EIF6 accumulates in the cells of patients with SDS and sbds-null zebrafish. Somatic mutations in EIF6 or its haploinsufficiency due to loss of 20q have been found in two-thirds of patients with SDS who have a decreased risk of developing myeloid malignancies (22, 23). To investigate the effect of EIF6 on the pathophysiology of SDS, we first observed an accumulation of EIF6 in lymphoblasts and PBMCs from individuals with SDS compared with controls (Figure 1, A and B, and Supplemental Table 2; supplemental material available online with this article; https://doi.org/10.1172/JCI187778DS1). Previously, we reported increased Eif6 protein levels in sbds-null zebrafish larvae at 10 dpf (4). Here, we discovered that EIF6 accumulation occurred at earlier stages of embryonic development (Figure 1C). Eif6 accumulation occurred as early as 5 dpf in sbds–/– zebrafish, which coincided with depletion of Sbds protein. Next, we determined the location of this accumulation by larvae at 10 dpf. We performed IHC, which detected a cytoplasmic accumulation of Eif6 in sbds-null fish compared with their WT siblings (Figure 1D).

Eif6 accumulation in patients with SDS and sbds-KO zebrafish. Western blots showing EIF6 accumulation in SDS lymphoblasts (A), SDS PBMCs (B), and sbds-KO fish from 5 dpf to 10 dpf (C). CTL1, healthy control no. 1. (D) IHC images of Eif6 expression showing accumulation in the liver and digestive tract (DT) of 10 dpf zebrafish. Original magnification, ×40.

EIF6 is essential for zebrafish embryonic development. We hypothesized that the accumulation of EIF6 contributes to the pathogenesis of SDS. To address this, we used CRISPR/Cas9 editing to create an eif6-mutant allele with a 1 bp deletion (eif6lri110, hereafter denoted as eif6–) that produced a premature termination codon (Figure 2A). Reverse transcription quantitative PCR (RT-qPCR) revealed that eif6 transcripts were significantly decreased in the eif6+/– and eif6–/– strain compared with WT siblings (Figure 2B). Western blotting revealed an absence of Eif6 expression at 5 dpf in the eif6–/– fish, and the heterozygotes showed half the amount of protein compared with their WT siblings (Figure 2, C and D). By 5 dpf, they showed severe defects, such as cardiac edema, failure of the swim bladder to inflate, a smaller head and eyes, and a decreased number of neutrophils compared with their eif6+/– and eif6+/+ clutchmates (Figure 2, E and F). No eif6–/– fish survived beyond 10 dpf (Figure 2G).

Eif6 is essential for zebrafish embryonic development. (A) Amino acid alignment of WT and eif6 KO zebrafish; the insertion of 1 bp causes early truncation. (B) mRNA eif6 levels by RT-qPCR on the tp53+/+ and tp53M214K/M214K backgrounds at 5 dpf. (C) Western blots showing the absence of Eif6 protein in the eif6–/– zebrafish larvae at 5 dpf. Note the dose effect in the eif6+/– fish compared with their WT siblings. (D) Western blot quantification. (E) eif6 KO showed a significantly lower number of neutrophils than the WT siblings at 5 dpf. (F) Sudan black staining for neutrophils counts (original magnification, ×2.5 and ×8). (G) Survival percentages for eif6+/+, eif6+/–, and eif6–/– siblings at 5, 8, and 10 dpf. *P < 0.05, **P < 0.01, and ***P < 0.001, by ANOVA. Data represent the mean ± SEM.

RNA-Seq was performed on whole zebrafish larvae from eif6+/+ and eif6–/– at 5 dpf. Hierarchical clustering showed a high overlap between the replicates within each genotype while clearly separating the triplicates into 2 groups. RNA-Seq analysis at 5 dpf larvae revealed 1557 upregulated and 668 downregulated genes in mutants. To gain deeper understanding of the molecular processes and signaling pathways affected in eif6–/–, Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was conducted on genes that were either upregulated or downregulated in eif6-null fish. The analysis revealed that differentially expressed genes (DEGs) were enriched in a variety of biological processes and signaling pathways. Specifically, these included ribosome and oxidative phosphorylation pathways, highlighting their marked involvement in the genetic alterations observed in the eif6-null phenotype (Table 1).

DEGs involved in ribosome and oxidative phosphorylation pathways

Low levels of Eif6 are sufficient for survival. When targeting eif6 exon 2 to create the eif6-null fish analyzed above, we obtained 2 other founders with different mutations, 1 with a deletion of 6 bp in-frame that caused a deletion of 2 amino acids (eif6lri111, hereafter denoted as eif6del2aa) and another with 4 bp in-frame missense mutation that resulted in a change in 2 amino acids (eif6lri112, hereafter denoted as eif6ms) (Figure 3A). Surprisingly, both mutants survived to adulthood and were fertile, allowing us to create maternal zygotic (MZ) lines. We evaluated the eif6 mRNA and protein levels of these eif6-mutant lines. At 5 dpf, mRNA levels of the eif6del2aa/del2aa mutant were similar to those of the WT fish, but in the eif6ms/ms mutant, the mRNA levels were upregulated 4-fold (Figure 3B). We next determined protein expression levels in each mutant strain at 5 dpf. The eif6del2aa/del2aa mutant made 20%–30% and the eif6ms/ms mutant made 5%–10% of the protein expressed by the WT fish (Figure 3, C and D). Decreased protein levels of mutant EIF6 were reported in human tissues (22, 23). Because there was no evidence for transcriptional repression, these results suggested that mutant Eif6 caused protein instability (see below), degradation, or translational inefficiency.

Low levels of Eif6 are enough for survival to adulthood, but only an absence of eif6 KO affects the levels of ribosomal proteins. (A) Amino acid alignment of the 2 eif6 mutants (eif6ms/ms and eif6del2aa/del2aa) compared with the Eif6 WT sequence. (B) mRNA levels in all 3 eif6 mutants and WT zebrafish at 5 dpf. (C) Western blots showing Eif6, Sbds, and RP levels in the different eif6 mutants. (D) Western blot quantification. (E) Western blots showing Rpl23 and Rps3 in eif6+/+, eif6+/–, and eif6–/– siblings at 5 dpf. (F) Western blot quantification relative to actin. (G) Western blotting showing Rpl23 and Rps3 expression in eif6+/+, eif6del2aa/del2aa, and eif6ms/ms zebrafish at 5 dpf. (H) Western blot quantification relative to actin. (I) Polysome profiles of eif6 WT and mutants at 5 dpf. *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA. Data represent the mean ± SEM.

Since there were decreased ribosomal protein transcripts in the eif6 KO (Figure 3, A and B), we explored how the different eif6 mutations affect Sbds and other ribosomal proteins. Sbds protein levels were not affected in any of the 3 mutants (eif6del2aa/del2aa, eif6ms/ms, and eif6–/–) compared with WT at 5 dpf (Figure 3C). Rps3 and Rpl23 were decreased in the eif6–/– mutants. Eif6 interacts with Rpl23, located close to the surface of the 60S subunit (27). Rpl5 and Rpl11 can activate the tumor suppressor protein Tp53 pathway by binding to Mdm2 (28). This process is a key component in the context of ribosomal stress. Additionally, somatic mutations in RPL5 and RPL22 have been identified in patients with SDS. This suggests that mutations in regulators of the nucleolar signaling pathway, like RPL5 and RPL22, may interfere with nucleolar stress–induced stabilization of TP53 (29). Interestingly, Rpl5 was significantly reduced in the eif6–/– and eif6ms/ms mutants, the eif6-KO, and the lowest Eif6 protein mutant respectively, and Rpl11 only in the eif6-KO mutant (Figure 3, C–H). Rpl22 did not changed in any of the mutants. Thus, Eif6 levels affected the expression of other ribosomal proteins. Altogether, the changes in Rpl23, Rpl5, Rpl11, and Rps3 levels suggested more global effects due to loss of Eif6 and merit further investigation.

We performed polysome profiling to study how alterations in Eif6 levels might affect translation. Polysome profiles showed a significant reduction in the monosomal and 40S ribosomal subunits ratio between eif6+/+ and all eif6 mutants at 5 dpf. Interestingly, loss of Eif6 did not result in any free ribosomal subunits and showed a significant reduction in the levels of 80S ribosomes, while eif6del2aa/del2aa and eif6ms/ms mutants showed a polysome profile similar to that of the eif6 WT (Figure 3I). In summary, the absence of Eif6 led to aberrant ribosome structures, which could explain the early lethality observed in the eif6-KO zebrafish line. These changes led to an earlier demise than the loss of Sbds (4).

Tp53 pathway activation in eif6 mutants does not affect survival rates. We previously reported that deletion of Sbds in zebrafish caused activation of the Tp53 pathways involving tp53, cdkn1a, mdm2, cdkn2ab, bax, puma, and ccng1 (4). Since Sbds and Eif6 are part of the same ribosomal assembly pathway, we hypothesized that these pathways were also affected in the eif6 KO at 5 dpf. Our results demonstrated a significant upregulation of Tp53 targets (cdkn1a, mdm2, bax, ccng1, and casp9) in the eif6-null fish (Figure 4A). To determine whether Tp53 is involved in the dysregulation of its own pathway, we created a new eif6–/– zebrafish on the tp53M214K background (eif6+/– p53M214K/M214K). When we compared with WT eif6 versus eif6–/–, we observed that eif6 mRNA level differences were similar to those observed on the tp53 WT background (Figure 2B). As expected, all Tp53 targets previously evaluated (bax, puma, mdm2, ccng1, and casp9) were not upregulated (Figure 4B).

Tp53 pathway activation in eif6 mutants does not affect survival rates. RT-qPCR analysis of tp53 and its targets of WT and eif6 KO on the (A) tp53+/+ and (B) tp53M214K/M214K backgrounds. (C) Scheme of the double-heterozygote crosses (eif6+/– tp53+/M214K). (D) Percentage of survival for double-heterozygote crosses of eif6+/– tp53+/M214K at 8 and 15 dpf. (E) RT-qPCR analysis of tp53 and its targets (*P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA), and (F) UPR markers in WT and eif6ms/ms zebrafish at 5 dpf (*P < 0.05 and ***P < 0.001, by 2-tailed t test). Data represent the mean ± SEM.

To investigate the effect of Tp53 activation on survival rates in the eif6 mutants, we incrossed double heterozygotes and assessed their progeny at 8 and 10 dpf (Figure 4, C and D). At 8 dpf, all 9 of the predicted genotypes were found at the expected Mendelian ratio. However, at 10 dpf, we did not detect any eif6–/– zebrafish alive independently of the tp53 genotype (Figure 4D). Thus, the tp53M214K/M214K mutants did not rescue the Eif6 deficiency. We recently reported that tp53M214K/M214K mutants also did not rescue Sbds deficiency (4, 17).

Next, we sought to determine whether the Tp53 pathway was also affected in the eif6del2aa/del2aa and eif6ms/ms. Interestingly, we found that the Tp53 targets cdkn1a, cdkn2ab, and ccng1 were upregulated only in the eif6ms/ms mutant. These findings suggested that the quantity of Eif6 protein played a crucial role in regulating Tp53 (Figure 4E). Since we had observed that eif6 mRNA levels were upregulated in this mutant, we also checked some markers for the unfolded protein response (UPR). The eif6ms/ms mutants exhibited an upregulation of some UPR markers (chop, bip, and atf4b) (Figure 4F), which could potentially lead to protein degradation and result in the lower levels of Eif6 protein we observed.

Erythrocyte and neutrophil counts are unaffected in eif6 mutants. Using Sudan black and O-dianisidine staining, we found that all zebrafish expressing low levels of Eif6 produced neutrophils and RBCs, with no differences compared with WT (Supplemental Figure 1, A and B). While eif6–/– zebrafish also produced these cells, substantial developmental defects hindered comparisons with their WT siblings.

Partial rescue of sbds–/– occurs with 1 copy of the WT eif6 allele. Since we observed an accumulation of Eif6 protein in the sbds–/– zebrafish (Figure 1, C–E), we investigated whether reduction of Eif6 would be beneficial for them, as has been suggested in patients with SDS (22, 23). We created double heterozygotes (eif6+/– sbds+/–) and assessed survival rates at 15 dpf (Figure 5, A and B). The only sbds–/– fish that survived were those that harbored 1 copy of WT eif6 (eif6+/–). This suggested that the Eif6 dosage may be crucial for somatic genetic rescue. However, the eif6+/– sbds–/– zebrafish did not survive to adulthood. Next, we measured Eif6 protein levels by Western blotting. We detected an accumulation of Eif6 in the eif6+/– sbds–/– fish compared with their eif6+/– sbds+/+ siblings (Figure 5, C and D). This partial rescue in survival indicated that some accumulation of Eif6 could be toxic and that other ribosomal stress responses persist.

Partial rescue of sbds KO with 1 copy of the WT eif6 allele. (A) Scheme showing the incrossing of 2 double-heterozygotes (eif6+/– sbds+/–). (B) Genotype percentages at 15 dpf of an incrossing of double-heterozygotes. (C) Western blots showing Eif6 accumulation in sbds KO fish with only 1 copy of the eif6 WT allele (eif6+/–). (D) Western blot quantification of Eif6 protein expression. (E) Sudan black staining to detect the number of neutrophils in 10 dpf siblings. Scale bars: 200 μm and 50 μm. Data represent the mean ± SEM.

We previously reported that sbds-null zebrafish have a lower number of neutrophils than do their WT siblings. To determine whether the amount of Eif6 protein affects the neutrophil, we measured the number of neutrophils in zebrafish on the eif6-null background. The number of neutrophils was lower on the sbds–/– background in both eif6+/+ and eif6+/– zebrafish compared with their sbds+/+ siblings (Figure 5E). Thus, the neutropenia could not be rescued.

Low Eif6 levels decrease Tp53 pathway activation and partially rescue survival but do not rescue neutropenia in sbds–/– zebrafish. We showed that sbds-KO fish with only 1 WT eif6 allele (eif6+/–) could survive longer (Figure 5A), so we next incrossed double-heterozygotes carrying the eif6 missense mutation (sbds+/– eif6+/ms). Similar to our previous observations, we found that sbds-KO fish with slightly higher survival rates were those on the eif6+/ms and eif6ms/ms backgrounds (Figure 6A).

Correlation of Eif6 levels with tp53 activation in sbds-KO zebrafish. (A) Scheme of sbds+/– crossings in the WT and Eif6 missense mutants and percentage of genotypes observed at 20 dpf. (B) Schemes of the 3 different crosses to obtain sbds WT and KO zebrafish on the eif6+/+, eif6+/ms, and eif6ms/ms backgrounds. (C) Immunoblot showing Eif6 accumulation in sbds-KO zebrafish. (D) Neutrophil counts at 10 dpf. *P < 0.05, **P < 0.01, ***P < 0.001, by 1-way ANOVA with Dunnett’s test. (E) Protein levels in sbds+/+, sbds+/–, and sbds–/–, in the eif6+/+ and eif6ms/ms mutant backgrounds, along with protein quantification. (F) RT-qPCR results for sbds and eif6 (G) and the tp53 pathway. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001, by 1-way ANOVA with Tukey’s test. Data represent the mean ± SEM.

In a dihybrid cross (eif6+/– sbds+/–), the expected Mendelian ratio is 1 WT (sbds+/+ eif6+/+) or 1 double-mutant (sbds–/– eif6–/–) of 16 fish (6.25%). This low frequency makes collecting these specific genotypes challenging, especially for RNA extraction purposes. Unlike the eif6-KO fish, the eif6ms/ms mutants were viable. To further study how low Eif6 levels affect sbds KO, we created an sbds-mutant line in the eif6ms mutation (sbds+/– eif6ms/ms). In this case, the expected Mendelian ratio is 1 of 4 sbds WT/mutant fish on the eif6ms/ms background (25%). To study how Eif6 levels affect sbds KO at 10 dpf, we performed 3 different crosses: (a) incross of sbds+/– on the eif6 WT background (sbds+/– eif6+/+); (b) incross of sbds+/– on the eif6 missense mutation background (sbds+/– eif6ms/ms); and (c) outcross of sbds+/– on the eif6 WT background (sbds+/– eif6+/+) with a sbds+/– on the eif6 missense mutation background (sbds+/– eif6ms/ms). With these 3 crosses, we determined differences between eif6+/+, eif6ms/ms, and eif6+/ms in the sbds WT and sbds-KO fish, respectively (Figure 6B). We then determined Eif6 protein levels in fish on the sbds WT background at 5 dpf and found that the protein levels in the heterozygotes were approximately 50% of the levels in WT fish (Figure 6C).

Previously, we demonstrated that sbds-KO zebrafish had substantially lower numbers of neutrophils compared with their WT siblings (Figure 4D) (4). At 10 dpf, we determined the number of neutrophils and found that sbds–/– fish had significantly lower numbers of neutrophils than did the WT fish, regardless of eif6 genotyping (Figure 6D). These results are consistent with those previously observed in eif6+/– fish (Figure 5E), suggesting that Eif6 levels did not affect neutrophil counts in our zebrafish models.

Next, we assessed Eif6 levels in eif6 WT and missense mutants and found an accumulation of Eif6 in fish on the sbds-KO backgrounds at 10 dpf. We chose this time point on the basis of our previous studies, which showed that sbds–/– fish had no detectable Sbds protein at 10 dpf (4). However, this accumulation was significantly lower in eif6ms/ms fish than in eif6+/+ fish (Figure 6E). Additionally, we analyzed sbds and eif6 mRNA levels across all genotypes and found, as expected, that sbds mRNA levels were decreased in sbds-KO zebrafish, regardless of the eif6 background (Figure 6F).

Subsequently, we examined markers of tp53 pathways. Surprisingly, when Eif6 levels were low (eif6+/ms) or lower (eif6ms/ms), sbds-KO fish did not exhibit a pronounced activation of the Tp53 pathway compared with those on the eif6 WT background (eif6+/+), in which cdkn1a, cdknd2ab, bax, puma, and casp9 expression levels were strongly upregulated (Figure 6G). Hence, low levels of Eif6 reduced the activation of the tp53 pathway, which could serve to mitigate the role of Eif6 accumulation in TP53-mediated stress responses and apoptosis.

Tp53 is upregulated in SDS patient–derived cell lines. Because we observed that the Tp53 pathway plays an important role in our zebrafish models of SDS, we sought to determine whether human SDS–derived tissues exhibited the same level of activation. We assessed TP53 and p21 protein levels in SDS patient–derived lymphoblastoid cell lines (LCLs) and bone marrow mononuclear cells (BM-MNCs) and detected a significant 4-fold increase in SDS LCL cells compared with the controls (Figure 7A). We confirmed this finding in primary BM-MNCs freshly isolated from patients with SDS, in which SDS cells had more extensive increases in TP53 levels (Figure 7B).

Tp53 activation in SDS patient–derived cells. Western blots and quantification of (A) LCLs and (B) BM-MNCs. EIF6 mRNA (C) and protein (D) levels in 3 different siRNAs. (E) EIF6 mRNA levels in patients with SDS with and without siRNA (F) TP53 and CDKN1a mRNA levels are decreased after siRNA in patients with SDS. (G) Western blots showing decrease in TP53 levels after EIF6 knockdown using 3 different siRNAs. (H) TP53 protein quantification. Data represent the mean ± SEM. HD, healthy donor; NC, negative control.

Knockdown of EIF6 in SDS-derived LCL decreases TP53 and CDKN1A mRNA levels. Having observed increased EIF6 protein levels in LCL cells derived from patients with SDS (Figure 1B), we hypothesized that decreased EIF6 levels could mitigate the activation of the TP53 pathway. We first tested 3 different EIF6 shRNAs (Figure 7, C and D). We chose siRNA no. 2, which showed the highest silencing efficacy, inducing a 5.7-fold reduction in EIF6 mRNA expression and a 9-fold reduction in protein levels (Figure 7, D and E). Corresponding to the decreased EIF6 protein levels, TP53 and CDKN1A levels in these cells were significantly decreased from approximately 5.6-fold to approximately 2.7-fold (Figure 7F). Additionally, in SDS-deficient cells, TP53 protein levels decreased following the reduction of EIF6. In contrast, TP53 levels in healthy controls remained unaffected (Figure 7, G and H). Hence, low EIF6 levels mitigated the activation of the TP53 pathways, alleviating the cellular stress in SBDS-deficient cells as in sbds–/– zebrafish and offering a mechanism for somatic cell rescue.