Therapeutic T cells navigating spatial tumor niches
02 / SPACE

Decode the tumor’s geography.

Engineered T cells do not fail in isolation. They encounter cellular neighborhoods, metabolic constraints, and molecular gradients that shape fate across space.

Enter the program ↓
niche → interaction → mechanism
The premise

A question with
engineering consequences.

Where are therapeutic T cells, which neighbors and cues surround them, and which interactions cause functional or dysfunctional fates?

A tumor microenvironment is a structured ecosystem rather than a uniform barrier. We combine spatial measurement with causal perturbation to distinguish descriptive neighborhoods from actionable mechanisms.

02 / CAUSAL LOOP

Question to evidence,
without a missing step.

01Question

Where does therapeutic fate diverge?

Locate functional and dysfunctional engineered T cells within structured tumor neighborhoods.

02Measure

Read cells, niches, and metabolism

Combine spatial cell states, tissue composition, and inferred metabolic programs in the same geography.

03Mechanism

Prioritize causal interactions

Test which ligand–receptor and metabolic relationships drive divergent TCR-T-cell states.

04Engineer

Design for the hostile niche

Convert actionable spatial mechanisms into rescue strategies and therapeutic targets.

05Evidence

Let evidence drive design

Connect tissue-resolved maps to causal mechanisms, engineered interventions, and the publications that establish each step.

Tissue-resolved discovery

See the tissue.
Read the future.

These spatial maps reveal how cellular neighborhoods and metabolic programs shape engineered T-cell fate in pancreatic tumors. Presented in the laboratory’s original scientific palettes, they anchor a research program that moves from tissue-resolved observation to causal mechanism and therapeutic design.

Tissue neighborhoodsTCR-T-cell stateMetabolic programsAdoptive cell therapy
FIGURE 01 / DISCOVERY MAP

TCR-T-cell state across pancreatic tumor neighborhoods

TLS-like nichesFibrotic tumor coresFunctional TCR-T cellsDysfunctional TCR-T cells
Spatial maps of pancreatic tumor tissue. The upper map identifies TLS-like regions and fibrotic tumor cores. The lower map shows engineered T-cell locations, with dysfunctional cells in blue and functional cells in orange.ORIGINAL DATA COLORS
Fig. 01Spatial mapping reveals TLS-like niches and fibrotic tumor cores with distinct cellular composition. Functional and dysfunctional TCR-T-cell states localize differently across these regions, motivating causal analysis of the ligand–receptor and metabolic interactions that drive divergent TCR-T-cell fates.
02
PROGRAM LOGIC

Observe. Perturb.
Build.

01

Map therapeutic neighborhoods

Locate engineered T cells and resolve the native immune, stromal, and tumor cells that define their local context.

02

Perturb in place

Use in vivo Perturb-seq and spatial perturb-omics to measure how a genetic change affects both a T cell and its neighbors.

03

Resolve metabolic geography

Infer and validate spatial nutrient and stress programs, including lipid and iron metabolism, that condition T-cell behavior.

Connected scholarship

The papers behind the program.

Each connection is curated to show how published evidence informs the questions, methods, or engineering logic of this research program.

01 · T-cell dysfunction in the tumor microenvironment
Nature

The ubiquitin ligase KLHL6 drives resistance to CD8+ T cell dysfunction

Open publication
02 · Environmental cue conversion
PNAS

A TGF-betaR/IL-2R immunomodulatory fusion protein transforms immunosuppression into T cell activation to enhance adoptive T cell therapy

Open publication
Explore the complete publication archive
Experimental toolkit

The methods serve
the mechanism.

01Spatial transcriptomics02In vivo Perturb-seq03Cell–cell interaction modeling04Spatial metabolic inference05Tumor models
DESIGNED OUTPUTSCausal niche mapsRescue opportunitiesTherapeutic targets
NEXT / FROM PROGRAM TO PROOF

See how a systems map
became a mechanism.

Explore the KLHL6 discovery path ↗