Precision immunotherapy

BiSpec Pairwise AI: guiding the selection of bispecific antibody target combinations with pairwise learning and GPT augmentation

PURPOSE

Bispecific antibodies (BsAbs), capable of targeting two antigens simultaneously, represent a significant advancement by employing dual mechanisms of action for tumor suppression. However, how to pair targets to develop effective and safe bispecific drugs is a major challenge for pharmaceutical companies.

METHODS

Using machine learning models, we refined the biological characteristics of currently approved or in clinical development BsAbs and analyzed hundreds of membrane proteins as bispecific targets to predict the likelihood of successful drug development for various target combinations. Moreover, to enhance the interpretability of prediction results in bispecific target combination, we combined machine learning models with Large Language Models (LLMs). Through a Retrieval-Augmented Generation (RAG) approach, we supplement each pair of bispecific targets' machine learning prediction with important features and rationales, generating interpretable analytical reports.

RESULTS

In this study, the XGBoost model with pairwise learning was employed to predict the druggability of BsAbs. By analyzing extensive data on BsAbs and designing features from perspectives such as target activity, safety, cell type specificity, pathway mechanism, and gene embedding representation, our model is able to predict target combinations of BsAbs with high market potential. Specifically, we integrated XGBoost with the GPT model to discuss the efficacy of each bispecific target pair, thereby aiding the decision-making for drug developers.

CONCLUSION

The novelty of this study lies in the integration of machine learning and GPT techniques to provide a novel framework for the design of BsAbs drugs. This holistic approach not only improves prediction accuracy, but also enhances the interpretability and innovativeness of drug design.

Extracellular Matrix-Trapped Bioinspired Lipoprotein Prolongs Tumor Retention to Potentiate Antitumor Immunity

The immunomodulatory effects of many therapeutic agents are significantly challenged by their insufficient delivery efficiency and short retention time in tumors. Regarding the distinctively upregulated fibronectin (FN1) and tenascin C (TNC) in tumor stroma, herein a protease-activated FN1 and/or TNC binding peptide (FTF) is designed and an extracellular matrix (ECM)-trapped bioinspired lipoprotein (BL) (FTF-BL-CP) is proposed that can be preferentially captured by the TNC and/or FN1 for tumor retention, and then be responsively dissociated from the matrix to potentiate the antitumor immunity. The FTF-BL-CP treatment produces a 6.96-, 9.24-, 6.72-, 7.32-, and 6.73-fold increase of CD3+CD8+ T cells and their interferon-γ-, granzyme B-, perforin-, and Ki67-expressing subtypes versus the negative control, thereby profoundly eliciting the antitumor immunity. In orthotopic and lung metastatic breast cancer models, FTF-BL-CP produces notable therapeutic benefits of retarding tumor growth, extending survivals, and inhibiting lung metastasis. Therefore, this ECM-trapping strategy provides an encouraging possibility of prolonging tumor retention to potentiate the antitumor immunity for anticancer immunotherapy.

Biomimetic Tertiary Lymphoid Structures with Microporous Annealed Particle Scaffolds for Cancer Postoperative Therapy

Immunotherapy plays a vital role in cancer postoperative treatment. Strategies to increase the variety of immune cells and their sustainable supply are essential to improve the therapeutic effect of immune cell-based immunotherapy. Here, inspired by tertiary lymphoid structures (TLSs), we present a microfluidic-assisted microporous annealed particle (MAP) scaffold for the persistent recruitment of diverse immune cells for cancer postoperative therapy. Based on the thermochemical responsivity of gelatin methacryloyl (GelMA), the MAP scaffold was fabricated by physical cross-linking and sequential photo-cross-linking of GelMA droplets, which were prepared by microfluidic electrospraying. Due to the encapsulation of liquid nitrogen-inactivated tumor cells and immunostimulant, the generated MAP scaffold could recruit a large number of immune cells, involving T cells, macrophages, dendritic cells, B cells, and natural killer cells, thereby forming the biomimetic TLSs in vivo. In addition, by combination of immune checkpoint inhibitors, a synergistic anticancer immune response was provoked to inhibit tumor recurrence and metastasis. These properties make the proposed MAP scaffold-based artificial TLSs of great value for efficient cancer postoperative therapy.

Multistage Self-Assembled Nanomaterials for Cancer Immunotherapy

Advances in nanotechnology have brought innovations to cancer therapy. Nanoparticle-based anticancer drugs have achieved great success from bench to bedside. However, insufficient therapy efficacy due to various physiological barriers in the body remains a key challenge. To overcome these biological barriers and improve the therapeutic efficacy of cancers, multistage self-assembled nanomaterials with advantages of stimuli-responsiveness, programmable delivery, and immune modulations provide great opportunities. In this review, we describe the typical biological barriers for nanomedicines, discuss the recent achievements of multistage self-assembled nanomaterials for stimuli-responsive drug delivery, highlighting the programmable delivery nanomaterials, in situ transformable self-assembled nanomaterials, and immune-reprogramming nanomaterials. Ultimately, we perspective the future opportunities and challenges of multistage self-assembled nanomaterials for cancer immunotherapy.

Tumor-TME Bipartite Landscape of PD-1/PD-L1 in Endometrial Cancers

The bipartite landscape of tumor cells and stromal cells determines a tumor's response to treatment during disease management. In endometrial cancers (ECs), the mechanistic contribution of PD-L1/L2 and PD-1 signaling of the host's tumor microenvironment (TME) (CAF and immune cells) in the context of the tumor cells is elusive. To understand the tumor-stroma-immune crosstalk, we studied the compartmental pattern of PD-L1/L2 and PD-1 expression in EC tissues and their matched CAFs. Over 116 surgically resected tumors (T) and the tumor-adjacent normal tissues (N) were obtained from consented unselected consecutive patients. IHC was performed in T, N-epi-thelium, and the stromal mesenchymal environment (SME; mesenchyme) in the T and N tissues. The staining intensity and distribution patterns of PD-L1/L2 and PD-1 in the FFPE sections of T and N were evaluated by a pathologist using a standard scoring system of TPS and CPS. We tested the PD-L1/L2 and PD-1 immune landscape of tumor-TME pair and normal epithelial-stromal mesenchyme pairs from patients with different grades of disease vis-à-vis their CAF PD-L1 levels. We used qRT-PCR to determine the expressions of mRNAs, while the flow cytometry and ICC determined the level of expression of proteins. We observed higher levels of PD-L1 mRNA and protein expression in primary CAFs from the resected tumor tissue compared to the tumor-adjacent normal tissues. We also determined the expression of patients' soluble PD-L1/L2 as peripheral readouts of PD-L1/L2 and PD-1. As we evaluated the results in the context of their pathological parameters, such as grades, stages, lymphovascular invasion, percentage of myometrial invasion, and dMMR in patients, the dominance of PD-L1 expression in TME was positively correlated to the higher pathological grades of tumors, and its relationship with the dMMR. Since the neutralization of CD8-positive cytotoxic T-cells is PD-L1-dependent, our data indicate that irrespective of the PD-L1 positivity of tumor cells, the PD-L1-positive CAFs can play a critical role in bringing out an additional load of PD-L1 for an effective engagement of PD-1 within a tumor mass.