Revealing the Structural Fingerprints of Disease: Q&A with Faraz Choudhury

Immuto CEO Faraz Choudhury spoke with Pharmaceutical Executive about his company’s work with focusing on Surface Protein Conformations (SPCs) for target discovery. According to him, traditional target discovery methods can overlook pathologic differences in protein structures, while this new method can provide more accurate results.

Pharmaceutical Executive: How are traditional target discovery methods not meeting their full potential?
Faraz Choudhury: Traditional target discovery has largely focused on sequence and abundance. Genomics identifies mutations. Transcriptomics and proteomics measure expression levels. Those approaches have been extraordinarily powerful, but they leave out an important dimension of biology: structure.

In many diseases, especially cancer, the protein sequence is unchanged. The protein is wild type. What changes is its conformation, how it folds, how it assembles into complexes, and how it is presented on the cell surface under disease-specific stress and signaling conditions. When we define targets only by mutation or overexpression, we miss these structural distinctions.

That limitation has practical consequences. Many oncology targets are present on both tumor and healthy tissues. As a result, therapies often face narrow therapeutic windows, dose-limiting toxicities, and high clinical attrition. In other words, the field has been trying to extract selectivity from differences in quantity, when in many cases the more meaningful difference is in shape.

To reach their full potential, target discovery methods need to incorporate structure directly in disease-relevant contexts. That is the gap we set out to address.

PE: How do Surface Protein Conformations work?
Faraz Choudhury: Surface Protein Conformations refer to disease-specific three-dimensional forms of proteins that appear on the cell surface under pathological conditions.

Proteins are not static entities. Their conformation depends on the cellular environment, including signaling activity, metabolic state, oxidative stress, trafficking dynamics, and interactions with other proteins. In a tumor microenvironment, those pressures are fundamentally different from those in healthy tissue. The same protein, with the same amino acid sequence, can adopt a different structural state.

When that conformational shift occurs at the cell surface, it can expose epitopes that are not accessible in the healthy state. These disease-associated conformations effectively become structural fingerprints of pathology.

Importantly, the underlying sequence does not need to change. We are not relying on mutation. Instead, we are detecting a structural state that is tied to disease biology. That structural state can then be selectively engaged by a therapeutic binder designed to recognize the disease conformation but not the normal form.

PE: How does Immuto’s platform work?
Faraz Choudhury: Our platform integrates structural proteomics in living, disease-relevant systems with AI-driven modeling and antibody design.

We begin with patient-derived or clinically relevant disease models. Rather than purifying proteins, we interrogate them in intact cellular contexts. Using a proprietary workflow, we label and analyze surface-exposed regions of proteins in living models, then use high-resolution mass spectrometry as the readout. This allows us to detect structural differences across thousands of proteins simultaneously.

The output is a structural map of the surfaceome, not just which proteins are present, but how their conformations differ between diseased and healthy states. That dataset is high dimensional and information-rich, so AI plays a central role. We use machine learning to identify statistically robust conformational differences, prioritize targets, and model the three-dimensional features that define disease specificity.

Once a disease-specific conformation is identified, we transition to therapeutic design. Structural constraints derived from the proteomic data guide antibody generation. We design and screen binders directly against the conformational epitope, and we counter-screen against the normal form to minimize cross-reactivity. This creates a direct line from structural discovery to conformation-selective therapeutics.

PE: How does Immuto define the fingerprint of a disease?
Faraz Choudhury: We think of a disease fingerprint as a reproducible structural pattern that distinguishes diseased cells from healthy counterparts.

In practice, that means identifying conformational features that meet several criteria. First, the structural state must be consistently observed across multiple patient-derived samples. Second, it must be stable under disease-relevant conditions, not an artifact of sample handling. Third, it should demonstrate absence or minimal presence in matched healthy tissues.

Because we work in native or near-native systems, we are able to see how proteins behave under the actual stresses that define pathology. In cancer, for example, hypoxia, metabolic shifts, and chronic signaling reshape surface protein architecture. When those structural features recur across heterogeneous samples, they form a conformational signature.

That signature becomes the disease fingerprint. It is not defined by genotype alone, but by how proteins are configured and presented in the pathological state. This allows us to define precision in structural terms rather than only in genetic terms.

What successes have Immuto seen with its method?
Faraz Choudhury: We are rapidly advancing five pipeline programs. Our most advanced internal example is in acute myeloid leukemia. Instead of selecting a target based on overexpression or mutation, we examined structural differences directly in patient-derived samples. We identified a conformationally distinct surface feature that appeared consistently across AML samples and was not tied to a specific genetic alteration.

When we generated binders against that disease-specific conformation, we observed selective engagement of the AML-associated form and minimal binding to the corresponding normal form in early assays. In xenograft models, we observed strong in vivo activity, and in preclinical safety assessments we did not see evidence of toxicity to healthy hematopoietic stem cells under the tested conditions.

More broadly, across programs, this approach has changed how we prioritize targets. We routinely identify structurally differentiated proteins that would not have emerged from mutation or expression analysis alone. At the same time, structural validation has allowed us to deprioritize candidates that do not maintain disease specificity in more complex models.

The consistent theme is that measuring structure in native disease contexts yields a different class of targets. Those targets are defined by conformational selectivity, which we believe is central to improving specificity, widening therapeutic windows, and reducing downstream attrition in drug development.