Aberrant Expression of Human Endogenous Retrovirus K9-Derived Elements Is Associated with Better Clinical Outcome of Acute Myelocytic Leukemia

Background: Acute myelocytic leukemia (AML) is a common hematological malignancy in adults. Although several risk stratifications based on cytogenetic and molecular abnormalities are available to decide the indication of allogeneic hematopoietic cell transplantation (allo-HCT), planning treatment strategies for AML without them remains challenging. In this study, using transcriptome datasets, we investigated the association of event-free survival (EFS) of intensively treated AML cases and the aberrant expression status of endogenous retrovirus (ERV)-derived open reading frames (ORFs), which have been reported to be associated with the pathophysiology of various malignancies and have the potential to become neoantigens in specific cancers.

Methods: The RNA-seq data were obtained for 173 AML cases from The Cancer Genome Atlas (TCGA-LAML) and the Gene Expression Omnibus (GEO; GSE49642), as well as for 29 normal hematopoietic stem cells from GEO (GSE111085 and GSE114922). All FASTQ files were mapped to the human genome (hg38), and then the reads were counted with the GTF-formatted annotation file of the human ERV ORF definition obtained from the gEVE database version 1.1. DESeq2 was used to calculate statistical differences in ERV expression between AML and normal HSCs, and those with an adjusted p value <0.05 were extracted as differentially-expressed ERVs (DE-ERVs). We conducted gene-set enrichment analysis (GSEA) of DE-ERVs based on Repbase ERV family annotations to annotate the DE ERV families in AML cells. Quantitative PCR was performed for validation of ERV-derived mRNA expression in AML-derived cell lines (THP1, KG1, HL60, and K562). Furthermore, immunopeptides presented by human leukocyte antigen (HLA) class I of these cell lines were collected using immunoprecipitation with an anti-HLA class I antibody, and their amino acid sequences were identified using mass spectrometry (immunopeptidomics) to investigate the AML specific neoantigens derived from DE-ERVs. Clinical information of 151 AML cases registered in TCGA-LAML was obtained from TCGA website for prognostic analyses.

Results: Access three independent DE analyses (TCGA-LAML vs. GSE111085, TCGA-LAML vs. GSE114922, and GSE49642 vs. GSE111085), a total of 557 DE-ERV ORFs were annotated as AML-characterizing DE-ERV ORFs. According to GSEA of the DE-ERV ORFs, a total of 10 ERV families were annotated as DE-ERV families, with total of 155 core-enrichment DE-ERV ORFs. Immunopeptidomics suggested that immunopeptides derived from seven of those 10 DE-ERV families were presented by HLA class I as neoantigens. Among those seven DE-ERVs, the mRNA expression values of human ERV family K9 (HERVK9) ORFs were found to be associated with EFS of patients treated with 3+7-based intensive chemotherapies (median EFSs of HERVK9-high and low expressors were 32.7 and 9.6 months, respectively; p=0.00479). This survival predominance in higher HERVK9 expressors was independent of their conventional risk stratifications (NCCN2017 and ELN2017) and gene mutations known to contributing to risk stratifications according to cox proportional hazards model (p=0.01971). Furthermore, according to GSEA, it was revealed that AML cells with higher expression of HERVK9 activated antigen processing and presentation, accompanied by excess expression of genes associated with responses to adaptive immune reaction and apoptosis, indicating that aberrant expression of HERVK9 may initiate an antineoplastic immune response against themselves via excess antigen presentation.

Conclusions: While ERVs are aberrantly expressed in AML cells, HERVK9 expression induces an anti-neoplastic immune reaction via excess antigen presentation and is associated with better EFS in cases treated with intensive chemotherapies, independent of known risk classifications. Clinically, measurement of HERVK9 expression at initial diagnosis may help the dicision-making of indication of allogeneic hematopoietic cell transplantation, especially in the cases with good- or intermediate-risk stratification. In addition, HERVK9-derived neoantigens presented by HLA class I may have the potential to be efficient target of cell therapies.