Prevalence and impact of molecular variation in the three-prime repair exonuclease 1 TREX1 and its implications for oncology

Abstract

Background: The three-prime repair exonuclease 1, TREX1, degrades cytosolic DNA to prevent aberrant immune activation. Its inactivation results in DNA accumulation in the cytosol and induction of the cGAS-STING DNA sensing pathway, interferon signaling, and inflammation. Germline pathogenic TREX1 mutations are known to lead to hereditary autoimmune and autoinflammatory disorders, whereas the consequences of TREX1 mutations in cancer remain poorly understood.

Results: To assess the importance of human TREX1 amino acid variants, we analyzed protein sequences of the functional TREX1b isoform from 168 mammalian species and integrated available data on TREX1 sequence and copy number alterations in hereditary autoimmune and autoinflammatory disorders, cancer, and in human populations. While the entire TREX1b protein was conserved in placental mammals, egg-laying mammals and marsupials had their own unique C-terminal regions, with each predicted to contain a transmembrane domain. We modeled human TREX1 variants occurring in autoimmune disease and cancer samples at 12 protein positions to evaluate their predicted impact on protein stability and function.

Conclusions: Our findings provide novel insight into the role of TREX1 molecular variation in cancer, where genetic or epigenetic loss of TREX1 activity may improve susceptibility to treatment. However, TREX1 gene deletion in tumors was associated with unfavorable patient outcomes, most likely due its frequent co-occurrence with the loss of the entire 3p chromosomal arm, which contains known cancer-related genes.

Keywords: Autoimmune disease; Cancer; Copy number alteration; Innate immunity; Sequence variant; TREX1.

Fig. 1

A workflow representing collection, integration, and analysis of TREX1 data. Detailed information about individual steps is provided in the Methods. CNA, copy number alterations; COSMIC, the Catalogue of Somatic Mutations in Cancer; dbVar, database of human genomic Structural Variation; Decipher, DatabasE of genomiC varIation and Phenotype in Humans using Ensembl Resources; gnomAD, the Genome Aggregation Database; LOVD, Leiden Open Variation Database; PCAWG, Pan-Cancer Analysis of Whole Genomes; TCGA, The Cancer Genome Atlas Program; UCSF, University of California, San Francisco

Fig. 2

Phylogenetic tree inferred using the maximum likelihood (ML) for 159 TREX1b protein sequences from placental mammalian species. The tree is presented as a circular cladogram and was rooted using Atlantogenata, consistent with earlier studies [99]. Species of Atlantogenata are indicated by the purple bar. Human TREX1b sequence is shown by the red arrow. Numbers indicate bootstrap support for the tree nodes with support ≥ 60%, out of 1000 bootstrap replications. The scale indicates the number of amino acid substitutions per site

Fig. 3

Short 39 nt palindromic DNA sequence in the 3ʹ coding part of koala TREX1 gene encoding its C-terminal domain. A 39 nt koala DNA sequence, which was found to be conserved among marsupials and was also found in invertebrate, plant, and bacterial species (Table S2). B Predicted alternate DNA hairpin structures. C Predicted alternate RNA hairpin structures

Fig. 4

Frequency of TREX1 DNA sequence variants in TCGA. Percent of samples with TREX1 variants in 17 TCGA cancer types. The remaining cancer types had no samples with TREX1 variants. Types of TREX1 variants are indicated with different color stacked bars

Fig. 5

TREX1 mutation effect modeling. A FoldX predicted ΔΔG of folding (i.e., − ΔΔG of unfolding) vs. protein sequence position. B INPS-3D predicted ΔΔG of folding vs. protein sequence position. C AlphaMissense pathogenic probability score vs. protein sequence position. D FoldX predicted ΔΔG of folding vs. AlphaMissense pathogenic probability score. E INPS-3D predicted ΔΔG of folding vs. AlphaMissense pathogenic probability score. F Heatmap of Pearson correlation coefficients between scores from the different prediction methods

Fig. 6

TREX1 structural interactions analysis. A Analysis of contacts between TREX1 monomer and DNA fragment. Pink highlights denote sites at the interface of the TREX1 monomer and a complexed DNA molecule determined with UCSF Chimera contact analysis: D18, M19, E20, A21, G23, L24, P25, F26, A80, A81, I84, T85, H124, N125, R128, Y129, I156, K160, R164, S176, Y177, S178, L179, H195, D200. G23 and R164 are known sites of disease-associated variants. B Analysis of contacts between TREX1 monomers in a homodimer. Yellow highlights denote sites at the interface between two TREX1 monomers, determined with UCSF Chimera contact analysis: E33, H40, C42, A43, R62, V63, V64, D65, K66, L67, S68, L69, C70, V71, T85, G86, L87, V91, L92, A94, H95, G96, R97, Q98, C99, D101, N103, L104, L107, A110, F111, R113, R114, Q115, P116, H195, T196, E198. L87, C99, and D101 are known sites of disease-associated variants

Fig. 7

Fraction of samples with co-incident copy number loss in TREX1 and chromosome 3p or 3q in TCGA studies. A Samples with TREX1 loss and co-incident 3p loss. B Samples with TREX1 loss and co-incident 3q loss. Detailed number of samples per TREX1 status and 3p or 3q status subgroups can be found in Tables S15 and S16