Lipid metabolism transcriptomic signature of defective erythropoiesis in Diamond Blackfan anemia syndrome

Key Points

• •The variable anemia in Rpl5^Skax23-Jus/+ mice is triggered by intrinsic/extrinsic stress.
• •Rpl5 haploinsufficient murine and human erythroid progenitors exhibit a lipid metabolism signature with downregulation of Scd1/SCD.

Visual Abstract

Abstract

Analysis of Diamond Blackfan anemia syndrome (DBAS) cohorts and animal models has not revealed a potential mechanism for the variable anemia phenotype, a key feature of this disease. Here, we used an established Rpl5^Skax23-Jus/+ murine DBAS model to study this dynamic erythropoiesis deficiency. These haploinsufficient mice exhibit variably penetrant craniofacial and cardiac defects mimicking the phenotypes of patients with DBAS bearing RPL5 mutations. We additionally discovered that this specific heterozygous splicing mutation is pathogenic and leads to partial intron retention. By examining the transcriptome of fetal liver erythroid progenitors at embryonic day 12.5 (E12.5), we demonstrate downregulation of erythroid differentiation pathways consistent with the DBAS phenotype. We also identified dysregulated transcription of lipid metabolism genes with significant reduction in Scd1 expression in the subset of E12.5 mutant embryos at risk for erythroid failure. Stearoyl–coenzyme A desaturase 1, a key enzyme that converts saturated to monounsaturated fatty acids, has not been previously linked to erythropoiesis or DBAS. When anemia was induced in adult Rpl5^Skax23-Jus/+ mice, mutant mice exhibited delayed erythroid recovery, whereas pretreatment with an stearoyl–coenzyme A desaturase 1 inhibitor resulted in improved erythropoiesis in both wild-type and mutant mice. This analysis suggests a potential role of lipid metabolism in the variable anemia penetrance in DBAS and highlights a previously unappreciated pathway that requires further study as a potential target for drug development.

Introduction

Diamond Blackfan anemia syndrome (DBAS) is a heterogeneous genetic disorder mainly caused by heterozygous ribosomal protein (RP) mutations.^1,2 This disorder was previously diagnosed in individuals with the classical presentation of macrocytic anemia in infancy, but with increasing recognition of variable phenotypes, the term DBAS was more recently adopted. In addition to a defect in erythropoiesis, a subset of patients will present with defects in other hematopoietic lineages, leading to thrombocytopenia, neutropenia, or pancytopenia.³ Some patients have birth defects, and there is an increased risk of developing cancer.⁴ Developmental defects such as craniofacial malformations and cardiac abnormalities occur more frequently in RPL variants, especially RPL5 (uL18) and RPL11 (uL5), for unclear reasons.^2,5,6 Treatment-independence, previously termed remission, occurs in ∼20% of individuals who previously required steroids or red cell transfusions. In some patients, the period of treatment independence can be long, lasting for many years. Other patients develop periods of anemia cycling with treatment independence but remain at risk of developing malignancy.⁷ The underlying mechanisms of this phenomenon remain unknown. We focused our studies on RPL5, because individuals with this genotype have very severe disease manifestations, specifically a higher rate of congenital anomalies and a lower chance of developing treatment independence.² In the Ulirsch et al² study, only 5% of RPL5 individuals attained treatment independence, compared to 8% to 36% in the other genotypes.

The initial discovery that most individuals with DBAS have an RP mutation was intriguing and fostered increased interest in ribosome biology.⁸ A key question that emerged was whether RP haploinsufficiency leads to downstream developmental defects through reduced protein synthesis and/or possibly through extraribosomal functions of RPs.^9,10 Many animal models and clinical cohorts have been examined to determine the underlying mechanism(s) and to develop new therapies.¹¹ These studies have demonstrated the role of nucleolar stress, heme toxicity, aberrant inflammatory pathways, and reduced translation of erythroid-specific factors as major drivers of erythropoiesis failure in this disorder.¹² Patients with severe anemia require red blood cell (RBC) transfusions, but a subset of patients respond to long-term steroid treatment.⁷ The finding of nonribosomal mutations such as GATA1 and reduced GATA1 translation in DBAS has provided new avenues for broad therapies for patients, such as gene therapy.13, 14, 15 However, despite tremendous progress, the mechanism for variable anemia among individuals with an identical RP genotype or within the same family remains unclear.

In our prior studies, we found a transient erythroid differentiation block at the colony-forming unit-erythroid (CFU–E)/proerythroblast progenitor stage in the embryonic day 12.5 (E12.5) fetal livers (FL) of Rpl5^Skax23-Jus/+ mice, which began to resolve by E14.5.¹⁶ Newborn pups had macrocytic anemia and increased mortality, but mice that survived weaning had no anemia or evidence of a hematopoietic stem/progenitor cell defect. We also demonstrated the variable presence of a ventricular septal defect (VSD) and a kinked tail phenotype with poor growth and increased mortality. Here, to investigate the mechanism that led to the erythroid differentiation defect, we examined in detail FL erythroid progenitors using an unbiased whole-transcriptome analytical approach. This strategy revealed dysregulation of lipid metabolism genes including downregulation of Scd1 in mutant mice. Treatment of adult Rpl5 mutant mice with a known stearoyl–coenzyme A desaturase 1 (SCD1) inhibitor (SCD1-i) did not induce anemia. Rather, SCD1 inhibition appeared to have a protective effect when anemia was induced with the hemolytic agent phenylhydrazine. We therefore propose modulation of lipid metabolism and/or SCD1 as a possible underlying mechanism leading to variable anemia penetrance in DBAS.

Methods

Mice

We used Rpl5^Skax23-Jus/+ mice (mouse genome informatics; 3046775), which we previously characterized.¹⁶ For simplicity, we will refer to Rpl5^Skax23-Jus/+ mice as Rpl5^+/−. Timed matings were performed to obtain E12.5 FL cells and embryo sections were analyzed for the presence of VSD or craniofacial malformation as previously described.¹⁶ CD71⁺/Ter119⁻ cells were sorted to obtain early erythroid progenitor cells (BD FACS ARIA III or Sony MA900 Cell Sorter). Total RNA was extracted using RNeasy Micro Kit (Qiagen; catalog no. 74004) and used for bulk RNA sequencing (RNA-seq) analysis. Analysis was performed using DESeq2 and gene ontology (GO) enrichment analysis. Additional details are provided in the supplemental Methods, supplemental Tables 1 and 2.

RNA-seq data processing

Both library preparations, next-generation sequencing (NGS) runs, and data preprocessing were conducted by the University of Michigan Advanced Genomics Core. Paired-end 151–base pair raw reads retrieved from the experiments were trimmed using Cutadapt v2.3¹⁷ and mapped to the GRCm38 assembly (ENSEMBLE) using STAR v2.7.8a.¹⁸ Gene counts from the alignments were estimated using RSEM v1.3.3.¹⁹ Alignment options followed ENCODE standards for RNA-seq. Quality controls were reported by FastQC v0.11.8,²⁰ Fastq Screen v0.14,²¹ and Multiqc v1.7.²² Gene information was reannotated with AnnotationDbi²³ and org.Mm.eg.db to retrieve the Entrez Gene identifiers (IDs) and gene names based on their ENSEMBL IDs.

Statistics

GraphPad Prism 10 was used to graph data and conduct the Student unpaired t test to determine statistical significance. Error bars represent the standard deviation from the mean (∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001 for all figures).

Institutional Animal Care and Use Committee approval was obtained for this study.

Results

Rpl5^+/− mice have evidence of a cleft palate

We previously described Rpl5^+/− mice, with a heterozygous intronic Rpl5 mutation that leads to Rpl5 haploinsufficiency.¹⁶ Mutant mice are small, with partially penetrant kinked tails, macrocytic anemia, and a VSD but no craniofacial defect at birth. The severe phenotype in this model aligned closely with human patients with DBAS with RPL5 mutations.² However, it was unclear whether these mutant mice developed craniofacial defects, which is a major phenotype of RPL5 mutations in humans.⁵ Further embryo characterization at E15.5 revealed that 20% of Rpl5^+/− mice had a VSD (supplemental Table 3) and 30% of Rpl5^+/− mice showed lack of palatal fusion (Figure 1A; supplemental Table 3). The lack of cleft palate and/or VSD in adult animals indicates that severe morphological defects might be associated with early embryonic/perinatal mortality. Additional histological observation of the facial structures by hematoxylin/Alcian blue was performed at E14.5, which again demonstrated defects in palate elevation and tongue descent (Figure 1B). We performed staining for proliferation (pH3) and cell death (TUNEL) at E14.5 (Figure 1C). The mutant mice had significantly higher pH3⁺ cells but a trend toward lower numbers of TUNEL-positive cells compared to wild type (WT; Figure 1D). Analysis of E12.5 embryos showed a decreased nucleic p53 foci in the first branchial arch in Rpl5^+/− mice (Figure 1E-F). This indicates dysregulation of normal patterns of cell growth and p53-mediated apoptosis during craniofacial embryogenesis caused by Rpl5 haploinsufficiency.

Figure 1.

Rpl5^+/− mice have a cleft palate during embryonic development. (A) Histological characterization of 4 litters at E15.5 (WT, n = 11; Rpl5^+/−, n = 13), with representative histology sections from 4 mutant mice. Mutant “3-3” had no cleft palate, whereas the other mutants demonstrated lack of palatal elevation and/or fusion (scale bar, 300 μm). (B) Analysis of E14.5 embryos (WT, n = 3; Rpl5^+/−, n = 3) after Alcian blue staining showed a delay in normal palate elevation in mutant mice. (C) Further staining (WT, n = 4; Rpl5^+/−, n = 5) indicated an increase in proliferation (pH3) and a decrease in apoptosis (TUNEL) in Rpl5^+/− cells (2 representative images presented; scale bars, 500 μm and 100 μm [enlarged area]). (D) Quantification of PH3 and TUNEL in palatal shelf, tongue, and Meckel cartilage. (E) Analysis of E12.5 embryos (WT, n = 3; Rpl5^+/−, n = 3) after p53 staining showed a decreased nucleic p53 foci in the first branchial arch in Rpl5^+/− mice. (F) Quantification of nucleic p53 in the first branchial arch. DAPI, 4′,6-diamidino-2-phenylindole.

The c.3+6T>C intronic mutation results in alternative splicing

Our prior work demonstrated that the ENU-induced Rpl5 mutation (c.3+6T>C) was at a highly conserved intronic site and resulted in Rpl5 haploinsufficiency.¹⁶ Concordantly, RNA-seq of E12.5 CD71⁺ Ter119⁻ FL cells in this work also showed a shift away from the canonical isoform in mutant cells (supplemental Figure 1A). To test this splicing disruption as the underlying mechanism, we performed a minigene assay using the WT and mutant alleles of the Rpl5 first exon. After transfection in HEK-293T cells and reverse transcription polymerase chain reaction, we analyzed the spliced minigene transcript product by gel and NGS (supplemental Figure 1B-C). In the WT construct, the splice donor immediately following the start codon was predominantly used, resulting in most in-frame transcripts (∼92%). By contrast, in the mutant construct, use of this donor was almost entirely abolished (∼0%).

E12.5 Rpl5^+/− FL erythroid progenitors have reduced erythroid differentiation and lipid metabolism gene expression

In these mice, we previously reported a severe erythroid differentiation block at E12.5. To further investigate the cellular mechanism of this defect, we sorted E12.5 erythroid progenitors in the CD71⁺ Ter119⁻ gate and performed mRNA-seq analysis (Figure 2A). Due to the extremely low cellularity of this progenitor cell population in mutant animals, we used pooled embryos to generate WT and mutant samples. The differential expression (DE) analysis identified 220 highly variable genes with adjusted P values ≤.005 and absolute log₂ fold changes (LFC) ≥0.99 (Figure 2B), among which 142 of the genes were downregulated and 78 were upregulated in the mutant compared to WT mice (Figure 2C). GO enrichment analysis highlighted 271 significant terms from all variable genes (q ≤ 0.05) and 52 from the downregulated ones (supplemental Figure 2). Based on these top significant terms and our analyses of previously published human data²⁴ (supplemental Figures 3 and 4), we selected erythrocyte-, lipid-, and oxidative stress–related terms for further investigation. Nine erythrocyte- and lipid-related terms demonstrated significant enrichment among all significant and downregulated genes (Figure 2D). Gene set enrichment analysis using all expressed genes further supports the observations related to erythropoiesis (Figure 2E). We found 7 downregulated genes involved in erythrocyte development, differentiation, and homeostasis networks, consistent with the erythroid differentiation defect we previously described (Figure 2F). The DBAS gene Ada was also significantly upregulated but fell below our LFC cutoff (LFC, 0.76; P = 7 × 10⁻⁶). We also found 2 upregulated and 6 downregulated genes involved in several lipid metabolism networks, including the neutral lipid, glycerolipid, triglyceride, and acylglycerol biosynthetic/biosynthetic processes (Figure 2G).

Figure 2.

Transcriptomic analysis of E12.5 FL erythroid progenitors. (A) We sorted E12.5 FL and performed bulk RNA-seq using mRNA from CD71⁺ Ter119⁻ cells (WT, n = 3; Rpl5^+/−, n = 3 using pooled embryos [II]). (B) We performed differential gene expression analysis and data visualization via heat map with the variance stabilizing transformation log-scaled expressions for all the significant genes. The rows and columns of the heat map are automatically clustered. The corresponding LFC is also visualized along with the main heat map. Genes with top LFC (>3.5 or less than −2) are highlighted. (C) Volcano plot of gene expression with thresholds (dashed lines) set at adjusted P value < .005 and absolute LFC ≥0.99 (genes colored red are significant for P value and LFC; genes colored blue are significant for P value; and genes colored green are significant for LFC). Names of the top 35 differentially expressed genes are presented. (D) GO enrichment analysis results for all significant and downregulated genes with terms related to erythrocyte, lipid, and oxidative stress. The dot size reflects the gene numbers, and colors represent the q-values of these terms. Triangle dots indicate no significance (false discovery rate [FDR] q > 0.05). (E) The gene set enrichment analysis (GSEA) results of the erythrocyte-related GO terms. (F-G) Gene interaction networks related to erythrocyte and lipid regulation. The dot colors indicate the fold changes of the gene expression (blue, negative; red, positive). NS, not significant; NES, normalized enrichment score. Figure created with biorender.com. Min J. (2025) https://biorender.com/5mvwxqc.

Scd1 is significantly downregulated in Rpl5^+/− erythroid progenitors

We also performed RNA-seq using FL erythroid progenitors from single E12.5 embryos, comparing mutant and WT mice. We separated mutant samples into 2 groups based on the number of erythroid cells recovered, under the hypothesis that mutants with significantly lower cellularity in the erythroid progenitor gate (M-low) were likely to succumb to erythroid failure compared to mutants with higher cell counts (M-high) (Figure 3A). M-low embryos had lower numbers of cells in both quadrants II and III. To investigate the potential pathways involved in driving the cellularity differences between mutant groups, we performed DE analysis among 4 types of comparisons: (1) M-low + M-high and WT (M-all vs WT); (2) M-high and WT; (3) M-low and WT; and (4) M-low and M-high. We identified 390, 231, and 407 DE genes in the WT comparisons (Figure 3B), with no significantly dysregulated noncoding RNAs. They were enriched in GO terms relating to erythrocyte development and oxidative stress (supplemental Figures 5 and 6). Five genes, including Scd1, were identified from the M-high vs M-low comparison (Figure 3C-D). Notably, Scd1 was consistently downregulated across all comparisons, including the prior RNA-seq experiment presented in Figure 2 (Figure 3E). The same downregulation of SCD (human homolog) was also observed in the previously published human data²⁴ (LFC of less than −3.9; adjusted P < 1e-5; supplemental Figure 7), further suggesting potential involvement of Scd1/SCD in DBAS.

Figure 3.

Mutant E12.5 FL with lower cellularity shows significant downregulation of the lipid metabolism gene Scd1. We performed bulk RNA-seq using total RNA from sorted population II (CD71⁺ Ter119⁻) cells from E12.5 FL using single embryos (WT, n = 3; Rpl5^+/−, n = 6). Mutants were analyzed together and separately using differences in FL erythroid cell counts obtained from sort (M-low, n = 3; M-high, n = 3). (A) M-low embryos had significantly less erythroid progenitor cells in populations II (CD71⁺Ter119⁻) and III (CD71⁺Ter119⁺) compared with M-high. (B) The heat map visualizes the variance stabilizing transformation log-scaled expressions for all the significant genes unioned from the 4 types of comparisons (M-high vs WT, M-low vs WT, M-all vs WT, and M-low vs M-high). Heat map rows are automatically clustered when the columns are manually sorted. Corresponding LFC for the 4 comparisons are also visualized. Genes with top LFCs (less than or equal to −2.5 or ≥5) from the 4 comparisons are labeled. The shapes after the gene names indicate which comparisons they show significant differences in and whether they have top LFCs. Scd1 is highlighted because it is the only gene that shows significance across all four comparisons. (C) Volcano and (D) summary plots showing the comparisons between mutants and wild-types, highlighting the positions of the 5 significant genes in the M-low vs M-high comparison. (E) Summary of common genes involved in the significant GO terms across all comparisons of the 2 RNA-seq experiments. Circles indicate significance (FDR q < 0.05). Triangles indicate no significance (FDR q > 0.05). (F) Validation of significant genes in population II by reverse transcription quantitative polymerase chain reaction. Abs., absolute.

We validated dysregulated genes by reverse transcription quantitative polymerase chain reaction and demonstrated Scd1 downregulation in E12.5 FL mutant erythroid progenitors (Figure 3F). Patients with DBAS have elevated fetal hemoglobin, thought to be a marker of stress erythropoiesis. We found that Hbb-bh1, which uses one of the β-like embryonic chains was significantly more elevated in M-low than M-high (Figure 3F). This suggests that the elevation of Hbb-bh1 is a signature of stress erythropoiesis in M-low mice, which we hypothesize are the group likely to have failure of erythropoiesis. SCD1 is known to catalyze the conversion of saturated to monounsaturated fatty acids.²⁵ However, to our knowledge, the role of SCD1 and lipid metabolism in erythropoiesis and DBAS has not been previously described.

Rpl5^+/− adult mice have delayed erythroid recovery after stress

Our prior work in Rpl5^+/− mice demonstrated that postnatal macrocytic anemia leads to early mortality after birth in some mice, but surviving adult/aged mice exhibited no anemia or bone marrow failure.¹⁶ Before further analysis of adult mice, we analyzed the bone marrow by NGS and found no differences in mutant/WT ratios (ie, mutant mice had ∼50% of the WT sequence compared to 100% in WT mice), indicating that there were no Rpl5 clonal escape mutations or uniparental disomy as reported in some cases of DBAS and other bone marrow failure conditions.26, 27, 28, 29 We hypothesized that the lack of anemia in adult animals might be partly due to the lack of hematological challenges (ie, absence of infections/inflammation and/or environmental/dietary toxins) in normal mouse husbandry conditions, as seen in Fanconi mouse models, which do not develop bone marrow failure in the absence of stress.^30,31 To test this hypothesis, we first induced inflammatory stress with polyinosinic:polycytidylic acid (poly(I:C)), a synthetic double-stranded RNA that simulates a viral infection (supplemental Figure 8A). We found that poly(I:C) induced significant anemia in both WT and Rpl5^+/− mice by day 7. However, WT mice started to recover after day 7 despite being administered additional doses of poly(I:C), whereas Rpl5^+/− mice slowly recovered to baseline by day 28 (supplemental Figure 8B-E). We next tested a standard hematological stress condition in adult mice by injection of phenylhydrazine (Phz), which causes hemolysis (supplemental methods; Figure 4A). After Phz administration, mutant mice showed an enhanced nadir and delayed recovery of RBC counts after the first and second treatments, indicating that erythropoiesis in adult mutant mice was more severely affected by this hematological challenge (Figure 4B; supplemental Figure 9). In this experiment, 3 of 4 male Rpl5^+/− mice treated with Phz died, whereas there were no deaths in other groups (supplemental Figure 10).

Figure 4.

Rpl5^+/− mice show delayed erythroid recovery after treatment with Phz. (A) Adult (aged 1.5-3.5 months) mice were injected with Phz (60 μg/g IP) or PBS on days 0, 2, 21, and 23, and weekly blood count was obtained until recovery (n = 8 for each group; 50/50 male-to-female ratio). (B) Rpl5^+/− Phz-treated mice had a more significant macrocytic anemia with lower RBC after both Phz treatments compared with WT. WT and Rpl5^+/− both showed recovery by days 21 and 49; however, Rpl5^+/− mice experienced slower recovery of RBC counts after Phz treatment, and death occurred in 3 mutant mice. (C-D) Analysis of erythropoiesis in the spleen (C) using CD71 and Ter119 on day 4 with quantification (D; n = 6 for each group; 50/50 M:F ratio). (E-F) Analysis of CFU-E and pre–CFU-E in the spleen (E) on day 4 with quantification (F). IP, intraperitoneal; PBS, phosphate-buffered saline.

In mice, stress erythropoiesis occurs predominantly in the spleen, which rapidly induces erythroid output in response to anemic stress.³² We initially performed an analysis of peripheral blood counts on days 3 to 5 after Phz administration and found that the RBC nadir occurred on day 4 (supplemental Figure 11). When we examined the spleen erythroblasts at day 4, CD71⁺Ter119⁺ cells in the spleen were reduced significantly in mutant mice (Figure 4C-D). We represented these data using total cells, absolute cells per gram body weight, and percentage of live cells (Figure 4D; supplemental Figure 12A) and showed a statistically significant decline in this progenitor population in all 3 analyses. There was an increase in the percentage of live CD71⁻ Ter119⁻ population (population I) in Rpl5^+/− mice, but this was not statistically significant with absolute cell numbers. Bone marrow erythroid progenitors showed no change after treatment with Phz (supplemental Figure 12B). Next, we quantified early and late erythroid progenitors in the spleen after treatment with Phz (supplemental Figure 13A). In WT mice, the numbers of pre-megakaryocyte/erythroid progenitor, pre–CFU-E, and CFU-E progenitors were elevated as early as 4 days after initial treatment with Phz (Figure 4E-F; supplemental Figure 13B-E), demonstrating efficient regeneration response. Although pre-megakaryocyte/erythroid progenitor and pre–CFU-E numbers were not significantly impaired in Rpl5^+/− adult spleen compared to WT spleens after Phz administration (Figure 4E-F; supplemental Figure 13C-D), the CFU-E population was profoundly diminished in mutant mice (Figure 4E-F; supplemental Figure 13B). This is the likely cause of delayed production of CD71⁺Ter119⁺ stress erythroid progenitors and delayed mature RBC production in the circulation after treatment with poly(I:C) or Phz in mutant mice.

An SCD1-i improves erythropoiesis in adult mice after anemic stress

We next addressed whether the aberrant pathways described in E12.5 FL mutant erythroid progenitors also played a role in adult stress erythropoiesis in our model. The specific question was whether Scd1 downregulation drives the failure of erythropoiesis or whether Scd1 downregulation is a compensatory mechanism to augment erythropoiesis. We pretreated mice with an available SCD1-i (5 mg/kg; supplemental Methods) or vehicle control (dimethyl sulfoxide) via oral gavage for 14 days and then administered Phz (Figure 5A). The mice showed some weight loss but no change in peripheral blood counts with SCD1-i pretreatment (day −14 to 0; supplemental Figure 14A-B). There was a slight increase in bone marrow cellularity in WT mice treated with SCD1-i, but spleen cellularity was similar in all groups at day 7 (supplemental Figure 14C). Mutant mice treated with vehicle control showed a significant decrease in hemoglobin and RBC compared with WT, whereas SCD1-i–treated mice had no or significantly less differences in hemoglobin and RBC, respectively (Figure 5B). However, we terminated analysis after day 7 due to significant ocular toxicity from the drug.

Figure 5.

Treatment with SCD1 inhibitor improves erythropoiesis. (A) We pretreated adult mice (female, n = 6 per group; aged 2-4 months) with SCD1-i (CAY10566) or dimethyl sulfoxide (DMSO) for 2 weeks (days, −14 to 0) and then administered Phz on days 0 and 2. (B) Analysis of blood counts at day 7 showed a slight improvement in RBC counts in mutant mice after SCD1-i treatment. (C-D) Analysis of hematopoietic stem/progenitor cell subsets by flow cytometry at day 4 showed a significant increase in CFU-E and a decrease in pre–CFU-E progenitors in WT mice. (E-F) Analysis of erythroid precursors at day 4 by CD71 and Ter119 showed a decrease in earlier precursors (II) and an increase in late precursors (IV) in WT mice. (G-H) Further quantification of terminal erythropoiesis (Ter119⁺ gate) using CD44 vs FSC shows significant improvement in erythropoiesis in WT animals after SCD1-i treatment. BM, bone marrow; SP, spleen.

To explore the effect of SCD1 inhibition on erythropoiesis, we analyzed hematopoietic stem/progenitor cells by flow cytometry. WT mice treated with SCD1-i showed a significant increase in CFU-E number in the bone marrow and decreased pre–CFU-E in both the bone marrow and spleen at day 4, compared to WT mice treated with dimethyl sulfoxide (Figure 5C-D). There was no corresponding change in Rpl5^+/− mice. Similarly, when we analyzed later erythroid progenitors, we saw improvement in erythropoiesis in WT mice treated with SCD1-i, with a significant decrease in early CD71⁺Ter119⁻ (II) progenitors and an increase in late CD71⁻Ter119⁺ (IV) progenitors in bone marrow (Figure 5E-F). We did not observe a corresponding change in spleen erythroid progenitors. Analysis of terminal erythropoiesis (CD44 vs FSC (forward scatter)) showed a significant increase in mature bone marrow erythroid cells (IV + V) and an insignificant upward trend in Rpl5^+/− mice (Figure 5G-H).

We next tested whether the lack of erythroid progenitor effect after SCD1 inhibition in mutant mice was due to drug toxicity or a lag in drug effect in mutant animals. Analysis at day 7 (Figure 6A) now showed a significant increase in spleen cellularity with the lower 1.25 mg/kg SCD1-i dose in mutant mice but not at the higher doses (2.5 and 5.0 mg/kg). There was a nonsignificant CFU-E expansion at the 1.25 mg/kg dose (Figure 6C-D). However, further analysis of erythropoiesis showed a significant increase in CD71⁺Ter119⁺ erythroblasts in mutant mice after the 1.25 mg/kg SCD1- i dose (Figure 6E-F). This indicates that erythropoiesis in Rpl5 haploinsufficient mice is responsive to SCD1 inhibition but at lower doses, with a longer response time, and may affect a different progenitor stage than WT mice. Here, we also found that WT bone marrow but not spleen showed improved erythropoiesis with SCD1-i. We think this finding indicates that high-dose SCD1 inhibition works predominantly on bone marrow erythroid progenitors in WT mice, but lower-dose SCD1 inhibition improves spleen progenitors in mutant mice, which may indicate that bone marrow and spleen erythroid progenitors are metabolically distinct. It is equally possible that the SCD1-i also has a non–cell autonomous effect on erythropoiesis and may have different effects in the bone marrow and spleen microenvironment.

Figure 6.

SCD1 plays a role in stress erythropoiesis in Rpl5^+/− mice. Adult mice were pretreated with SCD1-i (1.25 mg/kg, 2.5 mg/kg, or 5 mg/kg) for 2 weeks and were then given Phz at days 0 and 2. (A-D) Spleen cellularity was obtained at day 7 (A), and quantification of spleen progenitors was obtained by flow cytometry (B-D), which demonstrated a trend toward increased CFU-E and pre–CFU-E in mutant animals with SCD1-i at the 1.25 mg/kg dose. (E-F) Further analysis demonstrated improvement of erythropoiesis in mutant animals with an increase in CD71⁺ Ter119⁺ erythroid cells (III).

Discussion

This work has further characterized Rpl5^+/− mice, which we initially described as an accurate model of the variable erythroid phenotype in DBAS. Here, we describe additional salient features of this mouse that align with the phenotype seen in the human disorder. Patients with DBAS with RPL5 and RPL11 mutations have been noted to have a higher incidence of congenital malformations including craniofacial defects.² We did not initially detect a postnatal craniofacial defect in the mutant mice, but in subsequent analysis of embryos, we identified the presence of a subset of mice with abnormal palate development. The craniofacial defects did not always occur in the VSD setting, and surviving adult mice had no evidence of either defect. These data indicate that these morphological defects are markers of a severe disease phenotype and suggest that Rpl5 haploinsufficiency may affect neural crest cells and/or downstream progenitors in a stochastic manner.^33,34 We further found an elevation of proliferating cells and a reduction in the number of apoptotic cells in Rpl5^+/− mice compared to controls. One of the functions of RPL5 is to activate p53 functions by suppressing MDM2, an inhibitor of p53. In prior work, we demonstrated that cells haploinsufficient for Rpl5 do not show aberrant basal p53 activation and had a p53-independent cell cycle defect, consistent with a nucleolar stress model in which RPL5 plays a crucial role in the 5S ribonucleoprotein particle complex.^35,36 Thus, the elevation of proliferation and reduction of cell death in Rpl5 haploinsufficient mice are expected outcomes due to the suppression of normal p53 regulation. Retinoic acid–induced cleft palate, a major known causative reason for cleft palate, also showed an elevation of proliferating cells.³⁷ These data suggest that embryos with Rpl5 haploinsufficiency may be at risk for development of craniofacial anomalies, such as cleft palate and lip, through abnormal proliferation and cell death caused by suppression of normal p53 functions during embryogenesis. Further work is needed to determine the signaling pathways that drive the variable development defects in RPL5-haploinsufficient DBAS, as well as in other genotypes that exhibit p53 activation.

Analysis of DBAS cohorts has revealed a wide spectrum of genetic variants leading to ribosome biogenesis defects from canonical loss-of-function, missense, or splice site mutations in RPs or ribosome-associated protein genes such as HEATR3 and TSR2.² Approximately 20% of patients do not have a known RP variant. Some of these patients have been subsequently found to have noncanonical splice or deep intronic mutations, whereas others have nonribosome variants, such as in GATA1 or TP53-activating mutations, which share elements of the defective pathways seen in RP-haploinsufficient DBAS. The mutation seen in Rpl5^+/− mice is at a common splice site, which should be identified by exome sequencing. However, deep intronic mutations can affect gene function and may be missed by typical exome sequencing. Whole-genome analysis and/or functional analysis of ribosomal levels may be warranted in those cases but have limited availability to clinicians.^38,39 Others have proposed use of minigene assays to further characterize variants of unknown significance, which occur frequently in clinical genetic testing of rare diseases.⁴⁰ Here, we demonstrate that the pathogenesis of this variant can also be detected by intron retention bioinformatic analysis of RNA-seq data.

Several mechanisms have been proposed to explain the erythropoiesis defect in DBAS, which are not mutually exclusive.⁴¹ However, these mechanisms do not offer an explanation for the spontaneous occurrence of treatment-independent periods or a lack of anemia in some patients. In our model, this resolution of anemia occurred quite rapidly within the same litter during the period from E12.5 to weaning, which argues against major acquired genetic/clonal processes and environmental factors but may indicate a role of metabolic and/or epigenetic factors.⁴² Because some patients with DBAS have anemia recurrence during stressful periods such as pregnancy and viral infections,⁷ we sought to replicate this disease aspect in the current animal model. We demonstrated that adult mutant mice without anemia had reinduction of anemia with additional stress and exhibited a delay in erythroid recovery, modeled here with both poly(I:C) and Phz. This argues that although DBAS is a de novo anemia syndrome, perhaps those without current anemia (ie, treatment-independence) may still exhibit defective or delayed recovery from erythropoietic stress. Analysis of the pathways that support anemia resolution and/or that augment stress erythropoiesis are therefore critical to further understand this disease aspect, which may aid in rational therapeutic strategies.

Advances in “omics” have recently facilitated our understanding of the role of metabolic regulation in erythropoiesis. Mitochondria, amino acid, iron/heme, glycolysis, and lipid metabolism are all involved in the stepwise process of erythroid lineage commitment, differentiation, and maturation.⁴³ Among these, the regulation of lipid metabolism during erythropoiesis has not been well studied. Metabolomic analysis of cells undergoing terminal erythropoiesis revealed the key role of PHOSPHO1 in regulating phosphocholine metabolism, ATP production, and amino acid supply during erythropoiesis,⁴⁴ expanding the knowledge of the link between lipid composition and erythropoiesis. Prior work in DBAS cohorts and models have described various metabolic abnormalities. For example, metabolic fingerprints of dried blood spot samples identified increased inosine in patients with DBAS compared with controls and elevated α-tocopherol levels in treatment-independent patients with DBAS⁴⁵; metabolic profiles revealed increased amino acid metabolism or nucleotide pool depletion in different individuals with DBAS with distinct RPL9 variants.⁴⁶ However, further validations of these metabolic disturbances were lacking in those studies. Here, we identified the downregulation of Scd1 through unbiased analysis and verified the role of lipid metabolism in promoting erythropoiesis with a known SCD1-i. These findings shed light on the association between lipid metabolism and erythropoiesis.

SCD1 is a critical enzyme that desaturates saturated fatty acids to monounsaturated fatty acids, thereby maintaining homeostasis of lipid metabolism.⁴⁷ Aberrant SCD1 expression can cause lipid accumulation, promoting metabolism-related diseases such as obesity and diabetes. In addition, SCD1 plays a role in oncogenesis, and SCD1 inhibition is currently being examined in clinical trials.²⁵ Despite a large body of prior work in this area, the role of SCD1 in erythropoiesis was not previously described. Here, we found significant downregulation of Scd1 in M-low embryos, which we propose are the mutant group likely to succumb to failure of erythropoiesis. A key question here was whether Scd1 downregulation was the cause or consequence of these erythropoietic defects. By examining the effect of SCD1 inhibitor treatment in the Phz-induced anemic mouse model, we demonstrated that SCD1 inhibition promoted erythroid maturation, suggesting that downregulation of Scd1 is likely a compensatory mechanism to augment erythropoiesis. Interestingly, this effect required a lower dose and a longer response time in Rpl5-haploinsufficient mice, possibly due to the already low expression level of Scd1 in these mutants. Our future work will focus on understanding the specific role of lipid metabolism and SCD1 during normal and defective erythropoiesis in DBAS, which we expect will increase our fundamental understanding of erythropoiesis and guide therapeutic approaches.

Conflict-of-interest disclosure: J.D.E. has served as a consultant for and is a shareholder in Imago BioSciences. The remaining authors declare no competing financial interests.