A Combinatorial Transcription Factor Screening Platform for Immune Cell Reprogramming
ON DEMAND: Dr. Carlos-Filipe Pereira | April 16, 2026
📺 [Watch the full talk here →] 📄 Read the paper → (Cell Systems, Vol. 17, Issue 1 — January 14, 2026)
💬 Key Quote
“We can now start thinking about whether it’s possible to do the same for autoimmune disorders — where you also want to induce a long-term response — but it’s the opposite of what we induce with cancer reprogramming. The idea is to reprogram beta cells directly into tolerogenic cells — so you have the islet antigens in the very cell that is going to present them in a tolerogenic fashion.” — Dr. Carlos-Filipe Pereira
🔬 Foundational Insights as They Apply to T1D
The dominant paradigm for treating T1D immunologically has been to remove, block, or redirect the immune cells that are doing damage — with teplizumab’s approval representing the most significant advance yet. But this paradigm operates downstream of a more fundamental problem: the immune system does not know it should be tolerant of beta cell antigens in the first place, and we have no reliable way to teach it. Tolerogenic dendritic cells have long been theorized as the answer — the gatekeepers that could present beta cell antigens in a context that induces regulatory T cells rather than destructive ones — but generating them consistently, scalably, and with the right antigen specificity has remained a major practical barrier.
Dr. Carlos-Filipe Pereira’s work arrives from a completely different direction. His lab has spent two decades decoding the transcription factor grammar of immune cell identity — starting with the discovery that just three transcription factors (PU.1, IRF8, BATF3) are sufficient to directly convert fibroblasts and cancer cells into functional type 1 conventional dendritic cells. That foundational work, published in Science and now underpinning the clinical program at Asgard Therapeutics for intratumoral dendritic cell reprogramming in cancer, has now been systematized into REPROcode: a combinatorial, barcoded, single-cell screening platform capable of identifying the minimal transcription factor combinations needed to generate any immune cell identity from scratch.
The T1D implications are not peripheral to this work — they are the explicit next chapter. Dr. Pereira’s lab has already identified transcription factor codes for mregDC (mature regulatory dendritic cells, known activators of regulatory T cells), plasmacytoid DCs, NK cells, and multiple DC subsets with tolerogenic potential. A postdoc in his lab is now specifically targeting T1D as the first proof-of-principle autoimmune application. The proposed experiment — reprogramming beta cells themselves into tolerogenic antigen-presenting cells so that islet antigens are presented in a tolerance-inducing context from within the target tissue — is among the most conceptually novel approaches to T1D immunotherapy described anywhere in the current literature.
🎯 Core Premise
REPROcode — a combinatorial, barcoded transcription factor screening platform operating at single-cell resolution — can systematically identify the minimal transcription factor combinations required to reprogram any cell type into a defined immune identity. Applied to immune cells, this platform has already decoded the transcription factor logic of multiple DC subsets, NK cells, and regulatory immune populations. Applied to T1D, it offers a route to engineering tolerogenic antigen-presenting cells on demand — either from patient-derived starting cells or, more radically, by reprogramming beta cells themselves into cells that present islet antigens in a tolerance-inducing context.
🌟 Why This Talk Matters to T1D Scientists and Clinicians
For scientists: REPROcode addresses a fundamental bottleneck in immune cell engineering: the field has lacked a systematic, combinatorial method to discover which transcription factor combinations drive specific immune identities. Classical approaches — iterative N-1 subtraction, purely computational prediction, or single-perturbation screens from iPSCs — each miss a critical dimension of combinatorial logic. REPROcode resolves this by expressing barcoded pools of transcription factors in fibroblasts, then using single-cell RNA sequencing to simultaneously read out which cells were reprogrammed, into what identity, by which combination of factors, at what stoichiometry, and with what fidelity. The entire experiment that previously took years can now be done in nine days. The platform is openly available — 408 barcoded immune transcription factors deposited in Addgene — and a ranked-list website allows any researcher to query their target immune cell type and receive prioritized candidate transcription factors. This is infrastructure the field can use immediately.
For clinicians: The clinical translation pathway Dr. Pereira has mapped for cancer immunotherapy — intratumoral delivery of a replication-deficient adenoviral vector encoding three transcription factors, with a clinical trial in melanoma and head and neck cancer beginning imminently — provides a tested delivery framework that could, in principle, be adapted for pancreatic or lymph node administration in T1D prevention. More immediately relevant to the T1D clinical community is the tolerogenic DC angle: a scalable, transcription-factor-defined method for generating patient-specific tolerogenic DCs loaded with islet antigens could address the manufacturing and reproducibility failures that have limited tolDC clinical trials to date. The Bart’s trial in 2022 demonstrated safety and durable reduction in T cell autoreactivity with islet antigen-pulsed tolDCs — but manufacturing remained the bottleneck. REPROcode is precisely the kind of platform that could break that bottleneck.
3️⃣ Big Takeaways
1. REPROcode solves the combinatorial transcription factor discovery problem at scale — in a single nine-day experiment. Previous methods for identifying reprogramming transcription factors required sequential, labor-intensive subtraction of individual factors from candidate pools — a process that could take years and still miss optimal stoichiometries or synergistic combinations. REPROcode uses barcoded lentiviral libraries, controlled multiplicity of infection (five to ten transcription factors per cell), and single-cell RNA sequencing to simultaneously identify the right combination, the right levels, the right stoichiometry, and auxiliary fidelity-enhancing factors — all in one experiment. Validation with the known CDC1 triad (PU.1, IRF8, BATF3) confirmed the platform works: these three factors emerged as the top hit from a pool of nine, then twenty-two, then forty-two transcription factors. Applied to the full 48-factor DC-enriched library, the platform generated not just CDC1s but also CDC2s, plasmacytoid DCs, macrophages, innate lymphoid cells — and constructed a transcription factor hierarchy tree mapping the branching logic of myeloid and lymphoid identity decisions.
2. The platform has already identified tolerogenic and regulatory immune identities directly relevant to T1D — including mregDCs and NK cells. Among the immune identities generated in the 48-factor screen, two drew specific attention for autoimmune applications. First, mregDCs — mature regulatory dendritic cells with documented capacity to activate regulatory T cells — emerged as a distinct population. Second, NCR1-positive NK cells with a cytotoxic program were generated, validated by degranulation assays, and characterized by single-cell profiling. Both are relevant to T1D: mregDCs for tolerance induction toward islet antigens, NK cells as the very regulatory effectors shown by Dr. Kevin Bode’s concurrent work to be essential for beta cell-autonomous immune protection. The convergence between Pereira’s reprogramming platform and Bode’s NK cell mechanism is striking — and points toward a potential future in which engineered tolerogenic DCs and regulatory NK cells are deployed together for combinatorial immune correction in T1D.
3. The most audacious T1D application is direct beta cell reprogramming into tolerogenic antigen-presenting cells — beta cells that teach the immune system to stand down. The concept Dr. Pereira described in discussion is genuinely novel: rather than manufacturing tolerogenic DCs externally and infusing them, use in vivo transcription factor delivery to reprogram residual beta cells in the pancreas — or beta cells in an SC-islet transplant — into mregDC-like cells that present islet antigens in a tolerogenic context directly at the site of autoimmune attack. Because the beta cell already contains all the relevant islet antigens, this approach would be inherently antigen-specific without requiring any separate antigen loading step. It is early-stage and the in vivo delivery and safety questions are substantial — but as a conceptual framework it is the most direct possible route to antigen-specific, site-directed immune tolerance in T1D.
❓ Key Questions from the Discussion
How stable are reprogrammed tolerogenic immune identities under inflammatory challenge? This question — raised by the host and directly echoing the same concern in Dr. Bode’s talk on RNLS-deficient beta cells — applies with equal force here. Tolerogenic DC states are context-dependent and can be overridden by pro-inflammatory signals. For in vivo applications in T1D, where the pancreatic microenvironment is chronically inflamed, the durability of transcription-factor-induced tolerogenic identity will be critical. Dr. Pereira noted that BCL6 expression appears to regulate the CCR7-positive migratory tolerogenic phenotype — suggesting that the stability of tolerogenic identity may itself be transcription-factor-tunable, and that REPROcode could be used to identify the factors that lock cells into a durable tolerogenic state rather than a plastic one.
Will reprogramming efficiency and fidelity hold in patient-derived cells with a dysregulated immune background? Dr. Pereira acknowledged that embryonic fibroblasts are used for initial screens precisely because of their epigenetic plasticity. Transition to adult patient-derived cells — the relevant starting material for autologous T1D therapy — adds variables including epigenetic memory, inflammatory conditioning, and patient-to-patient variability. His reassuring data point is that CDC1 reprogramming has been validated in over 70 different cell types from cell lines to patient samples, with efficiencies that vary but are not linearly determined by cell type. The key is that once the right transcription factor combination and stoichiometry are known, polycistronic delivery (all factors in a single vector) substantially improves efficiency — reducing the problem to a delivery and optimization challenge rather than a discovery one.
Could this platform be used to engineer exhausted autoreactive T cells toward regulatory phenotypes, rather than creating new cells from scratch? This question — whether transcription factor combinations could be delivered to existing autoreactive T cells to shift them from effector toward regulatory states — was raised in discussion and acknowledged as conceptually possible but technically more complex than APC reprogramming. T cell identity involves TCR rearrangement and developmental checkpoints that make full lineage conversion challenging. However, Dr. Pereira noted that using REPROcode to modulate T cell subsets and states — for example, shifting TH1 effectors toward Treg-like phenotypes — without full lineage conversion is a more tractable near-term application. This framing positions the platform not just as a cell manufacturing tool but as a state-modulation tool applicable to endogenous immune cells in vivo.
🔗 3 TSS Talks That Connect With This One
1. Jeff Bluestone, PhD — Sonoma Biotherapeutics Ask the Expert Dr. Bluestone’s foundational work on regulatory T cells and tolerance induction in T1D — and his clinical program at Sonoma developing Treg-based cell therapies — is the most direct clinical complement to Dr. Pereira’s platform. REPROcode’s ability to generate mregDCs that activate Tregs positions it as a potential upstream manufacturing tool for exactly the kind of tolerance-restoration strategies Dr. Bluestone has pioneered. The conversation between engineered tolerogenic DCs and engineered Tregs is one the T1D field urgently needs to have. ▶️ Watch here
2. Kevan Herold, MD & Matthias Von Herrath, MD — Yale and UCSD Ask the Expert Drs. Herold and Von Herrath represent the leading edge of clinical immune intervention in T1D — teplizumab’s approval, anti-CD3 mechanisms, and the broader landscape of immunotherapy trials. Their discussion of what immune modulation can and cannot achieve in established T1D provides the clinical context that frames exactly where a tolerogenic DC reprogramming approach could fit: in the prevention window, in combination with existing agents, or as a complement to beta cell replacement. The question of whether antigen-specific tolerance induction could extend the window of teplizumab benefit is directly relevant. ▶️ Watch here
3. Le Cong, PhD — Stanford University TSS Think Tank Dr. Cong’s work on CRISPR-based gene circuit engineering for T1D applications shares the same conceptual foundation as REPROcode — using defined genetic programs to re-engineer cell identity and behavior in disease-relevant contexts. His perspective on the design principles, delivery challenges, and regulatory considerations of in vivo genetic programming is directly applicable to the in vivo dendritic cell reprogramming approach Dr. Pereira is developing for autoimmune applications. ▶️ Watch here