Tissue self-aggregation properties for building novel 3D in vitro models to study human T-cell–dependent B-cell immune responses

The germinal center (GC) is a microanatomical structure that nucleates from B-cell follicles in secondary lymphoid organs upon encounter of a foreign antigen. Within the GC, antigen-specific B cells undergo iterative cycles of somatic hypermutation, clonal expansion, and antigen-driven selection, favoring the output of long-lived plasma cells and memory B cells expressing high-affinity antibodies. These processes rely on spatially-confined interactions between GC B cells and immune and stromal cells, including follicular dendritic cells (FDCs) and T follicular helper (Tfh) cells, delivering survival and differentiation signals to ensure the positive selection of GC B cells expressing the highest-affinity B cell receptors (BCRs). While murine models have provided fundamental insights into the understanding of the basic principles underlying GC biology, species-specific differences and the use of model antigens limit their translational relevance for humans. This limitation underscores the need to develop standardized human in vitro systems which faithfully recapitulate the basic processes underlying initiation, maintenance and resolution of a GC B cell response to human-relevant immunogens. In this study I address this problem building on self-aggregation properties of human tonsillar tissue to establish two complementary three-dimensional (3D) in vitro models reproducing key aspects of human GC physiology. The first, is based on an air liquid interface (ALI) transwell system allowing dissociated whole tonsillar cell suspensions to spontaneously aggregate and persist over time. This configuration enabled phenotypic and clonotypic characterization of individual 3D cellular aggregates by flow cytometry and immunoglobulin heavy chain (IGH) gene rearrangement analyses. Single-aggregate analysis revealed prolonged viability and preferential expansion of GC-like B cells, as well as preservation of B cell subsets with memory phenotype. Notably, aggregates were formed by a mixture of B cells and CD4+ T-helper cells resembling the requirements for the build-up of a GC reaction. The second model is based on the seeding of tonsillar cell suspensions in porous gelatin scaffolds, which provide a tunable and scalable 3D matrix where scaffold porosity and size can be adjusted to optimize cell density and interactions. This setting allows longitudinal monitoring of B–T cell contacts and spatial reconstruction of GC-like structures through histological analyses of fixed scaffolds. In this model, I tested the importance of providing tonsillar cells with the FDC-immortalized cell line YK46 expressing membrane-bound CD40 ligand and soluble IL-21 to guarantee the prolonged expansion of B-T cell aggregates. Flow cytometry and immunofluorescence analyses revealed that within the aggregates, B cells lined along internal scaffold trabeculae, forming clusters within pores, while T cells progressively disappeared over time. Immunophenotypic profiling of B cells revealed dynamic changes in subpopulations, culminating with the preferential expansion of IGD-negative CD38+ GC-like B cells. Overall, the ALI- and scaffold-based 3D self-aggregation platforms represent versatile and complementary tools to reproduce key events of the GC reaction in vitro. They enable longitudinal monitoring of human B-T cell interactions, B cell activation and terminal differentiation. They also provide a translational framework to expedite the development of improved vaccines and immunotherapies.