The use of T-cell based immunotherapy has become a valid strategy for the treatment of a number of different malignancies. The FDA has approved monoclonal antibody therapies for cancer patients based on the following immunological targets: PD-1 (e.g., nivolumab, pembrolizumab), PD-L1 (e.g., avelumab, atezolizumab), and CTLA4 (e.g., ipilimumab). Even though these treatments have been successful for many, a large percentage of patients do not show a response to these antibodies.
To gain a better understanding of the genes that may be relevant to the tumors that manage to evade these immunotherapies, Nicholas Restifo, MD, a senior investigator at the Center for Cancer Research, NCI, and colleagues undertook a study that utilized a CRISPR-Cas9-based model to determine which mutations in genetically modified melanoma cells conferred resistance when treated with the CTLA4-targeting antibody ipilimumab (Nature 2017;548:537-542).
Somatic mutations that often arise during the course of cancer in a patient can render the disease vulnerable to a number of targeted therapies, including T-cell based immunotherapies based on immune checkpoint blockade; however, the genetic landscape within a tumor is a dynamic, evolving one, and frequently mutations will arise that will render the disease no longer susceptible to the previously efficacious treatment. When asked if there were genes that have mutations associated with resistance to immunotherapy, Restifo stated that “loss-of-function mutations have been noted in β2-microglobulin (B2M) and Janus kinases (JAK1 and JAK2) in patients unresponsive to immunotherapies.”
In previous studies of cancer cells, CRIPSR-Cas9 technology has been utilized to identify genes relevant to metastasis, proliferation, and drug resistance. In discussing the use of that technology in this study, Restifo noted, “We developed a ‘two-cell type’ (2CT)-CRISPR assay consisting of human T cells as effectors and melanoma cells as targets to identify the genes in tumors that were essential for the ‘effector function of T cells’ (EFT). We sought to understand how genetic manipulations in one cell type can affect a complex interaction with another cell type.”
Correlation analysis was performed on gene expression data from melanoma patients who had been treated with ipilimumab to determine the cell types and genes relevant to immunotherapy.
“Consistent with previous reports, we found that intratumoral cytolytic activity was strongly correlated with CD8+ T cell tumor infiltration and with the expression of genes involved in the major histocompatibility complex (MHC) class I antigen processing/presentation pathway, but weakly correlated with interferon-g (IFNg) signaling genes,” Restifo explained. “Reduction in the patients' overall survival was significantly associated with loss of B2M and TAP1 (transporter associated with antigen processing 1) expression in tumors biopsied before ipilimumab treatment.” As a result of these associations, CD8+ T cells and MHC class I genes were utilized to develop the 2CT-CRISPR assay system.
2CT CRISPR-Cas9 Assay Development
In describing the immune cells used in this 2CT model, Restifo commented, “This assay utilized genetically modified CD8+ T cells to specifically target an antigen expressed in a human leukocyte antigen (HLA)-class I-restricted manner; the genetically altered primary human T cells expressed a T-cell receptor (TCR) specific for the NY-ESO-1 antigen that we had previously reported to mediate tumor regression in patients with melanomas and synovial cell sarcomas.”
As an initial test for this 2CT-CRISPR model, the antigen presentation genes TAP2 and B2M were targeted in NY-ESO-1+ Mel264 melanoma cells using three unique single-guide RNAs (sgRNAs) via cloning with the lentiCRISPRv2 lentiviral vector.
“In this initial test, we wanted to see whether the loss of antigen presentation genes can directly compromise T cell-mediated cell lysis of human cancer cells using our 2CT-CRISPR assay,” Restifo said. When asked why the sgRNAs used targeted the tumor cells instead of the T cells, he noted, “In cancer, the tumor cells can show great genetic variability, even within different tumors in the same patient, thus we decided to focus on the genetic changes within those cells that render them resistant to elimination by the patient's T cells. Thus, in our initial studies, it made the most sense to investigate those cells, which showed the greatest genetic diversity.”
The researchers showed that the B2M-targeting lentivirus CRISPR knocked down protein levels by 95 percent using fluorescence-activated cell sorting. “Significant resistance to cell lysis was noted for the melanoma cells genetically modified with the B2M sgRNAs and the TAP2 sgRNAs against the cytolytic activity of ESO-targeting T cells,” Restifo stated. “These results showed that the loss of key MHC class I genes can promote evasion of T cell-mediated lysis in the melanoma cells we used in our optimized 2CT-CRISPR assay.”
Genome-Wide CRISPR-Cas9 Screen
To identify which genes may play a role in permitting tumor cells to evade T cell-mediated elimination, a genome-wide screen using the described 2CT CRISPR-Cas9 assay was undertaken. The Mel264 melanoma cells were treated with a genome-scale knockout library consisting of 123,411 sgRNAs that targeted 19,050 protein-coding genes, 1,864 microRNAs, and approximately 1,000 “non-targeting” control sgRNAs. The Mel624 cells were treated with the lentivirus at a multiplicity of infection (i.e., ratio of lentiviruses to cancer cells) of less than 0.3. When asked about this low ratio, Restifo explained, “In our system, each lentivirus inserts a single sgRNA into the cancer cell; by maintaining a lower ratio of virus to cancer cell, we are hoping to have a greater chance of one cancer cell being transduced by only one lentivirus. As a result of this, only one gene per cell would be altered per cancer cell affected.”
To determine which sgRNAs had an effect on tumor cell survival, deep sequencing analysis was performed on the sgRNA library representation in tumor cells prior to and after co-incubation with the NY-ESO-1-targeting T cells. “Those tumor cells that undergo T cell-mediated lysis do not show up when deep sequencing is performed, thus, only those sgRNAs which conferred resistance to the tumor cells will show up in the post-T cell-treatment analyses,” Restifo clarified. “We observed that the distribution of the sgRNA reads in T cell-treated samples versus controls was significantly altered in the screens having a higher number of T cells, indicating that the efficiency of this 2CT-CRISPR assay was dependent on the selection pressure applied by T cells.”
CRISPR Genome-Wide Assay Results
The candidate genes' consistent enrichment was quantified using three independent ranking methods. “There was a significant degree of overlap between the genes included using these three ranking systems,” Restifo noted. “Based on our initial optimization of the 2CT-CRISPR assay, we expected that genes directly associated with MHC class I antigen processing and presentation would be enriched in our screens; correspondingly, we found that HLA-A, B2M, TAP1, TAP2, and TAPBP were among the most highly enriched genes in our results. Curiously, many genes without an established connection to the EFT were ranked amongst the top 20 enriched genes in this genome-scale analysis, such as SOX10, CD58, MLANA, PSMB5, RPL23, and APLNR,” he also observed.
Biological Roles for Enriched Genes
T cells, as part of their effector mechanism, can induce transcriptional modification of the tumor microenvironment by secreting cytokines like IFNg or tumor necrosis factor-α (TNF-α), which can enhance both the recognition and lysis of tumor cells.
“We intersected the gene expression profiles of cytokine-induced genes with the candidate genes in order to assess whether any of these genes induced by effector cytokines are likely to have a functional role in the modulation of EFT,” Restifo explained. “We found that 13 IFNg-induced genes and three TNF-α-induced genes were captured in our 554 top gene candidates, bolstering the functional relevance of these cytokine-induced genes in tumor cells for EFT.”
The use of gene ontology and pathway analyses showed that, in addition to antigen presentation and IFNg signaling, the genes enriched in the 2CT-CRISPR screen have a number of different functions, including eukaryotic initiation factor 2 (eIF2) signaling, endoplasmic reticulum stress, apoptosis, assembly of RNA polymerase II, TNF receptor signaling, and protein ubiquitination pathways.
“These data showed that previously unrecognized signaling circuitry exist in tumors, the loss of which can dampen EFT, which clearly warrants further investigation,” Restifo said.
2CT Results With Human Tumor Genomes
In the next phase of the study, the investigators sought to determine whether the loss of the candidate genes, which were identified in the 2CT-CRISPR screen, had any association with the loss of cytolysis in cancer patients. To accomplish this, the gene expression profiles for 11,409 human tumors across 36 different tumor types were obtained from The Cancer Genome Atlas (TCGA) database.
“We measured the correlation between candidate genes and cytolytic activity in these datasets; with this approach, we were able to generate a list of genes that were associated with cytolysis for each TCGA cancer type and were also enriched in our 2CT screens,” Restifo explained. “Then, using hierarchical clustering, we identified a set of 19 genes that were correlated with cytolytic activity across most of the 36 cancer types; of these 19 genes, 10 were inducible by IFNg, indicating that these genes may be upregulated in cancers owing to an increased infiltration of T cells.”
The loss of expression of these 19 genes in tumors could either reduce or eradicate the following vital functions: tumor antigen presentation (relevant genes: HLA-A, HLA-F, B2M, TAP1, and TAP2); T cell co-stimulation (relevant genes: ICAM1, CLECL1, LILRA1, and LILRA3); or cytokine production and signaling (relevant genes: JAK2 and STAT1). Consequently, the loss or reduction of these functions that drive the infiltration and activation of T cells in the TME could serve as the primary mechanism for immune evasion.
In Vitro Validation of Top Candidates
In the next phase of the study, the researchers attempted to validate 17 genes identified in their screens that had no prior known association with the antitumor activity of T cells. In describing these studies, Restifo explained, “We included sgRNAs targeting CTAG1B (encoding tumor antigen, NY-ESO-1) and TAPBP (encoding tapasin involved in MHC class I antigen processing) as positive controls.” Each gene was targeted using four different sgRNAs in NY-ESO-1+ Mel624 and A375 melanoma cells, and resistance to ESO-targeting T cells was quantified.
When discussing the results for this part of the study, Restifo noted, “Fifteen of the 17 genes tested showed significant resistance to T cell-mediated cytolysis with at least one sgRNA in these cells; additionally, nine genes showed resistance using two different sgRNAs in both A375 and Mel624 cells.”
The nine genes that were confirmed by two unique sgRNAs each were subsequently tested in a model that utilized melanoma cells that expressed a different T-cell receptor antigen, MART-1 (termed MART-1/MLANA+ Mel264 cells). This was done to determine if the results obtained previously with the NY-ESO-1 antigen were generalizable to tumor cells expressing other T-cell receptor antigens. “Here, we introduced sgRNAs targeting several of these genes into MART-1/MLANA+ Mel624 cells and then co-cultured them with MART-1-targeting human T cells,” Restifo explained.
To show validity for the MART-1 model, control tumor cells were modified with non-targeting sgRNAs. Upon incubation with MART-1-targeting T cells, the non-target sgRNA Mel624 cells were completely eradicated. The tumor cells transduced with sgRNAs targeting each of the nine candidate genes showed increased resistance to the MART-1 T cells.
“The genes that we validated across different melanoma cell lines and different antigen-TCR combinations include cellular cytoskeleton genes (COL17A1 (collagen type XVII alpha 1)) and TWF1 (twinfilin-1); a microRNA (hsa-mir-101-2), and a 60S ribosomal subunit (RPL23). These findings highlight the role that these biological processes play in EFT modulation by tumor cells,” Restifo noted.
Another candidate gene identified in the genome-wide 2 CT-CRISPR assay was APLNR, the gene that encodes the apelin receptor. Mutations in this gene had been previously noted in some cancers.
“We wanted to see if there had been any loss of function mutations in this gene present in the tumors of patients that had prior treatment with T cell-based immunotherapies,” Restifo stated. “To do this, we mined available whole exome sequencing datasets from patients with metastatic melanoma and lung cancer previously treated with checkpoint blockade therapies, including anti-CTLA4 (ipilimumab) and anti-PD1 (pembrolizumab or nivolumab) antibodies.”
In these analyses, seven non-synonymous mutations were identified. Additionally, whole exome sequencing was performed on a resected metastatic melanoma lung lesion from a patient with disease that had shown resistance to both anti-PD-1 (nivolumab) and anti-CTLA4 (ipilimumab) therapies. This sequencing yielded two non-synonymous mutations in APLNR (T44S and C181S).
In a follow-up experiment, four non-synonymous APLNR mutations (T44S, C181S, P292L, and G349E) were selected for testing to see if they limited EFT.
“We re-introduced either wild-type (WT) APLNR or the specifically mutated APLNR using lentiviral transduction into APLNR-knockout A375 melanoma cells,” Restifo explained. “Although re-introduction of WT APLNR, or APLNR with C181S or P292L mutations rescued the sensitivity of APLNR-knockout cells to T cell-mediated cytolysis, the T44S and G349E mutations only resulted in partial rescue, indicating that these mutations may attenuate EFT. The presence of these loss-of-function mutations in tumors that show a lack of response to immunotherapy suggests that the functional loss of APLNR in tumors could be associated with immunosuppression in vivo.”
Mechanisms for APLNR Regulation of EFT
As an initial step to determine how APLNR may affect EFT, the researchers utilized RNA sequencing techniques to examine the transcriptome of APLNR-knockout cells. “We did not find any substantial differences in mRNA transcript levels of genes involved in antigen presentation, T cell inhibition, or co-stimulation; consequently, we examined whether APLNR regulates EFT by modulating protein signaling,” Restifo commented.
Previous studies have shown that APLNR (the protein) interacts with 96 different intracellular proteins, and of these JAK1 was the most enriched in this 2CT-CRISPR screen. “We performed immunoprecipitation experiments which confirmed that APLNR binds to JAK1 in both A375 and HEK293T cells,” Restifo stated.
The IFNg-driven phosphorylation of JAK1 has been demonstrated to stimulate the JAK–STAT signaling cascade, which then augments both antigen processing and presentation in tumors. This, in turn, enhances the recognition and cytolysis of tumor cells by T cells.
“As a first step to explore this signaling mechanism, we tried to alter the phosphorylation of JAK1 by treating tumor cells with apelin, the endogenous peptide ligand for the G-protein-coupled apelin receptor; however, we noted no changes in JAK1 phosphorylation with these treatments,” Restifo noted. “Next, we tested whether the activation of APLNR using peptide ligands ([Pyr1]apelin-13, apelin-17, and apelin-36) or ML233 (a small-molecule agonist) on tumor cells alters EFT. We did not observe any significant effect of these treatments, which suggests that the regulation of EFT by APLNR may be independent of its canonical G-protein-coupled receptor signaling.”
When asked about the findings of these studies, Restifo commented, “We have used a two-cell type CRISPR screen to discover both well-established and novel genes in cancer cells that regulate EFT. Our findings have direct clinical implications, as these data may serve as a functional blueprint for how to study the emergence of tumor resistance to T cell-based cancer therapies. Many of the genes that were uncovered in our studies, like APLNR, were not even on our radar in terms of being expected to participate in EFT.”
In discussing the collaborative nature of this study, Restifo singled out the contributions of three individuals. “All of this work was done in collaboration with Feng Zhang from the Broad Institute/MIT, one of the great innovators of CRISPR. This work was also done in close collaboration with Neville Sanjana from New York Genome and NYU, who is the co-first author of the paper. Finally, the paper comprises the PhD thesis work of a talented graduate student named Shashank Patel. This paper was essentially his thesis for a PhD from Georgetown University.”
Concerning APLNR, Restifo had the following comments, “Our data reveals a novel role of APLNR in regulating the antitumor response of T cells via modulation of the JAK-STAT signaling in target cells.” The cytokines IFNg and TNFα have recently been shown to exert an antitumor effect via the alteration of blood vessels' endothelial cells (i.e., where APLNR is highly expressed) to induce ischemia in the tumor stroma.
“Given our finding that APLNR interacts with JAK1 to augment IFNg response, APLNR may further increase the sensitivity of tumor blood vessels to IFNg and thus improve the antitumor efficacy of T cells,” Restifo said. “Hence, we speculate that APLNR might have a role beyond direct tumor cell recognition in vivo, and its expression on tumor blood vessels and stromal cells should be investigated in future studies.
“A careful evaluation and validation of mutations in these genes on a personalized basis in immunotherapy-resistant patients may allow identification of novel mechanisms of immune escape and speed the development of new drugs that circumvent these escape mechanisms.”
Richard Simoneaux is a contributing writer.