Marker-free T cell phenotyping - towards a label-free diagnostic and prognostic platform

Abstract

T cells are one of the key players in cell-mediated immunity. The peripheral blood consists of an intricate balance between various T cell subpopulations, which vary in their differentiation state and memory potential. Many of these T cell subpopulations are associated with diseases, and numerous studies highlight the importance of monitoring the frequencies of T cell subsets in peripheral blood for determining the immune competence status of a person. Particularly elderly persons show a wide spectrum of immunosenescence stages that need to be taken into account in targeted medicine. Methods that are used up to date to identify T cell subsets are based on the expression of an enormous and increasing number of surface markers that depend on cells’ differentiation and activation state. This is often insufficient since T cells display a great variety of differentiation states. In this thesis, a novel marker-free, fast and low-cost approach for T cell subpopulation identification is presented that is based on changes in adhesion- and migration-related parameters during T cell differentiation. We hypothesized that these fine phenotype shifts can be visualized with Reflection Interference Contrast Microscopy (RICM) imaging and suffice to allow clear classification of T cell subpopulations. To first prove the sensitivity of our approach we imaged CD4+ T cells isolated from human peripheral blood at different time points during in vitro differentiation. Several morphological parameters correlated with memory T cell marker expression and were distinct from less differentiated states, e.g. an increased fractal dimension (Df) parameter that describes the cell´s membrane topology or increased projected and adhesion cell area. In a next step we determined adhesion parameters and fractal dimensions from RICM images of T cell subsets that were differentiated in vivo. We separated naive, central memory and effector memory CD4+ T cells from human CD4+ populations. For all determined parameters, naive T cells showed significant lower values compared to memory cells. Using principal component analysis, the parameters that described the data variance of subsets best were identified to be area, adhesive area, excess perimeter and fractal dimension of contour/topology. These parameters were then used for generating cluster centres for each subpopulation with k-mean clustering. T cell subsets isolated from three different donors were then designated to these clustering points. With this approach, 80 % of T cells were assigned to their correct subtype. In accordance to in vitro differentiation experiments, we found an increase in the complexity of membrane topology and an increase in cell area from naive to central memory and to effector memory cells. These results suggest that all differentiation states of T cells may be continuously represented in adhesion maps. This visualisation approach will allow including T cell subsets that cannot be considered by conventional surface marker sets but are highly representative for distinct immune competence states. To apply our concepts of a marker-free adhesion-based assay to discriminate between a healthy CD4+ T cell “fingerprint” and one in disease state, we used an in vitro Graftversus- Host-Disease (GVHD) murine model. In this approach we did not evaluate adhesion properties on static RICM images but tracked adhesion-dependent intestine T cell homing on migration chips. For this, the chemokine CCL25 was anchored to 2D gold nanoparticle distance gradients and ICAM-1 was covalently linked to functional PEG molecules in between. CD4+ T cells with a GVHD phenotype were able to respond to decreasing CCL25 distances by increasing cell velocity, which was not observed for control CD4+ T cells from healthy mice. Furthermore, GVHD phenotype CD4+ T cells displayed a lower degree of complexity of cell trajectories than control cells. Our approaches of imaging T cell adhesion and/or in vitro homing is very promising for future marker-free immunodiagnostics that read out the immune competence status of a patient within short time, which can be applied in therapy monitoring and advising primary personalised treatment.