Skip to main content

System for Immunological Modelling as an Ultimate tool to Link Adjuvant function To Adaptive immune responses

Final Report Summary - SIMULATA (System for immunological modelling as an ultimate tool to link adjuvant function to adaptive immune responses)

The immune system uses specific molecular receptors to decide with what kind of pathogen it is dealing. For instance, this recognition system can distinguish whether the invading pathogen is a virus, a fungus or a bacterium. Much progress has been made in understanding how exactly this system recognises the different pathogens and which molecular structures from the pathogens are important. The discovery of such immune-stimulatory molecules from pathogens has opened the possibility to design vaccines that use this property of the immune system to help generate immune responses to vaccine antigens. One important compound among this group of immune-stimulatory molecules is monophosphoryl lipid (MPL), which is a de-toxified derivative of the bacterial molecule lipo-polysaccharide (LPS). MPL is currently used in several vaccines. When molecules such as MPL are used in vaccines to enhance immune responses, they are referred to as 'vaccine adjuvants'. It is important to better understand how these adjuvants work: what is the mechanism by which they enhance immune responses, e.g. production of antibodies or generation of T cells. So far, much of the work in this field has been carried out in experimental animals. However, the immune systems from experimental animals might be different from the human immune system, so it may not always be easy to interpret such results. On the other hand, the analysis of vaccine adjuvants in humans is not so straightforward either, as it is constrained by ethical considerations, genetic diversity and costs of human clinical trials. Therefore, we aimed to build an in vitro human immune system, in which we can study the behaviour of vaccine adjuvants and antigenic proteins in a test tube. In order to mimic the actual vaccine injection site, we designed a three-dimensional (3D) in vitro in vitro vaccination system. The design of this model, called SIMULATA, is loosely inspired by work published by G Randolph and co-workers (Randolph et al., Science 282:480, 1998). The in vitro model has three different components: (1) a collagen layer, (2) a thin layer of endothelial cells on top of the collagen (mimicking the wall of a blood vessel) and (3) peripheral blood mononuclear cells (PBMC). The basic assumption is that an important cell type in the PBMC population, monocytes, will migrate through the endothelial cell layer into the collagen. It is expected that these monocytes will migrate through the endothelial cell layer. In the process, monocytes are hypothesised to pick up antigenic proteins and to be exposed to any adjuvants that are present in the system. This process imitates that migration of monocytes from blood vessels into infected tissues or tissue in which the vaccine has been injected and then back into the blood vessel or into the lymphatics. Our results show that indeed monocytes migrate into the collagen and then back to the supernatant (where they came from). In the process, we observed that these monocytes differentiated into so-called dendritic cells (DC). DCs are antigen-presenting cells: they pick-up proteins (antigens) from pathogens or vaccines and process these into smaller peptides. These peptides are presented by DCs to other immune cells, such as T cells. We have repeated this process in our in vitro system by harvesting the monocytes, after they migrated back to the supernatant, and incubating them with T cells. This process indeed resulted in the activation and proliferation of the T cells. Next, various antigens and/or adjuvants were added and the impact of adjuvants on monocyte migration, DC activation and T cell responses was measured. We found that several immune-stimulatory molecules increased the migration of monocytes from the PBMC in the supernatant into the collagen. These migrating cells then acquired DC properties, indicating that they underwent at least partial differentiation in the process of migrating, encountering adjuvant and picking up antigen. Using this system, we measured T cell responses in healthy donor PBMC to different viral and bacterial proteins, including several different subsets of CD4 T cells, some of which are difficult to detect with conventional methods (e.g. so-called Th17 cells). Based on these observations, we conclude that our 3D in vitro system has sufficient promise to warrant further investigation.

Summary description of project context and objectives

The context in which antigens are presented to the immune system controls the immunological outcome of antigenic challenge, i.e. the precise nature of the immune response that follows exposure to antigens. The innate immune system uses a specific set of receptors that recognise molecular patterns from pathogens, known as pattern recognition receptors, to decode the nature of the antigen (e.g. viral, bacterial, fungal) and to translate this into an appropriate adaptive immune response. Different microbial compounds are recognised by toll-like receptors (TLR) but there are also compounds that can activate the immune system in a TLR-independent manner. The use of immune-stimulatory molecules that engage pattern recognition receptors as vaccine adjuvants opens the door towards the rational design of vaccines that induce defined and optimised immune responses. Thus, an important objective for adjuvant research, both from the efficacy and safety perspectives, is to better understand the links between the recognition of adjuvants by the innate immune system and the nature of the subsequent adaptive response. In other words: How does the innate immune system use the signals through pattern recognition receptors to generate a specific type of immune response? So far, much of the work in this field has been carried out in animal models such as mice, but these data are not readily extrapolated to human responses. On the other hand, the analysis of vaccine adjuvants in humans is constrained by ethical considerations, genetic diversity and costs of human clinical trials. Therefore, the in vitro re-creation of a human immune response would be ideal to begin bridging the gap between pre-clinical animal models and human clinical trials. One emerging concept of adjuvant-stimulated immune activation is that blood monocytes are recruited to the injection site and differentiate into either macrophages or DCs after which the latter migrate to the draining lymph nodes to activate adaptive immunity.

The overall objective of our proposal was to construct a 3D in vitro model to recapitulate this sequence of events and address fundamental questions related to vaccine adjuvants and adaptive immune response outcomes. The basic assumptions of our model derive from the work published by Randolph et al (Science 282:480, 1998). In short, PBMC or CD14+ blood monocytes are cultured on top of a collagen matrix that is covered by a monolayer of human umbilical cord endothelial cells (HUVEC). Monocytes will then migrate through the HUVEC monolayer into the collagen matrix. This migration through the HUVEC layer triggers differentiation into either macrophages or DCs. These DCs will then migrate back from the collagen layer towards the supernatant ('reverse transmigration') and will have a different activation status. DCs can be exposed to immune-stimulants and antigens in the system. To generate adaptive immune responses, antigen-loaded and immunostimulant exposed DC can be used to stimulate T cells. Based on the work from Randolph et al (Science 282:480, 1998) and with an emphasis on automation, a somewhat simplified system, referred to as MIMIC, was further developed in a commercial vaccine-testing setting by the Florida-based company Vaxdesign (Higbee et al, Altern Lab Anim 37:19, 2009; Ma et al, Immunology 130:374, 2010). Moreover, Gaucher et al (J Exp Med 205:3119, 2008) used an in vitro immunology system to show, using microarray RNA expression analysis, that clear overlaps existed between early immune responses induced by the YFV-17D vaccine in the in vitro model versus in vivo. However, these authors used a highly simplified in vitro system which was more reminiscent of in vitro stimulation of PBMC. A major advantage of the in vitro immunisation systems, such as the ones described by Randolph and coworkers (Science 282:480, 1998), Higbee et al (Altern Lab Anim 2009) and Gaucher et al (J Exp Med 205:3119, 2008) was that different vaccine formulations can be tested using PBMC from a single donor. Briefly, in the MIMIC model, PBMCs are isolated by leucopheresis from consenting adult healthy donors. To set up the culture system, a collagen matrix is created in a flat bottom 96 well plate. Then, a HUVEC monolayer is grown on top of the collagen matrix and PBMCs (open and red circles) are added for 2h to the culture to allow monocyte (red circles) to migrate. After the 2 hours, cultures are washed to remove non-migrated lymphocytes and monocytes and incubated for 2 - 5 days after which transmigrated DCs are harvested. Addition of antigen and / or adjuvants into the system results in activation of the transmigrating inflammatory DCs which can then either prime an autologous naïve CD4 T cell (blue circles) response or stimulate a pre-existing memory response.

The first objective of the current project was to set up an in vitro immunisation system and explore the parameters that control its functionality. Amongst others, we studied the behaviour of the HUVEC monolayer and its impact on the ensuing immune responses. Next, we studied the behaviour of monocytes and PBMC when cultured on the HUVEC / collagen layer and the impact of immune stimuli / adjuvants on monocyte migration was analysed. Different adjuvants and vaccine formulations were tested, focusing on Alum, AS01, AS02, AS03 (Garcon & Van Mechelen, Exp Review Vaccines 10:471, 2011) and zymosan.

Subsequently, we studied activation profiles of migrating monocytes as well as production of inflammatory cytokines. Finally, the capacity of DCs to take up antigens was analysed. Using fluorescently labelled antigen, it was found that cells carrying the fluorescent label could be detected in the supernatant, suggesting that antigen had been taken up. Thus, in objective 1, we established and optimised the basic parameters of the system. The second objective was to evaluate the impact of vaccine adjuvants on T cell quality and magnitude of primary CD4 T cell responses to a candidate human immunodeficiency virus (HIV) vaccine in the in vitro model. The question addressed here is whether reverse-transmigrated monocyte-derived cells (i.e. antigen-carrying and activated DCs) were capable of triggering primary CD4 T-cell responses. In addition, the quality of these responses was studied as a function of adjuvant. However, we elected to not assess the immune responses of healthy donors to HIV candidate vaccine proteins (because this test could then be mistaken for an HIV-positivity test). Instead, we used the Mycobacterium tuberculosis candidate vaccine antigen formulated with different adjuvants. The vaccine consists of a fusion protein named M72 containing part of the Mtb32a and Mtb39a proteins from M. tuberculosis (Leroux-Roels et al, Vaccine 2012, Epub ahead of print) and formulated with Alum, AS01, AS02, AS03 or zymosan. Cell cultures were established and the vaccine was administered to the system using the optimised procedures described above. Monocyte-derived cells were harvested and cultured with autologous PBMC. Responses were measured after 4 weeks of culture by stimulating cells with a set of overlapping peptides spanning the vaccine antigen, followed by intracellular cytokine staining (ICS). We analysed production of IFN, TNF and IL-17 in combination with CD40L by multiparameter flow cytometry. This setup provides the unique opportunity to compare different adjuvant formulations in the same donor, since multiple cultures will be set up using the same PBMC batch and established how adjuvant impacts on the cytokine profiles and magnitude of the primary response. Finally, the third objective was to assess the impact of adjuvanted priming on recall responses to drifted influenza virus vaccines. For reagent availability reasons, we decided to first focus experiments on the question whether adjuvant could affect the cytokine profile and the magnitude of the secondary response to antigens from Varicella Zoster virus (VZV), Hepatitis B virus (HBV), S. aureaus and S. pneumoniae bacteria. The Ge protein from VZV (Dendouga et al, Vaccine 30:3126, 2012; Leroux-Roels et al, J Inf Disease 2012, Epub ahead of print) and the HBs protein from HBV (Vandepapeliere et al, Vaccine 26:1375, 2008) were used in the in vitro system. To study the behavior of anti-bacterial T cells, antigens from both S. aureaus and S. pneumonia were tested in the system as well. For reasons of confidentiality, these proteins will be referred to as S. aureus and S. pneumoniae proteins 1 and 2. We analysed the immune responses to these candidate vaccine antigens formulated with different adjuvants in individuals likely to have pre-existing CD4 T-cell immunity to VZV, HBV, S. aureus and S. pneumoniae. To confirm pre-existing immunity, healthy donors were first screened for a detectable recall responses to VZV Ge, HBV surface antigen (HBs) and the S. aureus and S. pneumoniae proteins, using a long term in vitro PBMC restimulation culture with overlapping peptides spanning the antigens. The best responders were selected and used in the in vitro system. The main findings from this work were (1) that CD4 T-cell responses can be detected and that adjuvants can affect these responses, and (2) that the system revealed Th17-type responses with S. aureus proteins. The identification of S. aureus specific Th17 responses is consistent with a recent report (Zielinski et al, Nature 484:514, 2012). Interestingly, inclusion of IL-2 in the culture medium prevented expansion/induction of Th17 responses, consistent with the plasticity of the Th17 phenotype (Zielinski et al, Nature 484:514, 2012). Induction of such Th17 responses was not an artifact of the system, since this was not observed with other antigens.

Description of the main s&t results/foregrounds

Objective 1

(1) Construction of the in vitro SIMULATA system (set-up)

The first challenge of this project was to set up an in vitro immunisation system when only a few research papers on the topic were available (Randolph et al, Science 282:480, 1998; Qu et al, J Imm 182:3650, 2009; Higbee et al, Altern Lab Anim 37:19, 2009; Ma et al, Immunology 130:374, 2010). To get the system operational, we had to optimise multiple system parameters such as tissue culture plates, endothelial cells (HUVEC), timing of the different steps and quantitation of antigens and adjuvants.

Optimisation yielded a protocol in which 96 well plates were used and collagen was added. The HUVEC cells were seeded on top of the collagen matrix. The original research paper describing the in vitro system published by Randolph and colleagues (Science 282:480, 1998) demonstrated that 2 days was sufficient to get DC reverse trans-migration and differentiation. Here we tested this two day incubation period and also longer incubation times of monocytes in the collagen matrix (2 and 5 days). Finally, a titration of the dose of adjuvants added to the system was done to avoid HUVEC cell mortality and get good DC viability. The optimised system will be referred to as SIMULATA.

Overall, in the project, several different vaccine adjuvants or combinations of immune stimuli were used:

- Alum: Al(OH)3 or AlPO4, used to bind antigens; also activates the innate immune system
- Zymosan: a fungal compound that is a ligand for TLR2 and dectin-1
- LPS: a bacterial compound that is a ligand for TLR4
- MPL: detoxified LPS and also a ligand for TLR4
- AS01 (adjuvant system 1): MPL and QS21 in a liposome formulation
- AS02 (adjuvant system 2): MPL and QS21 in an oil-in-water emulsion
- AS03 (adjuvant system 3): an oil-in-water emulsion containing tocopherol
- AS15 (adjuvant system 15): MPL, QS21 in a liposome formulation and CpG oligonucleotides

More details on the adjuvant systems can be found in the following publications:

- Garcon & Van Mechelen, Exp Review Vaccines 10:471, 2011 (review on Adjuvant Systems)
- Morel et al, Vaccine 29:246, 2011 (AS03)

(2) Evaluation of the impact of vaccine adjuvants on monocyte migration

Intra-muscular injection of adjuvanted vaccines is known to recruit innate immune cells such as monocytes from the blood through the induction of chemokine expression. We therefore decided to assess the impact of various vaccine adjuvants on monocyte migration using the SIMULATA system. 10µg / ml of antigen (from HBV amongst others) formulated with or without adjuvant or the adjuvant alone (Alum, AS01, AS03, AS15 or 1 % zymosan) (see Garcon & Van Mechelen, Exp Review Vaccines 10:471, 2011 and Morel et al, Vaccine 29:2461, 2011 for an overview on the adjuvant systems AS01 and AS03) were at a 1/100 dilution. After reaching HUVEC confluency, PBMC were added on top of the HUVEC monolayer. The frequency of monocytes in the PBMC was determined by FACS using CD14 staining.

After migration, non-migratory cells, i.e. those cells that remained in the supernatant, were extensively washed and kept for an additional CD14 staining in order to calculate the percentage of monocyte migration through the endothelium using the following formula: % of monocyte migration = % CD14 + monocytes in the PBMC - % CD14+ after migration % CD14+ monocytes in the PBMC

The results were as follows: we showed that addition of the TLR2 ligand zymosan (Randolph et al, Science 282:480, 1998) increased the migration of monocytes thought the endothelium compared to the condition where nothing was added to the system (None conditions). In contrast, no impact of Alum, AS01 and AS03 on monocyte migration was observed.

(3) Evaluation of the impact of vaccine adjuvants on reverse transmigrating cell differentiation and activation

Even though the original research paper describing the in vitro system published by Randolph and colleagues (Science 282:480, 1998) demonstrated that 2 days was sufficient to get DC reverse trans-migration and differentiation, we decided to test longer times of incubation of monocytes . After migration through the endothelium, monocytes were left in the collagen matrix to allow the cells to reverse transmigrate. The phenotype of the reverse transmigrating cells was analysed by flow cytometry using different surface markers.

Reverse transmigrating cells collected after incubation in the SIMULATA system showed higher expression of DC activation markers, reflecting a 'DC like' phenotype. Moreover, these cells seemed to mature more efficiently when the immunostimulant zymosan was included. Indeed, a proportion of cells showed reduced expression of monocyte markers and increased expression of activated DC markers.

Another way to assess the activation of antigen-presenting cells (APC) (i.e. the monocytes and DCs) following contact to adjuvant is to measure their production of pro-inflammatory cytokines. Hence, cytokine production in the medium overlying the endothelial cell layer (the supernatant) was measured by multiplex cytokine analysis (luminex) after 5 days of differentiation. No effect of AS03 or AS01 on the production of several pro-inflammatory cytokines (such as IL-6, GMCSF, TNF, MCP-1 or IL-1) was observed. The cytokine IL-8 was induced by AS03 in the in vitro system. Zymosan was able to induce the production of all cytokines mentioned above by monocytic cells. Even though zymosan promoted migration of a higher proportion of monocytes, this increased migration alone probably does not account for the difference in cytokine production observed in these experiments. IFN production (at very low levels) was only found using zymosan activated APCs. No IL-18 or IL-22 was detected in any of the conditions.

The capacity of adjuvants to increase reverse transmigrating cell maturation can also be evaluated through their capacity to increase antigen uptake. Fluorescent HBV surface antigen (HBs antigen) was used to detect cells that would take up antigen. Thus, fluorescent HBs antigen was added to the SIMULATA system and after 5 days of differentiation, we quantified antigen uptake by flow cytometry. The presence of AS01 in the system led to an increase of antigen uptake compared to the condition with fluorescent Ag alone. There was no impact of zymosan on antigen uptake. Cells that had taken up higher amount of proteins displayed a more mature phenotype with higher expression of the DC activation marker HLA-DR.

The conclusion from this part of the project was that monocytes can migrate through the HUVEC monolayer into the collagen and then reverse transmigrate back into the supernatant. In this process, the monocytes acquire several properties, such as expression on DC markers, and production of cytokinesthat suggest that they differentiated into bona fide antigen-presenting cells. However, to confirm that, a functional test is essential, in which the capacity of the reverse-transmigrated cells to activate T cells is tested.

OBJECTIVE 2

Evaluation of the impact of vaccine adjuvants on primary T cell quality and magnitude

Having established that the in vitro systems reproduced key aspects of the innate immune response, the next step was to assess whether the reverse transmigrating cells, which were assumed to have differentiated into professional APCs, i.e. DCs, were capable of inducing CD4 T cell responses. In addition, the phenotypic characteristics (the 'quality') of the CD4 response were measured as a function of adjuvant.

To achieve this, we originally planned to use the GSK candidate HIV vaccine antigen (Leroux-Roels et al, Vaccine 28:7016, 2010) formulated with different adjuvants (Alum, AS01, AS02, AS03, zymosan at 2 different dilutions in collagen). However, we changed that and used the M. tuberculosis candidate vaccine antigen, the fusion protein M72 (Leroux-Roels et al, Vaccine 2012), because we concluded that our experiments in this system could not be used or be misinterpreted as a surrogate HIV test. We assumed that by using healthy donor PBMC, from donors with low probability of being exposed to TB, we would be able to assess primary T-cell responses using the M72 antigen.

We set up the system with the M72 antigen and the different adjuvants and adjuvant systems in the system. After differentiation, reverse transmigrating cells were collected. These cells would comprise the differentiated DCs containing antigen and that had been exposed to the adjuvants. These cells were then co-cultured with autologous PBMC (1:20 ratio) in order to assess the capacity of the M72/adjuvant exposed DCs to stimulate a T cell response. This culturing was done for 4 weeks in the absence of IL-2 to avoid any skewing of the immune response. Cells were then re-stimulated overnight with either medium or M72 peptides, in order to identify responding T cells and intracellular cytokine expression was assessed by flow cytometry. Thus, the 4 week culturing step evaluates the capacity of the DCs harvested from the collagen/HUVEC system to stimulate T cells. The short (overnight) incubation with peptides serves to visualise the responding T cells, such that they can be measured by flow cytometry. In these experiments, M72 specific responses induced by antigen-loaded reverse transmigrating cells were compared with responses obtained from 'simple' PBMC stimulation with M72 protein. We observed that M72-loaded APCs harvested from the in vitro SIMULATA system were better at stimulating a specific primary CD4 T cell response than PBMC that were simply incubated with the M72 protein. Formulation of M72 with AS02 and zymosan further increased the capacity of the monocyte derived APCs (from SIMULATA) to prime a specific CD4 T-cell response. While the magnitude of the CD4 T-cell response was affected by the adjuvants, the cytokine profiles (quality) of these cells were similar.

The conclusion from this part of the project was that the reverse transmigrating monocytes (that differentiated into DCs) behaved as antigen-presenting cells as evidenced by their capacity to stimulate naïve T cells.

Objective 3: Evaluation of the impact of vaccine adjuvants on recall T cell quality and magnitude

(A) Responses specific for VZV Ge and HBV surface antigen

We next focused on detecting secondary CD4 T-cell responses, i.e. by studying responses in the in vitro system using antigens for which immune memory was already present. Thus, here, we addressed whether existing memory T cells could be re-activated in the in vitro system. For this, we used PBMC samples from healthy donors with defined memory responses against VZV antigen Ge and HBV surface antigen HBs. Healthy donor PBMC samples were first screened for their Ge and HBs memory responses. This was done as follows: PBMC were stimulated for 5 days or overnight with Ge or HBs peptides, respectively. For the long term culture, cells were re-stimulated overnight with either medium or peptides. Ge and HBs responders were selected based on their antigen-specific CD4 T-cell responses as measured by flow cytometry.

To test whether the in vitro system was capable of activating memory CD4 T cells, the VZV Ge antigen was formulated with different adjuvants (1/10 dilution for all adjuvants except for zymosan 1 / 100) and these were added to the system. Then, PBMC from healthy donors that were VZV memory CD4-positive in the first screening test were added to the collagen/HUVEC wells and incubated. After differentiation, reverse transmigrating cells, were collected. These cells were co-cultured with autologous PBMC during 8 days in the presence of interleukin-2 (IL-2) and IL-7. In this set-up, the harvested cells, harbouring antigen-presenting cells that picked-up antigen and that were exposed to adjuvant, present the antigen to T cells in the PBMC preparation. Responses were measured by ICS after overnight stimulation of the cells to peptides spanning the Ge protein followed by flow cytometry. Antigen-specific responses induced by antigen-loaded and adjuvant-exposed reverse transmigrating cells were compared with responses obtained from PBMC stimulated with peptides covering the selected antigens. Cells were re-stimulated overnight with either medium or overlapping peptide pools and intracellular cytokine expression was assessed by flow cytometry. The results show that antigen-loaded transmigrating cells were better than overlapping peptides at stimulating specific CD4 memory responses. In addition, formulation of the antigen with either AS03 or zymosan further increased the capacity of the 'DC like' cells to activate a specific a CD4 T cell response. From these results, it was concluded that the SIMULATA system was capable of re-activating CD4 memory T cells.

Next, memory CD4 T cell responses specific for the HBV surface antigen HBs were assessed using the in vitro system. To do this, HBs antigen was formulated with different adjuvants and Adjuvant Systems and added to the system. After differentiation, reverse transmigrating cells were collected from the SIMULATA cultures and co-cultured with autologous PBMC during 8 days in the presence of IL-2 and IL-7. After this incubation, peptides spanning the HBs protein sequence were added to the PBMC cultures to allow flow cytometric detection of responding T cells. Antigen-specific responses induced by reverse transmigrating cells (i.e. the cells harvested from the in vitro system) were compared with responses obtained from PBMC stimulated with peptides covering the selected antigens. Cells were re-stimulated overnight with either medium or overlapping peptide pools and intracellular cytokines expression was assessed by flow cytometry. In contrast to the above-described results with the VZV Ge antigen, antigen-loaded transmigrating cells were not better than overlapping peptides at stimulating specific CD4 memory responses in the setting of the HBs antigen. This result suggested either that recall of HBs-specific memory T-cell responses does not need SIMULATA-generated antigen- loaded APCs or that the amount of HBs protein added to the system was not sufficient to get an optimal loading of the antigen on the APCs. Nevertheless, formulation of the antigen with AS03 was able to further increase the capacity of the 'DC like' cells to activate a specific CD4 T-cell response.

From this part of the project it was concluded that the in vitro system is capable of picking up antigens and presenting these to memory CD4 T cells. Whereas for one antigen, the VZV Ge protein, it appeared that the in vitro system led to better CD4 T cell expansion as compared to long-term peptide stimulation, this was not observed for the second antigen, the HBV HBs antigen. It is unclear why these two antigens behave different. Nevertheless, it can be concluded that the SIMULATA system achieved T-cell expansion for both antigens, suggesting that it can be used to assess recall responses.

(B) Responses specific for S. aureus and S. pneumonia

A newly described CD4 T cell subset expressing the cytokine IL-17 has been shown to play a key role in host protection against bacterial infections such as Staphylococcus aureus and Streptococcus pneumoniae. This novel CD4 T-cell lineage is now known as 'Th17'. During life, exposure to these bacteria is common and most people are likely to harbour immunological memory specific for these pathogens. This memory could comprise Th17 cells. Indeed, in a different in vitro expansion system, Sallusto and coworkers recently described Th17-type CD4 memory responses specific for S aureus in healthy donor PBMC (Zielinski et al; Nature 2012). We were interested to evaluate how the SIMULATA system would detect such potential Th17 responses.

Healthy donors were screened for their S. aureus and S. pneumoniae memory T cell responses. To this end, PBMC were stimulated for 8 days in presence of IL-2 and IL-7 with either inactivated bacteria, the protein antigens 'protein 1' and 'protein 2' or peptide pools spanning these proteins. Cells were then re-stimulated overnight with either medium or overlapping peptides and intracellular cytokine expression was assessed by flow cytometry. Most of the healthy donors displayed S aureus specific Th1-type memory CD4 T cell responses. No IL-17 expression was observed in the experiment, indicating that PBMC stimulation (note: this is not in the SIMULATA system) with inactivated bacteria, proteins or peptides does not readily reveal specific Th17 responses, despite the fact that these were shown to exist for at least S aureus (Zielinski et al; Nature 2012). There are several potential explanations for this. First, it is possible that the donor samples used do not contain Th17 memory cells specific for the bacterial antigens tested. Alternatively, it is possible that the presence of IL-2 during the co-culture could have blocked Th17 responses. Indeed, IL-2 has recently been shown to inhibit the production of IL-17 by memory Th17 cells (Zielinski et al; Nature 2012). The same experiment, i.e. direct PBMC stimulation with different antigens, was therefore repeated in the absence of IL-2 and IL-7 but this gave similar results.

To determine whether the SIMULATA system could reveal potential pre-existing Th17 memory responses against S. aureus, two bacterial proteins, protein 1 and protein 2, were formulated with different adjuvants or Adjuvant Systems (1 / 10 dilution for all adjuvants except for zymosan 1 / 100), with protein only (no adjuvants) as a control. These formulations were added to the SIMULATA system. After incubation, reverse transmigrating cells were collected and co-cultured with autologous PBMC (1:20 ratio) during 8 days in the presence or absence of IL-2. Antigen-specific responses induced by antigen-loaded reverse transmigrating cells were compared with responses obtained with PBMC stimulation with overlapping pool of peptides. After this expansion period, cells were re-stimulated overnight with pools of peptides to allow flow cytometry detection by ICS. As before, stimulation of PBMC with overlapping pools of peptides resulted in a Th1 response. Interestingly, incubation of PBMC with antigen-loaded reverse transmigrating cells from the SIMULATA system revealed antigen-specific Th17 responses, characterised by the production of IL-17 in the ICS assay. Some effects of the adjuvant systems ASO1 and AS02 and of the TLR2 adjuvant zymosan were observed on both Th1 and Th17 responses. Interestingly, the Th17 profile was clearly different when zymosan was added to the system: under these conditions, no antigen-specific IL17+ IFN+ cells were observed.

No IL-17 production was detected when IL-2 was added to the co-culture suggesting that IL-2 might either amplify the Th1 response to the expense of Th17 responses and/or inhibit Th17 cell expansion as described by Zielinski et al; Nature 2012).

The SIMULATA system was also used to analyse potential pre-existing Th17 memory responses specific for S pneumoniae. By screening an expression library containing > 96 % of predicted pneumococcal proteins, a recent report identified two S. pneumoniae antigens recognised by Th17 cells from human PBMC (Moffitt et al, Cell Host Microbe 2011). We produced these proteins SP0128 and SP0148 and formulated them with various adjuvants and Adjuvant Systems (1/50 dilution for all adjuvants except for zymosan 1 / 100) and added these to the system. After 5 days in the system, reverse transmigrating cells, were collected and co-cultured with autologous PBMC (1:20 ratio) during 8 days in the absence of IL-2. Responses were compared with responses obtained with PBMC stimulation with protein (i.e. no SIMULATA). Cells were then re-stimulated overnight with the protein and intracellular cytokine expression was assessed by flow cytometry. While no IL-17 was detected with formulated SP0148 protein, SP0128 loaded APCs generated in the SIMULATA system were able to stimulate specific Th17 responses. As observed for S aureus antigens, no Th17 or Th1 specific responses were observed after classical PBMC stimulation with protein SP0128, further emphasizing the capacity of SIMULATA-derived APCs to specifically stimulate, or rescue, such Th17 responses.

From the results generated in this part of the project, it was concluded that the in vitro SIMULATA system has the capacity to expand pre-existing Th17 cells that are specific for S aureus or S pneumoniae. It is important that such Th17 responses were not detected in parallel cultures in which PBMC were stimulated directly with the proteins, i.e. without the DC differentiation step in the SIMULATA system. This suggest that the in vitro system, in which APCs are generated by monocytes that migrate through HUVEC monolayers, does capture a specific, as yet undefined, property that is important to expand Th17 cells. This is of clear interest for the further study of Th17-type immune responses.

Discussion

We were able to set up an in vitro immunisation system, referred to as the SIMULATA system, after optimising multiple parameters, such as HUVEC cell handling, PBMC and timing. We measured the capacity of monocytes to migrate through the endothelial monolayer into the collagen matrix. We observed that the TLR2 agonist and immune-stimulant zymosan had a clear positive effect on monocyte migration in this system. This was not as evident with other adjuvants or with Adjuvant Systems tested.

We demonstrated that inclusion of adjuvants, starting with zymosan, in the SIMULATA sysatem induced reverse transmigrating cell differentiation and maturation. We also observed the production of pro-inflammatory cytokines in the culture system when zymosan was added to the system. IL-8 production was observed when the Adjuvant System AS03 was added to the system. We also assessed the capacity of adjuvants to affect reverse transmigrating cell maturation through their aptitude to increase antigen uptake. Additional work is needed to further characterize reverse transmigrating cells and to conclude on the impact of adjuvants on antigen uptake by these cells.

Perhaps most importantly, we evaluated the functional capacity of reverse transmigrating cells to stimulate a primary CD4 T cell response. We demonstrated that in contrast to M72 stimulated PBMC cultures, antigen-loaded APCs generated in the SIMULATA system were able to prime an antigen-specific CD4 T cell response that was further increased by the presence of adjuvants in the system.

In the next step, we assessed the capacity of SIMULATA-generated antigen-loaded APCs to stimulate CD4 memory T cell 'recall' responses to VZV or HBV antigens. We observed that Ge loaded APC were capable of stimulating a memory responses and that both zymosan and AS03 had an adjuvant effect on these antigen-specific memory T cell responses. Here, the SIMULATA system performed better that 'simple' PBMC / antigen stimulation. A different observation was made with the HBV surface antigen. In this setting, SIMULATA-generated DCs were capable to expand HBs-specific memory CD4 T cells. However, in contrast to the VZV Ge experiment, the SIMULATA-generated APCs were not better than simple PBMC stimulation with peptides covering the HBs. Antigen-specific T cell responses had a Th1 phenotype with no IL-17 detected by intracellular staining.

Based on our increasing understanding of the potential role of Th17 responses in host protection against bacterial infection, we aimed to extend our preliminary data on secondary CD4 T cell responses using antigens from S. aureus or S. pneumoniae. Given the recently published inhibitory role of IL-2 on memory Th17 responses (Zielinski et al; Nature 2012), co-culture experiments were done in the presence or absence of IL-2. Interestingly, we demonstrated that SIMULATA-derived antigen-loaded APC were able to reveal S. aureus and S pneumoniae -specific memory Th17 responses only in the absence of IL-2. The capacity to induce Th17 responses was unique to SIMULATA-derived APCs and was not observed after long term PBMC co-culture with S. aureus or S. pneumoniae proteins or with overlapping peptide pools, even in the absence of IL-2.

As a next step and to better recapitulate the intramuscular vaccination injection site, it would be of interest to evaluate the impact of incorporating of (apoptotic) antigen-carrying cells in the collagen matrix (Qu et al, J Immunol 182:3650, 2009) on phenotypic and functional properties of SIMULATA derived APCs.

Related documents