Explore

On-Demand Recordings

Full-length webinars featuring TROVO collaborators presenting their research.

Dr. Brandon Simone, Gill Lab, University of Pennsylvania

Dr. Brandon Simone presents work combining a CRISPR/Cas9 epigenetic modifier screen with single-well functional tracking on TROVO to identify genetic perturbations that improve CART19 cell fitness and effector function. Most pooled screens stop at enrichment — which edits won the population — without distinguishing proliferation-driven hits from those genuinely driving cytotoxicity or persistence. This session covers how single-well imaging over time separates those effects, links CRISPR screen hits to functional phenotype at single-well resolution, and recovers high-performing clones directly for downstream characterization.

Highlight Clips

Dr. Won Jin Ho, Johns Hopkins University

Dr. Won Jin Ho presents the DEFINE workflow — linking real-time functional killing kinetics directly to high-plex Imaging Mass Cytometry (IMC) protein data. By performing in situ proteomic profiling of rare clones within hydrogel microwells, DEFINE answers a question flow cytometry cannot ask: not just what a cell is, but what it did. This session covers myeloid heterogeneity across PDAC tumor sites, how specific myeloid subsets suppress or enhance T cell-mediated killing, and how TROVO enables the functional co-culture assays that make this indexing possible.

Highlight Clips

Dr. Xiujian Ma, City of Hope

Scientists at City of Hope demonstrate how TROVO supports CAR-T clone selection in both 2D and 3D brain tumor co-culture models, including glioblastoma. This session covers the challenge of CAR-T clone selection for solid tumors, how TROVO's real-time cytotoxicity imaging enables functional identification without predefined biomarkers, and how AI-driven kinetic analysis makes the approach scalable for CAR-T development programs.

Highlight Clips

TROVO in Action

Real workflows, real data — generated using TROVO in active research programs.

Katz Lab, Yale University

Identifying which patient TILs are capable of sustained tumor killing is one of the hardest problems in clinical immunotherapy. Surface markers alone don't answer this question. TROVO seeds patient TILs and tumor cells into thousands of microwells, tracks killing behavior longitudinally, and compares each co-culture against 50 matched tumor-only controls per microwell. The result is a statistically rigorous Z-score for each clone — derived from the patient's own sample, with no predefined markers required.

In data from the Katz Lab at Yale, TROVO successfully distinguished TILs from responding versus non-responding patients using as few as ~10,000 cells — a cell input level incompatible with traditional bulk functional assays.

Killer clone abundance (Z-score >2) — responding vs. non-responding patient TILs, two biological replicates

Killer and control microwell trajectories over 5 days — ranked by Z-score

• Native internal controls: Every microwell compared against 50 matched tumor-only control wells — no external reference needed.
• Statistically significant from ~10,000 cells: Rigorous potency metrics from clinical samples too scarce for traditional methods.
• Responder/non-responder discrimination: Killer clone abundance correlates with clinical response in Yale patient cohort data.
• Compatible with downstream characterization: Identified killer populations retrieved intact for NGS, expansion, or re-challenge.

Dr. Won Jin Ho's Lab, Johns Hopkins University

Hydrogel microwell array printed onto a chamber slide via TROVO, imaged in fluorescence

Myeloid cells — tumor-associated macrophages, MDSCs, neutrophils — shape prognosis and immunotherapy response across solid tumors, but they span a spectrum of pro- and anti-tumor states that surface markers alone can't resolve. Deep phenotyping tools can find candidate markers; without a functional readout, there's no way to know which of those markers actually drive outcomes.

Dr. Ho's team built DEFINE (Direct Evaluation of Functional Immune Networks via Ex vivo co-culture) to close that gap. Hydrogel microwells printed on TROVO hold a live 72-hour tumor–immune co-culture, then the same physical wells go directly into a 40-marker imaging mass cytometry panel — functional behavior and molecular phenotype come from the exact same cells, not two experiments reconciled after the fact.

Dr. Ho walks through the DEFINE workflow, start to finish

Tumor cell killing over 0–48h — myeloid co-culture vs. T cells alone, best vs. worst 50% killing microwells

Workflow

1. Print 3D hydrogel microwells onto a chamber slide using TROVO
2. Seed pancreatic cancer cells, antigen-specific CD8+ T cells, and CD11b+ myeloid cells into the same microwells
3. Image every 24 hours over 72 hours to track live tumor cell killing per microwell
4. Stain the same slide with a 40-marker antibody panel for imaging mass cytometry
5. Ablate via Hyperion XTi — 1μm "cell mode" for single-cell segmentation, 5μm "tissue mode" for higher-throughput validation
6. Cluster and differentially analyze myeloid and T cell phenotypes against each microwell's measured killing outcome

Myeloid marker expression by killing outcome — CSF1R, PD-L1, CD262, iNOS, and FcγRIIβ across best vs. worst 50% killing microwells

Kaplan-Meier survival probability in pancreatic cancer patients by CD262 expression (TCGA)

• Suppression tracks with myeloid state, not T cell count: Low-killing microwells had similar cytotoxic T cell abundance to high-killing ones, but significantly more MDSCs.
• CSF1R, CD262, and FcγRIIβ mark impaired killing: These myeloid markers were enriched specifically in low-killing microwells; iNOS-high MDSCs tracked with better tumor clearance instead.
• A functional marker with a clinical outcome behind it: CD262 expression correlated with patient survival in TCGA pancreatic cancer data.
• One microwell, two readouts: Live functional behavior and 40-marker molecular phenotype come from the same cells — not two experiments reconciled after the fact.

TROVO supports the complete upstream TIL workflow: tissue dissociation, microwell seeding, longitudinal co-culture imaging, functional clone selection, and live cell recovery — all on a standard 6-well plate, without microfluidics or predefined markers.

Phenotyping screening and enrichment — TIL vs. PBMC killing index comparison

Killing Index (KI₅₀) at 48h — TIL vs. normal control, with tumor killing progression over days

Workflow Steps

1. Dissociate tumor tissue to obtain TIL and tumor cell suspension
2. Seed cells into PEGDA hydrogel microwells — single cell encapsulation per well
3. Image longitudinally over 2–5 days using TROVO fluorescence and brightfield imaging
4. Rank microwells by Z-score killing index against 50 matched tumor-only controls
5. Select high-performing microwells for light-induced capture gel polymerization
6. Wash away uncaptured cells; enzymatically dissolve gel to release viable recovered cells
7. Proceed to NGS, expansion, or downstream re-challenge assays

Clone ranking by Z-score — killers vs. controls over time

T cell proliferation post tumor killing — Day 0 through Day 4

• CMO-compatible workflow: Designed for clinical manufacturing contexts where sample scarcity and process reproducibility are critical.
• Function-first selection: Cells selected based on observed killing behavior — not surface marker assumptions.
• Intact transcriptome on recovery: Fluidics-free capture means no shear stress — RNA integrity preserved for downstream sequencing.

Standard activation markers don't reliably predict which CAR-T or TCR-T constructs will sustain killing over time. TROVO tracks cytotoxicity and persistence kinetics live over 12 days — with tumor rechallenge every 3 days — to separate true persistence clones from short-term burst killers before any downstream investment.

Killer percentage and proliferation comparison across CAR-T constructs

Functional phenotype profiling — NR, P, K, K+P populations across constructs

• 12-day persistence assay: Track the same clones across multiple tumor rechallenge cycles — identify serial killers vs. early exhausters.
• 12,000–18,000 microwells per 6-well plate: Simultaneously monitor thousands of individual co-cultures without sacrificing resolution.
• Functional Z-score ranking: Objective, quantitative clone comparison based on direct behavioral data — not proxy markers.
• Light-induced capture: Retrieve specific viable clones directly for downstream NGS or expansion.

T cell (red) and tumor cell (green) co-culture — 12-day time course on hydrogel microwells

Capture gel workflow — selected microwells encapsulated, unwanted cells washed, viable cells released

T cells recovered from 12-day co-cultures maintained normal phenotypic expression and full functionality post-expansion — confirming that fluidics-free capture does not compromise cell health or downstream performance. This work was supported by NIAID R44AI147734 and NCI 75N91022C00061.

TROVO's per-clone kinetic data can be fit to Lotka-Volterra predator-prey equations to derive a quantitative killing rate constant (k_eff) for each individual clone. This separates intrinsic single-cell killing capacity from cooperative "wolf-pack" killing at higher effector densities — and the resulting in vitro kinetic profiles correlate directly with in vivo tumor control and survival outcomes.

Lotka-Volterra model — tumor observed vs. predicted across 3,000 microcultures, R² >0.95

Single-cell killing kinetics — intrinsic k_eff at E:T 1:10, differentiating constructs independent of cooperative effects

Single T co-culture — four distinct killing phenotypes

Collaborative killing at E:T 20:10 — population-level clearance and proliferation

In vivo validation — k_eff profiles correlate with NSG mouse survival outcomes

• 3,000 microcultures, 144-hour assay: R² >0.95 on held-out validation set — model fit across both fitting and validation sets.
• Intrinsic vs. cooperative killing separated: E:T 1:10 isolates individual clone capacity; E:T 20:10 reveals population dynamics.
• In vivo correlation: In vitro k_eff predicts in vivo tumor control and survival in NSG mouse models.

Single-readout B cell screens are prone to false positives from sticky targets or non-specific IgGs. TROVO enables a two-round multi-signal screening process — first against target peptide on beads, then against full-length native protein on cells — to identify "double-positive" clones that bind in both conformations before committing to sequencing.

PBMCs co-incubated with target beads, anti-IgG secondary staining, positive microwell selection and capture

Two-round screening — double-positive clones binding both target peptide and full-length native protein

• ~0.04% positive hit rate detected: Demonstrated sensitivity to rare secretors in primary PBMC repertoires across multiple seeding densities.
• Secretor detection in under 2 hours: From seeding PBMCs to identifying positive microwells in a single session.
• Native conformation specificity: Two-round screening ensures antibodies bind full-length protein in its native form — not just synthetic peptides.
• No microfluidics: Same resolution as nanopen-based platforms without the complexity, cost, or cell stress.

Screening Protocol

1. Immunize animals and obtain PBMCs
2. Coat streptavidin beads with biotinylated target peptide
3. Seed PBMCs and beads into microwells; pre-stain PBMCs with green fluorescence; incubate 2 hours
4. Remove excess media; add anti-rabbit IgG Alexa647 for staining; image with TROVO
5. Select microwells with high red fluorescent signal (positive secretors)
6. Add liquid capture gel; photo-crosslink in selected microwells; wash away uncaptured cells
7. Dissolve capture gel enzymatically; retrieve cells for RT-PCR, Sanger sequencing, or expansion

Preparing clinical tissue samples for single-cell sequencing is a bottleneck — most methods require large tissue input, risk clogging from debris, or damage fragile cells through high-pressure sorting. TROVO's image-guided live cell enrichment is designed for exactly the cases where standard bulk approaches fail.

Workflow: fresh or frozen tissue → digestion → microwell seeding → dead cell capture → live cell retrieval

% fraction reads in cells — TROVO-enriched vs. unenriched control across breast, lung, and colon tissues

• 4 hours for 3 samples: From raw tissue to single live cell suspension — batch processing compatible.
• Requires only 10⁴ cells: Designed for 5mm tissue pieces and rare clinical biopsies where input is limited.
• Fluidics-free at 4°C: No clogging, no shear stress, intact transcriptome — optimized for fragile edited lines and primary tissue.
• Proven data quality improvement: Higher % fraction reads in cells vs. unenriched controls across multiple tissue types in 10X Genomics scRNA-seq.
• Tissue-specific features preserved: Cell type clustering and tissue-specific distributions maintained in downstream analysis.

Katz Lab, Yale University

Nanobody MET CAR-T cells show efficacy in solid tumors — January 2026 preprint

In vivo tumor control — VHH2-CAR vs. scFv constructs in NSG mouse TNBC model

Per-clone killing kinetics across CAR-T constructs in microwell co-culture

Key Findings