Skip to main content

Kinetic analysis of molecular profiles during human induced regulatory T cell (iTreg) differentiation: an integrative study

Final Report Summary - ITREGDIFFERENTIATION (Kinetic analysis of molecular profiles during human induced regulatory T cell (iTreg) differentiation: an integrative study)

Regulatory T cells (Tregs) suppress other immune cells and are critical mediators of peripheral tolerance, preventing autoimmune diseases and excessive inflammatory reactions. The importance of Tregs is exemplified by the human disease IPEX, in which loss of Tregs leads to severe systemic autoimmune disease lethal at an early age. However, Tregs act as a double-edged sword in the immune system as they can also hamper anti-tumor immunity in certain settings.

Therapeutic manipulations of Treg number and function are therefore subject to numerous clinical investigations in autoimmune and inflammatory diseases as well as cancer. First in-man trials of adoptive Treg transfer to prevent graft-versus-host disease and to assess safety in treating type 1 diabetes showed very promising outcomes. Yet, the number of naturally occurring Tregs (nTregs) is minute, encouraging the complementary approach of inducing Tregs (iTregs) from naive T cell precursors. Reinforcing this concept, there are several studies in mice indicating that iTreg transfer may be superior to nTreg transfer, yet the stability of iTregs remains a concern. iTregs are generated in vivo and ample evidence corroborates that iTregs exert non-redundant functions in addition to nTregs to maintain health. However, the molecular mechanisms governing iTreg generation are incompletely understood and procedures for iTreg generation are controversial, in particular for human cells despite accumulating evidence for differences in murine and human Tregs.

Here we therefore first established and compared different protocols for human iTreg generation, using the cytokines interleukin-2 (IL-2) and TGF-beta in combination with other compounds, namely retinoic acid, rapamycin or butyrate. We found that human iTregs highly expressed the ”master” Treg transcription factor FOXP3 with specific patterns depending on the iTreg-generating protocol. FOXP3 expression was accompanied by expression of several other Treg signature molecules, such as high expression of CTLA-4 and IKZF4, while exhibiting low expression of the cytokines Interferon-γ, IL-10 and IL-17. Importantly, we identified a novel combination of TGF-beta, retinoic acid and rapamycin as a robust protocol to induce human iTregs with superior suppressive activity in vitro compared to currently established induction protocols. However, iTregs generated by these protocols did not stably retain FOXP3 expression, presumably due to absent DNA demethylation of the Treg-specific demethylated region (TSDR) in the Foxp3 gene. In line with the unstable phenotype of iTregs, they did not suppress in vivo in a humanized graft-versus-host-disease mouse model, highlighting the need for further research to attain stable, suppressive iTregs.

To further understand the regulation and induction of human iTregs, we performed deep molecular profiling over time during Treg induction, starting from naïve cells.
Since there is no gold standard protocol for human Treg induction and we aimed mainly to find generic Treg regulators independent of a specific protocol, we used four different Treg-inducing compound combinations alongside with control cells that were activated without Treg-inducing compounds. RNA and proteins were extracted from the same cell samples (a subset of samples for protein analysis) to enable true integrative analysis from the same system, during five time points of iTreg differentiation. We prepared samples from three male donors in a similar age range in three independent experiments, and the phenotype of the cells was controlled by flow cytometry and qRT-PCR analysis. As a further comparison, we also included nTregs from the same donors. Molecular profiling was performed using state-of-the-art techniques, namely RNA sequencing (RNAseq) to understand the transcriptome and high resolution mass spectrometry proteomics to comprehend the proteome of iTregs. The experimental design of the RNAseq and proteomics studies was performed in a way that enables control of potential batch effects.

The time-dependent transcriptomic and proteomic data, capturing molecular events during generation of iTregs, were subject to bioinformatics analysis. After quality control and normalization, we could observe that the samples strongly correlated by time point and iTreg condition, while there was little donor variation or experimental batch effect. We detected about 30000 transcripts and 10000 proteins in total. In addition, some samples were subjected to deeper sequencing (45 to 70 Million reads/sample) to enable splice isoform detection. Known Treg signature molecules, such as FOXP3, showed the expected behavior of increased expression in all iTregs compared to the control group over time, as well as high expression in nTregs. Next, differential gene expression was modeled over time using three different methods (maSigPro and two variants of DESeq2 LRT). Around 10000 genes were differentially expressed over time in at least 2 of 3 methods, and 1300 to 2300 genes were differentially expressed according to those criteria in a specific iTreg condition compared to the control stimulated cells. Of these, about 350 genes were shared between all iTreg conditions, and thus were chosen as a basis to select novel “candidate molecules” that might be regulators of Treg differentiation. Differential protein expression was analyzed using the limma method and uncovered about 4000 proteins differentially expressed over time and about 200 to 1000 proteins specifically expressed in certain iTreg conditions compared to the control. Integrating differentially expressed genes and proteins and sub-selecting transcription factors revealed a short list which included many of the known important regulators of Tregs, such as BACH2, GATA3, IKZF4 and SATB1, highlighting the successfulness of our method. Importantly, along with these transcription factors well-established in Treg biology, several novel transcription factors appeared and were chosen as “candidate molecules” for validating their functional role in Treg differentiation.

Based on the data analysis strategies described above, we performed detailed expression pattern analysis and literature survey on individual molecules and picked ~40 candidate molecules to study regarding their functional role in Treg differentiation. To this end, we established a method of lentiviral transduction of primary human T cells to knock down genes of interest using small hairpin RNAs (shRNAs) while at the same time leaving the cells in a sufficiently naive state that still allows for iTreg differentiation. Using a targeted shRNA screen, we found that individually knocking down many of the selected “candidate molecules” had an impact on iTreg differentiation, namely resulting in decreased FOXP3 induction. These molecules will be followed up in more detail in the future to confirm their role in FOXP3 regulation and in this case, regarding their molecular mechanisms of action and their functional role in iTreg differentiation as well as in nTregs. Furthermore, potentially altered expression or function of these molecules in T cells from patients with autoimmune diseases or cancer could be of interest in the future.

In conclusion, our integrative analysis presents the first global time-resolved molecular profiling during iTreg differentiation. Integrating two different data types, transcriptomics and proteomics, enables even more detailed analyses of specific molecules acting during Treg differentiation, and in addition provides a high-resolution molecular scaffold enabling deeper analyses, for example to resolve molecular events independent and dependent of established factors such as FOXP3. Our study will provide an important data resource for other researchers in the field of T cell immunology as well as for bioinformaticians with an interest in analysis of time-resolved molecular data and integration of different data types.
Furthermore, based on functional validation, we identified new candidate molecules to be involved in regulation of Treg differentiation which will be subject to future studies. These molecules might include novel drug targets to be exploited in the future.
Together, our interdisciplinary analysis of the molecular events ruling human iTreg generation may have important implications for our understanding and ability to treat cancer, autoimmune and inflammatory diseases.