IRF2 loss is associated with reduced MHC I pathway transcripts in subsets of most human cancers and causes resistance to checkpoint immunotherapy in human and mouse melanomas.

Background: In order for cancers to progress, they must evade elimination by CD8 T cells or other immune mechanisms. CD8 T cells recognize and kill tumor cells that display immunogenic tumor peptides bound to MHC I molecules. One of the ways that cancers can escape such killing is by reducing expression of MHC I molecules, and loss of MHC I is frequently observed in tumors. There are multiple different mechanisms that can underly the loss of MHC I complexes on tumor and it is currently unclear whether there are particular mechanisms that occur frequently and, if so, in what types of cancers. Also of importance to know is whether the loss of MHC I is reversible and how such loss and/or its restoration would impact responses to immunotherapy. Here, we investigate these issues for loss of IRF1 and IRF2, which are transcription factors that drive expression of MHC I pathway genes and some killing mechanisms. Methods: Bioinformatics analyses of IRF2 and IRF2-dependent gene transcripts were performed for all human cancers in the TCGA RNAseq database. IRF2 protein-DNA-binding was analyzed in ChIPseq databases. CRISRPcas9 was used to knock out IRF1 and IRF2 genes in human and mouse melanoma cells and the resulting phenotypes were analyzed in vitro and in vivo. Results: Transcriptomic analysis revealed that IRF2 expression was reduced in a substantial subset of cases in almost all types of human cancers. When this occurred there was a corresponding reduction in the expression of IRF2-regulated genes that were needed for CD8 T cell recognition. To test cause and effect for these IRF2 correlations and the consequences of IRF2 loss, we gene-edited IRF2 in a patient-derived melanoma and a mouse melanoma. The IRF2 gene-edited melanomas had reduced expression of transcripts for genes in the MHC I pathway and decreased levels of MHC I complexes on the cell surface. Levels of Caspase 7, an IRF2 target gene involved in CD8 T cell killing of tumors, were also reduced. This loss of IRF2 caused both human and mouse melanomas to become resistant to immunotherapy with a checkpoint inhibitor. Importantly, these effects were reversible. Stimulation of the IRF2-deficient melanomas with interferon induced the expression of a functionally homologous transcription factor, IRF1, which then restored the MHC I pathway and responsiveness to CPI. Conclusions: Our study shows that a subset of cases within most types of cancers downregulates IRF2 and that this can allow cancers to escape immune control. This can cause resistance to checkpoint blockade immunotherapy and is reversible with currently available biologics.

Cancer Immune Evasion Through Loss of MHC Class I Antigen Presentation.

Major histocompatibility class I (MHC I) molecules bind peptides derived from a cell's expressed genes and then transport and display this antigenic information on the cell surface. This allows CD8 T cells to identify pathological cells that are synthesizing abnormal proteins, such as cancers that are expressing mutated proteins. In order for many cancers to arise and progress, they need to evolve mechanisms to avoid elimination by CD8 T cells. MHC I molecules are not essential for cell survival and therefore one mechanism by which cancers can evade immune control is by losing MHC I antigen presentation machinery (APM). Not only will this impair the ability of natural immune responses to control cancers, but also frustrate immunotherapies that work by re-invigorating anti-tumor CD8 T cells, such as checkpoint blockade. Here we review the evidence that loss of MHC I antigen presentation is a frequent occurrence in many cancers. We discuss new insights into some common underlying mechanisms through which some cancers inactivate the MHC I pathway and consider some possible strategies to overcome this limitation in ways that could restore immune control of tumors and improve immunotherapy.

The stoichiometry of the outer kinetochore is modulated by microtubule-proximal regulatory factors.

The kinetochore is a large molecular machine that attaches chromosomes to microtubules and facilitates chromosome segregation. The kinetochore includes submodules that associate with the centromeric DNA and submodules that attach to microtubules. Additional copies of several submodules of the kinetochore are added during anaphase, including the microtubule binding module Ndc80. While the factors governing plasticity are not known, they could include regulation based on microtubule-kinetochore interactions. We report that Fin1 localizes to the microtubule-proximal edge of the kinetochore cluster during anaphase based on single-particle averaging of super-resolution images. Fin1 is required for the assembly of normal levels of Dam1 and Ndc80 submodules. Levels of Ndc80 further depend on the Dam1 microtubule binding complex. Our results suggest the stoichiometry of outer kinetochore submodules is strongly influenced by factors at the kinetochore-microtubule interface such as Fin1 and Dam1, and phosphorylation by cyclin-dependent kinase. Outer kinetochore stoichiometry is remarkably plastic and responsive to microtubule-proximal regulation.

Regulation of kinetochore configuration during mitosis.

Successful proliferation and function of an organism relies on the equal segregation of its genetic material during cell division. Duplicate sister chromatids need to accurately segregate at mitosis. Precise segregation depends on a multicomplex protein structure called the kinetochore. The kinetochore assembles at centromeres and attaches to microtubules to segregate sister chromatids. Even though the kinetochore structure was first observed nearly a century ago, many aspects of the regulation, function and assembly of this large 100 + protein structure remain to be determined. Improved microscopy and proteomics techniques over the years have helped to reveal the structure, composition and localization of sub-modules of the kinetochore. Recent work suggests that the configuration of the kinetochore is plastic, with extra submodules being added during anaphase to support microtubule tracking and chromosome segregation. We discuss our perspective of how this process might be regulated.

Chromatin assembly factor-1 (CAF-1) chaperone regulates Cse4 deposition into chromatin in budding yeast.

Correct localization of the centromeric histone variant CenH3/CENP-A/Cse4 is an important part of faithful chromosome segregation. Mislocalization of CenH3 could affect chromosome segregation, DNA replication and transcription. CENP-A is often overexpressed and mislocalized in cancer genomes, but the underlying mechanisms are not understood. One major regulator of Cse4 deposition is Psh1, an E3 ubiquitin ligase that controls levels of Cse4 to prevent deposition into non-centromeric regions. We present evidence that Chromatin assembly factor-1 (CAF-1), an evolutionarily conserved histone H3/H4 chaperone with subunits shown previously to interact with CenH3 in flies and human cells, regulates Cse4 deposition in budding yeast. yCAF-1 interacts with Cse4 and can assemble Cse4 nucleosomes in vitro. Loss of yCAF-1 dramatically reduces the amount of Cse4 deposited into chromatin genome-wide when Cse4 is overexpressed. The incorporation of Cse4 genome-wide may have multifactorial effects on growth and gene expression. Loss of yCAF-1 can rescue growth defects and some changes in gene expression associated with Cse4 deposition that occur in the absence of Psh1-mediated proteolysis. Incorporation of Cse4 into promoter nucleosomes at transcriptionally active genes depends on yCAF-1. Overall our findings suggest CAF-1 can act as a CenH3 chaperone, regulating levels and incorporation of CenH3 in chromatin.

Structural plasticity of the living kinetochore.

The kinetochore is a large, evolutionarily conserved protein structure that connects chromosomes with microtubules. During chromosome segregation, outer kinetochore components track depolymerizing ends of microtubules to facilitate the separation of chromosomes into two cells. In budding yeast, each chromosome has a point centromere upon which a single kinetochore is built, which attaches to a single microtubule. This defined architecture facilitates quantitative examination of kinetochores during the cell cycle. Using three independent measures-calibrated imaging, FRAP, and photoconversion-we find that the Dam1 submodule is unchanged during anaphase, whereas MIND and Ndc80 submodules add copies to form an "anaphase configuration" kinetochore. Microtubule depolymerization and kinesin-related motors contribute to copy addition. Mathematical simulations indicate that the addition of microtubule attachments could facilitate tracking during rapid microtubule depolymerization. We speculate that the minimal kinetochore configuration, which exists from G1 through metaphase, allows for correction of misattachments. Our study provides insight into dynamics and plasticity of the kinetochore structure during chromosome segregation in living cells.

Ligand-induced conformation changes drive ATP hydrolysis and function in SMARCAL1.

Mutations and deletions in SMARCAL1, an SWI2/SNF2 protein, cause Schimke immuno-osseous dysplasia (SIOD). SMARCAL1 preferentially binds to DNA molecules possessing double-stranded to single-stranded transition regions and mediates annealing helicase activity. The protein is critical for alleviating replication stress and maintaining genome integrity. In this study, we have analysed the ATPase activity of three mutations ­ A468P, I548N and S579L ­ present in SIOD patients. These mutations are present in RecA-like domain I of the protein. Analysis using active DNA-dependent ATPase A domain (ADAAD), an N-terminal deleted construct of bovine SMARCAL1, showed that all three mutants were unable to hydrolyse ATP. Conformational studies indicated that the α-helix and ß-sheet content of the mutant proteins was altered compared to the wild-type protein. Molecular simulation studies confirmed that major structural changes had occurred in the mutant proteins. These changes included alteration of a loop region connecting motif Ia and II. As motif Ia has been implicated in DNA binding, ligand binding studies were done using fluorescence spectroscopy. These studies revealed that the Kd for protein-DNA interaction in the presence of ATP was indeed altered in the case of mutant proteins compared to the wild-type. Finally, in vivo studies were done to complement the in vitro and in silico studies. The results from these experiments demonstrate that mutations in human SMARCAL1 that result in loss in ATPase activity lead to increased replication stress and therefore possibly manifestation of SIOD.