Bryan Yoo is co-founder and CEO of Nuanced Health. Bryan completed his PhD in Cellular and Molecular Neurobiology at Caltech in the Mazmanian Lab. Previously, he was a venture fellow at Flagship Pioneering and an operator at multiple early-stage startups. Bryan graduated from Stanford University with a B.S. in Biology and enjoys spending time with his daughter and following the 49ers.
Drug development is extremely time consuming, expensive, and high-risk. Biopharma companies spend tens or hundreds of millions of dollars to bring a single therapeutic candidate through clinical trials, yet only 12% of candidates that enter clinical trials achieve FDA approval. When failed programs are factored in, it takes 10-15 years and $2.6B to develop one new medicine. These failures represent enormous sunk R&D costs for biopharma and highlight the need for technologies that de-risk assets at earlier stages of development. Even among drugs that do secure FDA approval, response rates are remarkably low, plateauing around 50% for many blockbuster drugs. These inefficiencies in drug development, both in approval and response rates, indicate an industry-wide failure to translate extensive, costly preclinical research to the clinic.
Our first post outlined one strategy to bridge this so-called “translation gap” in preclinical research: generating animal models that more accurately represent human biology. Specifically, recent evidence of the central role of microbial symbiosis in human health and disease implicates a glaring need for preclinical models that accurately represent the human gut microbiome. Simultaneously, the enormous disease burden created by environmental risk factors and the challenges of large cohort human studies indicate a need for more efficient ways to interrogate the environmental determinants of health.
Generating preclinical research models which accurately represent the microbial heterogeneity seen in the clinic is an immediate way to address the “translation gap” and increase the efficiency of existing R&D workflows. In addition, representation of this complex symbiosis in animal models presents an exciting opportunity to study the environmental determinants of health, understand novel biology, and develop more effective therapeutics.
Clinical studies of human-to-human Fecal Microbiota Transplant (FMT) have shown that transferring a healthy or drug-susceptible microbiome to a diseased or drug-resistant patient can transfer the favorable phenotypes as well. For example, FMT from a patient responding well to immuno-oncology therapies into a previously non-responsive patient has been shown to induce clinical response in the recipient. Using the same principles as FMT, the Mazmanian Lab at Caltech and others have demonstrated that mice with “humanized” microbiomes (engrafted from a human stool donor) also possess the molecular and phenotypic profiles of that human donor. These seminal human-to-human and human-to-mouse FMT studies suggest that transferring the microbiome also transfers critical biological function.
In the early 2000s, the generation of mice with “humanized” immune systems revolutionized drug development by recreating human immune profiles in vivo. These models unlocked a novel way to study this complex system and facilitated the development of blockbuster immuno-oncology and inflammatory disease therapies. Two decades ago, humanized immune systems transformed drug discovery. Today, humanized microbiomes can do the same.
Nuanced Health systematically humanizes the mouse microbiome at scale, creating a novel drug discovery and development platform that leverages, rather than ignores, the heterogeneity of real-world populations.
It all starts with germ-free mice, a “blank slate” animal completely devoid of bacteria. Through rigorous quality control standards and protocols, we ensure that these mice are never exposed to any of the pathogens common on surfaces and in the air.
Nuanced Health sources human fecal samples from diverse, clinically relevant patient populations and engrafts these samples into germ-free mice. One human sample is engrafted into one male and one female mouse, effectively creating a mouse analogue of each human. This one-to-one engraftment transfers not only the individual bacterial species, but also community-level microbial interactions. Each male and female pair is then bred, creating a novel humanized mouse line harboring the same unique, human microbiome.
By repeating this process at scale with many human samples, we have generated a preclinical platform consisting of hundreds of humanized mouse lines. Individually, the mouse lines each recreate the complex biology of one unique human. Collectively, they recreate the enormous heterogeneity of the human population.
These methods and this platform have been extensively validated for robustness, stability, and reproducibility. Our high-fidelity humanization protocols ensure high engraftment rates of hundreds of microbial species, retaining not only the bacteria present but also their relative abundances. Additionally, the humanized microbiome is passed on through breeding: our humanized mice have stable microbiomes for over five generations with very minimal drift. After five generations, fecal samples from humanized mice or from the original human donors can be used to re-start a given humanized line from germ-free, effectively allowing the infinite propagation of a given humanized line.
Nuanced Health’s humanized mice are highly similar to their human donors, both molecularly and phenotypically. The relationships shown in the Venn diagrams below hold true whether analyzing microbes themselves, microbially derived secondary metabolites, circulating small-molecules, proteins, gene expression, or transcriptome sequences.
Additionally, humanized mice represent the diversity of human biology in never-before-seen fidelity and dynamic range. The principal component analysis below represents the microbiomes of five of our human fecal samples and the humanized mouse lines that they generated, as well as commercially available mice from leading vendors. The plot not only shows that Nuanced Mice better represent the taxonomical makeup of humans (overlap between the groups), but that they also better recreate the diversity of the human population (spread within groups).
By recapitulating human biology at never-before-seen fidelity and dynamic range, this in vivo platform can discover novel therapeutics, improve the efficacy of existing ones, and better predict response rates and adverse events in the clinic. Because of the microbiome’s systemic effects on our bodies, this platform is agnostic to indication and therapeutic modality.
Working with thousands of mice derived from hundreds of human donors, we have observed wide phenotypic diversity among our humanized mice in both cancer and in obesity. That is, when holding everything else constant, mice with microbiota engrafted from unique human donors have significantly divergent tumor growth or weight gain profiles. Our humanized mouse lines exist on spectra from fast to slow tumor growth and from fast to slow weight gain.
Additionally, when these mice are treated with first-line checkpoint blockade drugs or GLP-1 agonists, we observe similarly diverse drug response profiles: some humanized lines respond completely, while others do not respond at all. From our cancer and obesity programs, we have generated unique, massive, multimodal data sets and have identified candidate small molecule, peptide, and microbial therapeutics.
Through the acquisition of high resolution in vivo multi-omic data at scale, Nuanced Health’s platform is uniquely able to identify actionable insights based on human heterogeneity. We aim to add value to pre-IND drug candidates, agnostic of indication or therapeutic modality, in multiple ways:
First, the Nuanced platform is uniquely positioned for predictive biomarker discovery. There are over 20,000 more microbially derived small molecules and proteins shared between Nuanced mice and humans than between commercially available mice and humans. Additionally, we found that 60% of the circulating small molecules in mouse plasma are only present in mice with a gut microbiome, not in germ-free animals. These represent an untapped biological space which can only be studied preclinically through the Nuanced platform. By analyzing this novel molecular landscape with high fidelity and large sample sizes, Nuanced Health can identify key biomarkers related to both disease outcomes and drug responses.
Second, following from biomarker discovery, Nuanced mice can be used to inform patient stratification ahead of clinical trials and identify addressable patient populations. These mice show heterogeneous disease progression and drug efficacy phenotypes, similar to those seen in the clinic. In humanized mice, favorable microbial or metabolomic profiles stand out, and these profiles can be used to identify target populations for human trials.
Third, humanized mice predict clinical toxicities better than conventional mice, allowing researchers to de-risk drug candidates and prioritize preclinical pipelines. Predicting adverse events before clinical trials is critical for addressing the aforementioned “translation gap” and for preventing harm toward trial participants, and Nuanced mice contain many microbial enzymes and secondary metabolites that are known to cause or prevent drug toxicity in humans. Pre-IND toxicity identification can drastically increase the efficiency of drug development, giving researchers the opportunity to adjust their drug design, dosing, and DMPK profile before going to the clinic.
Finally, Nuanced Health aims to identify functional biomarkers that have potential as combination therapies for both internal and external therapeutic assets. Nuanced mice show a wide range of drug efficacy across various high-value indications and therapeutic modalities, implying that there are mechanistically relevant proteins and molecules that are differentially expressed among mice with different microbiomes. We acquire multi-omic data from Nuanced mice and use machine learning to model the predictiveness of tens of thousands of biological features. From our cancer program, we have identified 23 (85% hit rate) naturally occurring, innately safe metabolites with bioactivity relevant to checkpoint blockade efficacy.
Through biomarker discovery, patient stratification, toxicity prediction, and de-novo discovery, the Nuanced Platform can add significant value to Pre-IND drug candidates while de-risking their path to approval.
At Nuanced Health, we are laser focused on finding new ways to help patients who are not served by existing treatments. Over half of patients prescribed blockbuster drugs for everything from IBD to cancer will not respond, and we believe that the reductionist approaches taken to avoid heterogeneity in drug discovery and development are a critical cause of these response plateaus.
By reintroducing microbial diversity into preclinical animal models, we can begin to understand how our environmental and behavioral differences shape disease progression and drug response for individuals. We leverage human diversity, unlocking a new biological space previously inaccessible to preclinical researchers. The Nuanced platform recreates both the biology of individuals and the heterogeneity of populations, representing a paradigm shift for drug discovery and development. We utilize cutting-edge bioinformatic and machine learning algorithms to analyze diverse samples from our mice and their microbiomes, revealing novel therapeutic targets, identifying potential combination therapies, and predicting clinical outcomes. At Nuanced Health, we model all humans to create therapeutics that work for everyone.
Interested in learning more? We’d love to hear from you at info@nuancedhealth.com.
Led by Caltech co-founders Sarkis Mazmanian and Bryan Yoo, Nuanced Health uncovers the environment’s biological imprint on human health, fundamentally advancing our understanding of diseases and our ability to treat them. By systematically introducing microbial variability into preclinical models, we elucidate complex molecular, cellular, and physiological impact of the microbiome on health. These next-generation models recreate human biochemistry, physiology, disease progression, and drug response at never-before-seen fidelity and scale. By leveraging high-resolution, diverse multi-omic data, we develop therapeutics that are evolutionarily designed for humans, enhance the efficacy of existing therapeutic modalities, and maximize the efficiency of drug development paradigms. Nuanced was founded in 2021 with a transformative thesis backed by lead investment from 8VC and co-investors Freeflow, Gaingels, Open Field Capital, Heritage Group, Luma Bio-IT SPV, Presight Capital, Axial, January River, and other top investors, angels, and operators.
