First Look: Block I

Session Focus

Kingston opened by describing First Look as the format in which Mass General Brigham investigators present their most exciting ideas, giving the audience a front row seat to innovations emerging across the system. He noted the difficulty of narrowing the field to 12 presenters.

Each presentation runs roughly nine minutes and closes with a specific ask. Moderator introductions, summarized below alongside each talk, carried the investigator’s credentials and the nature of the partnership sought.

Targeting Reductive Stress — Vamsi Mootha MD

Kingston introduced Mootha as a member of the Molecular Biology Department and the Center for Genomic Medicine at Mass General, a MacArthur Prize winner, and an elected member of both the National Academy of Sciences and the National Academy of Medicine. The framing: the field long believed ATP was the key to mitochondria, and his work points elsewhere.

Mitochondria produce roughly 90 percent of cellular ATP by burning food with oxygen in a controlled way. More than 300 monogenic forms of mitochondrial disease exist, typically affecting children born with broken mitochondria. Secondary dysfunction appears in type 2 diabetes, inflammation and sepsis, and everyone loses a few percentage points of mitochondrial number and activity each decade after roughly age 20 or 30.

For more than 15 years his lab used metabolomics and proteomics to look for blood signatures of that dysfunction. They did not find signatures of ATP starvation. They found metabolites and proteins reflecting a buildup of electrons, which spill from mitochondria into the cell and then into the bloodstream. He calls this reducing equivalent stress, or reductive stress, in contradistinction to the oxidative stress the field has focused on.

The markers appear in rare mitochondrial disease, rise in sepsis, rank among the strongest predictors of future diabetes, drop rapidly after GLP-1 therapy and are among the strongest markers of aging itself.

To test causality the lab built tools using bacterial enzymes rather than gene editing. LbNOX catalyzes full reduction of oxygen to water, taking electrons from NADH and dissipating them cleanly. An engineered enzyme called LOXCAT targets circulating lactate, converting it to pyruvate using oxygen, also producing water. Both rescue disease phenotypes in cell culture models including proliferative defects. Delivered virally to mouse liver, LbNOX brought artificially elevated markers back down and improved impaired glucose tolerance in a diet-induced obesity model.

Next steps involve modern protein design including AI to reduce immunogenicity and extend circulatory half-life of these bacterial enzymes. The clinical plan starts with citrin deficiency, one of the most common rare mitochondrial diseases in East Asia and effectively a genetically pure form of reductive stress, then moves toward common conditions using biomarker-enabled trials that both identify appropriate patients and quantify therapeutic response.

The ask: partners to build the first entity targeting reductive stress, or an organization interested in licensing the technology.

Cytagonists — Lloyd Bod PhD

Kingston introduced Bod as a principal investigator in the Krantz Family Center for Cancer Research at MGH and an assistant professor of medicine at Harvard Medical School, whose team found that roughly half of pancreatic cancer patients with poor outcomes carry autoantibodies neutralizing interferon gamma.

The lab studies B cells and antibody responses across cancers with a focus on hard-to-treat disease. Immunotherapy has transformed care in cancers such as melanoma and is essentially ineffective in pancreatic cancer.

Interferon gamma is double-edged in a useful way: it boosts immune attack on cancer cells and acts directly on cancer cells to stop growth. The lab found that about half of pancreatic cancer patients produce antibodies against it, and that patients with high levels had poor survival.

The effect proved causal rather than correlative. Using an engineered reporter cell line that luminesces in response to interferon gamma, serum from short-survival patients suppressed that activity. In cancer cell culture, interferon gamma slowed growth, and patient serum containing the antibodies allowed growth to resume.

Rather than stopping at the finding, the lab designed peptide mimetic decoys, which it calls cytagonists, to intercept the harmful antibodies as appetizing baits and restore natural interferon gamma signaling. The first run produced four candidates across two peptides in two orientations. Cytagonist 4, the lead, significantly restored interferon gamma activity in the reporter system and slowed growth in patient-derived cancer cell cultures.

The platform logic runs in a loop: identify a harmful antibody, design a cytagonist against it using AI-guided modeling, neutralize, repeat. Three additional cytagonists have been designed against other pancreatic cancer targets, with validation underway in other cancer types and clear application to autoimmune, inflammatory and infectious disease, where pathogenic antibodies are abundant.

On market, the team estimates roughly 34,000 US pancreatic cancer patients would be affected, representing about $5 billion annually.

The ask: $7.5 million to form a new company and advance the lead cytagonist through in vivo validation toward IND filing. IP has been filed on the lead series.

Organs Should Wait for Patients — Korkut Uygun PhD

Uygun introduced himself as an engineer by training, and made his thesis plain: instead of an organ transplant waiting list, organs should wait for patients.

Roughly a third of all deaths in the United States could be significantly postponed with replacement organs. Transplantation is the treatment, and he called it the best longevity medicine currently in existence: a patient dying of organ failure receives a transplant and lives another one or two decades.

The problem is that transplantation has no supply chain. An organ must be transplanted within hours of procurement because no technology preserves it longer. That produces emergency surgery choreographed across five or six operating rooms in different locations, with teams and organs flown to the recipient hospital, at substantial cost to CMS and insurers. The current preservation technology is a bag of ice, in which the organ is slowly suffocating.

Machine perfusion over the last five to ten years provides an artificial body supplying oxygen and nutrients, and more than doubled liver transplant volume at Mass General Brigham. It extends preservation to perhaps a day or two, remains costly, and the organ still degrades.

The alternative is precision freezing, using sub-zero temperatures to arrest metabolism for weeks and months. The model is the Canadian wood frog, which freezes to minus 8 degrees Celsius for up to seven months and hops away in spring. The frog raises intracellular glucose before freezing, which acts as both lipid chaperone and metabolic suppressant, and exerts precise control over where and when ice forms: in vascular spaces, protecting the parenchyma from osmotic shock, homogeneously, and at temperatures just below freezing, producing soft slushy ice that does not rupture cell membranes.

Three technologies reproduce this. Machine perfusion infuses and removes cryoprotectants and handles quality assessment. A non-metabolizable glucose used clinically in insulin resistance studies enabled stable preservation at minus 5 degrees for human livers for about four days. The final breakthrough was Snowmax, a commercial ice nucleator approved by USDA for crop protection and familiar to anyone who has skied in recent decades, which allows discrete control of ice formation.

He showed a pig liver frozen four days, with team members removing the soft ice by hand without tools. The protocol integrates into clinical workflow, with the clinician seeing only normothermic perfusion at the last mile.

The work spans more than 10 years and 100 publications, funded through non-dilutive federal support. In April 2026 the team froze kidneys in Boston, flew them by private jet to Paris with collaborators at the Paris Transplant Institute, and transplanted one, with the kidney remaining healthy in the animal a week later. He was explicit that this was not a life-saving transplant and functional recovery was not the objective.

The ask: a $20 million seed round for the newly formed company to fund IND-enabling device work, large animal studies in a GLP lab and a clinical trial plan. Competing technologies currently claim around five days; this approach targets weeks and months. The beachhead market is sensitized patients, with xenotransplantation as the larger target.

Radiation-AMPlified (RAMP) Medicines — Miles Miller PhD

Kingston introduced Miller as an associate professor of radiology at Harvard Medical School and principal investigator in the Center for Systems Biology, where he directs the targeted delivery platform, and a recipient of the NIH Director’s New Innovator Award. The introduction noted that in models, confining activation to the radiation field allowed more than 20 times the payload dose possible with conventional linkers.

About half of all cancer patients receive radiation, delivered with high spatial precision to nearly any anatomical site. Many tumors are or become radioresistant. Clinicians combine radiation with systemic chemotherapy, which travels everywhere and exposes normal tissue to cytotoxic agents, producing dose-limiting and sometimes fatal side effects. Even targeted antibodies accumulate only fractionally at the disease site.

RAMP converts radiation’s spatial precision into control over drug chemistry. A patient receives a long-circulating, biologically inactive, chemically caged drug intravenously, which accumulates somewhat in tumor tissue. Standard localized radiation then deposits a form of energy not otherwise found in the body at high levels, triggering a chemical reaction that activates the payload within the tumor.

The advantages over conventional delivery are that it does not depend on endogenous biochemistry, requiring no receptor internalization, particular tumor pH or expressed protease, and relies instead on radiation physics. It remains compatible with standard ADC manufacturing, and because it circulates long, one infusion can in principle support activation across multiple treatment cycles, which matters because radiation is typically delivered daily over weeks.

The proof of principle is a RAMP ADC targeting EGFR and MET, using MMAE, a payload class found in roughly half of clinical ADCs. Where a conventional ADC relies on cathepsin cleavage, RAMP relies on radiolysis products: radiation splits water, generating reactive radicals that reduce an aryl halide and trigger an electron transfer cascade releasing the payload. Release was measured by mass spectrometry against clinically relevant radiation doses, and the linker remains compatible with standard bioconjugation and other payload classes including topoisomerase inhibition and immune receptor agonism.

In cytotoxicity assays and colony formation assays, the combination overcame resistance to radiation alone and to antibody alone. In a RAS-mutant xenograft model, RAMP ADC plus local radiation delayed progression and extended survival where monotherapies and the antibody-radiation combination did not. Biodistribution by mass spectrometry showed active payload selectively in irradiated tumor tissue. Compared against a conventional cathepsin-cleaved dipeptide linker, the platform showed mitigated weight loss, hepatotoxicity and myelotoxicity.

Two leads are in development sharing the same uncaging chemistry. An albumin-bound RAMP cytotoxin exploits serum albumin’s extended circulation and passive tumor accumulation. An antibody-based RAMP ADC uses de-risked components: fibroblast activation protein as the target and MMAE as the payload. Target indications are cancers already treated with radiation where radiation alone insufficiently controls disease, including locally advanced head and neck squamous cell carcinoma and unresectable lung cancers.

The ask: investment and strategic partners for parallel IND-enabling development covering candidate selection, CMC optimization, toxicology and first indication trial design. Linker chemistry is patent pending through Mass General Brigham Innovation.

TRACE — Gaurav Gaiha MD, DPhil

Kingston introduced Gaiha as a core member of the Ragon Institute of Mass General Brigham, MIT and Harvard, a gastroenterologist at MGH and associate professor of medicine at Harvard Medical School, framing the problem as protein AI now generating improvements to biologics faster than laboratories can test them.

Biologics are among the fastest-growing therapeutic classes, but discovering a promising one is only the beginning. Each must clear an optimization bottleneck spanning affinity, specificity, solubility and polyreactivity. That bottleneck sits within a $626 billion biologics market expected to double to $1.3 trillion over eight years, while only about 8 percent of lead candidates entering phase one reach FDA approval. Poor developability associates with more than 20 percent higher likelihood of non-approval, and lead optimization is already a $5 billion annual discovery services market.

Current optimization is slow, iterative and expensive: build libraries, express and display, screen, isolate hits, characterize, redesign, over cycles taking months. Optimizing one parameter such as affinity frequently compromises solubility or stability, forcing additional rounds.

TRACE stands for targeted reticulase assisted continuous evolution and combines two components. Targeted hypermutation uses a reverse transcriptase with more than 10,000-fold greater mutagenic potential than HIV reverse transcriptase, precisely diversifying triple adenine codons placed in a user-defined region, which produces focused diversity at rational sites rather than broad random mutagenesis. That is coupled with continuous selection through phage-based propagation, so fitness links to propagation and better-performing variants gain replication advantage and enrich across cycles.

Proof of concept used trastuzumab. Over five passages of roughly 18 hours each, totaling under four days, the platform selected an H91 variant in four of four replicates, a literature-validated mutant with 2.4-fold stronger binding affinity than wild type. A separate campaign targeting aggregation substantially reduced the multiple aggregate species visible in the wild-type size exclusion profile, producing a much cleaner monomeric peak, again in under four days.

The multiparametric run was the key test. In about 13 days the platform produced a 6 degree increase in melting temperature, a hydrodynamic radius decrease reflecting 1.3-fold lower self-association, 2.7-fold lower polyreactivity and a 6.7-fold affinity gain simultaneously.

The forward path positions TRACE as the experimental engine for AI protein design. Protein language models and structure prediction can nominate positions and generate thousands to millions of candidate mutations; TRACE tests and evolves them, producing AI-optimized biologics with experimental validation.

The platform has been applied to an HIV therapeutic biologic and is expanding into oncology and autoimmune targets.

The ask: approximately $10 million to accelerate automation, scale multiparametric TRACE, advance the first proprietary lead through IND-enabling studies and launch a new company.

Keeping Patients on Their Meds — Sherrie Divito MD, PhD

Divito introduced the problem as one that could affect anyone in the room and any therapy under development. Drug hypersensitivity reactions affect millions of patients annually in inpatient and outpatient settings, can affect any organ, range from mild to lethal, affect any age and any demographic, and can be caused by any drug including prescription, over-the-counter and supplements.

They can also be indistinguishable from reactions caused by CAR-T, stem cell therapy and immune checkpoint inhibitors. Because many patients take multiple potential culprits simultaneously and no reliable test exists, clinicians discontinue all of them. That means stopping cancer treatments and antibiotics and substituting less effective, less safe and more expensive alternatives, or leaving the underlying disease untreated where no alternative exists.

She framed it as a public health problem, not only an individual one: substantial cost to the system, a contribution to antibiotic resistance, and an impact on FDA approval for otherwise efficacious therapies.

Her clinical case illustrated the stakes. A 50-year-old woman with aggressive gynecologic cancer received a three-drug combination and reacted mildly, then reacted again on retreatment. Experts attributed it confidently to one agent and cleared the other two. She received them and the reaction progressed to a highly lethal form, nearly killing her. Oncology switched to second-line treatment and her tumor progressed. Experts again identified a culprit and cleared a third drug; she received it and reacted again, ending up hospitalized on IV immunosuppression. Her conclusion: clinical expertise is not enough, and a test is needed.

No FDA or CMS approved test exists in the United States. Alternatives have suffered poor sensitivity and specificity, and experts have low confidence in them, so they are rarely used.

The scientific basis rests on clonality. These reactions are T cell mediated, and each T cell expresses a highly unique T cell receptor she compared to a fingerprint. A drug-reactive T cell encountering that drug proliferates, and all daughter clones carry the same receptor, identifiable by sequencing. Nothing in T cell biology is more specific than reactivity and receptor sequence.

The test cultures patient samples with culprit drugs and controls, expands drug-reactive T cells and identifies them by receptor sequencing. Foundational work is published, with additional unpublished patient data supporting sensitivity, specificity and clonal persistence, meaning reactive clones remain detectable years after resolution.

In one patient, a clinical biopsy during active disease identified two pathogenic clones, which expanded in the assay against the culprit drug. The patient was switched to a chemically distinct treatment and reacted again; biopsy showed the original clones still present but not mediating the second reaction, which was driven by a distinct clone. Secondary assays visualizing T cell colonies and cytokine assays confirming activation support the approach.

Commercially there is a full US patent pending with no real competition, a pathway to LDT approval in roughly 18 months and FDA approval in three to four years, and favorable CPT benchmark analysis. The work received a near-perfect score on a first-submission R01 and won the 2024 Bright Futures Prize, with collaboration across Mass General Brigham Innovation, MGB Ventures and MGB Amplify.

Next steps are broadening the number of testable drugs, evaluating mimicker reactions and expanding a prospective study enrolling diseased patients and drug-tolerant controls. The primary goal is moving the test out of her lab into CLIA-certified space to enable LDT approval.

The ask: partnership and licensing opportunities, with the work particularly amenable to a sponsored research agreement.

Key Takeaways

1. Blood markers of reductive stress appear across rare mitochondrial disease, sepsis, diabetes risk and aging, with enzyme tools that lowered them and improved glucose tolerance in mice.

2. Roughly half of pancreatic cancer patients produce interferon gamma autoantibodies, and peptide decoys restored signaling and slowed tumor growth in patient-derived cultures.

3. Precision freezing holds organs for weeks rather than days, demonstrated by a Boston-to-Paris kidney transplant in April 2026.

4. Radiation can act as an activation switch, confining cytotoxic payload release to the treatment field and reducing systemic toxicity.

5. TRACE optimized four biophysical parameters of trastuzumab simultaneously in about 13 days, including a 6.7-fold affinity gain.

6. T cell receptor sequencing can identify the culprit drug in hypersensitivity reactions, where no approved test currently exists.