Advancing Immuno-Oncology Target Discovery

Anna Dostalova, Anneli Andersson, Jana Sponarova, and Philip Zimmermann
© NEBION AG. November 7, 2018


Suppressive immune checkpoint pathways are hijacked by tumors in order to evade the immune system. The immune-evading pathways are currently being clinically targeted, but the existing therapies are not effective in all patients and all cancer types. More efforts and efficient approaches are needed to identify how responsiveness can be predicted, and to discover novel targets as an alternative or combination treatment. In this study, we used GENEVESTIGATOR® in search of novel targets for cancer immunotherapy and their characterization. Several genes were identified as having notably similar regulation patterns as PD-1 (Programmed cell death protein 1), a well-known immunosuppressive checkpoint. Some of these are known targets for immunotherapy and confirm the method, while others are novel findings, which have not been described previously. These studies show how GENEVESTIGATOR® can effectively take advantage of the world's high-quality expression data, and help identifying new targets and characterize expression patterns of targets across cancers.


Immune checkpoint molecules offer a molecular target for modulating the immune response and a promising therapeutic target in numerous cancer indications. Monoclonal antibodies targeting the two best characterized immunosuppressive checkpoints, PD-1 and CTLA-4, have become essential immunotherapeutic agents with a broad clinical utility in recent years. However, the response rates remain limited and only a subset of patients can benefit from the current immune checkpoint blockade (ICB) therapies. Encouragingly, combined ICB of several molecular targets provides significant efficacy gains in the treatment of various cancer types (Ribas and Wolchok, 2018). Thus, there is an urgent need for novel target molecules. In this study, we used GENEVESTIGATOR® (Hruz et al., 2008) to analyze the expression of one of the best established ICB targets, PD-1 (Pdcd1, Programmed cell death protein 1) across perturbations from multiple experiments and to identify genes tightly co-regulated with PD-1.

Figure 1. Top-10 cell types with the highest expression of PD-1. The mouse RNA-seq compendium was selected and the Anatomy tool from the Condition Search toolset was used.

PD-1 gene expression across cell types and perturbations from multiple experiments

We first analyzed the cell type and tissue expression profile of PD-1 across the mouse RNA-seq compendium comprising 76 studies, 4889 samples, 122 tissues/cell types and 472 perturbations. The mouse RNA-Seq compendium was chosen due to its high number of studies focusing on tumor infiltrating immune cells and its high diversity of conditions. As expected, PD-1 showed highest expression in tumor-derived T-cells, followed by splenic T-cells and T-cells isolated from tumor-draining lymph nodes (Figure 1). In order to gain insight into the regulation of PD-1 expression in response to diseases and other stimuli, we used the GENEVESTIGATOR® Perturbations tool from the Condition Search toolset. We identified hepatocellular carcinoma, glioma, in-vitro T-cell activation, melanoma and lung neoplasm as the top-5 conditions causing PD-1 up-regulation (Figure 2).

Figure 2. Perturbations causing up-regulation of PD-1 expression. The mouse RNA-seq compendium was selected and the Perturbations tool from the Condition Search toolset was used, filtering for perturbations causing a fold-change in expression >1.5, P-value <0.01.

PD-1 co-regulated genes

To identify genes showing the same pattern of response to perturbations as PD-1, the Co-Expression tool from the Similarity Search panel was used (Figure 3), based on a selection of studies containing tumor infiltrating immune cells. Some of the top hits found are known immune checkpoint molecules (e.g. Lag3, Tigit, Litaf, Tnfrsf9), serving as a proof of the method, while others could represent novel potential candidates for immunotherapy targets. Among the top-50 co-expressed genes obtained, a network analysis was done by visualizing co-expression relationships beyond a chosen threshold (0.89) and circular clustering (the latter is done automatically in the Co-Expression tool). This analysis revealed the presence of several clusters. The cluster marked in red in Figure 3 contains six genes, mostly involved in the regulation of cytokine production and/or apoptosis. The cluster marked in yellow contains 11 genes, the majority of which are plasma membrane proteins. The cluster highlighted in blue is enriched in proteins involved in granulocyte chemotaxis.

Figure 3. Genes co-regulated with PD-1 in response to perturbations. The mouse RNA-seq compendium was selected and filtered for studies containing tumor infiltrating immune cells. PD-1 was set as the lead gene. Genes with a similar expression pattern were identified using the Co-expression tool across all perturbations in this compendium.

Validate co-regulation in CD8+ T-cells in immunotherapy and cancer studies

For a better visualization and characterization of the genes co-regulated with PD-1, we performed hierarchical clustering of the top-50 co-expressed genes across a set of perturbations involving T-cells (Figure 4). The clustering visualizes strong up-regulation of the PD-1 co-expressed genes in tumor derived CD8+ T-cells (hepatocellular carcinoma studies) as compared to naïve T-cells. Conversely, some of the genes are down-regulated in response to anti-PD-1 and/or anti-CTLA-4 immunotherapy in tumor derived CD8+ T-cells.

Figure 4. Regulation of PD-1 and the top-50 co-expressed genes in response to perturbations from studies containing tumor infiltrating immune cells. The list of genes co-regulated with PD-1 was generated as described in Figure 3. The mouse RNA-seq compendium was selected and filtered by anatomy to T-cell lineage. The Hierarchical clustering tool from the Similarity Search toolset in GENEVESTIGATOR® was used to order perturbations based on the similarity of the expression patterns across the genes co-expressed with PD-1. Both genes and perturbations were clustered. The most prominent groups of perturbations (disease, treatment) are highlighted on the left side, individual perturbations can be viewed in GENEVESTIGATOR®. Color coding: Red represents up-regulated, green represents down-regulated genes.

Identify cell type expression of PD-1 co-regulated genes

In order to identify T-cell-type specific differences in expression of the genes co-regulated with PD-1, we utilized the Hierarchical clustering tool again. This time we selected the option to cluster according to Anatomy (tissues and cell types). This analysis allowed for a rapid classification of PD-1 co-regulated genes with a rather uniform expression across various T-cell types (Litaf, Lgals3, Mgst1), as opposed to those with expression restricted to a specific cell type (Cnih3, Tnfsf4), e.g. tumor-derived CD8-positive T-cells (Figure 5).

Figure 5. Cell-type expression of PD-1 and the top-50 co-expressed genes. The list of genes co-regulated with PD-1 was generated as described in Figure 3. The mouse RNA-seq compendium was selected and filtered by anatomy to T-cell lineage. The Hierarchical clustering tool from the Similarity Search toolset in GENEVESTIGATOR® was used. Several genes with different patterns of expression are highlighted in blue.

Expression comparison between PD-1 and candidate target gene Lag3

Finally, we made use of the 2-Gene Plot tool for a detailed visualization of expression of PD-1 and one of the identified co-regulated genes in tumor derived CD8+ T-cells. Figure 6 shows expression of PD-1 and Lag3 in CD8+ T-cells isolated from hepatocellular carcinoma tumors and in healthy T-cells isolated from the spleen. Two clusters are apparent. Tumor-derived CD8+ memory T-cells (marked in yellow) along with other tumor-derived CD8+ T-cells (red) are clearly separated from healthy splenic CD8+ T-cells (blue).

Figure 6. PD-1 and Lag3 expression in splenic CD8+ T-cells and tumor-derived CD8+ T-cells. Data from a mouse hepatocellular carcinoma study (Philip et al., 2017) were selected. The 2-Gene Plot tool from the Similarity Search toolset in GENEVESTIGATOR® was used to plot expression of PD-1 against Lag3.


This study gives an impression of how GENEVESTIGATOR® can be used to advance immuno-oncology research. It illustrates the strengths of the platform in the identification of novel potential target molecules and their further characterization, e.g. in terms of cell-type specific expression and response to treatment. Besides the examples shown, numerous other functionalities of GENEVESTIGATOR® are available for researchers in the rapidly evolving field of immuno-oncology. Cell lines in which a gene of interest is highly or non-expressed can rapidly be identified using the Cell Lines tool. Conditions (cancers, treatments) affecting the expression of multiple genes of interest simultaneously can be easily identified by filtering for the sum of their Pi-scores in the Perturbations tool. As a result of deep and thorough curation of sample attributes, samples can easily be filtered, and differential expression analysis instantly performed based on a number of user-specified criteria (e.g. T-cell infiltration status, response to therapy or overall survival). In conclusion, GENEVESTIGATOR® provides a large, deeply curated database with a continuously growing coverage in the fields of oncology and immunology, in combination with a powerful search engine and visualization tool.

To try out GENEVESTIGATOR or to replicate/extend this analysis, please go here.


Ribas A, Wolchok JD (2018) Cancer immunotherapy using checkpoint blockade. Science 2018 Mar 23;359(6382):1350-1355. doi: 10.1126/science.aar4060  [Abstract]

Hruz T, Laule O, Szabo G, Wessendorp F, Bleuler S, Oertle L, Widmayer P, Gruissem W and P Zimmermann (2008) Genevestigator V3: a reference expression database for the meta-analysis of transcriptomes. Advances in Bioinformatics 2008, 420747 [Full Text]

Philip M, Fairchild L, Sun L, Horste EL, Camara S, Shakiba M, Scott AC, Viale A, Lauer P, Merghoub T, Hellmann MD, Wolchok JD, Leslie CS, Schietinger A (2017) Chromatin states define tumour-specific T cell dysfunction and reprogramming. Nature. 2017 May 25;545(7655):452-456. doi:10.1038/nature22367  [Abstract]