Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Mar 17;135(6):e183745.
doi: 10.1172/JCI183745.

Activating antiviral immune responses potentiates immune checkpoint inhibition in glioblastoma models

Affiliations

Activating antiviral immune responses potentiates immune checkpoint inhibition in glioblastoma models

Deepa Seetharam et al. J Clin Invest. .

Abstract

Viral mimicry refers to the activation of innate antiviral immune responses due to the induction of endogenous retroelements (REs). Viral mimicry augments antitumor immune responses and sensitizes solid tumors to immunotherapy. Here, we found that targeting what we believe to be a novel, master epigenetic regulator, Zinc Finger Protein 638 (ZNF638), induces viral mimicry in glioblastoma (GBM) preclinical models and potentiates immune checkpoint inhibition (ICI). ZNF638 recruits the HUSH complex, which precipitates repressive H3K9me3 marks on endogenous REs. In GBM, ZNF638 is associated with marked locoregional immunosuppressive transcriptional signatures, reduced endogenous RE expression, and poor immune cell infiltration. Targeting ZNF638 decreased H3K9 trimethylation, increased REs, and activated intracellular dsRNA signaling cascades. Furthermore, ZNF638 knockdown upregulated antiviral immune programs and significantly increased PD-L1 immune checkpoint expression in diverse GBM models. Importantly, targeting ZNF638 sensitized mice to ICI in syngeneic murine orthotopic models through innate IFN signaling. This response was recapitulated in recurrent GBM (rGBM) samples with radiographic responses to checkpoint inhibition with widely increased expression of dsRNA, PD-L1, and perivascular CD8 cell infiltration, suggesting that dsRNA signaling may mediate response to immunotherapy. Finally, low ZNF638 expression was a biomarker of clinical response to ICI and improved survival in patients with rGBM and patients with melanoma. Our findings suggest that ZNF638 could serve as a target to potentiate immunotherapy in gliomas.

Keywords: Brain cancer; Cancer immunotherapy; Epigenetics; Oncology; Virology.

PubMed Disclaimer

Figures

Figure 1
Figure 1. The retroviral silencing complex mediates the suppression of immunogenic RNA species in gliomas.
(A) ZNF638 acts as the master regulator of a retroviral silencing complex to silence retroelement expression via H3K9 trimethylation. Removal of HUSH-mediated repressive marks enhances antiviral immune responses through innate dsRNA signaling. Made with BioRender. (B) Bulk RNA-seq data from TCGA GBM (n = 617) and LGG (n = 516) cohorts and our institutional cohort (n = 71) were analyzed to conduct immune deconvolution and assess IFN-stimulated gene–related (ISG-related) pathways and functions. Made with BioRender. (C) The HUSH complex and ZNF638 transcripts are inversely correlated with CD8 immune cell infiltration (RTCGA = – 0.2017; Rinst = – 0.5409) based on data obtained from TCGA GBM (n = 617), and LGG (n = 516) database as well as from our institutional cohort (n = 71). (D) Correlation matrix demonstrates enrichment of ISGs with expression of several REs as well as a negative association between the HUSH complex and MDA5 signaling. Additionally, ZNF638 and the HUSH complex are directly correlated with increased inhibition of NK cell activation and Type 1–IFN signaling. Genes assigned to pathways based on the Reactome Pathways database and Gene Ontology analysis. TE, transposable elements.
Figure 2
Figure 2. dsRNA-sensing and IFN-signaling pathways are transcriptionally upregulated in high-ZNF638 regions.
Circular visualization of gene expression in regions of high and low ZNF638 from a sample from a patient diagnosed with IDH WT recurrent GBM. Utilizing the NanoString GeoMx Digital Spatial Profiler, gene expression involved for the NFkB signaling pathway (ngenes = 200), IFN-α signaling pathway (ngenes = 97), inflammatory response (ngenes = 200), JAK/STAT signaling pathway (ngenes = 87), and the HUSH complex (ngenes = 200), are shown. All gene sets obtained from the molecular signatures database (MSigBr). Central image shows the full tumor sample from the patient. Red, CD45; Green, Olig2.
Figure 3
Figure 3. ZNF638 suppresses total RE expression and dsRNA sensing.
(A) Single-cell RNA-seq and clustered analysis depicting ZNF638 expression and Neftel state classification (n = 18,400 cells) (NPC-like, neural-progenitor-like; OPC-like, oligodendrocyte-progenitor-like; AC-like, astrocyte-like; MES-like, mesenchymal-like). (B) Heatmap depicting reduced cellular expression of the retroviral silencing complex (ZNF638, SETDB1, PPHLN1, MPHOSPH8, TASOR) is associated with increased total retroelement expression and increased expression of genes involved in the dsRNA sensing pathway (RIG-I, IFIH1, TLR3, IRF3, IRF7, MAVS) (n = 18,400 cells, REs = 5,680). (C) Low ZNF638 expression is associated with increased lymphocyte expression in individual GBM tumors using unsupervised clustering. UMAPs from tumors with low (ntumors = 4, ncells = 17,535) and high (ntumors = 4, ncells = 20,517) ZNF638 expression show heterogenous and distinct enrichment of cell types. (D) Violin plots illustrate the expression levels of B cells, dendritic cells, differentiated-like cells, endothelial cells, fibroblasts, granulocytes, myeloid cells, oligodendrocytes, pericytes, proliferative stem-like cells, stem-like cells, and T cells in tumors with low versus high ZNF638 expression, based on unbiased cell type annotation. While a trend toward increased infiltration of T cells, B cells, myeloid cells, and oligodendrocytes is observed, this did not reach statistical significance, likely due to the limited number of samples. Single-cell data for panels AD were obtained from the European Genome-Phenome Archive under accession number EGAS00001005300.
Figure 4
Figure 4. ZNF638 expression is enriched in GBM and induces dsRNA signaling when knocked down.
(A and B) IHC staining and quantification for ZNF638 demonstrates marked overexpression of ZNF638 in GBM tumors compared with matched normal cortex (n = 43 versus 10, P < 0.0002). Results independently verified in biological replicate from 2 separate patient cohorts. (C) Western blot demonstrates that ZNF638 expression is enriched in GBM tumor samples compared with patient-matched adjacent normal brain (nT = 5 versus nC = 5). (D) Proteomic data from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) data portal for GBM (n = 12) and normal tissue (n = 99) corroborates significant enrichment of ZNF638, TASOR (FAM208A), MPHOSPH8, and SETDB1 in tumor tissue. (E) Western blot quantification validates that ZNF638 transient KD by siRNA reduces expression of HUSH via MPP8 and increases expression of RIG-I, MAVS, TBK1, pIRF3, and pSTAT1 in patient-derived GBM43 (1-way ANOVA, performed in technical triplicate, ****P < 0.0001, ***P < 0.001, **P < 0.01). (F) Knockdown of ZNF638 by siRNA decreases expression of SETDB1, PPHLN1, and MPP8 as well as increases expression of MAVS, TRAF3, TBK, MDA5, and TLR7 as measured by qPCR in A172 cells (performed in technical triplicate, 1-way ANOVA, ****P < 0.0001, ***P < 0.001, **P < 0.01).
Figure 5
Figure 5. ZNF638 KD increases dsRNA expression secondary to loss of H3K9me3 signature.
(A) ZNF638 knockdown by siRNA in A172 GBM cells demonstrates increased expression of dsRNA (yellow), decreased H3K9me3 (red), and increased expression of PD-L1 (red), as evidenced by quantitative immunofluorescence. Nuclei are stained blue with DAPI. (B) RNA-immunoprecipitation (J2 antibody) demonstrates increased pulldown of RE dsRNA with ZNF638 KD in A172 cells (performed in technical triplicate, 1-way ANOVA, ****P < 0.0001). (C) Flow cytometry demonstrates increased expression of dsRNA (anti-J2) and PD-L1 with ZNF638 KD. (D) ZNF638 knockdown elicits antiviral immune signaling via increased expression of pIRF3 (red) via quantitative immunofluorescence. Poly I:C (40 ng/mL) represents a positive control for pIRF3 signaling. Nuclei are stained blue with DAPI. (1-way ANOVA, ****P < 0.0001). (E) Knockdown of ZNF638 with siRNA results in loss of H3K9me3 in A172 cells based on Western blot (performed in technical triplicate, ****P < 0.0001, *P < 0.05). ZNF638-KD does not change H3K27 trimethylation.
Figure 6
Figure 6. ZNF638 knockdown upregulates retroelement expression and corresponding antiviral and immune programs.
(A) Volcano plot shows differential analysis of ATAC-seq and demonstrates that ZNF638 KD in a patient-derived GBM cell line results in opening of genomic regions associated with repeat elements (AluJb, Tigger15a) (log2fold changeAluJb = 2.69 PAluJb = 0.038, log2fold change(Tigger16A) = 2.57 PTigger16A = 0.036). (B) Volcano plot of differentially expressed genes from a patient-derived GBM43 cell line shows upregulation of transcripts related to the innate immune system (TMEM179B, log2fold change = 10.1, P = 0.033), actin regulation in activated T cells (CDC42SE2, log2fold change = 9.61, P = 0.016), and transcriptional activation (DNAJC2, log2fold change = 12.6, P = 0.0051) in ZNF638 KD. (C) ZNF638 KD in patient-derived GBM cell line results in upregulation of antiviral and immune pathways and programs. (D) ZNF638 KD in patient-derived GBM cell line results in upregulation of several retrotransposons, including LINE, LTR, and Alu elements (*P < 0.05).
Figure 7
Figure 7. dsRNA expression is associated with increased checkpoint expression and CD8+ cell infiltration in high grade gliomas.
(A) Locoregional dsRNA expression is associated with increased PD-L1 levels and tumor infiltrating lymphocytes in human GBM. Yellow, PD-L1; Green, dsRNA; Blue, DAPI; Cyan, CD8. (B) Increased expression of dsRNA stimulates a Type 1–IFN response to induce T cell activation and infiltration and PD-L1 upregulation. Made in BioRender. (C) Proteomic and transcriptomic data obtained from the CPTAC data portal validate a negative relationship between expression of ZNF638 and immune checkpoint markers: PD-L1 (P < 0.001), HLA-DRA (P < 0.01), and HLA-DRB1 (P < 0.001) (nlow = 50 and nhigh = 50 in each group).
Figure 8
Figure 8. ZNF638 knockdown potentiates ICI response in vivo.
(A) Diagram of in vivo study design with syngeneic murine GBM model. Made in BioRender. (B) Syngeneic GBM murine model with ZNF638 knockdown and PD-L1 inhibition demonstrates significantly improved survival relative to other treatment and control groups (n = 5 per group, P < 0.01). Multiple animals died on the same day in each group making it appear as if there are fewer animals than were included in each group. (C) ZNF638 knockdown and PD-L1 inhibition significantly reduces tumor volume relative to all other groups (1-way ANOVA, ****P < 0.00001, ***P < 0.0001, **P < 0.001, *P < 0.01). (D) ZNF638 knockdown + α-PD-L1 shows decreased expression of ZNF638 and increased expression of RIG-I, TLR3, NFK-β, and MERV (RLTR6) transcripts as measured by qPCR. Additionally, there is significant downregulation of IL-6, TNF-α (performed in biological triplicate, 1-way ANOVA, ****P < 0.00001, ***P < 0.0001, **P < 0.001). (E) ZNF638 knockdown with PD-L1 inhibition significantly increased expression of IFN-α and IFN-γ, as well as decreased expression of TNF-α (performed in biological triplicate, 1-way ANOVA, ****P < 0.0001, ***P < 0.001, **P < 0.01). (F) Proteomic cytokine profiler array with hierarchical clustering depicts distinct cytokine profiles between all treatment/control groups with the greatest difference between CTL+ α-PD-L1 mice and ZNF638 KD + α-PD-L1 mice (1-way ANOVA, *P < 0.05).
Figure 9
Figure 9. ZNF638 and dsRNA are biomarkers for ICI response in GBM.
(A) dsRNA expression correlates to increased CD8+ T cell infiltration and PD-L1 expression in patients with GBM receiving ICI. 60 year-old male patient with IDH-WT rGBM with temporal contrast enhancing intraaxial tumor concerning for tumor recurrence (MR T1CE) with low baseline CD8 infiltration (cyan) and PD-L1 expression (orange). Postoperative adjuvant ICI (anti-PD-1) resulted in increased enhancement in the tumor cavity 5 months after surgical resection, suggestive of immune pseudoprogression (immune cell infiltration and tumor necrosis). Postresponse multiplex immunofluorescence demonstrates increased dsRNA expression (green) associated with increased CD8 infiltration (cyan) and PD-L1 expression (orange). (B) Clinical responders to ICI with rGBM (R = 20, NR = 18, median = 8.238 versus 9.222 RPKM, P = 0.0034) and melanoma (R = 34, NR = 49, median = 3.855 versus 4.536 RPKM, P = 0.035) have markedly lower ZNF638 expression compared with nonresponders to ICI. (C) Low ZNF638 expression portends improved survival in recurrent GBM receiving immunotherapy (PD-1 or PD-L1) (Mantel-Cox, *P < 0.01).

Update of

References

    1. Tan AC, et al. Management of glioblastoma: State of the art and future directions. CA Cancer J Clin. 2020;70(4):299–312. doi: 10.3322/caac.21613. - DOI - PubMed
    1. Preusser M, et al. Prospects of immune checkpoint modulators in the treatment of glioblastoma. Nat Rev Neurol. 2015;11(9):504–514. doi: 10.1038/nrneurol.2015.139. - DOI - PMC - PubMed
    1. Medikonda R, et al. A review of glioblastoma immunotherapy. J Neurooncol. 2021;151(1):41–53. doi: 10.1007/s11060-020-03448-1. - DOI - PubMed
    1. Reardon DA, et al. Effect of nivolumab vs bevacizumab in patients with recurrent glioblastoma: The CheckMate 143 Phase 3 Randomized Clinical Trial. JAMA Oncol. 2020;6(7):1003–1010. doi: 10.1001/jamaoncol.2020.1024. - DOI - PMC - PubMed
    1. Chen R, et al. Endogenous retroelements and the viral mimicry response in cancer therapy and cellular homeostasis. Cancer Discov. 2021;11(11):2707–2725. doi: 10.1158/2159-8290.CD-21-0506. - DOI - PubMed