Insights and modulation of RNA polymerase–dependent R-loop and dsRNA in Fanconi anemia hematopoietic stem cells

The hematopoietic system is responsible for providing a continuous supply of various mature blood cell types, including erythrocytes, platelets, granulocytes, monocytes, and lymphocytes (1–3). Hematopoiesis depends on hematopoietic stem cells (HSCs), which can either self-renew or differentiate to multi-potent progenitors (1, 2). However, under certain disease conditions, a failure to maintain the HSC pool leads to BM failure (BMF). BMF is characterized by hypoplastic BM and/or peripheral cytopenia in one or more types of blood cells (4, 5). BMF can have serious consequences, including a predisposition to leukemia and an increased rate of death (5, 6). Fanconi anemia (FA) is the most common inherited BMF syndrome (7, 8). As such, understanding the etiology is important for diagnosing and treating BMF in FA. FA is characterized by progressive pancytopenia, as well as variable congenital anomalies and an elevated risk of developing myelodysplastic syndromes, leukemia, and solid tumors (9, 10).

Twenty-three FANC genes that cause FA when mutated have been identified (10–12). These correspond to distinct genetic complementation groups, including FA A–C, D1, D2, E–G, I–J, L–W, and Y. Although FANCB is X-linked, FA genes are typically recessive. FANCA is by far the most frequently mutated gene in FA, representing approximately 60% of cases (13, 14). There is some phenotypic heterogeneity among complementation groups (10, 15).

The FA nuclear core complex, which includes FANC A, B, C, E, F, G, L, and M, along with FANCT, has a ubiquitin ligase activity that monoubiquitinates FANCD2 and FANCI via a process termed the FA pathway (10, 14, 16–21). Monoubiquitinated FANCD2-FANCI form a DNA clamp (22, 23). Among other FA proteins are those not required for mono-ubiquitination of FANCD2 (or FANCI), including key homologous recombination proteins such as BRCA1/FANCS, BRCA2/FANCD1, PALB2/FANCN, RAD51/FANCR, RAD51C/FANCO, and XRCC2/FANCU (15, 24–30). Other proteins encoded by FA/FA-like genes include the nucleotide excision repair protein XPF/FANCQ, the SLX4/FANCP endonuclease, the REV7/FANCV trans-lesion polymerase, and RFWD3/FANCW (11, 31–34).

The different FA complementation groups display a shared cellular sensitivity and chromosome abnormality phenotype in response to agents such as mitomycin C (MMC) that cause DNA interstrand cross-links (10, 16, 35). It is, therefore, widely accepted that FA proteins have a role in interstrand cross-link repair. More recent work has suggested a broader role for FA proteins in responding to replication stress (12, 36–41). One way by which FA proteins ameliorate replication stress and protect genome stability is by mitigating R-loops, which are RNA-DNA hybrids that displace the complementary DNA strand (42–44). R-loops can arise from the collision of the transcription machinery and the replication fork and can be resolved, in part, by helicases and RNases (45). One place that the resolution of R-loops and transcription-replication conflicts may be particularly critical is in nucleoli, given that rRNA genes are among the most actively transcribed in cells (46).

Importantly, whether R-loops lead to BMF in FA, and if so by what mechanism, has remained elusive. Further, single-gene KO of FA genes, including Fanca, in mice does not lead to a clear BMF phenotype. Since RNA polymerase (RNAP) can indirectly generate R-loops by transcribing DNA into RNA, and because SETD2 restricts RNAP-dependent transcriptional elongation (47), we have generated Fanca and Setd2 double mutants in mice as a means to explore the role of R-loops in BMF in FA. In support of such an approach, SETD2 is essential for hematopoiesis, and disruption of SETD2 leads to leukemia (47). The SETD2 protein methylates histone H3 at lysine 36 (H3K36me3), thereby ensuring accurate transcription of genes necessary for blood cell development by promoting RNAP-dependent elongation of transcripts. Notably, RNAPs I, II, and III occasionally form R-loops during transcription that can generate dsRNA, which have a variety of transcriptional regulatory functions (46), including gene silencing through RNAi pathways (48). However, excessive accumulation of dsRNA can also lead to transcriptional dysregulation and genomic instability and can thereby contribute to diseases such as cancer and BMF.

RNAPs have specialized roles: RNAPI transcribes rRNA genes, RNAPII transcribes mRNA and certain small RNAs, and RNAPIII transcribes tRNA and the 5S rRNA gene (49). A variety of RNAP inhibitors have been developed that can be utilized to dissect the roles of distinct RNAPs in biological processes, including an RNAPI inhibitor, CX5461 (pidnarulex), which was in an early-stage clinical trial and is effective against tumors with BRCA1 and BRCA2 mutations (50). Juglone, 5-hydroxy-1,4-naphthoquinone, inhibits members of the parvulin PPIase family as well as RNAP I, II, and III (and is therefore a pan-RNAP inhibitor) (51). In addition, L-leucine is a branched-chain amino acid that can activate mRNA translation via mTORC1 activation. L-leucine has been used to improve the treatment of Diamond-Blackfan anemia in clinical trials (52).

In this study, we established an FA mouse model (Setd2^+/– Fanca^–/–), which displays severe BMF more similar to that observed in patients with FA than in single–FA gene KO mice. Further, we observed excessive R-loop and dsRNA accumulation and ribosomal biogenesis defects in cells from a patient with FA (FA-A complementation group) as well as HSCs from Setd2^+/– Fanca^–/– mice. Importantly, given that the pan-RNAP inhibitor Juglone ameliorated the ribosomal biogenesis defect and reduced mitotic errors in Setd2^+/– Fanca^–/– HSCs, our results support a key role for R-loops generated at rDNA loci, and potentially for dsRNA and a resulting ribosomopathy, as well as mitotic defects, as a driver of FA-associated BMF. Further, our results suggest that inhibition of RNAPs may provide a therapeutic alternative to BM transplantation to prevent and/or treat FA-related BMF.

R-loop and dsRNA accumulation results in ribosomal biogenesis defects in FANCA-deficient cells. Multiple reports have shown accumulation of R-loops in FA mutant cells (42-44). Thus, to better understand cellular defects associated with FA, we initially assessed the levels of R-loops in corrected (FANCA corrected) and uncorrected (FANCA deficient) pairs of FA patient–derived fibroblasts. Indeed, R-loops were expressed at higher levels in FANCA-deficient cells compared with FANCA-corrected cells (Figure 1, A and B). Additionally, when R-loops are formed in certain genes, they can trigger the production of another type of RNA, dsRNA (46). Strikingly, FANCA-deficient cells also had higher levels of dsRNA than FANCA-corrected cells (Figure 1, A and C). Given that R-loops were more abundant in FANCA-deficient cells, and since RNAPI-dependent rRNAs are among the most active sites of transcription in the cell (43, 53), we hypothesized that rRNAs might be compromised by deficiency for FANCA. As anticipated, all rRNAs transcribed in nucleoli, as well as RNAPIII-dependent 5S, were downregulated in patient-derived FANCA-deficient fibroblasts (Figure 1D).

Figure 1

Human FANCA-deficient cells from a patient with FA display accumulation of R-loops and dsRNA and errors in ribosomal biogenesis. (A) Visualization of R-loops detected by S9.6 mAb (red) and dsRNA detected by rJ2 mAb (green) and via immunofluorescence microscopy in FANCA-corrected or -deficient cells from a patient with FA. Nuclei shown by counterstaining with DAPI (blue). Original magnification is ×63. (B and C) Quantification of R-loops (B) and dsRNA (C) based upon the area integrated intensity measured by ImageJ (NIH). One-tailed Student’s t test; n = 37 for each condition. (D) Quantification of the expression of ribosomal genes using qRT-PCR. One-tailed Student’s t test; n = 3 for each condition. (E) DNA dot blots detecting R-loop levels using S9.6 antibody in FANCA-deficient cells reconstituted with 2 RNases (RNASEH1 or DICER1) or vector alone (Vec). (F) Quantification of R-loop levels shown in E. One-way ANOVA with Tukey’s multiple-comparison test; n = 3 for each condition. (G) RNA dot blots detecting levels of dsRNA in FANCA-deficient cells reconstituted with 2 RNases (RNASEH1 or DICER1) or Vec. n = 3 for each condition. (H) Quantification of dsRNA levels shown in G. One-way ANOVA with Tukey’s multiple-comparison test; n = 3 for each condition. (I) Quantification of specific rRNA forms in FANCA-deficient cells containing empty vector with or without overexpression of RNASE H1 or DICER1. One-way ANOVA with Tukey’s multiple-comparison test; n = 3 for each condition. (J) Visual summary. (K and L) For patient-derived cells, we utilized FANCA-deficient GM6935 cells (84). (K) Relative expression levels of SETD2, ALDH2, ADH5, and TP53 in healthy donors (n = 11) and FA patients (n = 21) were analyzed using normalized microarray data (GSE16334). Expression values presented as log₂-transformed intensities. One-tailed Student’s t test. (L) Reactome pathway enrichment analysis comparing HD and FA gene clusters performed using compareCluster function (fun = “enrichPathway”, pvalueCutoff = 0.05) from clusterProfiler and ReactomePA. Top 10 enriched pathways are shown in dot plot: color indicates adjusted P value; dot size represents gene ratio. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To address the role of R-loops and dsRNA in regulating transcription within nucleoli, we conducted ectopic expression experiments of RNASEH1 and DICER1 in FANCA-deficient cells (Figure 1, E–I). The expression of RNASEH1 led to a significant reduction in both R-loop and dsRNA levels, while DICER1 expression specifically resolved dsRNA in FANCA-deficient cells (Figure 1, E–H). Importantly, ectopic expression of both RNASEH1 and DICER1 resulted in upregulated expression of various rRNAs (Figure 1I). Taken together, these results suggest that resolving the accumulation of R-loops and, somewhat surprisingly dsRNA, may rescue ribosomal biogenesis defects associated with FA (Figure 1J). In support of this, levels of mono-ribosomes and polysomes were indeed decreased in FA patient cells (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.192126DS1). Additionally, we analyzed microarray data from low-density BM cells derived from patients with FA (GSE16334). This analysis revealed that, in addition to genes previously implicated in FA-associated BMF mouse models, such as ALDH2 and ADH5 (54), SETD2 expression was also significantly downregulated in FA patient samples (Figure 1K and Supplemental Figure 2), as were genes related to translation/ribosome biogenesis (Figure 1L).

Development of an FA mouse model with BMF. Single-KO mouse models of FA, such as the Fanca-KO model, do not fully recapitulate the human disease including an absence of BMF (55). However, since FA patient–derived fibroblasts displayed accumulation of R-loops and dsRNA, as well as ribosomal biogenesis defects, we hypothesized that enhancing RNAP-dependent elongation in mice might produce a BMF phenotype. Importantly, SETD2 restricts RNAPs (47), and SETD2 was downregulated in FA patient–derived low-density BM cells (Figure 1K). Thus, we generated a compound mutant mouse model: Setd2^+/– Fanca^–/–. Furthermore, we demonstrated that the Setd2^–/– mutant is embryonic lethal in our previous report (47). Indeed, Setd2^+/– Fanca^–/– mice exhibited phenotypes reminiscent of those seen in patients with FA, including defects of eyelids and eyeballs (Figure 2A). Importantly, as compared with Setd2^+/– and/or Fanca^–/– mice, Setd2^+/– Fanca^–/– mice exhibited further reductions in the total numbers of WBCs, neutrophils, monocytes, eosinophils, RBCs, platelets, BM cells, and HSCs and increased reticulocytes, mean corpuscular volume (MCV), and mean corpuscular hemoglobin (MCH) (Figure 2, B–M). For further insight into the effects of the combination of Setd2^+/– and Fanca^–/– mutations on HSC function, we evaluated the cell cycle status of HSCs from adult-stage mice. As compared with HSCs from control mice, those from Setd2^+/– Fanca^–/– mice, but not Setd2^+/– or Fanca^–/– single-mutant mice, displayed a decrease in quiescent (G0) cells accompanied by an increase in G1 cells (Figure 2N). Interestingly, while Fanca^–/– single-mutant HSCs had elevated DNA damage compared with WT HSCs, as measured by relative levels of γH2A.X, DNA damage was further elevated in Setd2^+/– Fanca^–/– HSCs to similar levels as in Setd2^+/– single mutants (Figure 2O). As a measure of replication stress (56), in addition to γH2AX, pRPA2, pCHK1, and pCHK2 were significantly upregulated in Setd2^+/– Fanca^–/– HSCs (Figure 2, O and P, and Supplemental Figure 3). In this context, it should be noted that hypersensitivity to DNA cross-linking agents, such as MMC, is a well-known characteristic of FA cells, including FA-A cells deficient for FANCA (57), and HSCs from FA single-gene KO mice (58). An examination of the MMC sensitivity of c-Kit–positive cells further supports the conclusion that the combination of the Setd2^+/– and Fanca^–/– mutations enhances FA phenotypes. Cells from Setd2^+/– and Fanca^–/– single-mutant mice exhibited increased sensitivity to MMC, as compared with WT control cells, and this was enhanced even further in cells from double-mutant Setd2^+/– Fanca^–/– mice (Figure 2Q). To further analyze HSC function in Setd2^+/– Fanca^–/– mice, we performed a competitive BM transplantation assay. The results indicate that, as compared with Setd2^+/– or Fanca^–/– mouse-derived BM cells (BMCs), Setd2^+/– Fanca^–/– mouse–derived BMCs exhibited a dramatic loss of engraftment capacity from the first month after BM transplantation (Figure 2R). Taken together, these findings suggest that the combination of Setd2^+/– and Fanca^–/– mutations results in more pronounced DNA damage and defects in HSC maintenance, leading to BMF.

Figure 2

Characterization of BMF in a mouse model for FA. (A) Representative photos of eyes in control and Setd2^+/– Fanca^–/– mice. (B–K) Peripheral blood (PB) analysis between control (black), Setd2^+/– (blue), Fanca^–/–(green), and Setd2^+/– Fanca^–/– (red) mice. WBCs (B), neutrophils (D), monocytes (E), eosinophils (F), RBCs (G), reticulocytes (H), MCV (I), MCH (J), platelets (K). One-way ANOVA with Tukey’s multiple-comparison test. (L) Total BM cell number in control (black, n = 3), Setd2^+/– (blue, n = 3), Fanca^–/– (green, n = 4), and Setd2^+/– Fanca^–/– (red, n = 4) mice. One-way ANOVA with Tukey’s multiple-comparison test. (M) Absolute HSC [lineage (–), c-Kit (+), EPCR (+), CD48 (–), and CD150 (+)] numbers for control (black, n =4), Setd2^+/– (blue, n = 3), Fanca^–/– (green, n = 3), and Setd2^+/– Fanca^–/– (red, n = 3) mice. One-way ANOVA with Tukey’s multiple-comparison test. (N) Cell cycle status by flow cytometry utilizing Ki67 to distinguish G0 cells from G1/S/G2/M cells with DAPI stain to measure DNA content. Graphs show distribution of cells to indicated cell cycle phases. One-way ANOVA with Tukey’s multiple-comparison test; n = 4 for each condition. (O) Levels of DNA damage in HSCs from control (black, n = 4), Setd2^+/– (blue, n = 3), Fanca^–/– (green, n = 3), and Setd2^+/– Fanca^–/– (red, n = 5) mice as measured by γH2A.X levels. One-way ANOVA with Tukey’s multiple-comparison test. (P) Levels of pRPA2 in HSC from control (black, n = 3), Setd2^+/– (blue, n = 3), Fanca^–/– (green, n = 3), and Setd2^+/– Fanca^–/– (red, n = 3) mice. One-way ANOVA with Tukey’s multiple-comparison test. (Q) MMC sensitivity for c-Kit⁺ cells from control (black), Setd2^+/– (blue), Fanca^–/– (green), and Setd2^+/– Fanca^–/– (red) mice. c-Kit⁺ cells were cultured with MMC for 3 days when cell numbers were counted using Celigo. Two-way ANOVA with Tukey’s multiple-comparison test; n = 4 for each condition. (R) BM transplantation of CD45.2⁺ control (black), Setd2^+/– (blue), Fanca^–/– (green), and Setd2^+/– Fanca^–/– (red) BMCs into CD45.1 mice. After recipient mice received lethal irradiation, BMCs from control, Setd2^+/–, Fanca^–/–, and Setd2^+/– Fanca^–/– were injected into recipient mice. PB was analyzed at 1, 3, and 5 months after transplantation. Two-way ANOVA followed by Tukey’s multiple-comparison test; n = 5 for each condition. Data shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Next, we investigated whether BMF and other FA-related phenotypes of HSCs in our Setd2^+/– Fanca^–/– mouse model might be associated with dysregulation of R-loops and rRNA levels since these were seen in FANCA-deficient fibroblasts (Figure 1). Indeed, similar to patient-derived FA-A cells, Setd2^+/– Fanca^–/– mice showed accumulation of R-loops and dsRNA that were elevated in comparison to HSCs from WT controls and Setd2^+/– and Fanca^–/– single mutants (Figure 3, A and B). Importantly, R-loop levels could be reduced in HSCs from Setd2^+/– Fanca^–/– mice via expression of various RNases (Figure 3C). Further, most rRNAs, such as 5S, 18S, 28S, and 45S, were significantly downregulated in Setd2^+/– Fanca^–/– mouse HSCs as compared with the WT or each single mutant (Figure 3D). Additionally, an immunofluorescent imaging analysis showed that the upstream binding factor, a marker for ribosomal loci (59), mostly colocalized with R-loops at nucleoli in Setd2^+/– Fanca^–/– HSCs (Figure 3E and Supplemental Figure 4A). Since upstream binding factor is related to RNAP I (60), and RNAP III localizes to nucleoli (61), accumulation of R-loops at nucleoli and resulting ribosome biogenesis defects in Setd2^+/– Fanca^–/– HSCs may be related to RNAPI and maybe also RNAPIII. Further, because R-loops accumulated at ribosomal loci, and there was a defect in various rRNA levels (Figure 3E), similar to that observed in cells from patients with FA, our Setd2^+/– Fanca^–/– double-mutant mice may serve as a model for FA-associated BMF driven by ribosomal biogenesis defects.

Figure 3

R-loop accumulation and ribosomal biogenesis defect in Setd2^+/– Fanca^–/– mouse model HSCs. (A) Levels of R-loop in HSC from control (black, n = 3), Setd2^+/– (blue, n = 3), Fanca^–/– (green, n = 3), and Setd2^+/– Fanca^–/– (red, n = 3) mice. One-way ANOVA with Tukey’s multiple-comparison test. (B) Levels of dsRNA in HSC from control (black, n = 3), Setd2^+/– (blue, n = 3), Fanca^–/– (green, n = 3), and Setd2^+/– Fanca^–/– (red, n = 3) mice. One-way ANOVA with Tukey’s multiple-comparison test. (C) Levels of R-loops in HSCs from control and Setd2^+/– Fanca^–/– mice, with or without treatment of various RNases, were analyzed by flow cytometry using S9.6 antibody. One-tailed Student’s t test; n = 3 for each condition. (D) Ribosomal gene expression in control, Setd2^+/–, Fanca^–/–, and Setd2^+/– Fanca^–/– HSCs measured utilizing qRT-PCR. Control (black, n = 4), Setd2^+/– (blue, n = 3), Fanca^–/– (green, n = 3), and Setd2^+/– Fanca^–/– (red, n = 4). One-way ANOVA with Tukey’s multiple-comparison test. (E) Measures of colocalization of R-loops (green, detected using S9.6 antibody) with RNA polymerase I (red, detected using Ubf antibody); DNA is counterstained with DAPI (blue). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Mitotic errors in Setd2^+/– Fanca^–/– HSCs. Since a failure to enter the quiescent-G0 phase may cause BMF in our Setd2^+/– Fanca^–/– mouse model (Figure 2N), and cell division defects have been reported in fibroblasts deficient for FA factors including FANCA (62), we conducted liquid culture experiments of HSCs to carefully examine their cell division process (Figure 4A). Notably, as compared with the control, Setd2^+/– Fanca^–/– double-mutant HSCs exhibited diminished proliferation seen as a decreased HSC population (Figure 4, B and C). Additionally, we analyzed the cell division history via single-cell culture (Figure 4D) and CytoTell green–based cell division history assay (Figure 4, E–H) with sorted HSCs from control or Setd2^+/– Fanca^–/– mice. This experiment indicates that almost 40% of Setd2^+/– Fanca^–/– HSCs underwent a first cell division within 2 days, whereas only 20% of control HSCs did so (Figure 4D). However, an examination of 5-day cultures revealed that, in comparison to the control, Setd2^+/– Fanca^–/– HSCs showed an arrest during their first cell division and failed to complete successive cell divisions (Figure 4, E–H). To clarify the basis for the arrested cell division in Setd2^+/– Fanca^–/– HSCs, we examined nuclear morphologies utilizing immunofluorescence microscopy. There was an increased number of abnormal binucleated and micronucleated cells in Setd2^+/– Fanca^–/– HSCs (Figure 4, I and J). These results suggest that mitotic errors underlie defects in Setd2^+/– Fanca^–/– HSC function, including decreased proliferation (Figure 2N and Figure 4, E–K) and decreased chimerism in BM transplants (Figure 2R).

Figure 4

HSCs from Setd2^+/– Fanca^–/– mice display cell division defects. (A) Schematic of the experimental setup for B–D. (B and C) Proliferative capacity (B) and HSC [lineage (–), c-Kit (+), EPCR (+), CD48 (–), and CD150 (+)] frequency (C) of control (WT) or Setd2^+/– Fanca^–/– HSCs, determined as number of cells after 5 days of liquid culture relative to number at day 1, and HSCs as percentage of total at day 5, respectively. Cultures of HSCs were counted with Celigo in B and were analyzed by flow cytometry in C; for each condition, n = 3 in B and n = 4 in C. One-tailed Student’s t test. (D) Counts by hand of the number of cells in each well at 2 days after establishing single-cell cultures of control or Setd2^+/– Fanca^–/– HSCs. One-tailed Student’s t test; n = 3 for each condition. (E–H) Analysis comparing the cell division histories of control versus Setd2^+/– Fanca^–/– HSCs in 5-day liquid cultures. HSCs were sorted and stained with CytoTell green, then cultured for 5 days. At day 5, HSCs were stained with surface markers, and cell division history was analyzed with flow cytometry based on the CytoTell green signal in various peaks. Low division, medium division, and high division numbers were based on CytoTell green signal associated with 1–3, 4–8, and 9 or more divisions, respectively. One-tailed Student’s t test; n = 3 for each condition. (I) Visualization of nuclei in HSCs from control and Setd2^+/– Fanca^–/– mice, as a measure of mitotic errors, after 5 days in liquid culture. (J) Quantification of cells with abnormal nuclei (binucleated or micronucleated), indicating mitotic errors. One-tailed Student’s t test; n = 3 for each condition. (K) Schematic depicting nuclear morphologies and mitotic errors in controls and Setd2^+/– Fanca^–/– HSCs during cell proliferation. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Inhibition of RNAPs rescues ribosomal biogenesis defects and survival of HSCs in Setd2^+/– Fanca^–/– mice. One approach to translating our findings of excessive cycling of Setd2^+/– Fanca^–/– HSCs, ultimately leading to mitotic errors and BMF into a potential therapy for BMF in patients with FA, is to reduce R-loop accumulation. Ideally, a drug treatment to reduce R-loop accumulation would suppress DNA damage and/or rescue HSC mitotic defects. Thus, for this purpose, we employed a stimulator of mRNA translation (L-leucine) (52), an RNAP I inhibitor (CX5461) (50), and a pan-RNAP inhibitor (Juglone) (51) in liquid cultures of isolated HSCs. Given that FA cells exhibit ribosomal defects (Supplemental Figure 1), activating mRNA translation through L-leucine, or inhibiting transcription and thus R-loop generation, may ameliorate HSC function. To validate these hypotheses, we treated Setd2^+/– Fanca^–/– hematopoietic stem and progenitor cells (HSPCs) in liquid culture with 200 μM of L-leucine, 1 pM of CX5461, or 0.1 μM of Juglone and measured R-loop (S9.6) and dsRNA (J2) levels. CX5461, L-leucine, and Juglone treatment reduced the levels of both R-loops and dsRNA (Figure 5, A and B). As a measure of whether this reduction in R-loop levels aids HSC function, we conducted single cell–based paired-daughter assays. Symmetric division maintains stemness, while asymmetric division leads to 1 HSC and 1 cell committed to division that eventually generates 4 myeloid lineages (47). Treatment with L-leucine and CX5461, and especially Juglone, increased symmetric cell divisions and decreased committed asymmetric cell divisions of Setd2^+/– Fanca^–/– HSCs (Figure 5, C–E). This suggests that low doses of L-leucine, CX5461, and Juglone help maintain the stemness of Setd2^+/– Fanca^–/– HSCs. Interestingly, Juglone treatment also improved symmetric division in control HSCs.

Figure 5

Inhibition of RNA polymerases reduces R-loops and cell cycle defects and rescues self-renewal capacity and BMF of Setd2^+/– Fanca^–/– HSCs. (A and B) The levels of S9.6 (R-loops) in untreated (black, control: n = 4, Setd2^+/– Fanca^–/–: n = 6) versus CX5461 (orange, control: n = 4, Setd2^+/– Fanca^–/–: n = 3), L-leucine (purple, control: n = 4, Setd2^+/– Fanca^–/–: n = 3), and Juglone-treated (green, control: n = 3, Setd2^+/– Fanca^–/–: n = 3) HSPCs (A) and J2 (dsRNA) in untreated (black, control: n = 3, Setd2^+/– Fanca^–/–: n =7) versus CX5461 (orange, control: n =4, Setd2^+/– Fanca^–/–: n = 3), L-leucine (purple, control: n = 4, Setd2^+/– Fanca^–/–: n = 3), and Juglone-treated (green, control: n = 4, Setd2^+/– Fanca^–/–: n = 5) HSPCs (B). One-way ANOVA with Tukey’s multiple-comparison test. (C–E) Paired-daughter assay with or without CX5461(C), L-leucine (D), and Juglone (E). One-way ANOVA with Tukey’s multiple-comparison test. (F–H) Cell cycle and apoptotic effects of CX5461, L-leucine, and Juglone treatment of control or Setd2^+/– Fanca^–/– HSCs are shown. (F) Ki67 staining after Juglone treatment as shown in FACS plots. (G) Frequency of apoptosis and of each cell cycle stage in HSCs. One-way ANOVA with Tukey’s multiple-comparison test; n = 3 for each condition. (H) Frequency of apoptotic or pre-apoptotic cells. One-way ANOVA, Tukey’s multiple-comparison test; n = 3 each condition. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To further analyze the mechanism of how L-leucine, CX5461, and Juglone facilitate the maintenance of Setd2^+/– Fanca^–/– HSC stemness, we performed cell cycle assays with treated cells. Unlike CX5461 and L-leucine, Juglone rescued survival of Setd2^+/– Fanca^–/– HSCs by reducing the number of apoptotic cells and restored levels of G0 and G1 cells to those similar to untreated controls (Figure 5, F–H). These results suggest that Juglone may have the potential to reverse BMF, as Juglone treatment was found to maintain HSC survival from Setd2^+/– Fanca^–/– mice.

The pan-RNAP inhibitor Juglone slows cell division, which ameliorates the maintenance of stemness in Setd2^+/– Fanca^–/– HSCs. To evaluate the effects of Juglone on the maintenance of HSCs, Setd2^+/– Fanca^–/– HSCs were cultured in vitro with Juglone. This resulted in reduced proliferation of both control and Setd2^+/– Fanca^–/– cells (Figure 6A) and significantly increased the frequency of control HSCs; there was a trend toward upregulation of the frequency of Setd2^+/– Fanca^–/– HSCs (Figure 6B). To further validate effects of Juglone on the rate of cell division, a single cell–based cell division assay was performed, which showed that Juglone treatment slowed the first cell division (Figure 6C). We then performed a CytoTell-based cell division history assay, which indicated that Juglone treatment reduced the number of Setd2^+/– Fanca^–/– cells with a medium division history and increased the number of both control and Setd2^+/– Fanca^–/– cells with a low division history (Figure 6, E–G). These results are consistent with Juglone treatment slowing cell division and allowing HSCs sufficient time to resolve R-loops and/or dsRNA, and thereby preparing for the next cell division. To determine whether Juglone might improve the proliferation of Setd2^+/– Fanca^–/– HSCs by reducing mitotic errors, liquid cultures were performed. Indeed, while Juglone treatment did not alter levels of mitotic errors in control HSCs, it significantly reduced mitotic errors by Setd2^+/– Fanca^–/– HSCs (Figure 6H). Finally, to affirm the hypothesis that R-loop accumulation contributes to BMF in FA, we performed a BM transplantation assay using treated or untreated control or Setd2^+/– Fanca^–/– c-Kit–positive cells (Figure 6, I and J). Notably, c-Kit–positive Setd2^+/– Fanca^–/– cells and not controls, treated with Juglone but not CX5461 or L-leucine, showed significantly higher engraftment (Figure 6J and Supplemental Figure 5B). Importantly, Juglone treatment enhanced long-term engraftment capacity for cells from Setd2^+/– Fanca^–/– mice, as confirmed by secondary BM transplant (Figure 6K). In addition to the in vitro Juglone treatment experiments, we performed in vivo Juglone treatments. Juglone was administered as a single i.p. injection to P5 neonatal mice. Four weeks after injection, we analyzed total BM cell numbers and HSC numbers. This analysis revealed that Juglone treatment significantly rescued both total BM cellularity and HSC numbers in WT controls and Setd2^+/– Fanca^–/– mice (Figure 6, M and N).

Figure 6

Treatment with Juglone slows cell division and facilitates the maintenance of stemness and improves engraftment of Setd2^+/– Fanca^–/– HSCs. (A and B) Proliferation capacity counted by Celigo (A) and HSC frequency analyzed by flow cytometry (B) of WT control or Setd2^+/– Fanca^–/– HSCs after 5 days of liquid culture. N = 3 each condition. (C) Single-cell culture of control or Setd2^+/– Fanca^–/– cells with or without Juglone (0.1 µM) for 2 days. Cell numbers per well counted at day 2. N = 3 for each condition. (D–G) Cell division history analysis of control and Setd2^+/– Fanca^–/– HSCs with or without Juglone (0.1 µM). HSCs were sorted, labeled with CytoTell Green, and cultured for 5 days before analysis by flow cytometry. Representative CytoTell plots (D) and quantification of high (E), medium (F), and low (G) division histories are shown. N = 3 for each condition. (H) Mitotic errors (binucleated or micronucleated cells) in Setd2^+/– Fanca^–/– HSCs after 5 days of Juglone treatment (0.1 µM) in liquid culture. N = 3 for each condition. (I) Experimental scheme of BM transplantation with drug-treated HSPCs. (J) BM transplantation of control (black) and Setd2^+/– Fanca^–/– (red) HSCs, with or without Juglone treatment. Engraftment was measured as the percentage of CD45.2⁺ PB cells at 4, 8, and 12 weeks. N = 5 for each condition. (K) Secondary transplantation of control (black) and Setd2^+/– Fanca^–/– (red) BMCs, with or without Juglone treatment; engraftment measured as CD45.2⁺ PB cells at 10 weeks. N = 4 for each condition. (L) Experimental scheme of Juglone in vivo treatment. (M and N) Total BMC numbers (M) and absolute HSC numbers (N) in control (black) and Setd2^+/– Fanca^–/– (red) mice with or without Juglone treatment. N = 3 for each point. Data are shown as mean ± SEM. Statistical significance was determined by 1-way ANOVA with Tukey’s multiple-comparison test unless otherwise indicated; panel J was analyzed by 2-way ANOVA with Tukey’s multiple-comparison test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Taken together, these results suggest that Juglone treatment slows cell division and gives HSCs enough time to resolve R-loops and dsRNA, and thereby suppress mitotic errors and rescue Setd2^+/– Fanca^–/– HSC stemness and engraftment capacity.

In this study, we began by demonstrating that an FA patient–derived cell line accumulates R-loops, as well as dsRNA, and has reduced expression of all rRNA genes, which underlies ribosomal biogenesis defects. R-loop and/or dsRNA accumulation, and rRNA expression, can be ameliorated by overexpression of either RNASEH1 or DICER1. Because single-gene KO for FA genes in mice does not display BMF (55), to explore the relevance of our observations to this important manifestation of FA, we built an FA mouse model based upon Setd2^+/– Fanca^–/– double mutants. Indeed, these mice displayed decreased BM cellularity and a greatly reduced function of HSCs in competitive BM transplants (Figure 2). Further, as compared with control (WT) mice and Setd2^+/– and Fanca^–/– single mutants, Setd2^+/– Fanca^–/– double-mutant HSCs displayed accumulation of R-loops and dsRNA. Correspondingly, Setd2^+/– Fanca^–/– double-mutant HSCs showed ribosomal biogenesis defects that could be rescued by treatment with the pan-RNAP inhibitor Juglone. Additionally, Juglone could rescue key features of Setd2^+/– Fanca^–/– HSCs including mitotic errors, decreased quiescence, and a reduced capacity for successive cell divisions, elevated levels of apoptosis, and ultimately, engraftment defects. Taken together, given that Setd2^+/– Fanca^–/– HSCs also have elevated levels of DNA damage and increased sensitivity to MMC, these results suggest the importance of R-loops and dsRNA as well as rDNA loci and ribosomal biogenesis defects in HSCs to BMF associated with FA. Notably, this work may have also identified a putative therapeutic option for treating BMF in patients with FA.

In FA research, a large body of work has focused on DNA damage, MMC sensitivity, and HSC transplantation as a basis for understanding the etiology of BMF and/or a predisposition to various malignancies that are associated with this disease (63). Additionally, more recent work has demonstrated that FA cells have increased levels of R-loops and decreased transcription of rRNA genes (64). However, the current study is the first, to our knowledge, to demonstrate increased levels of R-loops and dsRNA and decreased transcription of rRNA genes, which are largely nucleolar, in FA-related HSCs. In particular, we also reveal a role for RNAP-mediated transcription in this process. Further, based on results obtained utilizing our mouse model, we propose that elevated R-loops and dsRNA, including at rDNA loci, lead to increased levels of DNA damage and MMC sensitivity in HSCs, at least in part via transcription-replication conflicts. Thus, our study provides important insights for understanding FA and the BMF that is frequently associated with it.

One key limitation for the FA field is that Fanca^–/– and other single FA gene KO mouse models do not exhibit BMF-like phenotypes (55). Several double-mutant mouse models, such as Aldh2^–/– Fancd2^–/– and Adh5^–/– Fancd2^–/–, can induce severe BMF (54, 65), which suggests the importance of DNA damage in this process; while ADH5 directly detoxifies formaldehyde in the nucleus generated by various biochemical pathways, ALDH2 detoxifies acetaldehyde (a byproduct of alcohol metabolism). However, the role of R-loops and dsRNA in FA-related BMF was previously untested, to our knowledge; our Setd2^+/– Fanca^–/– double-mutant mouse model demonstrates the role of these factors in the etiology of BMF in FA and, additionally, may position R-loops and dsRNA at rDNA loci as an important upstream source of DNA damage that also contributes to BMF in FA. The Setd2^+/– Fanca^–/– FA model we have built accurately recapitulates cellular defects associated with patients with FA, including increased R-loops and dsRNA, as well as decreased transcription of rDNA and increased sensitivity to MMC (Figures 1–3). As such, this provides a relevant model for studying the disease mechanisms in, and potential treatments for, BMF in FA. The Setd2^+/– Fanca^–/– FA mouse model displays other phenotypes, including eye abnormalities, typically seen in patients with FA, and should thereby advance FA research on various fronts.

Since SETD2 is an epigenetic factor that constrains elongation by RNAPs and regulates ribosomal biogenesis (47, 66–68), our Setd2^+/– Fanca^–/– model appears to combine a source of increased R-loops and dsRNA with decreased resolution of these structures due to deficiency for Fanca. Here, we note that heterozygous KO of SETD2 results in transcriptional dysfunction (69). Additionally, although we have combined the Setd2^+/– mutation with a Fanca-KO mutation, our findings should be relevant to other complementation groups associated with components of the FA pathway, since they share a common DNA damage sensitivity when mutated (16).

On a separate note, SETD2 is downregulated under inflammatory stress, which subsequently leads to the downregulation of multiple FANC genes (data not shown). In patients with FA, genetic modifiers or frequent infections during infancy may lead to SETD2 downregulation, which could exacerbate the FA phenotype and contribute to the development of BMF. Thus, we posit that other modifiers, such as variant alleles or environmental stresses or stimuli, may contribute to the development of BMF in HSCs in patients with FA. In particular, the presence of SETD2 in mice could act as a genetic modifier that elevates FA gene expression and may thereby explain why single FA KO mouse models do not display FA patient–like phenotypes. In line with this possibility, SETD2 expression is decreased in BM cells from patients with FA compared with healthy donors (Figure 1K). Further studies may help answer this question.

In addition to R-loop–associated genomic instability, multiple mechanisms have been proposed to contribute to the pathogenesis of BMF in FA (70, 71). Defective DNA interstrand cross-link repair is a hallmark of FA and leads to persistent DNA damage responses, chromosomal instability, and p53-mediated apoptosis in HSPCs (72). Chronic activation of inflammatory cytokines such as TNF-α and IFN-γ further exacerbates stem cell loss by inducing oxidative and replicative stress (73). Mitochondrial dysfunction and elevated ROS production have also been implicated in the progressive depletion of the HSPC pool (74). Moreover, studies have shown that replication stress and impaired recovery from stalled replication forks promote premature senescence and exhaustion of FA HSCs (75). Together, these findings suggest that FA-associated BMF results from a combination of DNA repair deficiency, inflammatory signaling, and metabolic stress. Our results place R-loop accumulation within this multifactorial context, indicating that targeting transcription-replication conflicts may complement existing strategies to mitigate hematopoietic failure in FA (76).

Our study of the cellular mechanisms involved in BMF in Setd2^+/– Fanca^–/– mice has unveiled an apparent role for defects in the cell cycle and mitosis in FA-associated BMF. HSCs from these Setd2^+/– Fanca^–/– mice, aged 4 weeks, displayed limited maintenance of a quiescent G0 phase, more mitotic errors, and reduced proliferation when cultured ex vivo. Importantly, in patients with FA, BMF often occurs during the early teenage years or at an even younger age (77). The equivalent in mice is an age of approximately 3 to 4 weeks (78). At these ages in mice and humans, HSCs are still in the cycling phase and have not yet entered the quiescent stage (79). Although not all Setd2^+/– Fanca^–/– HSCs exhibit mitotic errors during cell division, the more rapid cell cycle in HSCs from younger mice increases the likelihood of mitotic errors. Importantly, the accelerated cell cycle and associated mitotic errors in HSCs from young Setd2^+/– Fanca^–/– mice, or in pediatric or adolescent patients with FA, may lead to differentiation or cell death, and may thereby likely contribute to severe BMF.

We found that Juglone, an inhibitor of transcription by RNAP I, II, and III (51), maintains HSC survival by reducing R-loops, thereby potentially resolving ribosomal biogenesis defects. Here, we utilized Juglone to manipulate R-loop levels, via effects on RNAP-dependent transcriptional elongation. Juglone is a small compound found in all parts of walnut trees and other plants in the Juglans family. In human fibroblasts, juglone decreases p53 protein levels and induces H2A.X phosphorylation and cell death because of DNA damage and inhibition of transcription (80). Juglone has previously been used as an anticancer agent, with effective concentrations typically ranging from 1 to 10 μM. In contrast, our study utilized juglone at a significantly lower concentration (100 nM) to inhibit pan-RNAPs. Notably, this dose is approximately one-tenth of the concentration commonly employed for anticancer activity. In short-term in vitro culture, treatment with 100 nM juglone maintained the survival of control mouse HSCs and improved the survival of Setd2^+/– Fanca^–/– HSCs (Figure 5, H and F). Furthermore, in long-term experiments, particularly after BM transplantation, juglone treatment rescued the chimerism of Setd2^+/– Fanca^–/– BMCs and also reduced the chimerism of WT BMCs at 1 month after transplant (Figure 6J). To advance this discovery toward clinical application, further studies are required to assess the potential toxicity of juglone under various experimental conditions. The single cell–based cell division assay showed that Juglone treatment slows down the first cell division (Figure 6C). Furthermore, the CytoTell-based cell division history assay indicated that Juglone treatment reduces the number of cells with a medium division history and increases the number of cells with a low division history (Figure 6, D–G). These results suggest that Juglone treatment slows cell division, allowing HSCs sufficient time to prepare for the next cell division and to resolve R-loop and dsRNA accumulation. To clarify the effects of Juglone on mitotic errors in Setd2^+/– Fanca^–/– HSCs, a liquid culture assay was performed. Juglone-treated WT control HSCs did not show any notable difference compared with untreated controls; however, in Setd2^+/– Fanca^–/– HSCs, Juglone treatment significantly reduced mitotic errors (Figure 6H). Although Juglone has beneficial effects in this context, it is important to note that Juglone is known to be a cytotoxic compound due to its redox-cycling activity and generation of ROS (81). Excessive ROS production can lead to oxidative stress, DNA damage, and cell death, particularly at higher concentrations or with prolonged exposure. Therefore, the dose and exposure duration of Juglone must be carefully optimized to minimize cytotoxicity while preserving its protective effects on HSCs. Nevertheless, in this context, it should be noted that doses of Juglone that increased BMC and HSC numbers in vivo appear to have been tolerated in mice (Figure 6, L–N).

To explore potential therapies for BMF in FA, we hypothesized that reduction of R-loop accumulation could restore stemness in Setd2^+/– Fanca^–/– HSCs. However, ectopic expression of RNASEH1, DICER, and SENATAXIN on Setd2^+/– Fanca^–/– HSPCs did not rescue HSPC function, indicating that simply reducing R-loop and dsRNA levels is not by itself sufficient to rescue the function of Setd2^+/– Fanca^–/– HSPCs (Supplemental Figure 3B). Indeed, treatments with L-leucine, low doses of the RNAPI inhibitor CX5461, and the pan-RNAP inhibitor Juglone all significantly reduced R-loop and dsRNA levels (Figure 5, A and B). However, only Juglone also improved cell survival in Setd2^+/– Fanca^–/– HSCs (Figure 5, F–H), suggesting that a balanced inhibition of RNAPs may be necessary to restore HSC function. Thus, modulating the activity of RNAPs could be a promising therapeutic strategy for patients with FA. Importantly, low HSC numbers and HSC defects may be a barrier to autologous gene therapy for patients with FA, and although allogeneic HSC transplant is the frontline treatment for FA-related BMF, it also poses significant risks. Our study suggests that oral drug therapies targeting RNAPs, which thereby reduce R-loop/dsRNA accumulation, may ultimately relieve ribosomal biogenesis defects and make autologous transplants a safer alternative. The promising results from our in vitro treatments of HSCs with Juglone warrant further preclinical/clinical investigation and validation, including in vivo treatments in mice. A therapeutic strategy to prevent or cure BMF in FA, based on pan-inhibition of RNAPs with a drug such as Juglone, via rescue of HSC defects, may also potentially apply to other BMF conditions. Even more broadly, this approach may also help maintain HSCs ex vivo or facilitate HSC editing for many other applications.

Sex as a biological variable. Sex was not considered as a biological variable in this study.

Animals. We bred Fanca^+/– and Setd2^+/– mice (47, 82) to generate Setd2^+/– Fanca^–/– offspring on a C57BL/6 background where Fanca-KO mice develop reduced numbers of HSCs and craniofacial defects (83). All animal experiments were performed according to the guidelines of the IACUC.

Cell preparation. Whole BM cells were obtained by flushing the femur and tibia bones with DMEM (Sigma-Aldrich) containing 10% FBS (Biowest). The collected BM cells were then washed once with Dulbecco’s PBS (Sigma-Aldrich) containing 2% FBS. For HSC sorting, BM cells were first stained with a microbead-conjugated antibody for c-Kit (Miltenyi Biotec), and c-Kit⁺ cells were initially separated using an autoMACS Pro Separator (Miltenyi Biotec) before further staining with fluorescence-conjugated antibodies, lineages CD4 (RM4-5, BioLegend) CD8 (53-6.7, BioLegend), CD3 (17A2, BioLegend), CD11b (M1/70, BioLegend), Gr1 (RB6-8C5, BD Biosciences), B220 (RA3-6B2, BioLegend), and TER119 (TER-119, BD Pharmingen), along with CD117 (2B8, BioLegend), EPCR (eBio1560, eBioscience), CD48 (HM48-1, BioLegend), and CD150 (TC15-12F12.2, BioLegend).

Cell culture. Sorted E-SLAM HSCs were cultured with StemSpan (STEMCELL Technologies) treated with mouse thrombopoietin (100 ng/mL, PeproTech) and stem cell factor (100 ng/mL, PeproTech). For MMC treatment, E-SLAM HSCs were treated with different concentrations of MMC (0.01, 0.1, 1, 10 ng/mL) and cultured for 2 days. After 2 days of culture, HSC numbers were counted using Celigo (Revvity). For cell division assays, sorted E-SLAM HSCs were stained with CytoTell-green (AAT Bioquest) and cultured for multiple days, varying for each experiment.

Single-cell paired-daughter assay. Individual E-SLAM HSCs were initially cultured in StemSpan medium supplemented with murine stem cell factor and thrombopoietin (each at 100 ng /mL) for 48 hours. Subsequently, cells were transferred to IMDM containing 10% FBS along with murine SCF, TPO, G-CSF, EPO, and IL-3. After 14 days of culture, individual clones derived from LSK-SLAM cells were harvested and analyzed for differentiation into myeloid lineages, including neutrophils, erythrocytes, macrophages, and megakaryocytes.

Transient transfections and lentiviral transductions. Lentiviruses were generated by cotransfecting HEK293T cells with lentiviral vector plasmids (pLenti-IRES-eGFP, pLenti-RNASEH1-IRES-eGFP, or pLenti-DICER-IRES-eGFP) and packaging plasmids using calcium phosphate and collected virus at a 24-hour interval. HEK293T cells were maintained in DMEM with 10% FBS in a humidified incubator at 37°C and 5% CO₂. c-Kit⁺ (CD117) cells collected from control or Setd2^+/– Fanca^–/– mouse BM were transduced twice with collected lentiviruses using RetroNectin (Takara Bio).

Antibodies for flow cytometry. The following monoclonal antibodies were used as surface markers for cell sorting and flow cytometric analyses: anti-c-Kit (2B8), anti-CD150 (TC15-12F12.2), anti-CD48 (HM48-1), anti-EPCR (eBio1560; eBioscience), anti–Sca-1 (E13-161.7), anti-CD45.2 (104), anti-CD45.1 (A20), anti-B220/CD45R (RA3-6B2), anti–Mac-1 (M1/70), anti-Gr-1 (RB6-8C5), anti-CD4 (RM4-5), and anti-CD8 (53-6.72) antibodies. All antibodies were obtained from BioLegend unless otherwise noted. For R-loop and dsRNA detection, we utilized anti-DNA-RNA hybrid antibody (S9.6) (Kerafest) and dsRNA antibody (J2) (Cell Signaling Technology). Flow cytometric analysis and cell sorting were performed by using FACSCelesta (Becton Dickinson), FACSAria III (Becton Dickinson), and FACSMelody (Becton Dickinson), respectively. Obtained data were analyzed using FlowJo software (v. 10. 6. 1).

Transplantation. As described previously (57), C57BL/6 Ly5.1 recipient mice were lethally irradiated (total 9.5 Gy). Within 24 hours after irradiation, 1 × 10⁶ whole BM cells (test cells; Ly5.2) were transplanted with 1 × 10⁶ or 1 × 10⁵ whole BM cells (competitor cells; Ly5.1). At 1, 3, and 5 months after the transplantation, the chimerism in peripheral blood was analyzed using flow-cytometric analyses. For secondary transplantation, 5 months after the first transplantation, BM cells were harvested, and 1 × 10⁶ BM cells were transplanted to lethally irradiated (total 9.5 Gy) C57BL/6 Ly5.1 recipient mice. Five months after the secondary transplantation, peripheral blood chimerism was analyzed using flow-cytometric analysis.

Confocal microscopy. FA patient–derived cells were seeded into a chamber plate (Nunc). For HSC seeding, we used fibronectin coating. After cells were attached to the bottom of the chamber plate, they were fixed with 2% paraformaldehyde (Thermo Fisher Scientific) for 10 minutes and permeabilized with 0.1% Triton X-100 (Thermo Fisher Scientific) for 15 minutes, both at room temperature. After permeabilization, samples were blocked with 3% BSA in Tris-buffered saline with Tween 20 (TBST; Sigma-Aldrich) for 30 minutes. Then, samples were stained with primary and secondary antibodies in 3% BSA in TBST for 30 minutes at room temperature. Primary antibodies included S9.6 (ENH001, Kerafast), J2 (76651, Cell Signaling Technology), and Ubf1 (ab244287, Abcam). After secondary antibody staining, samples were incubated with DAPI (Invitrogen) for 10 minutes in TBST and mounted with ProLong Diamond Antifade Mountant (Invitrogen). After enclosing with cover glasses, samples were observed with an ECHO spinning disk confocal microscope.

In vivo treatment with Juglone. Juglone was diluted in a PEG-based vehicle. Mice were i.p. administered Juglone at a dose of 100 μg/kg body weight in a total volume of 20 μL per mouse.

qPCR for ribosomal genes. For mouse HSCs, total RNA was isolated using the RNeasy micro kit (QIAGEN) and converted to cDNA using SuperScript VILO (Invitrogen). The cDNA was amplified using an Applied Biosystems Step One Plus thermal cycler (Applied Biosystems). Specific probes for each gene were used and are listed in Table 1.

Table 1

Probes for ribosomal gene detection

Statistics. Statistical analyses were performed using GraphPad Prism version 8.4.3. Data are presented as mean ± SEM unless otherwise indicated. Statistical significance was determined using a 1-tailed Student’s t test for comparisons between 2 groups or 1-way ANOVA with appropriate post hoc tests for multiple comparisons. A P value of less than 0.05 was considered statistically significant.

Study approval. All animal studies were conducted according to an approved IACUC protocol (20210066AR) and federal regulations at UT Health San Antonio.

Data availability. Values for all data points found in graphs are in the Supporting Data Values file.