Recent successes of immunomodulatory approaches for the treatment of serious diseases such as cancer have generated a significant growth in efforts aimed at the discovery of novel therapeutics in this area. Immunomodulatory agents are expected to affect interactions among cells and signaling molecules involved in regulating the immune system. Therefore, hit discovery campaigns need to deliver profiles of the impact of compounds on these complex interactions.

Human peripheral mononuclear cells (PBMCs) are a primary source of various immune cells, including but not limited to NK cells, B cells, and T cells of various subtypes and stages of differentiation. PBMCs are an excellent model system for studying effects of potential drugs on the immune system, as many of the modulatory effects of compound treatment can be recapitulated. In this complex cell mixture, activation or suppression of the immune response (immunomodulation) is often seen in concert with coordinated cytokine secretion patterns (Figure 1).

Figure 1
Figure 1

T cells can be activated by treatment with phytohemagglutinin (PHA), which will trigger proliferation of the T-cell population as well as modulate the cytokine secretion profile of a PBMC culture as a whole. T cells are identified by the surface marker CD3. A specific subtype, cytotoxic T cells (CTLs), will also express CD8. Compounds that alter the ability of PHA-stimulated T cells to proliferate or secrete certain cytokines might be candidate immunomodulatory compounds for further investigation.

Library biased toward immunomodulating compounds

Based on a literature search for substances having a differential effect on cytokine secretion, seven compounds were selected (imiquimod, mycophenolic acid, resveratrol, thalidomide, tomentosin, verapamil, and compound C3) and used as templates for running an extensive structure–activity relationship (SAR) expansion against the Exquiron Compound Collection of 260,000 compounds, based on a published protocol.1 After a data fusion step, a total of 1438 compounds were identified. These compounds were cherry-picked for testing at two different concentrations (16.7 and 83 µM).

No-wash screening workflow

PBMCs were batch-labeled with the MultiCyt® FL4 Cell Proliferation Dye before plating into 384-well plates containing compounds from the SAR expansion selection. Each plate also included four reference substances as dilution series (resveratrol, verapamil, dexamethasone, and mitomycin C). Following plating of cells, PHA was added and cells were incubated for 3 days under appropriate tissue culture conditions (Figure 2).

Figure 2
Figure 2

After incubation was complete, 10 µL aliquots were stamped from each treatment plate into a multiplex of immunophenotyping antibodies (anti-CD3-FITC and anti-CD8PE) and the MultiCyt FL3 Membrane Integrity Dye. A second stamp of 3 µL from the same motherplate was used for QBeads detection of interleukin-17f (IL-17f), IL-6, and tumor necrosis factor (TNF). Each plate was read on the iQue Screener immediately after staining, without wash steps. Each 384-well plate took about 25 minutes to read.

Data analysis and activity profile generation

Among all the data generated by the iQue Screener, 11 parameters were extracted based on their biological significance (Table).

table
Table

Data were normalized plate-wise to the PHA-activated control cell population, using a modified z score transformation, and activity profiles were generated.2 Profiles for resveratrol at various concentrations are shown below (Figure 3).

Figure 3
Figure 3

Identification of compounds inducing specific phenotypes

Calculation of the Euclidian distance between profiles and subsequent similarity search against the profiles of the reference substances allowed the identification of compounds displaying specific phenotypes (Cpds1–4 and Cpds5–9 induce dexamethasone-like and verapamil-like phenotypes, respectively) (Figure 4).

Figure 4
Figure 4

Identification of compounds displaying new phenotypes

Through clustering and subsequent visual inspection of the activity profiles, compounds eliciting new phenotypes (that is, phenotypes not covered by the controls or reference substances) were identified as well. Examples for compounds inducing TNF s ecretion is shown in Figure 5.

Figure 5
Figure 5

Summary

High-throughput, multiplex screening of compounds on primary cells generates information-rich multivariate compound activity profiles that can be used for identifying or prioritizing potential therapeutic candidates.

Application of advanced data mining techniques to these profiles allows for the rapid identification of compounds with activities similar to those of reference substances (potentially bridging the gap between phenotype and mechanism of action), but also identifies compounds eliciting new, potentially interesting phenotypes.

 

Serge P. Parel, PhD, is director, chemistry and research informatics, Laurent Brault, PhD, serves as project leader assay development and hit discovery, and Daniela Brodbeck, PhD, works as director, biological sciences at Evotec.

Tom Duensing, PhD, is chief technology officer and Zhaoping Liu, PhD, is senior scientist at IntelliCyt.

References
1. Bergner A, Parel SP. J. Chem. Inf. Model. 2013; 53: 1057–1066.
2. Riesen F, et al. , Assay Drug Dev. Technol. 2015; 13: 415–427.

This site uses Akismet to reduce spam. Learn how your comment data is processed.