We introduce a new way to discover antibodies that enables epitope specification and unprecedented speed (24 hours to sequences, 2 weeks to KD determination) using our generative AI model, Chai-2. We discuss advanced, lab-validated case examples for antibody design leveraging Chai-2.

Aberrant signaling of the tumor necrosis factor receptor family has significant detrimental effects in multiple diseases. Ligand competition impacts multiple pathways, causing numerous side effects, and is challenged by native potency and high local concentrations. Synthetic scaffolds were engineered to bind receptors (separately TNFR1 and DR5) and inhibit signaling and downstream processes without competing for native ligand binding. We present on mechanism and engineered stability and cross-reactivity.

Immune checkpoint inhibitor antibodies have shown great success in a subset of patients; however, many treated patients (>70%) do not benefit. Towards providing a more effective therapy for these patients, the Spangler lab has developed a multispecific anti-PD-L1 antibody-drug conjugate. This molecule kills cancer cells through 3 mechanisms: disruption of the PD-1/PD-L1 immune checkpoint, internalization and downregulation of PD-L1, and direct killing of cancer cells via the drug payload.

Most fatal diseases occur in nonleaky solid tissues difficult to specifically target and treat even with our best precision therapies. Poor passive transvascular delivery limits specific uptake, target access and therapeutic efficacy. Active tranendothelial shuttling of size-optimized bispecific antibodies via the caveolae pumping system can boost precision targeting and drug potency by orders of magnitude.

Matrix metalloproteinases (MMPs) and a disintegrin and metalloproteinases (ADAMs) are essential for extracellular matrix remodeling and cell signaling, but their dysregulation is implicated in diseases such as cancer, neurodegeneration, and gynecological disorders. To develop targeted inhibitors, we are pursuing two complementary strategies: engineering natural tissue inhibitors of metalloproteinases (TIMPs) and selecting synthetic single-chain variable fragments (scFvs). Using yeast surface display, FACS, and next-generation sequencing, we identified high-affinity binders. Machine learning and structural modeling are applied to uncover sequence-function relationships, enabling the rational optimization of TIMP and scFv variants as next-generation therapeutic agents targeting metalloproteinase activity with improved affinity and specificity.

The therapeutic index of radioimmunotherapy describes the balance between the accumulation and retention of cytotoxic radiation in the tumor, the blood half-life, and the clearance of the radioactive tracer. While the long plasma half-life of IgGs can cause hematological toxicities, the rapid renal clearance of smaller fragments often leads to nephrotoxicity. Antibody fragments (scFv-Fc) with hepatobiliary clearance can spare the radiosensitive kidneys. Fc-engineering to modulate pharmacokinetics reduces bone marrow toxicity. We have generated variants of an anti-prostate stem cell antigen (PSCA) scFv-Fc and use 89Zr-immunoPET to study the tumor targeting, biodistribution, and clearance in human PSCA knock-in mice.

• CD98hc vs transferrin receptor shuttling: pros/cons & applications  
• Impact of bispecific antibody properties on CNS delivery efficiency, selectivity, and retention  
• Tailoring antibody shuttles to specific CNS drug delivery applications: nucleic acids, proteins, enzymes, IgGs, and more   
• Outstanding challenges in biologics delivery to the CNS and future directions
  

• Safety-by-design mechanisms
• Potency vs. selectivity engineering​
• Manufacturability & developability funnel: what are some early predicters of a go/no go?​​

Tissue factor (TF) is a transmembrane glycoprotein that plays an important role in the extrinsic coagulation cascade. TF is aberrantly expressed in various cancers. XB371 is an anti-TF antibody-drug conjugate developed using the SMARTag platform and is designed to deliver a cytotoxic payload to TF-expressing tumors while minimizing adverse events related to the disruption of TF function, notably bleeding. XB371 is composed of a tandem-cleavage topoisomerase-inhibitor-based linker payload conjugated to a monoclonal antibody that binds to TF with high affinity and does not interfere with the clotting cascade. Here, we describe and present the preclinical characterization of XB371.

Antibody drug conjugate (ADC) based therapeutics have revolutionized the field of drug development by achieving therapeutic responses that were previously unattainable with small molecule drugs. However, these therapeutics are susceptible to various modifications and degradation pathways during circulation, which may impact their stability and efficacy. Moreover, the increasing complexity of the protein scaffold requires additional effort to understand and assess their stability liabilities. In this presentation, we explore a suite of various in vitro characterization assays, for identification of liabilities such as aggregation, cleavage, and non specific payload release. This screening of biologics and drug conjugates workflow, guides project teams to select the best candidates within a short timeline.

JK06 is a biparatopic antibody-drug conjugate targeting two non-overlapping 5T4 epitopes with tetravalent binding capacity. This design enhances internalization and cytotoxic payload delivery in 5T4-expressing solid tumors, including lung, breast, ovarian, and colorectal cancers. Preclinical studies demonstrated superior internalization versus mono-specific antibodies and potent anti-tumor activity in xenograft models. JK06 showed favorable safety in GLP toxicology studies. A Phase 1/2 clinical trial is ongoing.

InduPro’s MINT™ platform enables high-resolution mapping of cell surface protein microenvironments by combining photocatalytic proximity labeling with machine learning, uncovering co-localized antigen pairs for rational bispecific therapeutic design. Applying this approach to EGFR-expressing tumors, we identified a novel tumor-associated proximity antigen that enabled selective dual targeting. IDP-001, the resulting bispecific ADC targeting EGFR, demonstrates potent and selective anti-tumor activity across diverse preclinical models. These findings illustrate the translational potential of spatial proteomics to guide first-in-class therapeutic development with improved precision and therapeutic index.

This talk highlights how artificial intelligence can accelerate the design and optimization of antibody-drug conjugates (ADCs), with a focus on improving internalization efficiency. By integrating high-throughput screening data with AI-driven modeling, we demonstrate how predictive tools can guide the development of more effective ADCs, ultimately enhancing therapeutic performance. The approach offers a scalable framework for rational ADC engineering with broad applications across oncology and beyond.

SOTIO is advancing a pipeline of ADCs through collaborations leveraging multiple leading site-specific conjugation and linker-payload technologies. Our preclinical pipeline of next-generation bispecific molecules aspires to minimize on-target toxicity and to overcome tumor heterogeneity. The talk will highlight our discoveries on how to optimize the bispecific format for optimal effect on tumor cells with varying antigen expression and discuss how to translate functional in vitro data into in vivo efficacy.