Tonsil explants as a human in vitro model to study vaccine responses

Abstract

Introduction: Vaccination is one of the most effective infection prevention strategies. Viruses with high mutation rates -such as influenza- escape vaccine-induced immunity and represent significant challenges to vaccine design. Influenza vaccine strain selection is based on circulating strains and immunogenicity testing in animal models with limited predictive outcomes for vaccine effectiveness in humans.

Methods: We developed a human in vitro vaccination model using human tonsil tissue explants cultured in 3D perfusion bioreactors to be utilized as a platform to test and improve vaccines.

Results: Tonsils cultured in bioreactors showed higher viability, metabolic activity, and more robust immune responses than those in static cultures. The in vitro vaccination system responded to various premanufactured vaccines, protein antigens, and antigen combinations. In particular, a multivalent in vitro immunization with three phylogenetically distant H3N2 influenza strains showed evidence for broader B cell activation and induced higher antibody cross-reactivity than combinations with more related strains. Moreover, we demonstrate the capacity of our in vitro model to generate de novo humoral immune responses to a model antigen.

Discussion: Perfusion-cultured tonsil tissue may be a valuable human in vitro model for immunology research with potential application in vaccine candidate selection.

Keywords: antibodies; bioreactor; human; in vitro model; influenza; tonsil; vaccines.

Figure 1

Tonsil-derived and peripheral blood immune cells have a distinct immunophenotype. (A) Cell composition of immune cell types in freshly isolated tonsil cells and PBMC from blood donors. Frequencies were determined by flow cytometry (n=17). Frequencies of the main B cell (B) and CD4+ (C) and CD8+ (D) T cell differentiation states in tonsil vs PBMC determined by flow cytometry (n=17). (E) Representative H&E and immunohistochemistry for CD3, CD20, CD21 and Ki67 of ex vivo tonsils showing the architecture of the secondary lymphoid organ. In the top left H&E image, the black square indicates the region that is magnified in the other panels. Red arrows indicate the B cell follicles and yellow arrows the T cell area. Scale bar = 50 µm. Values were compared with two-way ANOVA with Bonferroni’s post hoc test. *p<0.005, ***p<0.0001, ****p<0.0001.

Figure 2

Cell survival and cellular fitness of tonsils in the 3D perfusion bioreactor vs static system. (A) Representative H&E images of tonsil tissues cultured in 3D perfusion bioreactor vs static system for 5 and 14 days. (B) Cell survival quantified by flow cytometry at day 0, day 5, and day 14 in unstimulated and quadrivalent influenza vaccine (QIV)-stimulated conditions (n=7). Plotted values are the mean +/- SD. (C) Metabolic cell activity (‘cellular fitness’) in an MTT assay at day 5 in unstimulated and QIV-stimulated conditions (n=5). Similar results were obtained with CpG stimulations (not shown). (D) Maintenance of the lymphoid organ structure was assessed in perfusion-cultured tonsils stained for the markers CD3, CD20, CD21, and Ki67 at day 5. (E) The CD3/CD20/CD21 positive area and the frequency of Ki67⁺ cells were assessed by automated image analysis (n=3). Statistical significance between groups was determined using two-way ANOVA with Bonferroni’s post hoc test. **p<0.001, ***p<0.0001, ****p<0.0001.

Figure 3

Adaptive immune responses to vaccines and single-protein antigens in tonsils cultured in 3D perfusion bioreactor vs static system. (A) Plasmablast differentiation in unstimulated, CpG- and influenza vaccine-stimulated tonsils (n=7) at days 5 and 14. Flow cytometry-based immunophenotyping. (B) Influenza-specific IgG antibodies (Luminex) in supernatants of tissue cultures (n=7) using bioreactors vs static system at days 5 and 14 in the indicated conditions. (C) Cytokine concentrations in the culture supernatants in response to CpG and influenza vaccine stimulation in tonsil tissues (n=5) cultured in 3D perfusion bioreactors. (D) Plasmablast differentiation and pneumococcal specific-IgG production in response to the pneumococcal conjugate vaccine PCV13 (n=4). The dotted line indicates the cut-off value for test positivity of the pneumococcal-specific IgG. (E) Plasmablast differentiation and H1N1 Cal 2009-specific IgG response (normalized to unstimulated condition) upon 5-day stimulation in perfusion-cultured tonsils (n=3). Two-way ANOVA with Turkey’s post hoc test and Friedman test with Dunn’s correction were used to determine the significance values in the graphs a-c and d-e, respectively. BAFF= B cell activating factor; adj = adjuvant (=LPS). *p<0.005, **p<0.001, ***p<0.0001, ****p<0.0001.

Figure 4

Multivalent vs monovalent influenza HA antigen in vitro immunization. (A) Plasmablast differentiation after 5-day stimulation was determined by flow cytometry (n=3). (B) Phylogenetic tree showing the antigenic relatedness of the influenza H3N2 strains included in vaccines between 2013/2014 and 2022/2023. The graph was generated using nextrain.org (49, 50). (C) Multiplexed, Luminex-based influenza strain-specific antibody measurements against the HA protein of ten influenza H3N2 strains. The y-axis indicates the antigenic stimulation condition used for in vitro immunization in the 3D perfusion bioreactor; the x-axis indicates the magnitude (fold increase) of the H3N2 strain-specific IgG/IgA/IgM responses normalized to unstimulated condition upon 5-day stimulation (n=3). Values were compared using Friedman test with Dunn’s correction. adj = adjuvant (=LPS).

Figure 5

Impact of the antigenic diversification in multivalent H3N2 in vitro immunizations. (A) Phylogenetic tree highlighting the specific clades and H3N2 strains selected for multivalent HA antigen stimulations. The indicated distinct multivalent stimulation conditions (mix 1-3) were tested. (B) Effect of the different H3N2 antigenic compositions on the in vitro immunization-induced GC B cells and plasmablast differentiation in tonsil and PBMC samples (n=5). (C) Frequency of activated memory CD4⁺ T cells induced by in vitro multivalent stimulation conditions based on the expression of HLA-DR and CD45RA (n=5). adj = adjuvant (=LPS) (D) Influenza strain-specific antibodies in the supernatant of the cultured tonsils and PBMC upon 5-day stimulation were quantified by multiplexed Luminex assay. The y-axis shows the stimulation condition, while the x-axis shows the relative specific IgG responses against ten H3N2 strains. Luminex-derived median immunofluorescent (MFI) data was normalized to the MFI of the unstimulated condition; (E) H3N2 Darwin 2021-specific IgG normalized to H3N2 Cambodia 2020 response in tonsils vs PBMC (n=5). Two-way ANOVA with Turkey’s post hoc test and Mann-Whitney test were used to compare several or two groups as appropriate. *p<0.005, **p<0.001, ***p<0.0001, ****p<0.0001.

Figure 6

De novo B cell responses to a neo-antigen. (A) GC B cell and plasmablast frequencies at 5 days following stimulation with KLH of tonsils and PBMC (n=5). (B) KLH-specific IgG, IgA, and IgM responses in tonsil and PBMC samples (n=5). Statistical significance was determined using two-way ANOVA with Turkey’s post hoc test. KLH = keyhole limpet hemocyanin; adj = adjuvant (=LPS). *p<0.005, **p<0.001.