Activation of the cGAS/STING Axis in Genome-Damaged Hematopoietic Cells Does Not Impact Blood Cell Formation or Leukemogenesis

Genome integrity is continuously challenged by DNA damage resulting from spontaneous hydrolysis of phosphodiester bonds, ionizing radiation, mutagenic chemicals or reactive oxygen species (1). Cells detect DNA damage and activate signaling cascades to halt the cell cycle and induce repair pathways (2). The transcription factor p53 orchestrates a multitude of cellular responses to diverse forms of stress, including genome damage, thereby acting as a major tumor suppressor (3). Depending on quality of DNA damage, signal intensity, and cell type, p53 mediates cell-cycle arrest, triggers DNA repair, activates different forms of programmed cell death, or drives cells into senescence (4–6).

In contrast to short-lived or postmitotic somatic cells, unrepaired DNA damage in long-lived stem and progenitor cells with high proliferative potential results in acquisition and propagation of potentially deleterious mutations (7).

Genome damage emerged as a major cause of cell-intrinsic type I IFN production (8–12). An important link between genome instability and innate antiviral immunity is the pattern recognition receptor cyclic GMP-AMP synthase (cGAS), which detects nucleosome-free double-stranded (ds) DNA (13, 14). Upon binding to dsDNA, mammalian cGAS catalyzes formation of the unique second messenger 2′3′-cyclic guanosine monophosphate–adenosine monophosphate (cGAMP) that in turn activates the sensor stimulator of interferon genes 1 (STING) to induce type I IFN and proinflammatory cytokine production (15). An important source of self-DNA that is sensed by cGAS in genome damaged cells are chromatin fragments (10, 16–19). In addition to a strong proinflammatory cytokine response, cGAS/STING signaling can also trigger senescence and cell death (11, 18, 20–22), potentially eliminating genome-damaged cells and thereby preventing cancer (11). A pivotal role of STING in controlling growth of implanted tumors (23) informed studies demonstrating that targeted activation of the STING pathway boosts antitumor immunity (24–26). While the effects of targeted cGAS/STING activation on established tumors have been extensively studied, the role of this pathway in controlling spontaneous malignant transformation of genome-damaged cells remains poorly understood. In addition to that, reports on STING-independent functions of cGAS during DNA repair resulted in controversial observations, with some groups reporting potentially oncogenic activities of cGAS (27, 28), while others observed tumor-suppressive effects of the DNA sensor (29, 30).

Ribonucleotide excision repair (RER) is a nonredundant DNA repair pathway in mammals (31, 32). RER depends on the activity of the trimeric RNaseH2 complex, which initiates removal of ribonucleotides misincorporated during genome replication (33). Failure to repair these lesions results in genome instability, micronucleus formation, chronic activation of cGAS/STING, type I IFN production and in embryonic lethality (10, 12, 31, 34, 35). Interestingly, the human RNASEH2B gene [formerly named deleted in leukemia (DLEU) 8] resides on chromosome 13q14, a region that is often deleted in hematopoietic tumors, most frequently in chronic lymphocytic leukemia (CLL; ref. 36). Indeed, a previous study reported monoallelic and biallelic loss of the RNASHE2B gene in 43% and 14% of CLL tumors, respectively (37).

Here, we report the development of a mouse model for hematopoietic RER deficiency to investigate the interactions between RER, DNA damage, and cGAS/STING signaling in hematopoiesis and in the development of hematopoietic malignancies. We find that genome damage ensuing from loss of RER severely compromised hematopoiesis and resulted in malignant transformation. Additional loss of p53 largely rescued blood cell production at the cost of further accelerated leukemogenesis. Inactivation of the cGAS/STING axis or of type I IFN signaling had no detectable impact on hematopoiesis and leukemia development in RER-deficient mice. In addition, exclusive loss of cGAS/STING did neither alter steady state nor stress hematopoiesis. Our findings argue against an important role of the cGAS/STING axis in response to DNA damage and prevention of malignant transformation in the hematopoietic system.

The following mouse strains were used in the study: Rnaseh2b^flox (32), Vav-Cre (38), Ifnar1^KO (39), Cgas^KO (40), Trp53^KO (41), Sting1^GT (42), and B6.CD45.1 (RRID:IMSR_JAX:002014). Male and female mice were used for experiments and housed in individually ventilated cages under specific pathogen-free environment at the Experimental Center of the Medical Faculty, TU Dresden. 5-fluorouracil (5-FU; 150 μg/g body weight, Applichem) was administered via intravenous injection, whole-body γ radiation was applied using a MaxiShot source (Yxlon). All experiments were in accordance with institutional guidelines on animal welfare and were approved by the relevant authority (Landesdirektion Sachsen, permit numbers TVV46/2019, TVV54/2017, TVV88/2017).

Long bones (tibiae, femora, pelvis) were crushed using mortar and pestle in PBS/2%FCS/2 mmol/L ethylenediaminetetraacetic acid (EDTA) and filtered through a 100 μm sieve. Erythrocytes were lysed by incubating in hypotonic NH₄Cl buffer for 30 seconds at room temperature. After washing, bone marrow (BM) cells were filtered through a 30 μm mesh.

Either thymus or spleen were gently rubbed through 40 μm cell strainer using PBS/2%FCS/2 mmol/L EDTA. After centrifugation, cells were subjected to erythrocyte lysis for 30 seconds, washed and filtered through a 30 μm mesh.

Peripheral blood (PB) was drawn retroorbitally into EDTA-coated tubes (Sarstedt). PB was diluted 1:5 in isotonic saline and PB counts were determined on a Sysmex XT2000i Vet hemacytometer. To obtain leukocyte suspension, PB was subjected to erythrocyte lysis two times for 5 minutes each and washed.

Absolute counts of all cell suspensions were determined employing a Miltenyi Biotec MACSquant analyzer.

Cell suspensions were incubated with pretitrated mAbs (see Supplementary Table S1) for 30–40 minutes. After washing, cell suspensions were acquired on Miltenyi Biotec MACS Quant 10 (RRID:SCR_020268) or BD LSR II analyzers (RRID:SCR_002159) or BD FACS ARIA II (RRID:SCR_018091) or ARIA III cell sorters. Flow cytometry data were analyzed using FlowJo (RRID:SCR_008520) software version 9/10. Gates were set with help of fluorescence-minus-one controls (see Supplementary Fig. S1 for gating strategies).

To isolate BM megakaryocytes (Mk), BM was flushed from one femur with 4 mL PBS/2% FCS/2 mmol/L EDTA, filtered through a 100 μm mesh, and 10⁷ cells were stained directly without performing erythrocyte lysis with anti-CD41-APC mAb for 30 minutes on ice. After washing, a secondary staining in 10 μmol/L Hoechst 33342 was performed for 1 hour at 37°C. Finally, cells suspensions were washed and acquired on a BD FACS Symphony A3 flow cytometer.

BM cell suspensions were depleted of lineage-positive cells using the mouse lineage cell depletion kit (Miltenyi Biotec) according to the manufacturer's instructions. Lineage-negative BM cells were stained and immunophenotypic hematopoietic stem cells (HSC) were sorted on BD FACS ARIA II/III cell sorters. Reanalysis revealed a purity of sorted cells >95%. HSCs were mixed with B6.CD45.1 whole BM cells and retroorbitally injected into lethally-irradiated (9 Gy) B6.CD45.1/CD45.2 recipients, which were generated by crossing C57BL/6JRj wt (Janvier, RRID:MGI:2670020) and B6.CD45.1 mice. For secondary transplantation, 5 × 10⁶ whole bone marrow cells (WBMC) were isolated from primary recipients and intravenously injected into lethally irradiated secondary recipients.

Single HSCs were deposited into 96-well round bottom plates (TPP) and cultivated for 10 days in IMDM (Gibco) supplemented with 20% FCS (Biochrom), 1% penicillin/streptomycin, 20 ng/mL rmSCF (Peprotech), 20 ng/mL rmTPO (Peprotech), 20 ng/mL rmIL3 (Peprotech), and 5 U/mL rhEPO (NeoRecormon, Roche). Cells were counted by light microscopy (PrimoVert, Zeiss) for the first 96 hours every 12 or 24 hours. After 10 days, the total cell number of each colony was determined on a Miltenyi Biotec MACSquant flow cytometer.

Thymus and spleen tissue were fixed overnight in 4% formalin fixative, followed by paraffin embedding, sections were cut using a Leica microtome. Staining was performed according to hematoxylin and eosin and May-Grünwald-Giemsa protocols. Sections were analyzed on a Keyence BZ-x710 microscope (RRID:SCR_017202).

Micronucleated erythrocytes were identified following the protocol of Balmus and colleagues (43). Briefly, 30 μL PB was mixed with 120 μL PBS and fixed with precooled (−80°C) methanol overnight and stored for <1 year at −80%. Fixed erythrocytes were stained with antibodies against CD71 and Ter119 and digested with 10 μL RNase A (10 mg/mL, Invitrogen) for 1 hour at 4°C. Samples were washed and after adding 1 μg/mL propidium iodide, 1–2.5 × 10⁶ events were acquired on BD LSR II or BD FACS Aria III flow cytometers. Representative gating of micronucleated erythrocytes is shown in Supplementary Fig. S1E.

A total of 10⁶ WBMCs were resuspended in 100 μL 2× Laemmli Buffer and boiled for 5 minutes. Proteins of approximately 2 × 10⁵ cells were separated on a 12% PAGE and subsequently blotted onto a membrane for 10 minutes using a Trans-Blot Turbo System (Bio-Rad, RRID:SCR_023156). Membranes were blocked in PBS (0.05% Tween-20, 3% BSA) and incubated with primary antibodies (Supplementary Table S1) over night at 4°C. If needed, membranes were washed three times and secondary antibodies were added for 2 hours at room temperature before detection using Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific) on an Odyssey FC Imager (LI-COR, RRID:SCR_023227).

Freshly isolated BM cells were stained with Zombie NIR Fixable Viability Dye (BioLegend) for 15 minutes. Cells were washed and blocked with anti-CD16/32 (Fc block) for 15 minutes. Cell surface staining was performed for 30 minutes by adding antibodies against CD117, Sca-1, and hematopoietic lineage. Cells were washed and fixed for 30 minutes with Fixation/Permeabilization buffer (BioLegend), followed by incubation with rabbit anti-phospho-histone H2A.X (Ser139, clone 20E3) for 30 minutes, washed twice before staining with Alexa Fluor 488 goat anti-rabbit IgG (H+L) (Thermo Fisher Scientific) for 30 minutes. After washing, cells were analyzed using a BD FACS Aria III.

Lineage-negative BM cells were prepared from RH2^hKO mice and RH2^hWT littermate controls (n = 4/genotype) and 30,000 to 100,000 GMPs (LE⁻K CD16/32^hiCD34⁺) were sorted. Total RNA was prepared using the RNeasy Micro Kit (Qiagen). RNA quality was analyzed using an Agilent 2100 Bioanalyzer (RRID:SCR_018043) and samples were frozen at −80°C. mRNA was isolated from 37 to 50 ng total RNA by poly-dT enrichment using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB) according to the manufacturer's instructions. Samples were then directly subjected to the workflow for strand-specific RNA sequencing (RNA-seq) library preparation (Ultra II Directional RNA Library Prep, NEB). For ligation, custom adaptors were used (Adaptor-Oligo 1: 5′-ACA CTC TTT CCC TAC ACG ACG CTC TTC CGA TCT-3′, Adaptor-Oligo 2: 5′-P-GAT CGG AAG AGC ACA CGT CTG AAC TCC AGT CAC-3′). After ligation, adapters were depleted by an XP bead purification (Beckman Coulter). Unique dual indexing was done during the following PCR enrichment (15 cycles) using custom amplification primers carrying the same sequence for i7 and i5 index (Primer 1: AAT GAT ACG GCG ACC ACC GAG ATC TAC AC NNNNNNNN ACA TCT TTC CCT ACA CGA CGC TCT TCC GAT CT, Primer 2: CAA GCA GAA GAC GGC ATA CGA GAT NNNNNNNN GTG ACT GGA GTT CAG ACG TGT GCT CTT CCG ATC T). After two more XP bead purifications (1:0.9), libraries were quantified using the Fragment Analyzer (Agilent). For Illumina flowcell production, samples were equimolarly pooled and sequenced on Illumina NovaSeq 6000 Sequencing System (RRID:SCR_016387) in 2×50 bp paired-end mode, resulting in at least 35 million fragments per sample. Sequencing data were uploaded to Gene Expression Omnibus (GEO, RRID:SCR_005012; accession number: GSE207761).

Total RNA was isolated from WBMCs and reverse transcribed into cDNA using oligo dT primer (see also quantitative PCR). An Rnaseh2b specific amplicon library was prepared using a modified protocol from Lange and colleagues (44). Briefly, first-round PCR [95°C/2 minutes − 30×(95°C/10 seconds − 58°C/30 seconds − 72°C/20 seconds) − 72°C/5 minutes] was performed using primers H2B_FLOX_NGS_F1 and H2B_FLOX_NGS_R1 (Supplementary Table S2), yielding a PCR fragment spanning the loxP-flanked exon 5 and binding sites for second-round primers. A total of 0.1 μL of the first-round PCR product were used as a template for the second-round amplification [95°C/2 minutes − 10×(95°C/10 seconds − 58°C/30 seconds − 72°C/20 seconds) − 72°C/5 minute] to add standard P5 and P7 Illumina adapters and sample-specific barcodes using primers MiSeq_IndexF1_01 to MiSeq_IndexF1_08 in combination with R_1 from Lange and colleagues (44). Amplicons were purified from an agarose gel and mixed in equimolar amounts. The resulting library was again purified from an agarose gel and sequenced on an Illumina MiSeq System (RRID:SCR_016379) using V3 chemistry (250 bp single end, forward). At least 4,000 reads were generated per sample and assembled to the reference sequence rnaseh2b-201-ensmust00000022499 using the standard assembler in GeneiousPrime 2022.1.1 (Bio Matters, RRID:SCR_010519) with the option “Re-trim sequences” enabled and “find short insertions and large deletions” set to 300 bp. Absolute number of reads with (“WT”) and without exon 5 (“del E5”) were counted per sample and displayed using GraphPad Prism, V9 (RRID:SCR_002798).

Fragments were aligned with GSNAP (45) in paired-end mode to the mouse mm10 reference genome and Ensembl gene annotation version 98 was used to support detection of splice sites. Afterward, fragments were assigned to genes (Ensembl v98) with featureCounts (46) and followed by normalization, exploratory, and differential expression analysis using DESeq2 (47). Unless otherwise stated, differentially expressed gene lists were sorted according to ascending P_adj. All transcripts with P_adj < 0.05 were subjected to gene set enrichment analysis (GSEA) and enriched gene set with a FDR ≤ 0.25 were considered significant (48). If not state otherwise, hallmark gene set collection (MSigDB) was used to identify enriched gene sets. To generate heat maps, all transcripts with P_adj < 0.05 were extracted from lists containing normalized read counts of all genotypes in the respective experiment and the resulting sublist was displayed using Morpheus (https://software.broadinstitute.org/morpheus).

Total RNA was isolated using NucleoSpin RNA Kit (Macherey-Nagel) and reverse transcribed into cDNA using PrimeScript RT Reagent Kit (Takara) or RevertAid Kit (Thermo Fisher Scientific) following the manufacturer's instructions. Quantitative RT-PCR using Luna Universal qPCR Master Mix (New England BioLabs) was performed with the following cycling conditions on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad, RRID:SCR_018057) or on a QuantStudio 5 (Thermo Fisher Scientific, RRID:SCR_020240): 10 minutes 95°C, 40 cycles of 95°C for 20 seconds, 60°C for 30 seconds. Employed qRT-PCR primers are listed in Supplementary Table S2. Transcript levels were normalized to the housekeeping gene Tbp1. All samples were run at least in technical duplicates.

Statistical analysis was performed using GraphPad Prism, V9 (RRID:SCR_002798) and applied tests are given in each figure legend (ns, not significant; *, P < 0.05–0.01; **, P < 0.01–0.001; ***, P < 0.001).

RNA-seq data generated in this study are publicly available in GEO (RRID:SCR_005012) with the accession number GSE207761. Other raw data are available from the lead contact upon reasonable request.

To investigate defense mechanisms against genome damage in the hematopoietic system, we generated mice with loss of RER throughout hematopoiesis by conditional inactivation of the Rnaseh2b gene. As expected, Rnaseh2b^fl/fl/Vav-Cre⁺ mice (RH2^hKO; refs. 32, 38) featured high numbers of micronucleated erythrocytes and histone 2AX-phosphorylated (γH2AX) BM cells (Fig. 1A–D; Supplementary Fig. S1), reflecting genome instability ensuing from unrepaired ribonucleotides contained in the genomic DNA (43, 49–51). Except for a minor growth retardation (Supplementary Fig. S2A), the mutants were macroscopically indistinguishable from Cre-negative control littermates (RH2^hWT) during the first 3 months of life. However, median survival was significantly reduced to about 13 months (Fig. 1E). Analysis of moribund mice at different age revealed a massive enlargement of the thymus (Fig. 1F and G). Histology showed a loss of normal thymic medulla and cortex architecture (Supplementary Fig. S2B) due to a massive expansion of CD4/8 double-positive (DP) thymocytes (Fig. 1H). In some of these animals, DP thymocytes disseminated to PB (Fig. 1I) and spleen (Supplementary Fig. S2B), suggesting T-cell acute lymphoblastic leukemia (T-ALL)-like disease as the most likely cause of death. Some moribund mice did not show signs of leukemia as judged by macroscopic inspection of organs or immunophenotyping of PB, BM, thymus, and spleen. These animals presented with very low hematocrit and leukocytes counts, suggesting hematopoietic failure or infection as the reason for lethality.

These data show that conditional inactivation of RER represents a suitable model for studying hematopoiesis under conditions of chronic DNA damage.

At a young age, that is, already before onset of malignancy, RH2^hKO animals featured massive alterations of the hematopoietic system. These included macrocytic anemia, almost complete absence of B cells, reduced T cell and granulocyte numbers, but elevated platelet counts (Fig. 2A). Analysis of RH2^hKO BM revealed that total cellularity was reduced to about one-third of controls (Fig. 2B). To reliably identify HSCs (see Supplementary Fig. S1A for gating), we substituted the marker Sca-1, which is upregulated by type I IFN signaling (52) with EPCR (CD201; ref. 53). Like HSCs, also absolute numbers of progenitor populations were massively reduced in RH2^hKO BM. All populations of B-lymphocyte development from pre-pro B cells to mature B cells (see Supplementary Fig. S1B for gating) were reduced to below 10% of normal numbers (Fig. 2C). Likewise, the thymi of RH2^hKO mice were severely reduced in weight (Supplementary Fig. S2C and S2D) and cellularity was only about 5% of controls (Fig. 2D). Thymocytes of the first stages of T-lymphocyte development [CD4/8 double-negative (DN) 1–3; see Supplementary Fig. S1C for gating] were reduced to about half of normal numbers and cells of later stages were almost completely absent (Fig. 2D). To address stem cell fitness, we cultivated single HSCs and found a reduced proliferative capacity of RH2^hKO HSCs (Fig. 2E and F). Moreover, we determined the functional potential of HSCs in vivo by competitive transplantation into lethally irradiated recipients and found a severe repopulation defect of RNaseH2-deficient HSCs (Fig. 2G–I; Supplementary Fig. S2E and S2F). In these recipient mice, we observed a significantly higher contribution of RNase H2-deficient donor cells to HSCs than to progenitors and mature granulocytes, which argues for a competitive disadvantage of RNase H2 knockout (KO) progeny during chimeric hematopoiesis (Supplementary Fig. S2G).

About 10% of the Rnaseh2b^fl/fl/Vav-Cre⁺ mice distinctly differed from the majority by ameliorated cytopenia (Supplementary Fig. S3A–S3D) and showed almost normal numbers of micronucleated erythrocytes (Supplementary Fig. S3E), but retained high Sca-1 expression (Supplementary Fig. S3F). Genotyping of PB leukocytes from these “rescued” RH2^hKO mice revealed the presence of the Rnaseh2b^flox allele (Supplementary Fig. S3G), which most likely reflected overgrowth of clones that had escaped Cre-mediated inactivation of Rnaseh2b. Targeted sequencing of Rnasehb2 transcripts in total RNA isolated from WBMCs revealed wild-type (WT) transcript levels in these RH2^hKO escapee animals comparable with RH2^hWT controls (Supplementary Fig. S3H).

Taken together, chronic genome damage in the hematopoietic system of RH2^hKO mice reduced numbers, proliferation, and repopulation capacity of HSCs. Impaired hematopoiesis resulted in anemia, lymphopenia, and reduced granulocyte numbers.

Inflammatory conditions confound the fidelity of hematopoietic stem and progenitor cell (HSPC) marker expression (52), but granulocyte-monocyte progenitors (GMP) are unambiguously identified by upregulation of Fcγ receptors (CD16/32). Moreover, genome-damage related to loss of RER primarily manifests in replicating cells (51), and GMPs are rapidly dividing cells. To elucidate how chronic genome damage leads to hematopoietic malfunction in RH2^hKO mice, we sorted GMPs from BM of mutant and control animals and sequenced their transcriptomes. We found 198 genes significantly upregulated in RH2^hKO GMPs compared with controls (Fig. 3A; Supplementary Table S3). Among the most upregulated genes were the interferon-stimulated genes (ISG) Oasl1, Mx1, Ifit1 (Fig. 3B; Supplementary Table S4) as well as Cdkn1a, encoding cell-cycle inhibitor p21 and other p53-induced genes, including Trp53inp1, Bax, and Mdm2 (Fig. 3C; Supplementary Table S5). Rnaseh2b was the most downregulated gene, validating the analysis. In addition, Ear1/2, Gzmb, and Tspan9 were among 20 significantly downregulated genes. These results suggested induction of type I IFN and p53-mediated DNA damage responses in the mutant cells, which was confirmed by GSEA (Fig. 3D). We also found upregulation of several ISGs in BM and spleen of RH2^hKO animals by qRT-PCR (Supplementary Fig. S4A and SD4B). In accordance with the ISG response on the transcriptome level, the type I IFN-induced surface protein Sca-1 was significantly upregulated on HSCs and PB lymphocytes from mutant compared with control mice as determined by flow cytometry (Fig. 3E; Supplementary Fig. S4C). Compatible with activation of a cGAS/STING response in RH2-deficient cells (10), GSEA also revealed induction of NFκB-dependent proinflammatory gene expression in the mutant cells. Because loss of RH2 causes Aicardi-Goutières syndrome via induction of chronic type I IFN signaling (54, 55), we asked whether RH2^hKO mice develop inflammation. However, histologic screening of various organs did not reveal abnormal immune cell infiltration or other inflammation-associated changes (Supplementary Fig. S4D).

We next investigated the impact of the activated p53 response on the phenotype of RH2^hKO mice and crossed the animals to Trp53^KO/KO mice (41). Heterozygous loss of Trp53 significantly ameliorated the hematopoietic deficits caused by absence of RH2 activity (Fig. 4A). Erythrocyte counts were significantly increased in RH2^hKOTrp53^KO/WT mice compared with the severe anemia of RH2^hKO controls. PB B cells, virtually absent in RH2^hKO animals, reached about two-third of control numbers, while PB T cells numbers were back to normal and thrombocytosis was significantly ameliorated in RH2^hKOTrp53^KO/WT mice. To further investigate thrombopoiesis, we analyzed Mk-biased CD41^hi HSCs and Mk progenitor numbers as well as numbers and ploidy of mature Mks (Fig. 4B; Supplementary Fig. S5A–S5E). While Mk-committed HSPCs (56, 57) appeared normal or reduced in RER-deficient mice, mature Mk frequencies and numbers were significantly elevated in RH2^hKOTrp53^WT/WT animals. The thrombocytosis upon RER deficiency was a direct consequence of p53 signaling as heterozygous loss of Trp53 reverted Mk numbers to control levels. Contrary to a previous report (58), p53 signaling did not impact Mk ploidy.

Whereas BM was still hypocellular, HSCs and progenitor populations of RH2^hKOTrp53^KO/WT animals were robustly increased in numbers compared with RH2^hKO BM, with multipotent progenitors (MPP) and HPC-1s reaching control numbers (Fig. 4C). Likewise, B- and T-cell development was partially rescued (Supplementary Fig. S5F and S5G). Competitive transplantation revealed that the additional loss of one Trp53 allele partially restored the repopulation deficit of RH2-deficient HSCs upon primary (Fig. 4D–G; Supplementary Fig. S5H) and secondary transplantation (Supplementary Fig. S5I–S5L). Heterozygous loss of p53 did not further increase the frequency of micronucleated erythrocytes in RH2-deficient hematopoiesis (Fig. 4H). PB T lymphocytes (Fig. 4I), WBMCs (Supplementary Fig. S5M) and HSCs (Supplementary Fig. S5N) isolated from RH2^hKOTrp53^KO/WT mice still showed ISG upregulation compared with RH2^hWT controls.

While most parameters of RH2^hKO hematopoietic function were significantly improved by attenuated p53 expression, we observed a dramatically shortened median survival of RH2^hKOTrp53^KO/WT animals in comparison with RH2^hKO animals (Fig. 4J). In addition to development of T-ALL–like leukemia, some moribund RH2^hKOTrp53^KO/WT animals featured grossly enlarged spleens with abnormal histology (Fig. 4K and L; Supplementary Fig. S6A). Flow cytometric analysis of BM, PB, and spleen of these animals showed a massive expansion of aberrant lin⁻CD117^hiSca-1⁻CD201⁻CD34⁻CD150⁻CD48^−/lo myeloid progenitor cells (Fig. 4M; Supplementary Fig. S6B), thus a condition resembling acute myeloid leukemia (AML). Some mice developed both, T-ALL—like and AML-like disease simultaneously (Fig. 4N), which was revealed by gross enlargement of thymus and spleen by DP thymocytes (Supplementary Fig. S6C) and CD117^hiSca-1⁻ AML cells (Supplementary Fig. S6D), respectively, and both cell types coexisted in PB (Supplementary Fig. S6E).

We additionally generated few RH2^hKOTrp53^KO/KO mice (n = 4), which started to develop leukemic disease already by 8 weeks of age (Supplementary Fig. S6F). PB, BM, and thymus phenotyping of these animals revealed a more pronounced rescue of hematopoietic defects than in RH2^hKOTrp53^KO/WT mice (Supplementary Fig. S6G–S6J). Micronucleated erythrocytes were still elevated in RH2^hKOTrp53^KO/KO animals, but slightly reduced in comparison with RH2^hKOTrp53^WT/WT mice (Supplementary Fig. S6K), while the Sca-1 upregulation persisted (Supplementary Fig. S6L).

In summary, attenuation of p53 signaling ameliorated the hematopoietic defects of RH2^hKO animals, but simultaneously fueled leukemia development.

Loss of RER in hematopoietic cells led to DNA damage, micronucleus formation, and activation of innate immune pathways and, ultimately, malignant transformation. As genome damage and micronuclear DNA were shown to drive type I IFN via cGAS/STING in other models of RER deficiency (10, 35), we crossed RH2^hKO mice to Cgas^KO/KO, Sting1^GT/GT or Ifnar1^KO/KO animals. In line with our expectations, the enhanced expression of the type I IFN-induced surface protein Sca-1 in RH2^hKO mice was abrogated by each of these additional KOs (Fig. 5A).

Activation of the cGAS/STING/IFN axis has antiproliferative effects and triggers cell death or senescence and potentially contributes to the suppression of hematopoiesis, elimination of damaged cells, and prevention of leukemia in RH2^hKO mice. Unexpectedly, however, we detected no effect of defective cGAS/STING or IFNAR signaling on blood cell production in RH2^hKO mice. All of the RH2^hKO double KO mouse strains developed macrocytic anemia, leukopenia, and thrombocytosis at young age, indistinguishable from RH2 single KO mice (Fig. 5B–E; Supplementary Fig. S7A). Analysis of B lymphocyte and thymocyte development revealed persistent hematopoietic defects in RH2^hKOSting1^GT/GT mice (Supplementary Fig. S7B and S7C). The repopulation defect of transplanted RH2^hKO HSCs was not rescued by additional loss of either cGAS or IFNAR (Fig. 5F–I; Supplementary Fig. S7D). Likewise, the proliferative potential of cultivated RH2^hKOSting1^GT/GT HSCs resembled that of RH2^hKO HSCs (Fig. 5J and K). Abrogation of either cGAS, STING, or IFNAR in RH2^hKO mice also had no effect on the frequency of micronucleated erythrocytes (Fig. 5L; Supplementary Fig. S7E and S7F). Collectively, in contrast to p53 signaling, the activation of cGAS/STING and type I IFN signaling in RH2^hKO mice is not a relevant factor impairing hematopoiesis.

In accordance with these findings, neither cGAS/STING nor type I IFN signaling contributed to prevention of malignant disease as RH2^hKO mice with an additional loss of either cGAS, STING, or IFNAR exhibited a similar median survival as the RH2^hKO animals (Fig. 5M). Analysis of few moribund RH2^hKO mice with additional loss of either cGAS, STING, or IFNAR revealed similar manifestations of leukemic disease as compared with RH2^hKO animals with intact cGAS/STING or IFNAR signaling (Fig. 5N).

Of note, analysis of BM for ISG upregulation by quantitative RT-PCR revealed that most of the quantified ISGs remained upregulated in RH2^hKO mice with additional loss of cGAS or STING (Fig. 5O). While all probed ISGs showed a trend toward lower expression in RH2^hKOCgas^KO/KO mice, only Ifi44 and Oasl1 were significantly downregulated. Western blot analysis confirmed the absence of cGAS and STING proteins in the mutant mice (Supplementary Fig. S7G). In contrast, RH2^hKOIfnar1^KO/KO displayed complete loss of ISG upregulation, confirming type I IFN signaling in RH2-deficient hematopoietic cells. Our analysis suggests that genome damage in RER-defective hematopoiesis activates type I IFN signaling via the cGAS/STING axis and additional ISG-inducing signaling pathways.

In summary, the cGAS/STING and type I IFN responses had no significant effect on the hematopoietic deficits or incidence and progression of leukemia in mice with severe chronic genome damage throughout the hematopoietic system.

Given that cGAS/STING signaling did not detectably impact on RER-deficient hematopoiesis, we next asked whether this pathway may be of relevance to control effects of spontaneous or acute DNA damage in mice with intact DNA repair. We first compared the size of hematopoietic cell populations between Cgas^KO/KO and WT control mice. We found no differences in HSPC numbers or mature blood cell counts in control versus Cgas^KO/KO mice, neither young (Supplementary Fig. S8A–S8D) nor old (Supplementary Fig. S8E and S8F). To address whether cGAS deficiency affects the proliferative capacity of HSCs, we cultivated single HSCs purified from young and old Cgas^KO/KO or Cgas^WT/WT control mice and observed a similar proliferation kinetics for both genotypes (Fig. 6A and B; Supplementary Fig. S8G and S8H). HSC functionality was also unchanged in vivo as demonstrated by repopulation experiments. Serial competitive transplantation of HSCs purified from either Cgas^KO/KO or Cgas^WT/WT donor mice did not reveal any effect of cGAS deficiency on HSC repopulation capacity (Fig. 6C–F; Supplementary Fig. S8I–S8K).

To address the relevance of cGAS-mediated responses to acute DNA damage, we investigated the recovery of the hematopoietic system from a single exposure to 2 Gy whole-body γ radiation. This acute genotoxic stress indeed resulted in cGAS/STING activation, as reflected by a sharp increase in expression of the type I IFN-induced surface protein Sca-1 that was largely blunted in the Cgas^KO/KO animals (Fig. 6G). As expected, irradiation caused a rapid loss of leukocytes, reticulocytes, and platelets (Fig. 6H–J). After 5 to 10 days, white and red blood cell parameters started to recover, indicating compensatory mechanisms counteracting the radiation-induced blood cell loss. Neither the initial cell loss nor the following recovery phase was altered in Cgas^KO/KO animals compared with control mice. Likewise, the numbers of surviving HSPCs, B- and T-lymphocyte progenitors (Supplementary Fig. S8L–S8O) after irradiation were not changed by lack of cGAS. Competitive transplantation of WBMCs isolated from Cgas^KO/KO and WT control animals 30 days after 2 Gy whole-body γ irradiation confirmed that survival and regeneration of HSCs was not influenced by cGAS deficiency (Fig. 6K–N). Moreover, in vitro proliferation of sorted HSCs irradiated with 2 Gy prior to cultivation was not altered by the absence of cGAS (Fig. 6A and B; Supplementary Fig. S8G and S8H), and also the repopulation potential of γ-irradiated HSCs, reduced compared with unirradiated HSCs as expected, was not affected by cGAS deficiency (Fig. 6C–F). In addition to γ irradiation, we also induced acute replication stress in cGAS-deficient and control mice by exposure to the myeloablative drug 5-FU. This stressor causes rapid loss of cycling stem and progenitor cells and surviving cells subsequently initiate hematopoietic regeneration. However, loss of cGAS had no impact on the hematopoietic recovery after 5-FU–induced myeloablation (Fig. 6O–R; Supplementary Fig. S8P–S8S).

A previous study (28) reported that cGAS suppresses homologous recombination in a STING-independent fashion, and that loss of cGAS accordingly enhanced the resistance of hematopoietic cells to ionizing radiation. As we had not observed cGAS-mediated effects on the loss of hematopoietic cells upon whole-body irradiation with a dose of 2 Gy, we repeated this experiment with the same dose and observation interval as in the study of Jiang and colleagues (28). However, cGAS deficiency did not protect BM cells from radiation-induced cell death evidenced by massive cell loss and similar numbers of HSPCs as well as mature blood cells in both genotypes 10 hours after 9 Gy whole-body γ irradiation (Fig. 6S and T; Supplementary Fig. S8T).

In summary, loss of cGAS had no effect on steady-state hematopoiesis or stem cell function in young or old mice. In vivo, cGAS appeared irrelevant for the survival, function, and recovery of hematopoietic stem and progenitor cells after acute genotoxic insults.

We study the relevance of cGAS/STING signaling in control of malignant transformation of genome-damaged cells. We use a new in vivo model of leukemia resulting from chronic genome damage due to defective RER. If not repaired, ribonucleotides are by far the most common lesion in postreplicative genomic DNA of mammals and therefore, loss of RER has been implicated previously as an oncogenic principle (12, 37, 59). Our study expands these observations to the hematopoietic system, which is of high relevance as loss of at least one allele of the RNASEH2B gene was detected in almost half of CLL tumors (37). RNASEH2B is part of the 13q14 genetic interval that is frequently deleted in different forms of blood cancers, most commonly in CLL (36). The whole interval spans approximately 3 megabases harboring other tumor suppressors like RB1, often resulting in codeletion of different tumor suppressor genes from this region (60). The close proximity to other DNA repair genes might explain why loss of RNASEH2B is more frequently reported than deletion of the genes encoding for the RNase H2 A and C subunits, which do not cluster with DNA repair genes or cell-cycle regulators. As CLL is predominately a malignancy of the elderly, and develops slowly over more than a decade, deletion or codeletion of RNASEH2B might represent a late event in CLL development as loss of RER confers a proliferative disadvantage in cells with intact cell-cycle control. This might explain the differences to our model, in which all hematopoietic cells lose Rnaseh2b during embryonic development, therefore resulting in more aggressive forms of leukemia like T-ALL or AML. Furthermore, our targeted deletion of only Rnaseh2b, ultimately establishes the gene as a tumor suppressor in the hematopoietic system.

Loss of RER caused massive DNA damage and subsequent activation of the p53 pathway, which delayed leukemogenesis at the expense of impaired mature blood cell regeneration. Accordingly, erythropoiesis, which accounts for the main share of blood cell production, was strongly affected in RH2^hKO mice. However, lymphopoiesis was even more severely suppressed, suggesting that DNA damage related to loss of RER is either more strictly controlled in lymphocytes, for example, by induction of cell death, or that lymphocytes favor error-prone repair by nonhomologous end joining (61) and thereby massively accumulate fatal genome damage. Interestingly, RH2^hKO mice developed thrombocythemia, while production of all other blood cell lineages was impaired. This seemingly contradictory finding may in part be explained by the observation that large numbers of thrombocytes can be generated by direct differentiation of HSCs into Mks, a process that requires only minimal cell division (57), while supply of all other mature blood cells heavily relies on cell divisions of stem and progenitors cells. p53 activation in response to DNA damage in the absence of functional RER was responsible for elevated Mk and platelet numbers, while cGAS/STING/type I IFN signaling did not contribute to thrombocytosis. The additional loss of p53 significantly rescued HSC function and blood cell generation in RH2^hKO mice, but also accelerated leukemogenesis, demonstrating that p53 is crucial for orchestrating the response to genome damage ensuing from RER deficiency. Similar observations were made in neurons (62), skin (12), and gut epithelium (59). Interestingly, the thrombocythemia of RH2^hKO mice was reversed by loss of p53, in line with reported roles for p53 signaling in regulating Mk endoreplication and platelet numbers (63, 64).

In addition to the p53-mediated DNA damage response, the cGAS/STING pathway was activated by RER deficiency and contributed to production of type I IFN and TNF. cGAS/STING signaling was shown to induce cell death (20–22, 65, 66) and to trigger senescence (18, 67) and is currently perceived as a new anti-oncogenic principle counteracting malignant transformation by removing or inactivating excessively genome-damaged cells (11, 17). In contrast, two studies proposed cancer and metastasis-promoting functions of cGAS/STING dependent inflammation (68, 69). However, inactivation of cGAS or STING ameliorated, and loss of IFNAR completely abrogated ISG induction, but did not rescue the hematopoietic defects of RER-deficient mice, demonstrating that cell death and proliferation block occurred independently of this signaling axis. Similar observations were made in a mouse model of neuronal RH2 deficiency, in which loss of cGAS dampened the inflammatory response, but did not rescue neuropathology (62). Likewise, analysis of RH2-deficient astrocytes derived from patients with Aicardi-Goutières syndrome suggested that neurotoxicity developed independently of type I IFN signaling (70). Most importantly, we found that incidence of leukemia and survival of RH2^hKO mice were not affected by additional loss of cGAS, STING, or IFNAR in spite of activation of this pathway by genome damage. We recently observed that STING also had no role in controlling the tumor-free survival of p53-deficent mice, which develop thymic lymphoma independently of RER deficiency (71). Collectively, these data suggest that in RER or p53-deficient hematopoiesis innate immune activation via cGAS/STING/IFNAR does not function as a relevant tumor suppressor.

Several studies claim STING-independent functions of cGAS in DNA replication and repair. Nuclear cGAS was reported to maintain genome integrity by stabilizing replication forks (29), which may explain that cGAS protected from colitis-associated colon cancer (30). In contrast, two studies claimed that nuclear cGAS suppresses DNA double-strand break repair by homologous recombination (27, 28), thereby sensitizing cells to programmed cell death after acute genotoxic stress as well as promoting genome instability and tumorigenesis. Accordingly, Jiang and colleagues (28) reported markedly improved survival of cGAS-deficient BM cells upon ionizing radiation. However, in an exact repetition of this experiment, we observed massive and unaltered cell death of hematopoietic cells in irradiated cGAS KO mice. We also observed that cGAS deficiency abrogated inflammatory signaling in response to acute DNA damage, but this did not improve hematopoietic regeneration after different forms of genotoxic stress. We conclude that cGAS activation by either acute or chronic DNA damage does not alter survival, proliferation, and differentiation of hematopoietic cells. This is in line with mechanisms that selectively silence cGAS in HSCs (72) and with reports of HSCs being refractory to massive type I IFN stimulation and thereby protected from overactivation and exhaustion (73, 74).

Collectively, we show that loss of RER in the hematopoietic system caused profound defects in blood cell development and leukemia. p53 activation in RER-deficient hematopoiesis resulted in impaired blood formation but also efficiently delayed leukemia development. While acute or chronic DNA damage elicited robust cGAS/STING activation, the capacity of the hematopoietic system to cope with genomic instability was neither altered by cGAS/STING induced cytokine responses nor signaling-independent functions of cGAS. Likewise, inactivation of the cGAS/STING axis did not affect the capacity of hematopoiesis to cope with radiation inflicted genotoxic stress. Future work will clarify whether cGAS/STING responses may represent a tissue-specific back-up surveillance pathway limiting neoplastic transformation.

R. Behrendt reports grants from Hoffmann-La Roche, ISD Immunotech, and IFM Therapeutics outside the submitted work. A. Gerbaulet reports grants from Fritz-Thyssen Stiftung during the conduct of the study. No disclosures were reported by the other authors.

N. Dressel: Data curation, investigation, visualization. L. Natusch: Data curation, investigation, visualization. C.M. Munz: Investigation. S. Costas Ramon: Investigation, methodology. M.N.F. Morcos: Investigation. A. Loff: Investigation. B. Hiller: Investigation. C. Haase: Methodology. L. Schulze: Methodology. P. Müller: Methodology. M. Lesche: Methodology. A. Dahl: Methodology. H. Luksch: Resources. A. Rösen-Wolff: Resources, funding acquisition. A. Roers: Conceptualization, funding acquisition, writing–original draft, writing–review and editing. R. Behrendt: Conceptualization, funding acquisition, investigation, writing–original draft, writing–review and editing. A. Gerbaulet: Conceptualization, supervision, funding acquisition, investigation, methodology, writing–original draft, project administration, writing–review and editing.

This study was funded by the Fritz-Thyssen-Stiftung (Az. 10.19.1.013MN to A. Gerbaut), Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 369799452 – TRR237 Nucleic Acid Immunity, project B17 to A. Roers, project B18 to A. Rösen-Wolff and project B19 to R. Behrendt. The authors thank Madeleine Rickauer for expert technical assistance. They thank Katja Blumenstock and Jonathan Schmid-Burgk for sequencing of the Rnaseh2b amplicons. L. Natusch was supported by the Else-Kröner-Promotionskolleg.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.