of PD-1, CCR7, IL-2, IFN-γ, and TNFα in CD19-CAR and CD19-cMyc^T58A-CAR T cells. Graphs show mean ± SEM of n=4-6 (Fig. 4C-E) or pooled from >3 independent donors (Fig. 4G-H). *p < 0.05, **p < 0.005, **p < 0.001, n.s.=non-significant, paired Student’s t-test, ***p < 0.001, **p < 0.01, *p < 0.05.

Fig. 5.. Rational GOF engineering enhances cell therapy efficacy in solid tumors.

(A-B) Profiling of Tex-term and Tex-prog markers of donor P14 cells transduced with RV-Ctrl, RV-cMyc, or RV-cMyc^T58A from mice bearing B16-GP (A) or KPC-GP tumors (B). T cells were harvested, isolated from tumors 8-12 days post-T cell transfer. (C) Tumor growth curves of B16-GP–bearing mice receiving RV-Ctrl or RV-cMyc^T58A transduced P14 cells. (D) Accumulation of RV-Ctrl, RV-cMyc, and RV-cMyc^T58A transduced P14 cells within B16-GP tumors shown in (A). (E) Tumor-residency phenotypes of RV-Ctrl, RV-cMyc, and RV- cMyc^T58A P14 cells from (A), assessed by co-expression of CD39 and CXCR6 or gating of CD69+ CD62L⁻. (F) Tumor-residency phenotypes of RV-Ctrl, RV-cMyc, and RV- cMyc^T58A P14 cells from KPC-GP tumors shown in (B), assessed by co-expression of CD39 and CXCR6 or CD69 and CD103. (G) scGOF-seq screening in B16-GP and orthotopic KPC-GP tumor models (left). Tumors were harvested 14 days after T-cell transfer, and normalized ORF representation was calculated for each model (right). (H) Gene-module enrichment scores for individual ORFs identified in (G). (I) Integrated analysis of combinatorial GOF perturbations assessed by spectral flow cytometry. Top, schematic of the in vivo combinatorial GOF strategy. CD45.1⁺ P14 cells transduced with RV-Ctrl, RV-cMyc^T58A, single TF GOFs (Blimp1, RUNX1, or BATF), or dual GOFs combining each TF with cMyc^T58A were pooled and transferred into separate cohorts of B16-GP bearing mice. Cohorts included Ctrl + single cMyc^T58A GOF, Ctrl + single BLIMP1 + dual GOF, Ctrl + single RUNX1 + dual GOF, and Ctrl + single BATF + dual GOF. Donor P14 cells recovered from all cohorts from day 10-11 post-transfer were analyzed via spectral flow. Bottom, UMAP projections of recovered donor P14 cells, colored by GOF identity. Right, feature plots showing relative expression of differentiation and exhaustion markers (Ly108, TIM3, TIGIT, CD39, CD62L, CD69). (J) Accumulation of donor P14 cells from (I), normalized to input frequencies to enable quantitative comparison of enrichment across conditions. Tumor-resident populations were quantified by CD69⁺CD62L⁻ gating. (L) B16-GP–bearing mice received 2 × 10⁵ P14 cells transduced with RV-Ctrl, RV-cMyc^T58A, RV-BATF, or RV-cMyc^T58A + RV-BATF, and tumor growth was monitored longitudinally. Graphs show mean ± SEM of n=3-14 (Fig. 5A-F) or pooled from >3 independent donors (Fig. 5K-M). *p < 0.05, **p < 0.005, **p < 0.001, n.s. =non-significant, paired Student’s t-test, ***p < 0.001, **p < 0.01, *p < 0.05. For scORF-seq screens in Fig. 5G, donor cells were pooled from n=8-10 mice.