Article Text

PDF

Mutations in EFL1, an SBDS partner, are associated with infantile pancytopenia, exocrine pancreatic insufficiency and skeletal anomalies in aShwachman-Diamond like syndrome
  1. Polina Stepensky1,
  2. Montserrat Chacón-Flores2,
  3. Katherine H Kim3,4,
  4. Omar Abuzaitoun5,
  5. Arnulfo Bautista-Santos2,
  6. Natalia Simanovsky6,
  7. Dritan Siliqi7,
  8. Davide Altamura7,
  9. Alfonso Méndez-Godoy2,
  10. Abril Gijsbers2,
  11. Adeeb Naser Eddin1,
  12. Talia Dor8,
  13. Joel Charrow3,4,
  14. Nuria Sánchez-Puig2,
  15. Orly Elpeleg9
  1. 1 Department of Pediatric Hematology, Oncology and Bone Marrow Transplantation, Hadassah, Hebrew University Medical Center, Jerusalem, Israel
  2. 2 Departamento de Química de Biomacromoléculas, Instituto de Química, Universidad Nacional Autónoma de México, Ciudad de México, México
  3. 3 Division of Genetics, Birth Defects and Metabolism, Ann & Robert H. Lurie Children’s Hospital of Chicago, Chicago, Illinois, USA
  4. 4 Department of Pediatrics, Feinberg School of Medicine, Northwestern University, Chicago, Illinois, USA
  5. 5 Nablus Speciality Hospital, Palestinian Authority, Nablus, Palestine
  6. 6 Department of Medical Imaging, Hadassah, Hebrew University Medical Center, Jerusalem, Israel
  7. 7 Istituto di Cristallografia, Consiglio Nazionale delle Ricerche, Bari, Italy
  8. 8 Pediatric Neurology Unit, Hadassah, Hebrew University Medical Center, Jerusalem, Israel
  9. 9 Monique and Jacques Roboh Department of Genetic Research, Hadassah, Hebrew University Medical Center, Jerusalem, Israel
  1. Correspondence to Nuria Sánchez-Puig, Departamento de Química de Biomacromoléculas, Instituto de Química, Universidad Nacional Autónoma de México, Ciudad de México, México; nuriasp{at}unam.mx and Professor Orly Elpeleg, Monique and Jacques Roboh Department of Genetic Research, Hadassah-Hebrew University Medical Center, Jerusalem 91120, Israel; elpeleg{at}hadassah.org.il

Abstract

Background For the final step of the maturation of the ribosome, the nascent 40S and 60S subunits are exported from the nucleus to the cell cytoplasm. To prevent premature association of these ribosomal subunits, eukaryotic initiation factor 6 (eIF6) binds the 60S subunit within the nucleus. Its release in the cytoplasm requires the interaction of EFL1 and SDBS proteins. In Shwachman-Diamond syndrome (SDS), a defective SDBS protein prevents eIF6 eviction, inhibiting its recycle to the nucleus and subsequent formation of the active 80S ribosome.

Objective This study aims to identify the molecular basis of an SDS-like disease, manifested by pancytopenia, exocrine pancreatic insufficiency and skeletal abnormalities in six patients from three unrelated families.

Methods Whole exome analysis was used for mutation identification. Fluorescence microscopy studies assessed the localisation of Tif6-GFP, the yeast eIF6 homologue, in yeast WT and mutant cells. Human and yeast EFL1 proteins, WT and mutants, were expressed in Saccharomyces cerevisiae BCY123 strain, and circular dichroism and small-angle X-ray scattering were used to assess the folding and flexibility of these proteins. Green malachite colorimetric assay was performed to determine the GTPase activity of WT and Efl1 mutants.

Results Four patients were homozygous for p.R1095Q variant and two patients were homozygous for p.M882K variant in EFL1. Residue R1095 and M882 are conserved across species. Neither the GTPase activity of the mutant proteins nor its activation by the SDBD protein or the 60S ribosomal subunit were affected. Complementation of efl1Δ yeast cells with the EFL1 mutants rescued the slow growth phenotype. Nonetheless, Tif6-GFP was relocalised to the cytoplasm in mutant yeast cells in contrast to its nuclear localisation in WT cells.

Conclusions Mutations in EFL1 clinically manifest as SDS-like phenotype. Similar to the molecular pathology of SDS, mutant EFL1 proteins do not promote the release of cytoplasmic Tif6 from the 60S subunit, likely preventing the formation of mature ribosomes.

  • Ribosome
  • exocrine pancreatic insufficiency
  • pancytopenia
  • Shwachman-Diamond syndrome

Statistics from Altmetric.com

Introduction

Haematological disorders accompanied by skeletal anomalies are characteristic of patients with Diamond-Blackfan anaemia (DBA) and Shwachman-Diamond syndrome (SDS). Common to these diseases are underlying defects in ribosome structure and function. The eukaryotic ribosome consists of small (40S) and large (60S) subunits,1 each composed of ribosomal RNAs (rRNA) and ribosomal proteins. The small subunit includes the 18S rRNA and 33 proteins, whereas the large subunit includes 5S rRNA, 28S rRNA, 5.8S rRNA and 46 proteins. During ribosomes biogenesis, nascent 40S and 60S subunits are exported to the cell cytoplasm to face the last maturation steps. In Saccharomyces cerevisiae, the pre-60S arrives into this compartment loaded with the eukaryotic initiation factor 6 (eIF6, or the yeast orthologue Tif6) which acts as a ribosomal antiassociation factor with the pre-40S subunit. eIF6 needs to be evicted from the 60S surface to allow the formation of active 80S ribosomes. This step is triggered by the joint action of the Shwachman-Bodian-Diamond syndrome (SBDS) protein and the GTPase Elongation Factor-like 1 (EFL1).2 3 Studies with the yeast orthologue suggest SBDS may act as a guanine nucleotide exchange factor (GEF) for EFL1 favouring a conformation of low affinity for guanosine diphosphate (GDP).4 5 Approximately 90% of SDS patients have pathogenic mutations in the SBDS gene and some SBDS missense mutants appear to have decreased affinity for EFL1 preventing the nucleotide exchange regulation the SBDS exerts on EFL1.4 6 While no human disease has yet been associated with mutations in EFL1, given the relationship between the SBDS and EFL1 proteins in ribosomal assembly, mutations in EFL1 appear to be an excellent candidate gene for patients with SDS-like phenotype but no identifiable mutation in SBDS.

Patients and methods

Patients

Two siblings, brother and sister, (individuals A-II-3 and A-II-4 in figure 1) from non-consanguineous parents of Mexican ancestry (family A) and four patients, two girls and two boys (individuals B-II-1, B-II-3, B-II-5, C-II-1 from families B and C in figure 2A), originating from two unrelated consanguineous Palestinian–Muslim families, were the subjects of this study. Their clinical and laboratory data are presented in table 1. All were born at term following an uneventful pregnancy. Patient A-II-3, the proband from family A was referred at 7 months of age for evaluation of developmental delay, small size and possible skeletal dysplasia. The neonatal period was complicated by hypotonia, subglottic stenosis, posterior laryngeal cleft and laryngomalacia, poor feeding and umbilical hernia. A skeletal survey revealed diffuse metaphyseal widening and irregularity, most prominent in the ribs and femurs (figure 1C,D). The patient experienced poor growth and intermittent diarrhoea and constipation. At 2 years of age, he was found to have pancreatic insufficiency with low pancreatic elastase and normal vitamin E and D levels but low vitamin A level and was prescribed pancreatic enzyme supplementation. The patient was hospitalised multiple times with severe respiratory infections. He was found to have an unspecified antibody deficiency with poor immune memory and therefore started on monthly intravenous immunoglobulin (IVIG) therapy. During one of the patient’s admissions with respiratory illness, he was found to have severe anaemia and neutropenia. The anaemia improved following resolution of his respiratory illness but neutropenia was persistent and of unclear aetiology. Bone marrow aspirate revealed myeloid and erythroid cells with orderly maturation and no overt dysplastic features. When seen for follow-up at 5 years of age, he had global developmental delay, severe myopia, mild hepatomegaly and mild bilateral genu varum and hindfoot valgus. The patient remained significantly small, with height of 86.3 cm (<1st percentile, −4.8 SD) and weight of 10.6 kg (<1st percentile, −4.1 SD). Head circumference remained normal at 52.4 cm (84th percentile). Negative test results obtained to that point included chromosome microarray, fragile X, sequence determination of COL10A gene sequencing for Schmidt type metaphyseal chondrodysplasia, DYM gene sequencing for Dyggve-Melchoir-Clausen syndrome, RMPP gene sequencing for cartilage hair hypoplasia, DEB chromosome breakage studies for Fanconi anaemia and SBDS.

Figure 1

Family A pedigree, M882K EFL1 genotype and skeletal anomalies. (A) Family pedigree; patients are represented by filled symbols and the genotype of the variants c. 2645T>A in the EFL1 gene is shown. (B) Photograph of the patient A-II-3. (C–D) X-ray imaging of the hip skeleton demonstrating irregularity of the metaphyses (arrows).

Figure 2

Family B and C pedigree, R1095Q EFL1 genotype and skeletal anomalies. (A) Families pedigrees and genotype. (B) short fingers. (C) X-ray of the pelvis and hips demonstrating irregularity of the femoral metaphyses (arrows). (D) Chest X-ray demonstrating irregularity of the humeral metaphyses (arrows) as well as cupping of the anterior ribs (arrowheads). (E) Imaging of the pancreas—transverse ultrasound image at the level of the upper abdomen demonstrating hyperechogenic pancreatic tissue (arrows) anteriorly to the splenic vein (sp) suggestive of fatty infiltration of the pancreas.

Table 1

Clinical course and haematological findings in six patients with SDS-like phenotype

Patient A-II-4, the younger sister of patient A-II-3 (figure 1A) was referred at 2 years of age due to concern of short stature, severe myopia, mild speech delay and bilateral genu varum. She had been growing at below the 5th percentile in height but did not have difficulty with weight gain and did not have a history of recurrent infections. At 3 years of age, her height was 84.6 cm (<1st percentile, −2.6 SD), weight was 14.3 kg (58th percentile) and head circumference was normal at 48.7 cm (52nd percentile). X-rays of the lower extremities and hips revealed metaphyseal changes similar to those seen in her brother. Laboratory studies confirmed pancreatic insufficiency, and she was started on pancreatic enzyme supplementation. Haematological studies noted intermittent neutropenia. Bone marrow aspirate revealed trilineage haematopoiesis with progressive maturation and no apparent dysplastic features or abnormal infiltrates. She had no evidence of the unspecified antibody deficiency characterised in her brother. Given her clinical history, imaging studies and laboratory findings, we felt she was affected with the same condition as her older brother but to a milder degree.

The four affected individuals from families B and C (figure 2A) had low birth weight (1400–2450 g, see table 1). During the first year of life, all four suffered from diarrhoea and steatorrhea secondary to exocrine pancreatic insufficiency, as indicated by low stool elastase levels, low blood lipase and elastase, decreased vitamin E and abnormal coagulation tests (prolonged prothrombin time/partial thromboplastin time (PT/PTT)). Physical examination revealed severe failure to thrive (FTT), generalised muscle hypotonia, microcephaly, high arch palate, low set ears, rhizomelic shortening of the limbs and short fingers (figure 2B). Patients B-II-1, B-II-3 and C-II-1 failed to achieve developmental milestones and died between 7 and 15 months of age. Patient B-II-5 was treated with pancreatic enzymes replacement since early infancy with a favourable response. At 15 months, his development was that of a 10-month old infant; he could sit unsupported, stand with support and transfer objects from one hand to the other; he had eye contact, repeated vowels and his communication skills reached joint attention. On examination, the muscle tone was normal, and there were neither pyramidal nor cerebellar signs. Laboratory investigations on families B and C patients revealed progressively profound neutropenia, progressive normocytic anaemia with low reticulocyte count and fluctuating thrombocytopenia, necessitating repeated blood and platelets transfusions. The bone marrow was hypocellular with trilineage haematopoiesis. Abdominal ultrasound revealed hyperechogenic pancreas, consistent with fatty infiltration (figure 2E). Skeletal X-rays demonstrated irregular metaphyses of the long bones and cupping of anterior ribs typical for chondro-dysostosis (figure 2C,D). Brain MRI of patient A-II-5 was normal.

Methods

Whole exome analysis

Informed consent for exome sequencing was obtained from the parents in the three families. Trio-exome analysis was performed in family A (unaffected parents and patient AII-3), whereas in families B and C exome analysis was performed for patient B-II-5 and for patient C-II-1. Exonic sequences were enriched using Agilent Sure Select XT Clinical Research Exome enrichment kit for family A trio analysis and using Agilent Sure Select 50 Mb V4 Kit for patients B-II-5 and C-II-1 (Agilent Technologies, Santa Clara, California, USA). Sequencing was performed using the HiSeq 2000 (Illumina, San Diego, California, USA) with 100 bp paired end reads. For family A trio, data analysis was performed using latest version of XomeAnalyzer (GeneDx, Gaithersburg, Maryland, USA). The general assertion criteria for variant classification are publicly available on the GeneDx ClinVar submission page (www.ncbi.nlm.nih.gov/clinvar/submitters/26957/). For the patients in families B and C, data analysis including read alignment and variant calling was performed by DNAnexus software (Palo Alto, California, USA), using the default parameters with the human genome assembly hg19 (GRCh37) as reference.

Sequence alignment

EFL1 orthologues from the model organisms Arabidopsis thaliana, Caenorhabditis elegans, Dictyostelium discoideum, Homo sapiens, Mus musculus, Ratus norvegicus, Gallus gallus, Saccharomyces cerevisiae, Schizosaccharomyces pombe and Drosophila melanogaster were identified in the Princeton Protein Orthology Database (P-POD, http://ppod.princeton.edu/). These sequences were used to extend our search and identify other EFL1 orthologues in the UniprotKB/Swiss-Prot database. The EF-2 and EF-G protein sequences were obtained from the annotated sequences in the UniprotKB/Swiss-Prot database. Multiple protein sequence alignment was performed using the program T-COFFEE. Sequence alignments were visualised using Jalview. 

Plasmid construction

Expression plasmids for the human EFL1-R1095Q and EFL1-M882K and S. cerevisiae Efl1-R1086Q and Efl1-L910K proteins were generated by single primer PCR mutagenesis using as template the plasmids described by García-Márquez et al.4 The vector pAMG1 was used for the complementation assay and consisted of plasmid pRS316 with the promoter and terminator regions of the ADH1 gene controlling the expression of the proteins. The S. cerevisiae Efl1 gene was cloned between the 5’ SalI and 3’ SpeI sites, while human EFL1 gene was cloned between the 5’ SalI and 3’ BamHI sites. Efl1 yeast deletion strain was generated by homologous recombination using appropriate PCR products to transform the yeast strain Y5538. For the genetic complementation assay, the EFL1/efl1::NatMX4 cells were transformed with the empty plasmid pAMG1 and those expressing yeast wild type Efl1 and mutants Efl1-R1086Q and L910K and were plated onto selective SD-Ura media for 3 days at 30°C. Transformants were sporulated in enriched sporulation media for 12 days at 23°C. Spores were cultured on solid medium that selects for the germination of MATa meiotic progeny carrying the indicated plasmid with either (1) no antibiotic, which enables germination of WT cells and (2) 0.1 mg/mL nourseothricin, which selects for efl1::NatMX4 cells as previously described by Baryshnikova et al.7 The plasmid pRS411-Tif6-GFP was used for the fluorescence localisation assays. The coding sequence of Tif6-GFP together with the 5’ and 3’ regulatory regions were amplified by PCR from genomic DNA obtained from the yeast strain Tif6-GFP (ThermoFisher Scientific). The insert was cloned into the SacI restriction site of plasmid pRS411 (CEN/MET15) and transformed in haploid efl1Δ yeast cells previously complemented with the corresponding pAMG1 plasmids. Cells were selected in SD-Ura-Met media. Fluorescent images were visualised using an Olympus BX51 Fluorescence microscope equipped with a charge-coupled device (CCD) camera and were processed using Adobe Photoshop CS6. Quantification of the fluorescence images was done using the ZEN 2 blue edition imaging software (Carl Zeiss MicroImaging) with a 54–72 square pixels range. Statistical data analysis was performed using a one-way analysis of variance (ANOVA) with the GraphPad Prism 7 program (GraphPad Software Company, La Jolla, USA).

Biomolecules expression and purification

Expression of recombinant S. cerevisiae SBDS protein was done in Escherichia coli C41 and purified as described in Shammas et al.8 Human and S. cerevisiae EFL1 wild-type and mutant proteins were expressed in S. cerevisiae BCY123 as described in Finch et al.2 After purification, the yield of the mutant EFL1 proteins corresponded to 2 mg protein/L of culture for yeast Efl1-R1086Q and 0.2 mg protein/L of culture for human EFL1-R1095Q. For the other protein variant, the yield of yeast Efl1-L910K was 0.07 mg protein/L of culture while human EFL1-M882K failed to express. These compare with the yield of the wild-type protein of 7 mg/L of culture for the yeast protein and 3.5 mg/L for the human EFL1. Ribosomal 60S subunits were purified from S. cerevisiae JD1370 as described in Acker et al.9

Circular dichroism and small angle X-ray scattering experiments

Circular dichroism (CD) wavelength scan measurements were followed at 25°C with a JASCO J-720 spectropolarimetre equipped with a Peltier temperature controller. CD spectra were recorded using a 1 mm cuvette and a protein concentration of 0.05 mg/mL for all the EFL1 proteins studied. Scan wavelength was followed from 260 to 195 nm, with an increase of 1 nm per step, an averaging time of 5 s and a spectral resolution of 1 nm. Small angle X-ray scattering (SAXS) experiments were performed at the beamline P12, PETRA DESY synchrotron (Hamburg, Germany). Data collection was carried out using a Pilatus 2M detector located 3 m from the sample at a wavelength of 1.24 Å. Samples (1.5–4 mg/mL) were filtered through 0.22 µM filters (Whatman) prior to data collection. Data analysis was done with the ATSAS suite and the program ScÅtter (www.bioisis.net).

Molecular dynamics simulations

Human EFL1 wild-type and mutant proteins were modelled by homology using the Modeller software and minimised. Coordinates corresponding to domains III–V of the full-length human EFL1 were used as template for the molecular dynamic simulation. Proteins were placed in a dodecahedron box and solvated with 13 000 TIP3P water molecules. The system was neutralised with potassium cations K+ and the system energy was minimised. After this, the system temperature and pressure were equilibrated at 300 K and 1 bar, respectively, during 150 ps. Subsequently, the molecular dynamics simulation trajectories were carried out at 300 K for 15 ns using the AMBER99SB-ILDN force field. Finally, the trajectory potentials obtained from each simulation were investigated using the GROMACS utilities such as g_rms to get the root mean square deviation (RMSD).

Activity assay

Phosphate released from guanosine triphosphate (GTP) hydrolysis mediated by the EFL1 proteins was monitored with the green malachite colorimetric assay.10 Reaction mixtures of 50 µL consisted of 2 µM of the corresponding yeast GTPase, 10 µM yeast SBDS, 0.2 µM yeast 60S ribosomal subunits and combinations thereof and 100 µM GTP in a buffer consisting of 50 mM Tris at pH 7.4, 5 mM MgCl2 and 300 mM NaCl. The reaction mixture was incubated for 40 min at 30°C. Reactions were stopped by the addition of 172 µL of 28 mM ammonium heptamolybdate in 2.1 M H2SO4 followed by 128 µL of a mixture containing 0.35% polyvinyl alcohol and 0.76 mM green malachite in a total volume of 500 µL. After 20 min at room temperature, absorbance was measured at 610 nm in a 96-well microplate reader VERSA max (Molecular Devices).

Results

For family A, exome analysis produced an average of 100 MB of reads per sample. Mean depth of coverage was X138 with >98% of targeted regions covered at least 10X. Data analysis revealed a homozygous variant of unknown significance, p.M882K (c.2645T>A: NM_024580.5) in exon 18 of EFL1 as a candidate gene possibly related to the patient’s features (figure 1). The p.M882K variant in EFL1 segregated with the disease in family A (figure 1A and see online supplementary figure 1) and was not carried by any of the 60 700 individuals whose exome analyses were deposited at ExAC (Exome Aggregation Consortium, Cambridge, Massachusetts, URL: http://exac.broadinstitute.org). The variant was further confirmed using Sanger sequencing.

Supplementary file 1

Exome analysis of patients B-II-5 and C-II-1 yielded 42.5 and 37.5 million mapped reads with a mean coverage of X67 and X75, respectively. Following alignment to the reference genome (Hg19) and variant calling, we performed a series of filtering steps under the hypothesis of a recessively inherited, homozygous, rare, causal allele. These included removing variants which were called less than X8, were off-target, heterozygous, synonymous, on the X chromosome, had MAF>1% at ExAC or MAF>3% at the Hadassah in-house database. Twenty-six and 10 variants remained, but given the identical clinical phenotype, we focused on hg19 chr15:g. 82422793C>T, NM_024580.5:c.3284G>A, p.(Arg1095Gln) (R1095Q) in the EFL1 gene. This variant segregated in the families, was carried by only two of the 60 700 individuals whose exome analyses were deposited at ExAC and was not present in our in-house database (~940 Moslem-Arab exome analyses).

The human EFL1 encodes a 1120 amino acid proteins, and along with the SBDS protein, is essential for the formation of mature functional ribosome. The EFL1 protein family is homologous to the elongation factor 2 (EF-2) family; human EFL1 and yeast Efl1 share 39% identity with each other and 28% and 30% identity with their respective EF-2 homologues. Alignment of EFL1 sequences showed that residue R1095 is contained within a RRRKGL consensus signature which is conserved among all EFL1 members suggesting an essential function for it (figure 3B). Alignment of representative members of EFL1, EF-2 and EF-G protein families showed that the position corresponding to residue M882 in human EFL1 only allows hydrophobic substitutions such as isoleucine and leucine but not basic substitutions as the mutation M882K reported here (figure 3A). Like EF-2, EFL1 is predicted to share a five-domain architecture with residues M882 and R1095 of human EFL1 located in domain IV of the protein (figure 3C). M882 equivalent residue in the yeast EF-2 paralogue is I676. Based on the crystal structure of yeast EF-2 (PDB 1N0U),11 residue I676 of EF-2 directly contacts R823 which corresponds to R1086 and R1095 in the yeast and human EFL1 orthologues, respectively (figure 3D). Assuming that the folding between the EFL1 and EF-2 family proteins is conserved, this suggests that the residues mutated in our patients interact with each other in the WT protein and that the mutations may disrupt this interaction. Furthermore, the C-terminus of Efl1 has been shown to be essential for its function such that deletions extending up to R1086 could only be rescued by extragenic mutations in Tif6.12 Suppressor mutations in TIF6 also developed in efl1Δ and sdo1Δ cells to alleviate the slow growth phenotype.3 13

Figure 3

Sequence alignment of representative EFL1 proteins, eukarya EF-2 and archaea EF-2 orthologues and prokarya EF-G neighbouring residues (A) M882 and (B) R1095 of human EFL1. The numbering corresponds to amino acid sequence of human EFL1. Colour indicates conservation of the physicochemical properties of the amino acid in the corresponding position. (C) Homology model of S cerevisiae Efl1: Grey—domain I, Red—domain II, Yellow—insertion, Cyan—domain III, Green—domain IV, Magenta—domain V, Light blue spheres—residue L910 (human M882), Dark blue spheres—residues R1086 (human R1095). (D) Close-up of the region surrounding residue I676 and R823 in S. cerevisiae EF-2 (equivalent residues to M882 and R1095 in human EFL1), PDB:1N0U.

To better understand the functional implications of these amino acid change, studies were undertaken using human and yeast EFL1 orthologues expressing the equivalent p.M882K and p.R1095Q variants. Genetic complementation of yeast efl1Δ cells with plasmids providing EFL1 L910K and EFL1 R1086Q as the only source of Efl1 was able to sustain normal growth of the cells (see online supplementary figure 2A). Western blot of the aforementioned complemented cells revealed comparable expression of the two mutant proteins with that of the wild type (see online supplementary figure 2B). Sequencing of the plasmids recovered from the complemented efl1Δ cells did not show any mutation in the coding sequence of Efl1 as previously observed for deletions in Efl1 C-terminus. These data suggest that the L910K and R1086Q mutations do not affect the expression level of the Efl1 protein and do not compromise cells growth. Since mutations in a given protein can potentially alter its fold, it was important to establish the impact mutations M882K and R1095Q might have on EFL1 structure. Determination of the folding and secondary structure content of the mutant proteins using CD revealed comparable far-UV CD spectra of the mutants and the WT EFL1 for the human and yeast orthologues (see online supplementary figure 3A–B). These findings suggest that the R1095Q and M882K mutations do not affect EFL1 folding. It is of note, however, that the spectra of the human and yeast EFL1s were not identical, indicating that small differences exist in the secondary structure content of the two orthologues. Since structural information of a protein is not limited to the canonical concepts of secondary and tertiary structure but includes also shape and flexibility, we studied conformational changes and long-range delocalised flexibility in solution using SAXS. Analysis of the SAXS data for the mutant R1086Q-Efl1 and WT-Efl1, by means of the normalised and dimensionless Kratky plot,14 a qualitative indicator of flexibility, revealed that they are both flexible compared with a compact protein such as bovine serum albumin (see online supplementary figure 3C). The calculated radius of gyration (Rg) corresponded to 6.85±0.1 nm for WT-Efl1 and 6.12±0.5 nm for R1086Q-Efl1 in the Guinier region of 0.5–1.3 Å−1. Only small differences in the flexibility between the WT and the mutant R1086Q-Efl1 were observed by visual inspection of the Kratky plot. Further flexibility analysis using the Porod-Debye fourth power law15 showed that the mutated Efl1 is less compact than its WT counterpart. The Porod exponent, which is a quantitative metric increasing with compactness, was 3.5 for the WT protein and 3.2 for the mutant one; of note, a Porod exponent of 3.7 suggests a fairly compact protein. These results suggest that both L910K-Efl1 and R1086Q-Efl1 proteins fold similarly to the WT, and that the R1086Q affects Efl1 flexibility. Additionally, using molecular dynamics simulations in explicit solvent we evaluated the possibility that, although folded, mutations in EFL1 may destabilise the hydrophobic core of the protein and promote a conformational change. The root-mean-square (RMS) of the WT structure indicates that domains III–V of human EFL1 are slightly flexible as the RMS fluctuates between 3–5 Å. This same behaviour was observed for the mutant R1095Q. In contrast, the hydrophobic core of mutant EFL1 M882K is less stable as indicted by the RMSD changing from 3 Å up to 9 Å during the simulation time (see online supplementary figure 3D). Together, these results suggest that EFL1 mutants are folded to a similar extent as the WT with no evident changes in their secondary structure content but are most probable less stable.

Supplementary file 2

Supplementary file 3

SDS mutations in the SBDS protein lead to the accumulation of 60S ribosomal subunits in the cytoplasm because they fail to recycle eIF6 from late cytoplasmic pre-60S ribosomes to the nucle(ol)us. The shuttling of eIF6 depends not only on the function of SBDS but also on that of EFL1.3 In the absence of Sdo1 (SBDS yeast orthologue) and Efl1, the slow growth phenotype of yeast cells is suppressed by mutations in Tif6 that debilitate the interaction with the pre-60S subunit.3 13 We examined the effect of the R1086Q and L910K mutations on the cellular distribution of Tif6-GFP in yeast efl1Δ cells. As previously reported, in WT EFL1 cells, Tif6-GFP localised predominantly to the nucleus.3 In contrast, when efl1Δ was complemented with a centromeric plasmid expressing R1086Q-Efl1 or L910K-Efl1 as the only source of Efl1, Tif6-GFP accumulated in the cytoplasm (figure 4A). We calculated the fluorescence intensity of Tif6-GFP using one-way ANOVA and noted the difference between the nucleus and the cytoplasm in cells containing Efl1 WT compared to mutants L910K and R1086Q to be significant, F(5, 204)=37, p=<0.0001. We further assessed the differences between the groups by multiple-comparison analysis using the Tukey’s and Bonferroni method. The results showed that the fluorescence intensity of Tif6-GFP in the cytoplasm of both mutants differed from that observed in the WT cells in a statistical significant manner (p<0.0001) (figure 4B). This same result was found when comparing the fluorescence intensity in the nucleus of the cells. Comparison of the Tif6-GFP fluorescence intensity of the nucleous and the cytoplasm of the mutant cells to each other showed no difference (p=0.063). As control, we also analyzed the fluorescence intensity of the dye 4′,6-Diamidino-2-phenylindole dihydrochloride which serves as a reporter of nucleic acids in the cellular nucleus. As expected, the multiple-comparison analysis showed statistical significant differences between the fluorescence intensity in the nucleus compared with the cytoplasm (p<0.0001) but not between the nuclei of the different yeast cells tested (p>0.996). This observation suggests that Tif6 is relocalised to the cytoplasm in the Efl1 L910K and R1086Q mutant cells most likely because Efl1 mutants fail to release Tif6 from the surface of the 60S subunit in the cytoplam.

Figure 4

Tif6-GFP localisation. (A) The distribution of Tif6-GFP in EFL1 wild-type, mutant EFL1 L910K and mutant EFL1 R0186Q yeast cells visualised by fluorescence microscopy. N, nucleus; C, cytoplasm; V, vacuole. (B) Quantification of cytoplasmic Tif6-GFP fluorescence.

Our current understanding of the mechanism of eIF6 release from the 60S subunit requires SBDS to trigger the exchange of GDP for GTP in EFL1.4 In the bound conformation to the sarcin ricin loop (SRL) of the 60S ribosomal subunit, EFL1 displaces eIF6 by competing for a common overlapping binding site.16 In addition, the binding of EFL1 to the SRL promotes GTP hydrolysis causing the dissociation of EFL1 and SBDS from the 60S subunit. The inability of R1086Q-Efl1 and L910K-Efl1 to release Tif6 from the surface of the 60S subunit could therefore be attributed to any of the following mechanisms: (1) a lack of binding of EFL1 to the SBDS protein and/or the 60S subunit, which would result in a failure to activate EFL1 GTPase activity,2 5 13 (2) impaired GTP hydrolysis or (3) failure of EFL1 to undergo the necessary conformational changes to expel eIF6. To address these mechanistic options, we first confirmed that R1086Q-Efl1 and L910K-Efl1 are still active as a GTPases by evaluating the release of inorganic phosphate with a colorimetric method using green malachite (figure 5). The mutant proteins were capable of releasing inorganic phosphate to similar extents as the WT protein. The interaction of R1086Q-Efl1 and L910K-Efl1 with SBDS and the 60S subunit resulted in an increase in the signal of released phosphate; this effect was additive when both biomolecules were present (figure 5). Notably, the low intrinsic GTPase activity of the mutant Efl1s and their activation by SBDS and the 60S subunit were comparable to those of the WT enzyme within experimental error (figure 5). These findings suggest that catalysis is not noticeably affected by the mutation. Since catalysis cannot occur without binding, the interaction of EFL1 with the guanine nucleotides, SBDS and the 60S subunit must be similar in magnitude to that of the WT protein.

Figure 5

SBDS-dependent and 60S-dependent stimulation of GTP hydrolysis by yeast wild-type Efl1 and mutants. Phosphate release was measured with the green malachite colorimetric assay. Each experiment was repeated three times and the average values with SD (bars) are presented.

The release of eIF6 from the surface of the 60S subunit is tightly linked to the hydrolysis of GTP by EFL1, and uncoupling of GTP hydrolysis from eIF6 release has been established as a cause for SDS.2 The results presented here suggest that Tif6 is retained on the surface of the 60S ribosomal subunit despite R1086Q-Efl1 and L910K-Efl1 being capable of hydrolysing GTP. As explained above, EFL1 undergoes a conformational change which enables its competition with eIF6 for an overlapping binding site on the 60S subunit surface. We speculate that both R1095Q-EFL1 and M822K-EFL1 proteins cannot couple these two processes and fail to transduce the conformational change triggered by GTP hydrolysis to release eIF6.

Discussion

Ribosomes are among the most complex molecular machines in the cell. Their synthesis is highly regulated by the coordinated action of several small nucleolar ribonucleoproteins and accessory proteins that are not part of the functional ribosome. As explained above, the maturation of the pre-60S and pre-40S ribosomal subunits requires their export from the nucleus to the cytoplasm, where accessory factors are removed and additional ribosomal proteins are incorporated. To prevent premature association of the ribosomal subunits in the nucleus, eIF6 (Tif6 in yeast) binds the 60S subunit. To release the eIF6, the GTPase EFL1 competes with it for a common overlapping binding site. SBDS facilitates a conformational change in EFL1 that triggers the exchange of GDP for GTP followed by GTP hydrolysis promoted by the sarcin-ricin loop of the 60S ribosomal subunit.

Disorders involving defects in the production or function of ribosomes are referred to as ribosomopathies. SDS is a highly variable, rare ribosomopathy caused by mutations in the SBDS gene and is characterised by haematopoietic abnormalities, exocrine pancreatic dysfunction and skeletal dysplasia. Intermittent or chronic neutropenia is the most common haematological manifestation, followed by anaemia and thrombocytopenia. Some patients progress to bone marrow failure, myelodysplastic syndrome and malignant transformation, with acute myelogenous leukaemia being the most common. Exocrine pancreatic dysfunction is generally the first presenting symptom in infancy. Short stature and metaphyseal dysplasia are the most frequent skeletal manifestations identified in SDS patients. Hepatomegaly and elevated liver enzymes are common but tend to normalise with age and rarely lead to chronic liver disease. Immunological abnormalities are also observed with patients experiencing recurrent opportunistic infections. SDS patients are also reported to have variable degrees of developmental delay and cognitive impairment. Other reported symptoms include congenital malformations of the eyes, ears, heart, genitourinary system, dental anomalies and endocrine dysfunction.17 18 As observed in the majority of SDS cases, our patients exhibited the typical combination of exocrine pancreatic dysfunction, haematopoietic abnormalities, short stature and metaphyseal dysplasia. Exome analysis was negative for SBDS gene mutations but revealed homozygosity for p.R1095Q in the ELF1 gene in 2 unrelated families and homozygosity for p.M882K in the EFL1 gene in another family. In the crystal structure of EF-2, residues I676 and R826 (equivalent to residues M882 and R1095 in human EFL1) interact with each other; this interaction is most likely conserved in the EFL1 protein family as it is also reproduced in a homology-based model of yeast Efl1. Here we show that although the mutations do not disrupt the folding of the EFL1 protein, Tif6 accumulates in the cytoplasm of mutant-Efl1 yeast cells rather than in the nucleus, as previously observed in Tif6 localisation studies of mutant SBDS protein. Uncoupling GTP hydrolysis by EFL1 from the subsequent release of eIF6 from the 60S surface has been established as a possible cause for SDS.2 Here we show that the EFL1 protein of patients with SDS-like symptoms cannot release eIF6 from the 60S surface despite its ability to hydrolyse GTP and respond to the activation by the SBDS protein and the 60S subunit. Since the eviction of eIF6 requires the movement of domains I–II and IV around domains III and V that act as a hinge in EFL1,16 we speculate that the EFL1 mutants cannot transduce GTP hydrolysis into the necessary conformational change to release eIF6. In turn, this may result in abnormal ribosomes with altered translating capabilities.

Due to the functional interplay between the EFL1 and SBDS proteins and the clinical similarity, we propose that the EFL1 variants described here are responsible for our patients’ phenotype. Similar to SDS, the initiation of pancreatic enzyme supplementation in our patients significantly improved their outcome. Patients with EFL1 mutations who survive infancy are in need of continuous monitoring for the appearance of haematological abnormalities, immune deficiencies and complications related to their skeletal dysplasia. Exome analysis or sequencing of EFL1 may identify additional mutations in the 10% of patients with SDS-like symptoms who are negative for mutations in the SBDS gene.

Acknowledgments

KK would like to thank GeneDx and Mingjuan Liao, PhD for sharing the Sanger sequencing data of the patients in family A. Beamtime was granted under projects ‘PGR2016 Messico’ with the contribution of the Ministero degli Affari Esteri e dalla Cooperazione Internazionale, Direzione Generale per la Promozione del Sistema Paese and ‘SAXS-528’ at beamline P12 in the PETRA III DESY synchrotron, Hamburg, Germany.

References

View Abstract

Footnotes

  • Contributors PS, KHK and MCF contributed equally. PS, OA, NS, ANE, TD and OE led patients’ recruitment, clinical phenotyping and genetic analysis. MCF, ABS, DS, DA, AG, AMG and NSP performed and analysed all functional studies. All the authors assisted in the assembly and editing of the manuscript.

  • Funding NSP acknowledges the financial support from DGAPA-PAPIIT project IN201615. PS and OA are supported by funding from the Deutsche Forschungsgemeinschaft (DECIDE, DFG WA 1597/4–1). This work was supported in part by the Trudy Mandel Louis Charitable Trust to OE.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval The Ethical Committee of Hadassah Medical Center.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.