One-Time Optimization of Advanced T Cell Culture Media Using a Machine Learning Pipeline.

The growing application of cell and gene therapies in humans leads to a need for cell type-optimized culture media. Design of Experiments (DoE) is a successful and well known tool for the development and optimization of cell culture media for bioprocessing. When optimizing culture media for primary cells used in cell and gene therapy, traditional DoE approaches that depend on interpretable models will not always provide reliable predictions due to high donor variability. Here we present the implementation of a machine learning pipeline into the DoE-based design of cell culture media to optimize T cell cultures in one experimental step (one-time optimization). We applied a definitive screening design from the DoE toolbox to screen 12 major media components, resulting in 25 (2k + 1) media formulations. T cells purified from a set of four human donors were cultured for 6 days and cell viability on day 3 and cell expansion on day 6 were recorded as response variables. These data were used as a training set in the machine learning pipeline. In the first step, individual models were created for each donor, evaluated and selected for each response variable, resulting in eight final statistical models (R 2 > 0.92, RMSE < 1.5). These statistical models were used to predict T cell viability and expansion for 105 random in silico-generated media formulations for each donor in a grid search approach. With the aim of identifying similar formulations in all donors, the 40 best performing media formulations of each response variable were pooled from all donors (n = 320) and subjected to unsupervised clustering using the k-means algorithm. The median of each media component in each cluster was defined as the cluster media formulation. When these formulations were tested in a new set of donor cells, they not only showed a higher T cell expansion than the reference medium, but also precisely matched the average expansion predicted from the donor models of the training set. In summary, we have shown that the introduction of a machine learning pipeline resulted in a one-time optimized T cell culture medium and is advantageous when working with heterogeneous biological material.

Induced miR-99a expression represses Mtor cooperatively with miR-150 to promote regulatory T-cell differentiation.

Peripheral induction of regulatory T (Treg) cells provides essential protection from inappropriate immune responses. CD4(+) T cells that lack endogenous miRNAs are impaired to differentiate into Treg cells, but the relevant miRNAs are unknown. We performed an overexpression screen with T-cell-expressed miRNAs in naive mouse CD4(+) T cells undergoing Treg differentiation. Among 130 candidates, the screen identified 29 miRNAs with a negative and 10 miRNAs with a positive effect. Testing reciprocal Th17 differentiation revealed specific functions for miR-100, miR-99a and miR-10b, since all of these promoted the Treg and inhibited the Th17 program without impacting on viability, proliferation and activation. miR-99a cooperated with miR-150 to repress the expression of the Th17-promoting factor mTOR. The comparably low expression of miR-99a was strongly increased by the Treg cell inducer "retinoic acid", and the abundantly expressed miR-150 could only repress Mtor in the presence of miR-99a. Our data suggest that induction of Treg cell differentiation is regulated by a miRNA network, which involves cooperation of constitutively expressed as well as inducible miRNAs.

Cleavage of roquin and regnase-1 by the paracaspase MALT1 releases their cooperatively repressed targets to promote T(H)17 differentiation.

Humoral autoimmunity paralleled by the accumulation of follicular helper T cells (T(FH) cells) is linked to mutation of the gene encoding the RNA-binding protein roquin-1. Here we found that T cells lacking roquin caused pathology in the lung and accumulated as cells of the T(H)17 subset of helper T cells in the lungs. Roquin inhibited T(H)17 cell differentiation and acted together with the endoribonuclease regnase-1 to repress target mRNA encoding the T(H)17 cell-promoting factors IL-6, ICOS, c-Rel, IRF4, IκBNS and IκBζ. This cooperation required binding of RNA by roquin and the nuclease activity of regnase-1. Upon recognition of antigen by the T cell antigen receptor (TCR), roquin and regnase-1 proteins were cleaved by the paracaspase MALT1. Thus, this pathway acts as a 'rheostat' by translating TCR signal strength via graded inactivation of post-transcriptional repressors and differential derepression of targets to enhance T(H)17 differentiation.

Adenoviral transduction of naive CD4 T cells to study Treg differentiation.

Regulatory T cells (Tregs) are essential to provide immune tolerance to self as well as to certain foreign antigens. Tregs can be generated from naive CD4 T cells in vitro with TCR- and co-stimulation in the presence of TGFß and IL-2. This bears enormous potential for future therapies, however, the molecules and signaling pathways that control differentiation are largely unknown. Primary T cells can be manipulated through ectopic gene expression, but common methods fail to target the most important naive state of the T cell prior to primary antigen recognition. Here, we provide a protocol to express ectopic genes in naive CD4 T cells in vitro before inducing Treg differentiation. It applies transduction with the replication-deficient adenovirus and explains its generation and production. The adenovirus can take up large inserts (up to 7 kb) and can be equipped with promoters to achieve high and transient overexpression in T cells. It effectively transduces naive mouse T cells if they express a transgenic Coxsackie adenovirus receptor (CAR). Importantly, after infection the T cells remain naive (CD44(low), CD62L(high)) and resting (CD25(-), CD69(-)) and can be activated and differentiated into Tregs similar to non-infected cells. Thus, this method enables manipulation of CD4 T cell differentiation from its very beginning. It ensures that ectopic gene expression is already in place when early signaling events of the initial TCR stimulation induces cellular changes that eventually lead into Treg differentiation.

Roquin paralogs 1 and 2 redundantly repress the Icos and Ox40 costimulator mRNAs and control follicular helper T cell differentiation.

The Roquin-1 protein binds to messenger RNAs (mRNAs) and regulates gene expression posttranscriptionally. A single point mutation in Roquin-1, but not gene ablation, increases follicular helper T (Tfh) cell numbers and causes lupus-like autoimmune disease in mice. In T cells, we did not identify a unique role for the much lower expressed paralog Roquin-2. However, combined ablation of both genes induced accumulation of T cells with an effector and follicular helper phenotype. We showed that Roquin-1 and Roquin-2 proteins redundantly repressed the mRNA of inducible costimulator (Icos) and identified the Ox40 costimulatory receptor as another shared mRNA target. Combined acute deletion increased Ox40 signaling, as well as Irf4 expression, and imposed Tfh differentiation on CD4(+) T cells. These data imply that both proteins maintain tolerance by preventing inappropriate T cell activation and Tfh cell differentiation, and that Roquin-2 compensates in the absence of Roquin-1, but not in the presence of its mutated form.