Biallelic mutations of three prime repair exonuclease 1 (TREX1) cause the lupus-like disease Aicardi–Goutières syndrome in which accumulation of a yet unknown endogenous DNA substrate of TREX1 triggers a cyclic GMP–AMP synthase-dependent type I IFN response and systemic autoimmunity. Products of reverse transcription originating from endogenous retroelements have been suggested to be a major substrate for TREX1, and reverse transcriptase inhibitors (RTIs) were proposed as a therapeutic option in autoimmunity ensuing from defects of TREX1. In this study, we treated Trex1−/− mice with RTIs. The serum RTI levels reached were sufficient to block retrotransposition of endogenous retroelements. However, the treatment did not reduce the spontaneous type I IFN response and did not ameliorate lethal inflammation. Furthermore, long interspersed nuclear elements 1 retrotransposition was not enhanced in the absence of Trex1. Our data do not support the concept of retroelement-derived cDNA as key triggers of systemic autoimmunity in Trex1-deficient humans and mice and motivate the continuing search for the pathogenic IFN-inducing Trex1 substrate.
Aicardi–Goutières syndrome (AGS) is a severe monogenic disorder in which chronic type I IFN production triggers lupus-like systemic autoimmunity (1, 2) resulting in destructive CNS inflammation (3). Biallelic loss-of-function mutations in any of the genes encoding three prime repair exonuclease 1 (TREX1) (4), the three subunits of the RNaseH2 complex (5), SAM and HD domain containing 1 (SAMHD1) (6) and adenosine deaminase acting on RNA 1 (7) as well as monoallelic gain-of-function mutations of the cytoplasmic dsRNA sensor MDA5 (IFIH1) (8, 9), have been found to cause AGS. Importantly, heterozygous mutations in TREX1 or RNASEH2A-C are associated with polygenic, sporadic systemic lupus erythematosus (10–12). Type I IFN–induced loss of self-tolerance represents a major pathogenic principle in systemic lupus erythematosus and AGS (2, 13).
Loss of Trex1 in mice results in lethal systemic autoimmunity, production of anti-nuclear autoantibodies, and multiorgan inflammation, most prominently in the heart (14–16). This phenotype is fully rescued by additional inactivation of the cytosolic dsDNA sensor cyclic GMP–AMP synthase (cGAS) (17, 18) or components of its signaling pathway, including STING (16, 19), TANK-binding kinase 1 (20), and IFN regulatory factor 3 (15). Likewise, crossing of Trex1−/− mice to type I IFN receptor knockout (KO) or to lymphocyte-deficient strains prevents pathology (15). Collectively, these studies suggest that in the absence of the DNase Trex1, an endogenous substrate accumulates and acts as a potent ligand for cGAS. Development of a therapeutic strategy that would prevent generation of the pathogenic cGAS ligand is highly desirable but hampered by our lack of understanding the physiological functions of Trex1, which was initially described as a DNA repair enzyme (21, 22). Loss of Trex1 results in a chronic DNA damage response and accumulation of ssDNA in the cytoplasm (23), which leaks out of the nucleus and induces production of IFN-β (24). The finding that Trex1 degrades cDNA from HIV-1 (25) and endogenous retroelements (15) as well as the increased frequency of cDNA sequences originating from endogenous retroelements in the cytoplasm of Trex1-deficient heart cells (15) supported the hypothesis that retroelement-derived cDNA represents a major endogenous substrate for Trex1.
Endogenous retroelements represent ancient retroviruses that infected the germ line and are now vertically transmitted in the population. They are divided into two major classes based on their genome organization and their modes of replication. Whereas LTR retroelements such as murine leukemia viruses and intracisternal-A particles (IAP) retained the genomic organization of simple exogenous retroviruses and reverse transcribe their genome in the cytoplasm, non-LTR retroelements such as long interspersed nuclear elements 1 (LINE1) lack genetic information for capsid and envelope proteins and undergo reverse transcription in the nucleus at the site of integration (26). In the absence of TREX1, reverse transcripts of endogenous retroelements might activate cGAS. As autoimmunity in Trex1−/− mice has been shown to exclusively depend on the intracellular DNA sensor cGAS (17, 18), blocking reverse transcription therefore seems an attractive approach to prevent spontaneous IFN production in Trex1-deficient cells. Intriguingly, prolonged survival and reduced inflammation of the heart was reported in Trex1−/− mice that received a combination of reverse transcriptase inhibitors (RTIs) consisting of the non-nucleoside RTI (non-NRTI) nevirapine (NVP; Viramune) and the NRTI combination Truvada, which provides a fixed ratio of tenofovir disproxil fumarate (TDF) and emtricitabine (FTC) (27). Whether this effect was the result of a reduced type I IFN response remains unknown and prompted us to further investigate the mechanism by which the treatment with RTIs could affect the pathology in Trex1−/− mice. Furthermore, we asked the question whether the absence of Trex1 promotes retrotransposition.
Materials and Methods
All mice were on a C57BL/6 genetic background or backcrossed in our facility for at least four generations. Breeding and housing was performed at the Experimental Center of the Faculty of Medicine Carl Gustav Carus, Technical University of Dresden, under specific pathogen-free conditions. Animal experimentation was carried out according to institutional guidelines and was approved by the Landesdirektion Dresden.
Treatment of mice
Retrotransposition of tetORFeus was induced by treatment with 0.4 mg/ml doxycycline (AppliChem) via the drinking water upon conception. Litters from doxycycline-treated breedings were maintained on 0.4 mg/ml doxycycline until 30 d of age.
Treatment of mice with RTIs was performed as outlined in the main text. In experiments addressing the effect of RTIs on systemic autoimmunity, parental mice were treated starting 4 wk before mating. Parents and litters were maintained on RTI therapy until the end of the experiment.
The following RTIs were used in this study: 157 μM NVP (Viramune; Boehringer Ingelheim), 606 μM TDF either as pure TDF or Viread (both Gilead Sciences), 375 μM FTC either as pure FTC or Emtriva (both Gilead Sciences), 0.4 mg/ml Truvada (189 μM TDF plus 324 μM FTC, Gilead Sciences). All substances were provided by Boehringer Ingelheim and Gilead Sciences.
Cells were suspended in PBS plus 0.5% FCS and stained for 30 min at 4°C. After washing, cells were processed in PBS containing 0.5% BSA and 1 μg/ml propidium iodide (Sigma-Aldrich) and measured using BD LSR II or FACSAria III flow cytometers. Data were recorded using FACSDiva (BD Biosciences) and analyzed using FlowJo software. T cells were identified with CD3e-PE (eBioscience), B cells with CD19-PE-Cy7 (eBioscience), and Sca-1 was stained with anti–Sca-1-V500 (BD Biosciences).
HeLa cells from retrotransposition reporter assays were resuspended in PBS containing 0.5% FCS and 1 μg/ml propidium iodide and analyzed on a Miltenyi Biotec MACSQuant flow cytometer. Data were analyzed using MACSQuant (Miltenyi Biotec) or FlowJo analysis software.
Tissues were fixed in 4% formaldehyde solution (SAV LP), embedded in paraffin, and sections were stained with hematoxylin (Merck) and eosin (Shandon). In heart sections, myocardial changes (0–3) and infiltrating cells (0–3) were quantified using an additive score ranging from 0 to 6 as reported previously (28).
Quantitative RT-PCR for IFN-inducible gene transcripts
RNA was extracted using the RNeasy mini kit (Qiagen) and reverse transcribed by RevertAid H Minus RT (Thermo Fisher Scientific). Quantification of transcripts was performed using SYBR Green/ROX quantitative PCR (qPCR) master mix (Thermo Fisher Scientific) as previously described (28). Primer sequences are available upon request.
Quantification of tetORFeus retrotransposition
Doxycycline-treated CMVrtTA-tetORFeus mice were sacrificed on day 30 after birth and genomic DNA was extracted from the indicated organs. Subsequently, the genomic DNA was purified using phenol/chloroform and 40 ng was used in a TaqMan qPCR. Quantification of integrates was performed using Maxima Probe/ROX qPCR master mix (Thermo Fisher Scientific). A scheme of the assay principle is depicted in Supplemental Fig. 2. Primer sequences are available upon request.
Generation of TREX1-deficient HeLa cells using CRISPR/Cas9
TREX1 in HeLa cells was inactivated by simultaneous targeting of two different positions in the single coding exon of TREX1 (ENSG00000213689) using CRISPR/Cas9. The two guide RNA target sites were 5′-CAGCTGTGCTGGCAGCGCAT-3′ and 5′-ATACAGTGTGGCTACTGCCA-3′. For each target site, cDNA oligonucleotides (Sigma-Aldrich) were cloned into the px459 plasmid (ID 48139; Addgene) as described previously (29). Transfected cells were selected in the presence of 4 μg/ml puromycin for 2 d, and after recovery from selection for another 24 h single clones were isolated by limiting dilution into 96-well plates. Frame shift mutations were identified by TOPO TA cloning (Thermo Fisher Scientific) and Sanger sequencing of a PCR product. The absence of TREX1 was verified by Western blot using the Ab TREX1-E6 (sc-271870; Santa Cruz Biotechnology).
In vitro retrotransposition reporter assays
HeLa cells (105) were seeded in DMEM complemented with 10% FCS, 1× nonessential amino acids (Biochrom), 100 U/ml penicillin (Biochrom), and 100 μg/ml streptomycin (Biochrom) into a single well of a six-well plate. On the next day, the cells were transfected with 0.5 μg of plasmid (with the L1 or IAP retrotransposition reporter plasmid or with a control GFP plasmid) using FuGENE HD (Promega) as described by the manufacturer. One day later, the medium was replaced by fresh medium containing RTI at the indicated concentrations. After 5 additional days of incubation, the frequency of GFP+ cells was determined on a MACSQuant flow cytometer (Miltenyi Biotec).
Treatment of primary mouse embryonic fibroblasts with Truvada
Embryonic day 14.5 embryos were extracted by cesarean section from females after timed breeding. Mouse embryonic fibroblasts (MEFs) were isolated and cultured for 1 wk in complete DMEM. To check for spontaneous IFN-inducible gene (ISG) transcription, 3 × 105 cells were dissolved in RLT buffer (Qiagen) for RNA isolation (day 0). For treatment, 3 × 105 Trex1+/+ and Trex1−/− MEFs were seeded in 3 ml of complete DMEM containing 3.33 μg/ml Truvada (1.6 μM TDF plus 2.7 μM FTC) into a cavity of a six-well plate. To dilute remaining type I IFNs and the ISG signature, the medium was exchanged every second day (days 2, 4, and 6) during the treatment, whereas Truvada was maintained in the cell culture throughout the whole experiment. Additionally, cells were trypsinized on day 4 of treatment, and 3 × 105 MEFs were reseeded. At day 8, total RNA was isolated using the Qiagen RNeasy mini kit (Qiagen).
Determination of RTI concentration in the sera of mice by liquid chromatography–tandem mass spectrometry
Fifty microliters of plasma was mixed with 100 μl of acetonitrile and centrifuged for 10 min at 16,000 × g at room temperature. The clear supernatant was transferred into a 1-ml conical autosampler tube and 10 μl was injected into the Quattro micro liquid chromatography–tandem mass spectrometry (LC-MS/MS) system (ABSciex). NVP, FTC, and TDF were separated by reversed phase chromatography using a binary gradient at a flow rate of 0.5 ml/min with solvent A (3/97/0.05, v/v/v) and solvent B (95/5/0.05, v/v/v) of a mixture of acetonitrile, 2 mM ammonium acetate, and formic acid. NVP, FTC, and TDF were determined using a Synergi 4 μm Hydro-RP 80 Å, 150 × 2.0 mm column with the mobile phase gradient: 0–1.5 min, 100% (v/v) solvent A; 1.5–3.2 min, a linear gradient was programmed to 0% (v/v) solvent A; 1.4 min, 0% (v/v) solvent A; 4.6–4.8 min, a steep linear gradient to 100% (v/v) solvent A; 4.8–7.5 min, 100% (v/v) solvent A. The temperature of column was constantly at 40°C. Determination of the analytes was performed using the multiple reaction monitoring mode with nitrogen as collision gas. For positive ionization, a capillary voltage of 4000 V (positive ion mode) was used. For quantification, 8-point external calibration curves in blank plasma at the range from 7.8 to 1000 ng/l for all three substances were freshly prepared and measured in parallel. Quantification was performed by the peak area method, and a weighted (1/x) regression of first order yielded the concentrations.
Inhibition of replication-competent HIV-1
In a U-bottom 96-well plate 80 μl of HIV-1 IIIB (50% tissue culture–infective dose of 104) was mixed with 20 μl of heat-inactivated mouse serum followed by addition of the C8166-45 target cells (100 μl at 5 × 104 cells/ml). Incubation and preparation of the cell lysate was performed as previously described (30). Three microliters of the lysate was used as a template for quantification of the relative provirus load by TaqMan qPCR on an MX3005 cycler (Stratagene) using 45 cycles, annealing at 55°C, and no elongation phase using primers SK68i (5′-GGARCAGCIGGAAGCACIATGG-3′) and SK69i (5′-CCCCAGACIGTGAGITICAACA-3′) in combination with the probe 5′-6Fam-TGACGCTGACGGTACAGGCCAGAC-BHQ1-3′ (31).
Differences among multiple conditions for multiple genotypes were tested by two-way ANOVA, means of two groups were compared using a two-tailed t test, and survival analysis was carried out using the log-rank test. A p value ≤0.05 was considered significant (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). Statistical analysis was performed using GraphPad Prism software.
No evidence for amelioration of autoimmunity in Trex1−/− mice by treatment with RTIs
To assess the efficacy of TDF, FTC, and NVP in inhibiting reverse transcriptases (RTs) of non-LTR and LTR retroelements, we quantified retrotransposition of plasmid-encoded retroelements into the genomic DNA of HeLa cells in vitro. The retroelement sequence contained a fluorescent reporter cassette that can be expressed only upon reverse transcription from the retrotransposed genomic copy (32). Plasmids encoding either a LINE1 (L1) GFP or an IAP GFP reporter element were transfected in the presence of different concentrations of NVP, FTC, or TDF or the combination of all three substances. TDF and FTC inhibited reverse transcription of LINE1 and IAP in a dose-dependent manner (Fig. 1A). In contrast, NVP did not block retrotransposition of the reporter-retroelements (Fig. 1A). Our observations are in line with previous studies showing allosteric binding of the non-NRTI NVP exclusively to the RT of HIV-1, but not to RTs of HIV-2, murine leukemia viruses (relatives of IAPs) and L1 retroelements, respectively (33–35), whereas several NRTIs, including TDF and FTC, were shown to be active against RTs of L1 retroelements (34, 35).
To test whether RT inhibition reduces the chronic IFN response and autoimmunity of Trex1-deficient mice (14), we first assessed whether relevant RTI concentrations and antiretroviral activity can be achieved in sera of mice treated with the three inhibitors. Pure TDF, FTC, or NVP or the respective pharmaceutical products Viread, Emtriva, and Viramune were added to the drinking water at concentrations of 0.1 mg/ml (157 μM), 0.15 mg/ml (606 μM), and 0.1 mg/ml (375 μM), respectively. Additional mice received 0.4 mg/ml of the combination drug Truvada, equivalent to 189 μM TDF and 324 μM FTC. After 3 wk of treatment, the concentration of each compound was quantified by LC-MS/MS. TDF and FTC reached plasma levels well above their respective concentrations required for the inhibition of L1 and IAP RTs in cell culture assays (Fig. 1B). We next asked whether antiretroviral activity could be detected in the serum of RTI-treated mice and quantified inhibition of HIV-1 replication in a cell culture assay. Addition of sera from RTI-treated mice significantly decreased proviral load compared with cells that were infected in the presence of sera from untreated mice (Fig. 1C). Additionally, activity of serum from RTI-treated mice against LINE1 reverse transcription was tested using the retrotransposition reporter assay described above. Sera from RTI-treated animals, but not sera from untreated animals, reduced L1 retrotransposition (Fig. 1D). We did not detect a synergistic effect of NVP in combination with TDF and FTC when compared with the treatment with Truvada (TDF and FTC) only. Collectively, TDF and FTC, but not NVP, blocked reverse transcription of endogenous retroelements. Thus, treatment of mice with RTIs via the drinking water resulted in plasma levels of TDF and FTC that almost completely blocked L1 and IAP retrotransposition in cell culture assays. Moreover, sera of RTI-treated mice exhibited significant antiretroviral activity against HIV and L1.
Next, we tested the effect of RTI treatment on autoimmunity in Trex1−/− mice and added 0.4 mg/ml Truvada (equivalent to 189 μM TDF and 324 μM FTC) into the drinking water. This treatment did not prolong survival of Trex1−/− mice compared with untreated Trex1−/− controls (Fig. 2A). Furthermore, inflammation of the heart as well as weight loss (Fig. 2A) were unaltered between the groups, suggesting that Truvada had no impact on the development of the disease. In a second experiment, we included also NVP and added all three drugs to the drinking water of Trex1−/− mice and wild-type control littermates at final concentrations of 375 μM for NVP, 157 μM for TDF, and 606 μM for FTC. These concentrations were shown to result in significant antiretroviral activity in the serum (Fig. 1C, 1D). Except for one animal that died, all control mice that received the drugs survived for 30 wk without any signs of disease, indicating that the drug concentrations administered were not toxic. In contrast, both untreated Trex1−/− mice as well as treated Trex1−/− animals developed disease, and >80% of each group died before the age of 20 wk. Thus, RTI treatment did not prolong the survival of Trex1−/− mice (Fig. 2B).
Spontaneous type I IFN response caused by lack of Trex1 is not affected by RTIs
To analyze effects of Truvada on the activity of the type I IFN system, we quantified expression of ISGs. Cell surface protein Sca-1 is upregulated by type I IFN (28, 36). As expected, flow cytometry showed increased Sca-1 expression on blood cells from Trex1−/− mice, whereas Sca-1 expression was not reduced in Trex1−/− mice that received Truvada (Fig. 3A). Next, we isolated total RNA from PBMCs of these mice and quantified ISG transcripts. Untreated Trex1−/− mice showed upregulation of all ISGs analyzed compared with control littermates (Fig. 3B). Upregulation of ISG transcripts was not affected by treatment with Truvada (Fig. 2B), suggesting that the drug was unable to block the spontaneous IFN production in Trex1−/− mice. Next, we tested whether Trex1-deficient primary MEFs display decreased ISG responses when cultured in the presence of Truvada. In line with previous studies (16, 28, 37), a transcriptional ISG signature was readily detectable in primary embryonic day 14.5 Trex1−/− MEFs (Fig. 3C). Treatment with 3.3 μg/ml Truvada (1.6 μM TDF, 2.7 μM FTC), a concentration that inhibited L1 and IAP retrotransposition (Fig. 1A), failed to suppress the increased transcription of ISGs in Trex1−/− MEFs (Fig. 3D). Collectively, we show that inhibition of reverse transcription with Truvada had no effect on the spontaneous type I IFN response caused by deficiency of Trex1.
Frequency of retrotransposition is not increased in the absence of Trex1
Trex1-mediated degradation of retroelement cDNAs could represent a mechanism controlling retrotransposition frequency. To investigate this further, we quantified retrotransposition in the absence of TREX1 using CRISPR/Cas9 technology to inactivate TREX1 in HeLa cells (Fig. 4A, Supplemental Fig. 1). We transfected the L1 retrotransposition reporter plasmid into TREX1-deficient and control cells and quantified retrotransposition events by flow cytometry. In all TREX1 KO clones, retrotransposition was not enhanced but, on the contrary, significantly decreased (Fig. 4A). This may be explained by the chronic antiviral response that is induced in the absence of TREX1. In support of this notion, in additional TREX1 KO clones from the same gene-targeting experiment, L1 retrotransposition was also significantly reduced and all clones displayed a spontaneous transcriptional ISG signature, indicating that the pathomechanism triggered by TREX1 deficiency is fully active in HeLa cells (Supplemental Fig. 1). In line with our findings, pretreatment of HeLa cells with IFN-β was shown to reduce L1 retrotransposition (38).
To address the effect of Trex1 on L1 retrotransposition in vivo, we made use of the inducible retrotransposition reporter strain tet-ORFeus (39). This mouse line carries two transgenes, the tet-ORFeus construct, which encodes a codon-optimized L1 element under control of a tetracycline-inducible promoter, and a reversible tetracycline transactivator (rtTA). Upon administration of doxycycline, tet-ORFeus is transcribed, reverse transcribed, and the cDNA is integrated as an additional genomic copy. Splicing of the tet-ORFeus RNA during this process removes an artificial intron, allowing for discrimination of the original transgene from newly retrotransposed copies by PCR (39). We crossed rtTA-tetORFeus mice with Trex1−/− mice and administered doxycycline via the drinking water starting from conception. Only breedings of mice that carried both transgenes and received doxycycline produced offspring with the characteristic “white spot” phenotype (39) and led to a PCR product originating from the intronless ORFeus (Supplemental Fig. 2), confirming that retrotransposition had occurred and that the system was not leaky. Because the activity of endogenous retroelements differs substantially between tissues (26), we isolated genomic DNA from various organs of doxycycline-treated rtTA-tetORFeus-Trex1−/− and rtTA-tetORFeus-Trex1+/+ littermates and quantified retrotransposition events by TaqMan qPCR (Supplemental Fig. 2). Retrotransposition events were readily detectable in all tissues; however, their mean frequency was not affected by the absence of Trex1 (Fig. 4B), suggesting that Trex1 does not restrict retrotransposition of the tetORFeus L1 element. Collectively, we show that retrotransposition of L1 reporter elements was not increased upon loss of Trex1 in vitro and in vivo.
We treated Trex1−/− mice with higher RTI concentrations compared with the study of Beck-Engeser et al. (27) (Supplemental Table I), who reported rescue of pathology by this treatment. Although we found significant antiretroviral activity in the sera of RTI-treated animals, we did not see a reduction of disease activity despite presumably higher RTI serum concentrations. However, because Beck-Engeser et al. (27) did not report serum concentrations, a direct comparison is difficult. We cannot exclude that systemic concentrations reached in our animals were not sufficient to fully inhibit the broad range of genome-encoded RTs with largely unknown characteristics. Addressing the intracellular activity of RTIs against endogenous RTs is hampered by the fact that as opposed to doxycycline-enforced LINE1 reporter retrotransposition, which we used in Fig. 4B, the activity of endogenous retroelements in C57BL/6 mice is low and restricted to germ cells or early embryonic development, resulting in somatic mosaicism (40). The effect of RTI treatment on the frequency of somatic and germ line de novo LINE1 insertions could be assessed by mouse retrotransposon capture sequencing (41).
Because the higher concentrations of TDF and FTC that we used did not cause weight loss or reduced survival in control mice (Fig. 2A), these concentrations are also unlikely to have caused adverse effects in Trex1−/− mice. Furthermore, using NVP in combination with TDF and FTC did not prolong survival of Trex1−/− mice (Fig. 2B). In accordance with this finding, we observed no inhibition of L1 or IAP retrotransposition in vitro (Fig. 1). Hence, our data suggest that NVP was dispensable in this setting. In an experimental setting availability of the drugs might be increased by changing the route of administration (42).
Different mean survival times were reported for Trex1−/− mice (14–16, 19, 27, 28), which may reflect modulation of disease expression by varying microbial flora. Given that microbiota can significantly influence the activation of endogenous retroelements (43), differences in housing conditions of Trex1−/− mice might pose a further limitation to direct comparisons of our study with the report by Beck-Engeser et al. (27). Recently, anti-inflammatory properties independent of RT inhibition were observed for NRTIs such as azidothymidine (AZT) and must be considered in the investigation of therapeutic effects of RTIs in inflammatory conditions (44). However, treatment with a high dose of AZT (1 mg/ml = 3742 μM) did not prolong survival of Trex1−/− mice (15).
Two recent studies demonstrated that increased L1 activity can result in spontaneous production of IFN-β and ISG responses (38, 45), which can be blocked by treatment with TDF (45). Also, NF-κB activation in naive B cells induces transcription of endogenous retroelements, which activate the cGAS–STING pathway and intrinsically prime the B cells (46). Intriguingly, we recently found the latter mechanism to be active irrespective of the presence of Trex1 (28), arguing against a role of Trex1 in the cell-intrinsic control of retrotransposition in B cells. In the present study, we show that simultaneous blockade of reverse transcription with two NRTIs does not alter the spontaneous activation of the type I IFN system in the absence of Trex1, suggesting that the IFN-inducing pathogenic nucleic acid in Trex1−/− mice is not generated by genome-encoded RTs, which are sensitive to TDF and FTC. In contrast to L1 elements, unleashed replication of endogenous retroviruses caused by deficiency of the endosomal TLRs did not induce a type I IFN response in mice (47). This suggests that replication intermediates of LTR retroelements do not engage type I IFN-inducing cytosolic nucleic acid sensors and may therefore not represent major substrates of Trex1.
Intriguingly, pathology in Trex1−/− mice is almost completely rescued by haploinsufficiency for cGAS or STING, suggesting that the IFN-inducing nucleic acid might accumulate at levels only slightly above the threshold for the activation of the DNA sensor (17, 18). Hence, even incomplete blocking of reverse transcription might have reduced the concentration of the ligand and in turn led to a reduction of the IFN response. However, we did not observe any effect of the RTI treatment on ISG transcript levels, in any of our experiments.
Trex1 was reported to suppress the type I IFN response against HIV by degrading aberrant cDNA replication byproducts (25). Additionally, overexpression of wild-type Trex1 was shown to block retrotransposition of both L1 and IAP reporter elements, whereas overexpression of mutant, dominant-negative Trex1 promoted IAP but not L1 retrotransposition (15), Furthermore, short hairpin RNA–mediated knockdown of TREX1 in HeLa cells was found to promote L1 retrotransposition in a plasmid-based assay (48), suggesting that under these conditions Trex1 metabolizes replication intermediates of retroelements. CRISPR/Cas9-mediated gene targeting to delete a substantial part of the single coding exon of TREX1 in HeLa cells resulted in the complete absence of the protein (Fig. 4, Supplemental Fig. 1). In contrast to Li et al. (48), in these cells we consistently found reduced numbers of L1 retrotransposition events. In support of our findings, retrotransposition of another L1 reporter transgene was not increased in a wide range of cell types in Trex1−/− mice (Fig. 4B). Collectively, our data suggest that TREX1 can in principle block retrotransposition of L1 elements but it does not contribute to their control under physiological conditions.
It is noteworthy that L1 elements reverse transcribe their genome directly at the site of integration by a process called target-primed reverse transcription, thereby largely avoiding formation of free cDNA during their life cycle. A recent study demonstrated that Trex1 also degrades ssRNA, an activity that could possibly interfere with L1 replication during the nuclear import of the genomic RNA (49). However, no evidence for destabilization of full-length L1 reporter RNA after overexpression of TREX1 in HEK293T cells (48) or accumulation of endogenous retroelement transcripts was found in dendritic cells and macrophages isolated from Trex1−/− mice (19) and TREX1−/− patient fibroblasts (50).
Type I IFNs impair the activity of L1 elements (38). In Trex1−/− mice rescued by additional genetic abrogation of the IFN response, one could expect a situation in which restriction of endogenous retroelements by Trex1 nuclease activity and antiviral IFN effects is eliminated, potentially resulting in unleashed retrotransposition and insertional mutagenesis (43, 47). However, enhanced frequencies of spontaneous neoplastic disease have never been reported so far in any of the rescued Trex1−/− cohorts (including our own Trex1/Sting and Trex1/Ifnar1 double-deficient mice; A. Roers and R. Behrendt, unpublished observations).
Taken together, our findings do not support the concept of endogenous retroelements as key factors in the pathogenesis of autoimmunity ensuing from defects in TREX1. This is highly relevant to translational efforts regarding treatment of patients with AGS. In an ongoing phase II clinical trial (NCT02363452), therapeutic effects of the NRTIs zidovudine (AZT), lamivudine (3TC), abacavir (ABC) are currently being tested in patients with mutations in TREX1, SAMHD1, or RNASEH2A-C. Indeed, RT inhibition could be an effective treatment of pathologies caused by defects of other AGS-associated enzymes, including RNASEH2 or SAMHD1, which, similar to loss of Trex1, result in a chronic activation of cGAS (51–53). Thus, increased L1 retrotransposition activity was observed in the absence of SAMHD1 (54, 55), and L1-derived transcripts were found more abundant in RNaseH2-deficienct mice compared with controls (52). Alternatively, retrotransposition activity of endogenous L1 elements was not increased in the CNS of an AGS patient with mutations in SAMHD1 (56), and patient fibroblasts lacking either SAMHD1 or functional RNaseH2 did not feature increased retroelement transcript levels (50). Transcription of L1 might not be affected by the absence of SAMHD1 or RNASEH2, but increased genome damage that is associated with defects of these enzymes might facilitate L1 integration (45) and the extranuclear abundance of their sequences (50). Therefore, further experiments are required to clarify in which situations retroelement-derived cDNA can cause type I IFN–dependent disease.
In the case of patients with mutations in TREX1, promising treatment options might include specific inhibition of the cytosolic DNA sensing pathway or blockade of type I IFN receptor signaling by JAK inhibitors. The latter strategy was reported to suppress the spontaneous type I IFN response in patients carrying gain-of-function mutations in STING (57–60).
We thank Jeff D. Boeke for providing the tetORFeus system and Tomas Lindahl for Trex1−/− mice. We thank Mathias Wabl for sharing unpublished data and protocols for RTI treatment of mice. We are grateful to Boehringer Ingelheim and Gilead Sciences for providing RTIs. We thank Christina Hiller, Barbara Utess, and Tobias Haering for excellent technical assistance.
This work was supported by Deutsche Forschungsgemeinschaft Grants LE 1074/4-1 (to M.A.L.-K.), LI 621/7-1 (KFO249), LI 621/11-1 (SPP1923), and LI 621/10-1 (to D.L.), and R0 2133/6-2 (KFO249) (to A.R.). R.B. was supported by a MeDDrive Start Grant of the Medical Faculty Carl Gustav Carus Technical University of Dresden.
The online version of this article contains supplemental material.
Abbreviations used in this article:
cyclic GMP–AMP synthase
liquid chromatography–tandem mass spectrometry
LINE1, long interspersed nuclear elements 1
mouse embryonic fibroblast
reverse transcriptase inhibitor
reversible tetracycline transactivator
SAM and HD domain containing 1
tenofovir disproxil fumarate
three prime repair exonuclease 1.
The authors have no financial conflicts of interest.