INTRODUCTION
Diamond–Blackfan anemia (DBA), a rare but heterogeneous bone marrow failure syndrome primarily affecting erythroid lineage development, is manifested by either normocytic, normochromic, or macrocytic anemia with decreased number of erythroid progenitors in an otherwise normocellular bone marrow. DBA has variable phenotypic expression and penetrance1 and is frequently associated with low birth weight and growth delay. Almost half of affected individuals present a variable spectrum of congenital skeletal abnormalities, such as malformed or absent thumb, craniofacial malformations, cleft palate, as well as cardiac defects and urogenital malformations.
Most cases are linked to mutations of ribosomal protein (RP) genes, with 25% of all affected individuals harboring a mutation within the small ribosomal subunit protein 19 (RPS19).2 These mutations may disrupt protein expression or interfere with RPS19 incorporation into pre‐40S ribosomal particles and block ribosome assembly and maturation. There also exist missense mutations that are found within and around α‐helix 3—in the DBA hotspot (residues 52–67 in human ortholog).3
The homozygous disruption of Rps19 in mice resulted in embryonic lethality, while heterozygous mice displayed normal growth, organ development, and erythropoiesis.4, 5 Mice with autosomal dominant Rps19Y54N mutation displayed DBA‐like phenotype with reduced growth, low weight, and anemia due to increased apoptosis of erythroid‐biased BM progenitors in response to increased p53 signaling.6 Similarly, the dominant Rps19R62W mutation resulted in early embryonic lethality, while ubiquitous expression of this variant in adult animals resulted in growth retardation, mild anemia with reduced numbers of erythroid progenitors, and significant inhibition of terminal erythroid maturation.7 A mouse model of Rps19 messenger RNA (mRNA) downregulation developed macrocytic anemia with leukocytopenia and variable onset thrombopenia.8 The severity of the phenotype was related to the level of Rps19 downregulation leading to HSCs exhaustion and was ameliorated by simultaneous homozygous deletion of tumor suppressor Trp53.8 The dysfunction of RPS19 in human CD34+ progenitor cells, erythroid progenitors, or erythroid cell lines led to decreased proliferation and to upregulation of p53 targets (cyclin dependent kinase inhibitors CDKN1A/p21 and CDKN1B/p27), resulting in G0/G1 cell cycle arrest and apoptosis.9, 10
We investigated how the Rps19 mutation in the α‐helix 3 hotspot could affect murine development and whether it would show similarities to DBA. Indeed, the mouse Rps19R67∆ model exhibits a relatively broad spectrum of hematopoietic and non‐hematopoietic phenotypes, with an overlap with DBA, also including defects in pre‐ribosomal RNA (pre‐rRNA) processing and activation of p53.
METHODS
Mice
Rps19 transgenic mice (Rps19em1Ph) were generated by transcription activator‐like effector nuclease (TALEN) mutagenesis,5 resulting in single amino acid deletion of the highly conserved Arg 67. The TALEN strategy was designed using TAL Effector Nucleotide Targeter 2.0 (https://tale-nt.cac.cornell.edu)11 and assembled by Golden Gate Cloning12 into ELD‐KKR backbone vectors as described previously.13 TALEN mRps19 exon4L containing 17 repeat‐variable di‐residues (RVDs) was paired with TALEN mRps19 exon4R containing 18 RVDs (Figure 1A, Supporting Information S1: Supplementary Data [SD]). TALEN mRNA was generated using mMESSAGE mMACHINE T7‐Kit (Ambion), followed by polyadenylation with Poly(A) Tailing Kit (Ambion).
TALEN mRNA was microinjected into zygotes from C57Bl/6N mice, and founders were crossed with wild‐type (WT) until G2. Trp53 knockout strain (B6.129S2‐Trp53tm1Tyj/J) was kindly provided by Dr. Winfried Edelmann (Albert Einstein College of Medicine, NY).
All animal experimentation was performed in accordance with European Directive 86/609/EEC and was approved by the Czech Central Commission for Animal Welfare (project no. 40/2015).
Flow cytometry
Single‐cell suspensions of bone marrow from 6‐ to 8‐week‐old mice were obtained by flushing tibias and femurs with 1 mL ice‐cold flow cytometry buffer (FCB, 1× phosphate buffered saline [PBS] with 2 mM ethylenediaminetetraacetic acid [EDTA] and 0.5% biotin‐free bovine serum albumin [BSA]) using 22‐gauge needles. For the identification of hematopoietic progenitors, 15–25 × 106 freshly isolated bone marrow cells (equivalent of one femur) were stained in FCB for 30 min with primary antibody cocktail (see Supporting Information S1: SD). Single‐cell suspensions of splenocytes were obtained by mechanical disruption of spleens in ice‐cold FCB, and 5–10 × 106 cells were used for further analyses.
Peripheral blood samples were collected from orbital sinus during inhalational anesthesia with isoflurane. Bone marrow, spleen, and peripheral blood samples were analyzed on BD Fortessa (Becton Dickinson, NJ, USA) flow cytometer, and data analysis was performed using FlowJo software (FlowJo, LLC).
The single‐cell RNA sequencing of lineage‐negative c‐Kit+ bone marrow cells
Bone marrow cells from 10‐week‐old WT control, homozygous Rps19R67∆ and Trp53−/− mutant, and double‐homozygous Rps19R67∆/R67∆Trp53−/− mutant males were collected into ice‐cold Dulbecco's modified eagle medium (DMEM) (w/o phenol red) with 20% fetal bovine serum (FBS) and stained with anti‐CD117‐BV421, anti‐Sca1‐PE, and anti‐lineage (anti‐Ter119, anti‐Gr1, anti‐Mac1, anti‐B220, and anti‐Cd3e) antibody cocktail. Bone marrows were enriched for lineage‐negative Sca1−c‐Kit+ (LS−K) and Lin−Sca1+c‐Kit+ (LSK) stem cell populations by fluorescence‐activated cell sorting using BD Aria II SORP cell sorter. Cell suspensions with viability > 90% were processed using Chromium Next GEM Single Cell 3ʹ Kit v3.1 (dual index) library preparation kit (10x Genomics, CA, USA).
Dual index libraries were paired‐end‐sequenced (138 bp sequenced: Index1/Read1 10 + 28 bp and Index2/Read2 10 + 90 bp) at an approximate depth of 50K reads per cell using NovaSeq. 6000 instrument (Illumina, CA, USA). Demultiplexed sequencing data were aligned to the mouse genome reference GRCm39 using the STAR aligner16 and processed with CellRanger software (10x Genomics, CA, USA).
Only protein‐coding genes and long‐noncoding RNAs were selected for the reference transcriptome (Ensembl 104). The analysis, including quality control, data filtering, doublet detection, sample integration, cell clustering, cell type annotation, differential expression analysis, and data visualization, was carried out using R/Bioconductor17 (http://r-project.org/http://bioconductor.org) and Seurat packages.18 Only single cells with feature content above 600 and transcript count above 800 were included in the integrated data set for the final analysis. Single‐cell data normalization and variance stabilization were performed using the R package sctransform, which uses regularized negative binomial regression.19
Data analysis and statistics
All data included in the manuscript were obtained from at least three independent measurements and were expressed as the mean ± standard error of the mean (SEM). The data analyses were carried out in R/Rstudio20 (https://www.rstudio.com) and visualized using the R package tidyverse.21 Error bars indicate SEM (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 and N.S./ns as not significant).
For additional methods (Reverse transcription‐polymerase chain reaction [RT‐PCR], immunoblotting, mouse erythroleukemic [MEL] culture/transfection/analysis), see Supporting Information S1.
RESULTS
Recessive Rps19R67∆ mouse model of Diamond–Blackfan anemia
To study molecular aspects of human DBA pathogenesis, we targeted the Rps19 gene and generated a single amino acid deletion of Arginine 67 (Figure 1A), which manifests both hematopoietic and wide range of developmental defects without early embryonic lethality.5 This approach targeted a region frequently mutated in DBA patients at the C‐terminus of amphipathic α‐helix 3 encoded within Exon 4 and resulted in a deletion of four nucleotides and insertion of guanine (c.495_498delCCGAinsG). This mutation preserved the reading‐frame of the Rps19 gene but resulted in the removal of a single conserved Arg67 residue (p.67delR) flanking the third α‐helix of RPS19 protein, hereafter referred to as Rps19R67∆ (Figure 1A). The mutation affected rather structural properties of RPS19, as it did not significantly alter the expression levels of RPS19 protein in bone marrow and spleen that were comparable in mutants and controls (Supporting Information S2: Figure 1A).
It was previously shown that the sidechain of Arg67 is critical for maintaining a cation‐π stacking interaction with the two adjacent nucleotides of 18s rRNA molecule14 (Figure 1B) and is potentially targeted for posttranslational modification by the Prmt family of arginine methyltransferases.22, 23, 24
Mice heterozygous for Rps19R67∆ mutation were born at the expected Mendelian ratio and did not present any obvious developmental anomalies and were indistinguishable from WT controls. Rps19R67∆ heterozygote parents produced normal‐sized litters with no signs of embryonic lethality in utero. The phenotype of Rps19R67∆ homozygotes was relatively heterogeneous within a continuum ranging from severe with multiple abnormalities and reduced survival to mild with normal survival.
The lower‐than‐expected proportion of Rps19R67∆ homozygous mice surviving to weaning age indicates early lethality in both sexes (Supporting Information S2: Figure 1B). Nearly 25% of Rps19R67∆ homozygotes that survived beyond weaning age deceased before reaching 8 weeks of age, with a median survival of 20 weeks (Figure 1C). Postnatal development of Rps19R67∆/R67∆ homozygotes is marked by slower weight gain and subsequent growth delay compared to their WT and heterozygous littermates, and this was more prominent in males whose weight gain was approximately 25%–33% lower than in the control male littermates (Figure 1D).
Rps19R67∆ mutants exhibited smaller stature with discernible defects in postnatal development in both sexes (Figure 1E) and a white spot or strip in the middle of the abdomen (Figure 1F). In addition to the delayed growth and increased mortality in Rps19R67∆/R67∆ mice, both sexes exhibited a wide range of other developmental defects, including skeletal, cardiovascular, urogenital, muscular, and neurodevelopmental malformations with variable penetrance that also resembled DBA (Supporting Information S1: Supplementary Results, Supporting Information S2: Figure 1C–L).
Defective erythropoiesis and hematopoietic stem and progenitor cell commitment
The Rps19R67∆/R67∆ mice displayed a wide array of hematologic defects, including deficient erythropoiesis, which we examined in more detail. The peripheral blood from Rps19R67∆/R67∆ mice had decreased red blood cell counts, hemoglobin content, and hematocrit, which were readily observed in a subset, approximately one‐quarter of mice with a strong phenotype (red rectangles, Figure 2A, Supporting Information S3: Figure 2A). However, although many Rps19R67∆/R67∆ mice exhibit only a mild deficit in the hematopoietic phenotype, their bone marrow is often hypoplastic with significantly diminished bone marrow cellularity (Figure 2B,C), morphologically defective erythroid precursors on cytologic slides (Figure 2D,E).
Spleen weight and cellularity were also significantly reduced in the Rps19R67∆/R67∆ mice, unlike in DBA patients (Supporting Information S3: Figure 2B,C), and showed an altered morphology and cytology consistent with the decreased hematopoietic spleen activity (Supporting Information S3: Figure 2D–F).
Bone marrow analysis of Rps19R67∆/R67∆ mutants by flow cytometry using well‐established erythroid differentiation markers25 revealed a defect in erythroid cell development during the course of erythroid precursor Stages S0 to S5, associated with a significant reduction in the number of CD45+ and Ter119− cells, which likely reflects a more broad defect also in earlier precursor Stage S0 as well as later Stages S2–S5 (Figure 2F,G and Supporting Information S3: Figure 2G–I). The erythroid development in the spleen of mildly affected Rps19R67∆/R67∆ mutants is comparable to WT mice, with 40% of all cells belonging to clearly defined erythroid maturation stages, while in strongly affected Rps19R67∆/R67∆ mutants, the spleens contain very low numbers of basophilic and orthochromatophilic erythroblasts (<0.5%), suggesting rather an absence of late splenic erythropoiesis (Supporting Information S3: Figure 2J–L).
Defective hematopoietic commitment affects erythropoiesis in Rps19R67∆ homozygotes
To study how homozygous mutation in Rps19R67∆ affects early hematopoietic development, we used flow cytometry analysis of bone marrow and spleen erythropoiesis. c‐Kithigh progenitor populations in the most severely affected homozygous Rps19R67∆ mutants are reduced by ~30% and 66%, respectively, indicating a possible early hematopoietic defect (Supporting Information S3: Figure 2M,N). The relative abundance of hematopoietic stem and progenitor cell populations (Lineage‐Kit+) in the bone marrow of the average Rps19R67∆/R67∆ mice seems comparable to WT controls (Figure 2H); nevertheless, significant differences can be discerned when bone marrows of these mutants are closely examined. There is ~30% decrease in the absolute number of LS−K cells as well as in LSK cells composed of quiescent, transcriptionally, and translationally dormant cells in Rps19R67∆/R67∆ mutants, indicating a substantial reduction in differentiating progenitors (Figure 2I,J).
The LSK cells include diverse populations of hematopoietic stem cells (HSCs) with long‐term (LT‐HSCs), short‐term repopulating activity (ST‐HSCs), as well as lineage‐biased and unbiased multipotent progenitors (MPPs).26, 27 We observed reduced CD34+ populations; however, CD34− LT‐HSCs were not reduced in the mutants (Figure 2K). The most significant difference is at the immediate levels of the hematopoietic hierarchy in the stage of short‐term repopulating CD34+CD150+ HSCs (multipotent progenitors 1, MPP1), myeloid‐biased MPPs (MPP2, CD34+CD48+CD150+), balanced MPPs (MPP3, CD34+CD48+CD150−), and lymphoid‐biased MPPs (MPP4, CD34+Flt3+CD48+) subpopulations (Figure 2L,M). Defective hematopoietic commitment is propagated all the way to LS−K cells, which are enriched for committed erythroid/myeloid Kit+ progenitors, and megakaryocytic‐erythroid progenitors (MEPs; LS−K IL7R−CD34−), which are reduced two‐fold in Rps19R67∆/R67∆, indicating that a bias against erythroid lineage commitment is already present in the early bone marrow progenitors (Figure 2N).
To address the function of true HSCs and progenitors from Rps19R67∆/R67∆ mutants, we tested their bone marrow reconstitution activity in lethally (8.5 Gy) irradiated Ly5.1 recipients in competitive transplantation assays. The transplantation of 3000 Rps19R67∆/R67∆ Lin−Sca1+c‐Kit+ cells resulted in ~20% engraftment one month after transplantation, which is far below the expected 50% for WT controls under the same conditions. Furthermore, we observed a decreasing trend toward ~10%–15% engraftment at Months 2, 3, and 4 after transplantation, indicating a loss of competitiveness and intrinsic HSC defect in Rps19 mutants (Figure 2O).
In conclusion, these observations hint to a suspected role of HSCs in the origin of Rps19R67∆ phenotype.28, 29, 30
Activation of the p53 signaling pathway in Rps19 mutants and the effect of Trp53 deletion
To analyze molecular phenotype in the Rps19R67∆/R67∆ mice, we initially performed a bulk transcriptional profiling of the bone marrow cells from 10‐week‐old mice that yielded a highly selective set of differentially expressed genes: zinc‐finger matrin‐type 3 (Zmat3), pleckstrin homology (PH)–like domain family A member 3 (Phlda3), and ectodysplasin A2 receptor (Eda2r) (Figure 3A). Further transcriptional profiling of hematopoietic development, this time using E14.5 fetal livers from WT control and Rps19R67∆/R67∆ mutant embryos, yielded additional differentially expressed targets of the p53 signaling pathway: cyclin‐dependent kinase inhibitor p21 (Cdkn1a), cyclin G1 (Ccng1), and proline and serine‐rich coiled‐coil 1 (Psrc1) (Figure 3B,C). These are all known components of the p53 signaling pathway, whose increased expression in Rps19R67∆/R67∆ mutants may contribute to defective erythropoiesis under specific circumstances.
Activation of p53 targets at the mRNA level associates with p53 accumulation on the protein level in the Rps19R67∆/R67∆ bone marrow cells but does not associate with increased p53 phosphorylation (Figure 3D, Supporting Information S4: Figure 3C). We observed p53 accumulation in some, but not all, Rps19R67∆/R67∆ samples, suggesting that the mild phenotype may result from incomplete p53 activation (Figure 3D). To determine whether deletion of Trp53 in Rps19 mutants can improve erythropoiesis parameters in the first trimester after birth, we used a model of Trp53 tumor suppressor deletion in mice.31 Although the compensatory role of Trp53 loss‐of‐function in DBA phenotype has already been established,6, 30, 32 the introduction of a homozygous Trp53‐null mutation in the background of Rps19R67∆/R67∆ mice provided a basis for the study of the genetic interaction and phenotypic compensation between the RPS19 and transcription factor p53 in a novel double homozygous mouse model.
The functional ablation of Trp53 in Rps19R67∆/R67∆Trp53−/− double homozygous mutants was sufficient to reverse the upregulated expression of p53 pathway targets in E14.5 fetal livers (Figure 3C, Supporting Information S4: Figure 3A,B), as well as decreased bone marrow cellularity and amount of LK, LS−K, and LSK progenitors caused by Rps19 deficiency in adults (Figure 3E–I). This trend was also observed in Rps19R67∆/R67∆Trp53−/+ mutants carrying a single active Trp53 allele (Figure 3E,F, and H). In accordance with the improved bone marrow hematopoiesis, the stem and progenitor cell repertoire as measured by flow cytometry (gating strategy in Supporting Information S4: Figure 3D) revealed that distinctive LSK hematopoietic progenitor cell pool sizes in Rps19R67∆/R67∆Trp53−/− or Rps19R67∆/R67∆Trp53−/+ mutants appeared comparable to WT controls, while the amount of LSK CD34− stem cells was more abundant in bone marrow of Trp53−/− mice (Figure 3E–I, Supporting Information S4: Figure 3E–K).
The impaired engraftment potential of Rps19 mutant HSCs is reversed by concomitant loss of Trp53
To address the effects of Trp53 inactivation on HSC function in Rps19 mutant mice, we compared the repopulation activities of Rps19R67∆/R67∆, Rps19R67∆/R67∆Trp53−/−, and Trp53−/− HSCs using competitive transplantation assays (Supporting Information S4: Figure 3L). The competitive transplantation of 1.5 × 106 BMCs from WT, Rps19R67∆/R67∆, Trp53−/−, and Rps19R67∆/R67∆Trp53−/− mice (Ly5.2/Cd45.2) alongside with 1.5 × 106 healthy Ly5.1 (Cd45.1) BMCs (in 1:1 ratio) into lethally (8.5 Gy) irradiated Ly5.1 recipients revealed that decreased engraftment capacity of Rps19‐mutant HSCs (~20%) was effectively rescued by a loss of Trp53 gene (Figure 3J). HSCs from Rps19R67∆/R67∆Trp53−/− and Trp53−/− mutants have slightly better engraftment capacity compared to WT controls (Figure 3J).
The Rps19R67∆/R67∆Trp53−/− donor BMCs contributed to >75% of the host's hematopoiesis, similarly to Trp53‐null mutants. These results were consistent with bone marrow analysis, where Trp53‐deletion increased the size of HSC populations in both Trp53−/− and Rps19R67∆/R67∆Trp53−/− mutants (Figure 3I). While Trp53−/− and Rps19R67∆/R67∆Trp53−/− HSCs were able to fully restore hematopoiesis, Rps19R67∆/R67∆ HSCs could not do so (Figure 3K).
The rescued hematopoiesis of Rps19R67∆/R67∆Trp53−/− mutants coincides with reduced survival that is comparable to that of Trp53−/− mutants and offers a slight improvement over the early mortality of some Rps19R67∆/R67∆ mice (Figure 3L, see statistics in the right panel, Supporting Information S4: Figure 3M,N). The introduction of Trp53‐null mutation into the Rps19R67∆/R67∆ background also corrected the growth delay (Supporting Information S4: Figure 3O) and all skeletal defects seen in Rps19R67∆/R67∆ mice (Figure 3M, Supporting Information S4: Figure 3P). Correction of the harmful anemia phenotype and skeletal and cardiovascular defects resulted in many positive fitness benefits, including prolonged survival (up to 10 weeks).
Nevertheless, these gains in fitness are offset by the negative and expected long‐term pro‐oncogenic effect of Trp53 loss‐of‐function33 (mice with p53 inactivation exhibit shorter lifespan than many Rps19R67∆/R67∆ mice with mild phenotype due to accelerated tumorigenesis and subsequent premature death, see Supporting Information S1: Supplementary Results and Supporting Information S4: Figure 3M,N). Interestingly, Trp53+/− background improved the survival of DBA Rps19R67∆/R67∆ mice as well. Rps19R67∆/R67∆Trp53+/− mice develop tumors starting much later at ~9 months, similarly to Trp53+/− mice, and despite the improvement of their DBA‐like phenotypes, their survival and lifespan remain constrained due to late‐onset tumorigenesis induced by Trp53 loss‐of‐heterozygosity (Supporting Information S4: Figure 3M,N).
Rps19 mutants display activation of the p53 signaling in erythroid progenitors
The bulk RNA sequencing in adult and fetal hematopoietic tissues revealed a limited set of p53 targets (Figure 3A,B) that were further studied and validated using a single‐cell RNA sequencing in the bone marrow cells from 10‐week‐old WT control, Rps19R67∆/R67∆, Rps19R67∆/R67∆Trp53−/−, and Trp53−/− mutant mice using a 10× platform in LS−K and LSK stem cells and progenitors (Figure 4A). The integrated analysis of 7 single bone marrow cell RNA transcriptomes yielded 36 distinct clusters representing HSCs and early progenitor populations and their lineage‐specific descendants (Figure 4B, Supporting Information S5: Figure 4A,B). By performing differential gene expression analysis and associated gene expression clustering, we identified differentially expressed genes specific for either erythroid, B‐lymphoid, T/NK‐lymphoid, myeloid, granulocyte, neutrophil precursor stages, or stem cell populations and lineages (Supporting Information S5: Figure 4C).
The cell populations belonging to distinct developmental stages of erythroid lineages were identified by expression of GATA‐binding factor 1 (Gata1) transcription factor, erythropoietin receptor (Epor), glycophorin A (Gypa), and hemoglobin genes (Hba‐a1, Hba‐a2, Hbb‐bt, and Hbb‐bs), erythroid‐specific genes, and many other hematopoietic markers (Figure 4C, Supporting Information S5: Figure 4D), which served as basis for the determination of developmental trajectories and the identification of distinct erythroid, myeloid, lymphoid, and hematopoietic stem/progenitor cell clusters (Figure 4D). Trajectory analysis of the Rps19R67∆/R67∆ HSPCs did not reveal any obvious defect/delay in the development of erythroid lineage (Supporting Information S5: Figure 4E). By comparison of expression profiles among the four experimental genotypes, we have validated five genes: Zmat3, Phlda3, Eda2r, Cdkn1a, and Ccng1, which are dependent on p53 signaling (Figure 4E).
Rps19 mutants display an early defect in rRNA biogenesis
To assess whether the molecular phenotype of HSCs had already been affected, we compared the expression pattern of Rps19R67∆/R67∆ with control cells using single‐cell RNA (scRNA) analysis focusing specifically on differences at the HSC level, MPPs down to the level of erythroid cells, and found that dysregulation at the level of erythroid progenitors originates at the very immature level of HSCs, where changes in mRNA can already be observed, including some p53 targets as well as some other mRNAs (Figure 5A,B and Supporting Information S6: Figure 5A). To validate that rRNA biogenesis is impaired in Rps19R67∆ mutants, we focused on detecting pre‐rRNAs, which include junctions for 18S, 5.8S, and 28S products, which we detected using RT‐PCR (see Supporting Information S1: Supplemental Methods). We show that in Rps19R67Δ/R67Δ mutants, there is overproduction of some (but not all) pre‐rRNA products, as can also be seen in Rps19R67Δ/R67∆Trp53−/− mutants, while the wild‐type and single Trp53−/− mutants do not show rRNA deregulation (Figure 5C).
In addition, deregulation of p53 targets can be observed alongside pre‐rRNA accumulation in the Rps19R67Δ/R67Δ mutants, while this is blocked upon Trp53 codeletion in Rps19R67Δ/R67∆Trp53−/− mutants (Supporting Information S6: Figure 5B).
Finally, we also evaluated protein translation to determine whether the Rps19R67∆ mutation had any effect on its rate. To achieve this result, we expressed a doxycycline‐regulated transgenic cassette encoding Rps19 and its mutation Rps19R67∆ in mouse erythroleukemic (MEL) cells. As shown in Supporting Information S6: Figure 5C, doxycycline induced strong expression of the Rps19 transgene cassette in both RPS19 and RPS19 R67∆ cells, as measured by dsRED2, which is co‐expressed with RPS19 from the same P2A‐linked transcript.
After 24 h of doxycycline‐induced expression, MEL cells expressing high levels of transgenic RPS19R67∆/dsRED2 displayed inhibition of global protein synthesis, as measured by L‐HPG incorporation and Click‐iT detection (see Supporting Information S1). Interestingly, induction of the WT RPS19 cassette affected protein synthesis to a lesser extent, although it also partially inhibited it. This phenomenon could be studied without inducing apoptosis, partly because MEL cells have a defect in p53 function (Supporting Information S6: Figure 5D).
In our manuscript, we show that deletion of R67 in the RPS19 protein leads to a heterogeneous phenotype, including anemia as well as skeletal, muscular, and cardiac phenotypes, and a neurophenotype. Loss of R67 in Rps19 leads to impaired rRNA processing and, consequently, to impaired translation in the Rps19 mutants. The phenotype in mice requires activation of p53 and its targets, but loss of p53 is unable to affect dysfunction in pre‐rRNA processing.
The result is ribosomal stress leading to suppression of erythroid cell proliferation and apoptosis. Interestingly, gradual activation of p53 target gene expression occurs already at the level of very immature blood progenitor cells. We are convinced that our model will contribute to research into the biological properties of Rps19, and it also appears that it could have an impact on the development of therapeutic strategies for DBA.
DISCUSSION
The mouse model presented in this study carries a single amino acid deletion of conserved Arg67 in the frequently mutated region of ribosomal protein Rps19 (Rps19R67∆). Although this mutation cannot be found in the database of DBA mutations, our structural predictions of Rps19R67∆ deletion and the high similarity of human and mouse Rps19 proteins15 suggest that removal of the Arg67 could directly interfere with the interaction of Rps19 and 18S ribosomal RNA while not markedly downregulating Rps19 level (Supporting Information S2: Figure 1A and Figure 3D). The disruption of Rps19 near R67 likely destabilizes the adjacent regions of Rps19 that physically interact with neighboring RPs Rps2 and Rps3, affecting ribosomal assembly and ribosomal particle stability.
Our mouse model with homozygous Rps19R67∆ mutation manifests many but not all the characteristic traits of human DBA symptomatology, including reduced erythroid progenitor maturation, as well as growth retardation combined with dysmorphologies (cardiac, diaphragmatic, and hydrocephalic) all accounting for decreased fitness and increased mortality of the mice at age of 6–8 weeks (while some mutant mice deceased also prior birth). However, like in DBA, the Rps19R67∆ mutant phenotype also has heterogeneous phenotypic outcomes. For example, half of the mice survive for a relatively normal time (Supporting Information S4: Figure 3N), which may be based on stochasticity; we speculate that the level of ribosomal stress was not sufficient to effectively accumulate p53 and activate its program in some individuals (Figure 3C,D).
It is also possible that the p53 program, which is activated as early as the fetal period (Figure 3B), was activated later, and therefore, overall, it is likely that such a disorder of erythropoiesis or in non‐hematopoietic tissues will be significantly weaker.
Our Rps19 Arg67 deletion model has certain similarities with previous models; specifically, constitutive expression of Rps19 R62W in mice led to a strong phenotype, namely embryonic lethality.7 However, conditional (Cre‐induced) downregulation of Rps19 in mice led to growth retardation, mild anemia, a reduction in the number of erythroid progenitors, and disruption of terminal erythroid maturation, mimicking some features of DBA.8 In both models, the p53 program was activated. Regarding the role of p53 targets in Rps19R67∆/R67∆ mutants, we show for the first time that the p53 program is activated already at the level of early progenitor cells in the bone marrow (see Figure 5A,B), suggesting that the hematopoietic defect initiates at a very early uncommitted stage. Our mouse model shows some differences to some previous Rps19 models that also imperfectly mimic DBA within the hematologic compartment4, 6, 7 and are lethal relatively early (examples are models of Rpl5 and Rpl11).34, 35, 36, 37, 38 Mouse embryos with a heterozygous mutation of Rpl11 were not viable, whereas the Rpl11 homozygous mutation leads to early postnatal death due to defective erythropoiesis and bone marrow failure; no additional defects were reported.39 The Rpl5 mouse model with defective splicing faithfully mimics the various DBA symptoms, but in contrast to our mutants, the defective erythroid maturation and increased mortality in Rpl5‐mutant mice occur prenatally.40
The presence of cardiovascular, skeletal, or craniofacial defects in patients diagnosed with DBA and in mouse models with ribosomal defects, including Rps19R67∆/R67∆ mice, could be attributed to significant defects of neural crest cell differentiation, as these cell lineages directly contribute to craniofacial and cardiac structures.41, 42 A recent report by Tiu et al directly implicates the role of augmented p53 signaling in the development of craniofacial defects in many developmental syndromes, including ribosomopathies such as DBA.43, 44 This is due to the negative impact of prolonged p53 activation on the cell survival, proliferation, and differentiation of neural crest progenitors and their descendants. In addition to the notable role of p53, many important cellular factors involved in nucleolar ribosome biogenesis, such as Tcof1 and Ddx21, are associated with cranial neural crest‐derived developmental malformations, confirming a strong connection between p53 signaling, ribosome biogenesis, anemia, and developmental malformations.6, 45, 46, 47, 48, 49 Consequently, these observations and the research done by other groups using zebrafish DBA models32, 50 provide the main rationale for using Trp53‐null models to rescue anemic phenotype associated with Rps19‐deficiencies.
Our work herein also suggests that the hematologic phenotype of Rps19 mutants also involves the level of HSCs.28, 29, 30 We observed that Rps19 mutant stem and progenitor cells failed to efficiently repopulate bone marrow using competitive transplantation assays (Figures 2P and 3I). Based on the analysis of the molecular phenotype using scRNA (Figure 5A,B), it can be observed that the activation of p53 targets may probably already initiate at the HSC level although we do not have any other functional tests focused on these cells to directly confirm that p53 is indeed accumulated in stem cells, mostly due to the very low number of these cells. Furthermore, the number of the earliest mutant HSCs is comparable to control samples, so if there is a disruption of the p53 program in them, this effect can be observed first at the MPP level in the early stages of differentiation (Figure 5A) and subsequently to a much greater extent during erythropoiesis.
As the competitive transplantation experiments show, even the fitness of HSCs has inferior parameters in Rps19 R67 mutants. It is unclear whether anemia in Rps19 mutants is caused by the mere loss of certain HSC subsets and/or specific dependence on Rps19 function during erythroid differentiation stages. Our work suggests that it is probably not the number of HSCs but rather their progressively strengthening defect that reveals its dependence on Rps19 in erythroblasts, where apoptosis is induced as a result.
Regarding the role of p53 targets possibly involved in defective hematopoiesis, analysis of differential expression of Rps19R67∆/R67∆ c‐Kit+ bone marrow cells, E14.5 fetal liver, and single‐cell transcriptomic analysis of Rps19R67∆/R67∆, Rps19R67∆/R67∆Trp53−/−, and Trp53−/− bone marrow cells together yielded several primary p53 targets: Zmat3, Phlda3, Eda2r, Cdkn1a, Ccng1, and Prsc1, whose functions in the pathology of DBA are currently only partially understood. Nonetheless, the double‐strand RNA binding protein ZMAT3 is a key zinc‐finger tumor suppressor downstream of p53 that enhances the stability of Trp53 mRNA and modulates alternative exon inclusion in the transcripts of Mdm2 and Mdm4, thus regulating the stability and cytoplasmatic/nuclear translocation of the p53 protein.51, 52 Interestingly, a p53‐regulated repressor of Akt signaling—single PH‐like domain containing protein PHLDA3—competes with PH domain of the protein kinase B isoform Akt1 for binding of phosphatidylinositol‐(3,4,5)‐trisphosphate second messenger downstream to multiple growth, survival, and differentiation signals in hematopoietic cells and has the ability to directly control cellular metabolism.53, 54 It is also a most prominent target for future therapeutical intervention as a small druggable target with well‐defined function in PI3K‐Akt signaling. A direct transcriptional target of p53, ectodysplasin A2 receptor (EDA2R), activates both canonical and non‐canonical NF‐κB and MAP kinase pathways and mediates a p53‐regulated anoikis pathway.55 Trp53 activation in Rps19 Arg67del mutants may be only transient and potentially stochastic.
This would also be consistent with the fact that some mice in the Rps19 R67 deletion model survive and have a minimal phenotype coincidently with lower accumulation of p53 (Figure 3C,D), while others, approximately half, have a strong phenotype ranging from severe congenital anomalies, anemia, to premature (and in some cases also prenatal) lethality.
The main feature of HSC function in bone marrow failure syndromes may be their tendency to differentially modulate p53 activity and simultaneously respond to the elevation of p53 targets in differentiated tissues. Our data indicate a defect in pre‐rRNA processing (Figure 5C) that probably leads to p53 activation within erythropoiesis. The requirement of the precise regulation of p53 activity in hematopoietic cells has already been reported previously.43 Reducing p53 protein or some of its targets could thus render these hematopoietic progenitors resistant to prolonged chronic disturbances in ribosome biogenesis, an approach potentially applicable in DBA.
On the other hand, it is also essential to consider that at least a part of the erythroid phenotype may not be directly influenced by p53 activity. Specifically, R67 deletion in Rps19 may also cause a progressive or intermittent translation disruption, which may ultimately contribute to the worsening of anemia, although these claims will need to be validated in mice, as they are based on experiments on the MEL cell line (Supporting Information S6: Figure 5C,D).
In conclusion, it can be stated that our proposed model of deletion of a single arginine at position 67 in the Rps19 gene leads to a phenotype that includes early as well as later stages of hematopoiesis, specifically a disorder of rRNA processing, which leads to activation of the p53 pathway and, as a result, anemia and other developmental disorders that exhibit a relatively heterogeneous phenotype ranging from mild to very severe outcomes.
AUTHOR CONTRIBUTIONS
Juraj Kokavec: Writing—original draft; writing—review and editing; visualization; formal analysis; investigation; methodology; data curation; validation. Tereza Turková: Investigation; formal analysis; writing—review and editing; methodology; validation. Björn Schuster: Investigation; formal analysis; methodology.
Jan Prochazka: Investigation; formal analysis; visualization. František Spoutil: Investigation; formal analysis; visualization. Kristína Jamrichová: Investigation.
Markéta Holečková: Investigation; formal analysis. Karel Chalupský: Investigation; formal analysis. Inken M.
Beck: Investigation. Jesús Ruberte: Investigation; formal analysis. Matilde Vale: Investigation.
Lukáš Čermák: Investigation; conceptualization. Tomáš Stopka: Funding acquisition; conceptualization; supervision; resources; project administration; writing—review and editing; methodology. Radislav Sedlacek: Funding acquisition; project administration; resources; conceptualization; supervision.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
Supporting information