|Home | About | Journals | Submit | Contact Us | Français|
Epigenetic regulation of gene expression, through covalent modification of histones, is a key process controlling growth and development. Accordingly, the transcription factors regulating these processes are important targets of genetic diseases. However, surprisingly little is known about the relationship between aberrant epigenetic states, the cellular process affected, and their phenotypic consequences. By chromosomal breakpoint mapping in a patient with a Noonan syndrome–like phenotype that encompassed short stature, blepharoptosis, and attention deficit hyperactivity disorder, we identified haploinsufficiency of the histone acetyltransferase gene MYST histone acetyltransferase (monocytic leukemia) 4 (MYST4), as the underlying cause of the phenotype. Using acetylation, whole genome expression, and ChIP studies in cells from the patient, cell lines in which MYST4 expression was knocked down using siRNA, and the Myst4 querkopf mouse, we found that H3 acetylation is important for neural, craniofacial, and skeletal morphogenesis, mainly through its ability to specifically regulating the MAPK signaling pathway. This finding further elucidates the complex role of histone modifications in mammalian development and adds what we believe to be a new mechanism to the pathogenic phenotypes resulting from misregulation of the RAS signaling pathway.
Regulation of gene transcription includes dynamic structural changes of the packaged DNA, the chromatin. Nucleosomes, the basic units of chromatin, are composed as histone octamers, containing 2 copies of each subunit, H2A, H2B, H3, and H4, wrapped 1.6 times by DNA (1). The surface of nucleosomes exposes the amino-terminal tails of histones, which are the main sites for posttranslational modifications, such as acetylation, phosphorylation, methylation, ubiquitylation, and sumoylation (2).
Histone acetylation at promoters is a hallmark of actively transcribed genes, and acetylation levels are correlated with transcription rates (2). Of note, histone acetylation is a dynamic process established by complexes of histone acetyltransferases (HATs) and histone deacetylases (3). Abnormal histone acetylation was found as a feature of human cancer and neurological diseases (4, 5), and the essential role played by histone acetylation during mammalian development is illustrated by the observation that mice lacking H4K16 acetylation, as a result of a mutation in the MYST HAT 1 (MYST1) gene, fail to develop beyond the blastocyst stage (6). Despite major progress in the identification of lysine residues of protein complexes targeted by acetylation, the effect of individual HATs on regulatory networks is still largely unknown (7).
In this report, we found the MYST4 gene, which encodes 1 out of 5 members of the MYST HAT family, disrupted by a balanced chromosomal translocation in a patient with clinical features resembling Noonan syndrome. Noonan syndrome is a common monogenic condition, occurring in 1:1,000–1:2,500 individuals (8). It is characterized by postnatal reduced growth, cardiac defects, distinctive facial dysmorphism, and variable cognitive deficits. The main underlying pathomechanism is hyperactivation of the RAS/MAPK transduction pathway, with a heterozygous germline mutation in PTPN11, SOS1, KRAS, RAF1, SHOC2, NRAS, BRAF, or MEK1 being detected in approximately 70% of patients with this disorder or a clinically related phenotype. These genes encode signal transducers that promote RAS/MAPK signaling, the majority regulated by complex autoinhibitory mechanisms that are impaired by mutations (8).
We present data to demonstrate that the HAT MYST4 specifically regulates the MAPK signaling pathway via H3 acetylation in vivo and in vitro in human and mice. We show that MYST4 deficiency leads to hyperphosphorylation of proteins of the MAPK signaling pathway and, like mutations of previously identified disease genes from this pathway, underlies a phenotype with clinical features overlapping Noonan syndrome. Such surprisingly specific effects of a single chromatin factor should prompt consideration of epigenetic factors, even in monogenic human disorders that typically have a characteristic pathology.
We identified a patient with an apparently balanced de novo chromosomal translocation t(10;13)(q22.3;q34) in our cohort of patients with Noonan syndrome–like features, in whom mutations in the known disease genes from the RAS/MAPK pathway were excluded. The boy was born at term with normal birth weight (3,500 g, –0.28 standard deviation score [SDS]) and length (52 cm, –0.25 SDS) as the second child to healthy, nonconsanguineous parents of European descent. During infancy, his growth and weight fell below the third centile. When referred at age 6.2 years, he presented with proportionately short stature (101 cm, –4.2 SDS; 14 kg, BMI 13.7; head circumference, 49 cm, –2.44 SDS), with marked retarded bone age, ligamentous laxity, attention deficit hyperactivity disorder with learning disability (IQ 75–80), and distinct facial features with blepharophimosis, ptosis, high arched eyebrows, low-set ears with overfolded helix and fleshy lobe, smooth philtrum, retrognathia, and a high arched palate (Figure (Figure1,1, A–D). A cerebral MRI scan at the age of 5.3 years was reported normal. Endocrinological evaluation was suggestive of growth hormone deficiency, but the mother declined further investigations. No disease-related lesions within the entire coding sequence of the known Noonan syndrome genes had been identified. While cardiac anomalies were not present, he had a suggestive facial appearance, short stature, broad thorax, and mild mental retardation, fulfilling the established van der Burgt criteria for Noonan syndrome (9).
Karyotyping using GTG banding in the patient and both parents revealed a balanced de novo chromosomal translocation t(10;13)(q22.3;q34) in the patient. Genome-wide molecular karyotyping using an Affymetrix GeneChip Mapping 6.0 array excluded relevant copy number changes, suggesting a disruption of a gene at one of the translocation breakpoints associated with the phenotype. We mapped the translocation breakpoints using FISH analysis and long-range PCR (Figure (Figure1,1, E and F) and found the breakpoint in 10q22.3 to disrupt the MYST4 gene within intron 3 after the first coding exon. A 50% reduction of mRNA expression levels of this gene was confirmed by quantitative RT-PCR in the patient’s leucocytes (Figure (Figure1G).1G). The breakpoint in 13q34 was not located within a gene previously associated with Noonan syndrome, and the TUBGCP3 gene, which is located 17.2-kb downstream, showed normal expression levels in the patient (Figure (Figure1G).1G). Western blot analysis (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI43428DS1) and gene structure prediction using GENSCAN (10) showed no evidence of an aberrant MYST4 protein.
Mutation scanning of the entire coding sequence of the MYST4 gene was then performed in the patient to exclude a second hit, suggesting haploinsufficiency as a possible pathogenetic mechanism. No disease-associated mutation was identified by sequencing MYST4 in an additional 131 subjects who had the clinical features associated with Noonan syndrome but without mutations in previously identified disease genes. This showed that intragenic mutations in MYST4 do not represent a common event in Noonan syndrome and further illustrated the genetic heterogeneity underlying the common clinical features of this syndrome.
MYST4 belongs to the family of MYST HATs characterized by a highly conserved lysine acetyltransferase domain (Figure (Figure2A)2A) shared by MOZ, Ybf2/Sas3, Sas2 and TIP60 (MYST3) (11–13). The human MYST family has 5 members, making it the largest family of lysine acetyltransferases (14, 15). Despite a high homology between MYST3 and MYST4 (16, 17), studies in mice and zebrafish revealed that they are not functionally redundant (11, 17–21).
To determine the histone acetylation effect in our patient, we generated a lymphoblastoid cell line and examined the level of histone acetylation (Figure (Figure2B).2B). In line with recent data indicating that MYST3 primarily acetylates histone 3, at least at Hox loci (22), we observed significantly reduced H3 acetylation but unchanged H4 acetylation in these cells lacking MYST4 (Figure (Figure2C). 2C).
To further investigate the role of MYST4 in acetylation, and to exclude secondary effects on the histone acetylation by immortalization of the patient’s lymphoblastoid cell line, we also performed siRNA knockdown experiments in HEK293 and in HeLa cell lines. We used 3 independent siRNAs to exclude off-target effects; each siRNA transfection experiment resulted in a reduction in the level of MYST4 mRNA of between 60%–80% (Figure (Figure3A).3A). The reduction in MYST4 levels was similar in both HEK293 and HeLa cells. Reduction in MYST4 levels showed a substantial reduction in global histone 3 acetylation levels, with HEK293 cells showing a slightly more pronounced effect than HeLa cells. The extent of the effect of MYST4 deficiency strongly correlates with the reduction of H3 acetylation (Figure (Figure3B).3B). No change in global histone 4 acetylation levels could be detected as a result of transfection of these siRNAs (Figure (Figure3C).3C). From these results, we conclude that MYST4, like the closely related MYST3 protein, primarily acts on histone 3.
Mice homozygous for a hypomorphic gene trap allele of Myst4, known as Querkopf mice (Qkfgt/gt mice), also showed a failure-to-thrive phenotype (Figure (Figure4,4, A–C), ventrally positioned ears, small eyes with frequently occluded nasolacrimal duct (Figure (Figure4,4, D and E), reduced muscle mass and body fat, and retarded suture closure as a sign of retarded bone age (17). In addition, compared with those of wild-type mice, mutant mouse brains were shorter and reduced in size, had small olfactory bulbs and paucity of large pyramidal cells, and showed a neuronal migration defect, resulting in reduced size of the cortical plate (17). Furthermore, QKF is required for the establishment and maintenance of adult neural stem cells and neurogenesis in the adult (23, 24). The Qkfgt allele is due to a transgenic insertion that reduces the mRNA level such that homozygous mice have approximately 10% of wild-type mRNA levels. In the light of the phenotype of our patient, we also documented proportionate short length and small processes of the lower jaws in the Qkfgt/gt mouse (Figure (Figure4,4, F and G). In addition, mice heterozygous for an N-terminal deletion in the MYST4 protein showed a 15% (P < 0.0001) reduction in body weight at 5 to 7 weeks of age. Analysis of the lacZ-reporter gene activity showed high expression in mesenchyme surrounding cartilage primordia of the developing skeletal system, the dorsal telencephalon, and the adult parietal cortex and hippocampus (Figure (Figure4,4, H–K). Subtle changes in the columnization of the growth plate seen in the histological presentation might explain the growth phenotype (Figures (Figures44 and and5).5). Thus, the Qkfgt/gt mouse resembles the human phenotype in growth retardation, facial dysmorphism, and developmental brain anomalies.
Since the NRH domain, which might influence the HAT activity, differs between the 3 MYST4 transcripts (16), developmental stage–specific and tissue-specific functions were suggested for the different isoforms (25). We therefore analyzed total MYST4 expression and the proportion of each isoform in various human tissues by transcript-specific real-time PCR (Supplemental Figure 2). In accordance with the phenotype in humans and mice and the published data from mouse in situ assays, we observed, by far, the highest expression levels in fetal whole brain extracts, while in adult tissues, whole brain, skeletal muscle, and pancreas showed the highest levels. The pattern of the isoform distribution changed in most of the analyzed tissues but again most remarkably in the skeletal muscle and the brain (Supplemental Figure 2, C and D).
To uncover the target genes and involved pathways of MYST4 histone acetylation, we investigated the Qkfgt/gt mouse as well as HEK293 and HeLa MYST4 siRNA cell lines by genome-wide expression studies and the latter also by ChIP-CHIP experiments. Although HATs might represent general modulators of transcriptional activation by ubiquitous acetylation of histones, of those genes with a fold change of less than –1.5 or more than 1.5, we found only 689 genes differentially expressed in the HEK/HeLa MYST4 siRNA approach (Supplemental Figure 3), 329 genes in the mouse Qkfgt/gt versus wild-type adult dorsal cortex, and 619 genes in the mouse Qkfgt/gt versus wild-type E12.5 dorsal telencephalon. With reference to the human cell lines, 510 (74%) differentially expressed genes were downregulated (Supplemental Table 1), while in the adult Qkfgt/gt mouse cerebral cortex only 110 genes (33%) were downregulated and, in the E12.5 Qkfgt/gt mouse dorsal telencephalon, 345 genes (56%) were downregulated (Supplemental Tables 2 and 3). The expression level changes in the mouse of up to 6.11 were comparable to those from the HEK/HeLa approach (up to 5.75). Consistent with the high MYST4 expression level in brain, we found the largest proportion of differentially expressed genes to be ordinarily predominantly expressed in the brain but also in placenta, epithelium, and bone marrow (Supplemental Figure 4A and Supplemental Table 4). Gene ontology analysis of the differentially expressed genes revealed similar distinct categories of intracellular component and molecular function in all cell and tissue comparisons (Supplemental Figure 4, B–G, and Supplemental Tables 5–10). We found the highest percentage of genes coding for proteins located in the cytoplasm (42%–44%), nucleus (26%–35%), and endoplasmatic reticulum (6%–20%) (Supplemental Figure 4, B, D, and F) and functionally involved in general binding (40%–56%), protein binding (26%–27%), and catalytic activity (16%–27%) (Supplemental Figure 4, C, E, and G).
We also analyzed the differentially expressed genes for groups of molecular interaction and reaction networks with reference to the Kyoto encyclopedia of genes and genomes (KEGG) pathway database (Supplemental Tables 11–16 and ref. 26). For the combined HEK/HeLa siRNA comparison, we found significant enrichment of genes in 10 KEGG pathways involved in signaling processes, such as insulin, MAPK, Wnt, and GnRH signaling pathways, but also in structural processes of actin cytoskeleton and glycan biosynthesis and in regulation of apoptosis (Table (Table1).1). These networks are part of basic regulatory mechanisms of brain development, cell assembly, and limb formation.
In the adult cerebral cortex of Qkfgt/gt versus wild-type mouse comparison, we discovered 4 significantly altered networks of neuronal development, namely the neuroactive ligand-receptor interaction, axon guidance, calcium signaling, and the MAPK signaling pathways (Table (Table2).2). In the dorsal telencephalon of the E12.5 Qkfgt/gt mouse, our data revealed 7 enriched KEGG pathways (Table (Table3).3). Of note, we detected the MAPK signaling pathway in all 3 approaches, indicating this pathway as a major target of MYST4. The insulin, Wnt, and GnRH signaling pathways were enriched in the HEK/HeLa approach and the dorsal telencephalon of the E12.5 Qkfgt/gt mouse, demonstrating an involvement of MYST4 in embryonic development, especially of the brain, which has been proposed before (17, 21).
In order to distinguish between primary and secondary target genes of MYST4, we performed genome-wide ChIP using an anti-MYST4 antibody on HEK293 and HeLa cell lines (Supplemental Table 17 and Supplemental Figure 5). Antibody specificity was confirmed by Western blotting, showing identification of protein bands consistent in size with the 3 known protein coding transcripts of MYST4 only (Supplemental Figure 1). Using Pathway-Express and DAVID, 4,540 genes showing a P value of less than 0.05 were used to determine involved pathways. Ten pathways were significantly enriched in the Pathway-Express analysis, namely MAPK signaling, mTOR signaling, regulation of actin cytoskeleton, gap junction, phosphatidylinositol signaling, long-term depression, focal adhesion, circadian rhythm, basal cell carcinoma, and thyroid cancer pathways (Supplemental Table 18). Of these, only the MAPK signaling and the thyroid cancer pathways were also enriched in the DAVID functional annotation of KEGG pathways, which, in contrast to Pathway-Express, does not take into account the probability of ChIP-CHIP binding for each region (model-based analysis of tiling arrays [MAT] scores) (Table (Table44 and Supplemental Table 19).
We repeated the pathway analyses using both programs with the 467 genes that were significantly differentially expressed in the human cell lines and in addition showed MAT score P values of less than 0.05 (Supplemental Table 20). Using these genes, both programs again revealed only the MAPK signaling and the thyroid cancer pathways as both being significantly regulated by MYST4 (Table (Table55 and Supplemental Tables 21–23). These results were independently confirmed by quantitative real-time PCR of 9 significantly downregulated genes and 2 unaltered control genes of the MAPK signaling pathway (Supplemental Table 24). Our finding that expression was not altered in all genes with MYST4 binding revealed by ChIP-CHIP may be explained by the fact that MYST4 is only 1 out of several proteins that modulate gene expression. On the other hand, many differentially expressed genes did not show MYST4 binding, indicating an indirect regulation by MYST4. The 467 differentially expressed genes with a high probability of MYST4 binding indicative of direct regulation showed a similar tissue expression pattern as that of the MYST4 gene (Supplemental Table 25 and Supplemental Figure 5C). Supporting a mainly transcriptional activation of MYST4 (Supplemental Figure 5B), 80.9% were downregulated.
To investigate whether MYST4 haploinsufficiency resulted in dysregulation of the MAPK signaling pathway, we explored the phosphorylation levels of MEK1/2 (MAP2K1/2) and ERK1/2 (MAPK1/3) in the patient’s lymphoblastoid cell line. As shown in Supplemental Figure 6, similar to what is generally observed for most Noonan syndrome–causing SHP2, SOS1 and RAF1 mutants, an increased phosphorylation of MEK1/2 and ERK1/2 was documented. In addition, we also found enhanced AKT phosphorylation (Figure (Figure7).7). To demonstrate that this effect is indeed MYST4 dependent, we transiently transfected the patient’s cell line with an MYST4 wild-type construct at the extent of mean MYST4 expression levels. This resulted in complete rescue of the increased phosphorylation to a phosphorylated to nonphosphorylated protein ratio, as observed in wild-type cell lines (Figure (Figure6). 6).
We identified disruption of the MYST4 gene as what we believe to be a novel, rare cause of a Noonan syndrome–like phenotype. Our extensive studies in vivo and in vitro in humans and mice reveal that the H3 histone acetylase MYST4 primarily regulates the MAPK signaling pathway. In general, MAPK activity influences many different developmental stages by regulation of cell proliferation, differentiation, migration, and apoptosis (27–29). Four main MAPK pathways have been recognized: ERK1/2, c-Jun, p38 (MAPK14), and ERK-5 (MAPK7). Mutations in the genes PTPN11, SOS1, KRAS, RAF1, SHOC2, NRAS, BRAF, and MEK1 from the classical RAS/MAPK transduction pathway (ERK1/2) are detected in 70%–75% of patients with Noonan syndrome or closely related phenotypes. These disorders share short stature, intellectual deficits, cardiac defects, and distinctive facial dysmorphism. Other associated features commonly include ectodermal and skeletal anomalies, cryptorchism, lymphatic dysplasias, mild bleeding diathesis, and, rarely, predisposition to childhood hematologic malignancies. The underlying biochemical defect is increased signal traffic through the MAPK cascade, and increased phosphorylation of MEK1/2 and ERK has been shown for patients with Noonan syndrome carrying a mutation in N-RAS (30). A similarly enhanced phosphorylation of MEK1/2 and ERK was confirmed in our patient with MYST4 haploinsufficiency, who also showed enhanced AKT phosphorylation as part of the RAS effector pathway (Figures (Figures66 and and7).7). We also strengthened the MYST4 effect on the phosphorylation of these proteins by transfecting the patient cell line with an MYST4 wild-type construct and confirming normalization of the phosphorylation ratio.
Investigation of genes with ChIP-CHIP binding and significant differential expression by MYST4 yielded 10 downregulated genes (AKT2, CACNA2D2, ELK1, MAP2K3, MAX, MRAS, PPP3CB, PRKX, SMAD4, and SRF) modulating the ERK1/2, c-Jun, and p38 branches of the MAPK signaling pathway (Supplemental Tables 23 and 24, and Figure Figure7).7). Notably, MRAS and PRKX, the latter highly expressed in adult and fetal brain, are modulators of the transcription factors ELK1 and SRF, which are part of the ERK1/2 pathway. Therefore, our data suggest that dysregulation of ERK1/2 signaling in general leads to Noonan syndrome–like features, as seen in our patient and the Qkfgt/gt mouse (Figure (Figure1,1, A–D, and Figure Figure4).4). MRAS is also involved in membrane targeting of the SHOC2-PP1C complex, which is required for efficient RAF1 translocation to the membrane and activation (31). AKT and ERK are major regulators on early limb bud formation (32–34), but MYST4 haploinsufficiency in our patient did not result in a specific limb phenotype as distinct from an overall reduction in skeletal growth. AKT2 and MAP2K3 are part of the p38 and JNK pathway that is involved in ELK1 and MAX regulation. MAP2K3 phosphorylates and activates p38 during skeletal muscle development, neuronal regulation, and chondrocyte differentiation (35, 36). The involvement of MYST4 in chondrocyte differentiation is shown by the aberrant columnar structure of the hypertrophic region of the Qkfgt/gt growth plate (Figure (Figure5).5). Interestingly, AKT2 also plays a key role in myoblast differentiation and interacts with p38 and PI 3-kinase, regulating cell cycle progression and apoptosis (37). Smad4 is involved in BMP-Smad signaling, which regulates stem cell renewal, cell proliferation, differentiation, migration, and apoptosis and controls embryo development and postnatal tissue homeostasis (38, 39).
Somatic, oncogenic transformations due to MYST3 or MYST4 transfusion genes caused by recurrent chromosomal translocations are observed in leukemia and uterine leiomyomata cells, but target genes are unknown (40–43). Here, the generation of a fusion protein with hyperactivated acetyltransferase activity seems to be the underlying cause. While oncogenic transformation was neither seen associated with MYST4 haploinsufficiency in our patient nor in the Qkfgt/gt mouse, increased tumor risk is a feature of Rubinstein-Taybi syndrome, the only known human phenotype with defective HAT, the CREB binding protein (5). Of note, both the Noonan syndrome–like syndrome reported here and the Rubinstein-Taybi syndrome result from haploinsufficiency; both have strong phenotypic similarities, presenting with short stature, intellectual disability, and facial dysmorphism, including ptosis.
Even though heterozygous Qkfgt mice with approximately 55% MYST4 mRNA show mild but recognizable phenotypical features, including reduced body size, the range of phenotypes seen in our MYST4 haploinsufficient patient is more clearly identifiable in the more severely affected homozygous querkopf mice. The mouse phenotype, which results from a hypomorphic allele producing approximately 10% normal mRNA, reflects the failure-to-thrive together with craniofacial and skeletal anomalies. Interestingly, cardiac defects were neither noted in our patient nor the Qkfgt/gt mouse. However, only 65%–86% of patients with PTPN11 mutations (44, 45), 60% of patients with KRAS mutations (46), and 76% of patients with SOS1 mutations show cardiac defects (47), indicating reduced penetrance for the heart phenotype. In addition, myeloproliferative disease, as described in the Noonan syndrome Ptpn11 (D61G/+) mouse, was neither observed in our patient nor in the Qkfgt/gt mouse. As MYST3 is involved in hematopoiesis and leukemia, any role that MYST4 might play in the hematopoietic system might be performed equally by MYST3 (11, 20). This might also explain morphological differences, as observed in the cerebral structural anomalies of Noonan syndrome and MYST4 mice models.
We note that inhibitors of lysine deacetylases, mimicking HATs, are used to induce pluripotent stem cells (48) and are widely emerging as important drug targets for the treatment of a range of cancer and neurodegenerative diseases (49, 50). With the knowledge of the acetylation targets and the pathways modulated, new applications for lysine deacetylases may evolve and resulting effects may become more predictable.
We conclude that the Noonan syndrome–like phenotype in the patient with MYST4 haploinsufficiency as well as in the Qkfgt/gt mouse is explained by disturbed MAPK signaling. Although disturbed epigenetic modifications might be expected to have a major general effect, our data indicate that failure of a distinct HAT, such as MYST4, acetylating the histone H3, may specifically affect only a certain pathway. Such epigenetic effects should therefore be considered in genetically heterogeneous conditions like Noonan syndrome and may represent a future target for novel therapeutic strategies.
Region-specific BAC clones for FISH mapping were selected from the NCBI and UCSC genome browsers. FISH analysis was performed using metaphase chromosomes prepared from the patient’s cell suspension and lymphoblastoid cell lines. Genomic BAC DNA was fluorescently labeled with Cy3-dCTP (Amersham Biosciences) by standard nick translation with the DIG-Nick-Translation Kit (Roche). The probes were blocked with Cot-1 DNA. Metaphase spreads were hybridized at 37°C overnight with the Cy3-labeled BAC probes and a specific FluoroX-labeled (Amersham Biosciences) subtelomeric control probe. After posthybridization washes, chromosomes were counterstained with DAPI (Serva). Images were captured on a Zeiss Axioplan 2 microscope with a CCD camera and processed with the Isis Software (Metasystems).
All long-template PCRs were performed with the Expand Long Template PCR-System (Roche Diagnostics) according to the manufacturer instructions, with primer pairs (480K16-F1-F ACCATTAGGTGTCCATTTTTAAGAAACA, 480K16-F1-R CAGCTTACGGAGAATGTAGGAGACTTAG, 88E10-F5-F TGCTTTAGCTAGAATAGGGACAGAAAGA, 88E10-F5-R TGCACTACAGTCACAGCACAATAAAATA) chosen from the genomic sequence of the breakpoint-spanning BAC clones. Purified long-template PCR-products (QIAquick PCR Purification Kit, Qiagen) were labeled with Cy3-dCTPs by standard nick translation as described above, and FISH was performed on metaphase spreads.
Primer pairs for breakpoint-spanning PCRs were chosen from the genomic sequence of the breakpoint regions on chromosomes 10 and 13. Different combinations of forward and reverse primers were used to amplify a breakpoint-spanning PCR product with the Expand 20kbPlus PCR-System, dNTPack (Roche Diagnostics), according to the manufacturer’s instructions.
HeLa and HEK293 cells were maintained in DMEM/HAM’s F12 (BIOCHROM AG), supplemented with 10% heat-inactivated fetal bovine serum (BIOCHROM AG) and 1% penicillin/streptomycin (10,000 U/10,000 μg/ml) (BIOCHROM AG) at 37°C, 5% CO2, 91% relative air moisture. Three independent RNAi siRNAs targeting MYST4 mRNA (MYST4 HSS118880, HSS177468 and HSS177469; Invitrogen) were validated by real-time RT-PCR. Cells were then transfected with 1 out of 3 different Stealth RNAi siRNA targeting MYST4 mRNA (HSS118879, HSS118880, HSS118881; Invitrogen) or Scrambled Negative Control Stealth RNA (Invitrogen) according to the manufacturer’s instructions. Medium was changed to growth medium without antibiotics 12 hours after transfection. Cells were harvested 51.75 hours after transfection. After evaluation of siRNA efficiency by quantitative RT-PCR and histone acetylation assay, HSS118880 was used for further studies.
Histone extraction and detection of global histone H3/H4 acetylation were performed using EpiQuik Global Histone H3/H4 Acetylation Assay Kits, according to the manufacturer’s instructions (Epigentek), with an input of 1 μg of histone proteins extracted from lymphoblastoid cell lines of the patient and 3 unrelated healthy controls and HEK293 and HeLa cells. Data analysis was performed according to the manufacturer’s instruction. Acetylation levels were calculated for the patient with reference to the 3 healthy controls and for the HEK293 and HeLa MYST4 siRNA cell lines with reference to Scrambled Negative Control Stealth RNA transfected cell lines.
Histone extraction and detection of HAT activity were performed using the EpiQuik HAT Activity/Inhibition Assay Kit, according to the manufacturer’s instructions (Epigentek). Briefly, histone substrate was stably spotted on wells where active HATs bind and acetylate histone substrate. The acetylated substrate was colorimetrically quantified after binding of high-affinity antiacetylated histone antibody.
Harvested cell pellets were washed using PBS. RNA was isolated using the QIAcube instrument in combination with the RNeasy Mini Kit, QIAshredder, and RNase-Free DNase Set, according to the manufacturer’s instructions (QIAGEN).
cDNA was synthesized using the SuperScript II Reverse Transcriptase Kit with random hexamer primers (Invitrogen) from RNA extracted from fresh blood samples collected in PAXgene tubes (Becton Dickinson) or from cell cultures.
For relative expression analysis of MYST4/MORF isoforms, real-time PCR primers and mgb probes were designed according to the manufacturer’s instructions (Applied Biosystems) to obtain primer/probe sets specific for each isoform: MORF forward (F), TGTCTGTAACCAGTGATGAAGGA; MORF probe (P), 6FAM-TCACCTGATACTGAAATAA; MORFα F, AAAAAGGTCTCTCAGAAACAGTCATG; MORFα P, 6FAM-TGTTGGCTACAGATACTGAA; MORFβ F, GAGCTTGACAGACGGAAGGATT; MORFβ P, 6FAM-CAGGATGATGATACTGAAATA; and all isoforms reverse, CATCTGCACTTTCTTGTTTGATGTT.
The tissue expression pattern was determined using 4 commercial available cDNA panels (Human Fetal MTCTM Panel and Human MTCTM Panel I [both from Clonetech]; C8244525 and C8234504 [both from BioChain Institute]). All real-time RT-PCRs were performed in quadruplicates in 384-well plates, with a final volume of 20 μl each on an ABI 7900HT, using the TaqMan Gene Expression Mastermix, according to the manufacturer’s instructions (Applied Biosystems). The expression levels of the individual isoforms were calculated using the ΔΔCt method with 2 endogenous controls (B2m, phosphoglyceratekinase 1 [PGK1]). Levels of significance were determined by a 2-sample equal-variance t test applied to the log-expression values.
All exon and flanking intron sequences of MYST4 gene, except the 3ι UTR, were amplified from patient DNA by PCR using intronic primers designed with the Primer3 software. The PCR products were purified with the AMPure Kit (Agencourt) on a Biomek NX instrument (Beckman Coulter). Sequencing reactions were performed on both strands using the BigDye Terminator Cycle Sequencing Kit v3.1 (Applied Biosystems) according to the manufacturer’s instructions. After purification using the CleanSEQ Kit (Agencourt), the products were analyzed on an ABI Genetic Analyzer 3730 (Applied Biosystems), and the traces were evaluated using the SeqPilot 18.104.22.168 software (JSI Medical Systems). Variants were explored against the SNP database (NM_012330.2; http://www.ncbi. nlm.nih.gov/projects/SNP/).
Harvested cell pellets were washed using PBS and frozen at –70°C for a minimum of 1 hour. After thawing on ice, pellets were resuspended in lysis buffer (0.01 M Tris; 0.15 M NaCl; 1% Triton-X, pH 7.5) and shook at 4°C for 20 minutes. Shearing of cells was performed using a 14-gauge needle. Lysates were stored at –70°C.
Western blot analysis using protein lysates extracted from lymphoblastoid cell lines of the patient, HeLa and HEK293 cells, respectively, transfected with either Stealth RNA siRNA targeting MYST4 mRNA (MYST4 HSS118880, Invitrogen) or Scrambled Negative Control Stealth RNA (Invitrogen) was performed. Electrophoresis was performed on a 3%–8% Tris-Acetate gel using the NuPage Novex system from Invitrogen. The KAT6B/MORF antibody (ab58823, Abcam) and the β-actin antibody (ab8227, Abcam) were used as loading controls. As secondary antibody, we used the HRP-conjugated goat anti-rabbit IgG antibody (170-6515, Bio-Rad).
Mice heterozygous for a mutation in the Querkopf (also known as Qkf, Myst4, Morf, and Kat6b) gene, Qkfgt/+ mice (17), were used to generate E15.5 and newborn Qkfgt/gt and Qkf+/+ control pups. Skeletal preparations were conducted as described previously (51), with modifications. In brief, bodies were eviscerated; fixed in ethanol and acetone for 4 days each; rinsed in H2O; stained in a solution of 0.015% Alcian Bue 8GX (Sigma-Aldrich), 0.005% Alizarin Red S (Sigma-Aldrich), and 0.05% acetic acid in H2O for 10 days; and then cleared in a solution of 1% potassium hydroxide and 20% glycerol in H2O, followed by further clearing through 40%, 60%, and 80% glycerol in H2O.
β-Galactosidase staining was conducted as described previously (52), and in situ hybridization was carried out as described previously (17). Embryos and pups were fixed in 4% paraformaldehyde and Bouin’s fixative, respectively, and processed for serial paraffin sections as described previously (53). Each block of 4 consecutive sections of the hind limb was stained with (a) hematoxylin and eosin, (b) toluidine blue and fast green, (c) safranin O and fast green, and (d) Masson’s trichrome to visualize specific cell types in developing bone and cartilage, cartilage matrix, ossifying cartilage, and bone structures.
The RNA quality was controlled by an Agilent Bioanalyzer (Agilent Technologies). Affymetrix GeneChip Human Genome U133 Plus 2.0 arrays (Affymetrix) were used for HEK293 and HeLa gene expression analyses, and Affymetrix GeneChip Mouse Exon 1.0 ST arrays were used for Qkfgt/gt and control mouse gene expression analysis. Probe synthesis, hybridization, and initial expression analysis of the Affymetrix GeneChip microarrays were performed according to the protocol recommended by Affymetrix. MYST4 siRNA silencing in HEK293 and HeLa cell lines was performed in 3 individual experiments and hybridized on separate arrays (i.e., 6 arrays were run for HEK293 and 6 arrays were run for HeLa cell lines). All data are MIAME compliant and were uploaded to Gene expression omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/). The 12 arrays were analyzed together. Data processing and normalization were performed using the Partek Genomic Suite Software Package (Partek Incorporated). For each gene, a 2-way ANOVA was calculated, and an F-test was conducted for an MYST4 effect, adjusting for differences between the 2 cell lines. Genes were determined as regulated by MYST4 if the false discovery rates were less than 0.05 and the fold change was greater than 1.5. Experiments for the Qkfgt/gt and control mouse expression were performed likewise. Qkfgt/gt versus wild-type adult dorsal cortex and Qkfgt/gt versus wild-type E12.5 dorsal telencephalon were analyzed in triplicates with 3 animals per genotype, 1 array per animal and tissue. The Qkfgt/gt arrays were analyzed using Bioconductor software ( http://www.bioconductor.org/). Data from all tissues were analyzed together. Expression values were normalized using the gcRMA algorithm. Gene-wise linear models were fitted using the limma software package. Empirical Bayes-moderated t-statistics were computed for each tissue to test for differences between the Qkfgt/gt mutants and normal controls (54). The t-statistics were adjusted for age differences and for correlation between sibs (55). P values were adjusted to control the false discovery rate using the method of Benjamini and Hochberg. Genes were determined as regulated by MYST4 if the unadjusted P value was less than 0.05 and the fold change was greater than 1.5.
HeLa and HEK293 cells were processed with the LowCell ChIP Kit (Diagenode) shearing the DNA with a Bioruptor (Diagenode) according to the manufacturer’s instructions, including positive and negative controls (10,000 cells/IP). Two micrograms of MYST4 (KAT6B/MORF) antibody (ab58823, Abcam) were used in parallel to 2 μg of negative control IgG from rabbit (Diagenode). The immunoprecipitates were amplified using the GenomePlex Complete Whole Genome Amplification (WGA2) Kit (Sigma-Aldrich) as described in the Farnham laboratory WGA protocol for ChIP amplicons ( http://www.genomecenter.ucdavis.edu/farnham/pdf/8-18-06WGA.pdf).
ChIP and control input DNA samples were hybridized in triplicates on separate Affymetrix human promoter 1.0 arrays, as described in the Affymetrix Analysis Technical Manual. All data are MIAME compliant and were uploaded to ArrayExpress (ArrayExpress accession, E-MEXP-2591; http://www.ebi.ac.uk/microarray-as/ae/). Enrichment values (ChIP/control input DNA) were calculated using the MAT algorithm implemented in the Partek Genomic Suite Software. RefSeq genes were assigned to the MAT regions with MAT score P values of 0.1, 0.05, 0.01, and 0.0001.
All real-time RT-PCRs were performed in quadruplicates in 384-well plates with a final volume of 20 μl each on an ABI 7900HT using the TaqMan Gene Expression Mastermix according to the manufacturer’s instructions (Applied Biosystems). The expression levels of the individual isoforms were calculated using the ΔΔCt method with 4 endogenous controls (B2m, PGK1, β-actin, and TATA-box binding protein). Levels of significance were determined by a 2-sample equal-variance t test applied to the log-expression values. Nine significantly differentially expressed genes of the MAPK signaling pathway were validated using the following gene expression assays (Applied Biosystems): SMAD4 (Hs00929647_m1), MAX (Hs00231142_m1), PRKX (Hs00746337_s1), AKT2 (Hs01086102_m1), MRAS (Hs00171926_m1), ELK1 (Hs00428286_g1), SRF (Hs00182371_m1), PPP3CB (Hs00184176_m1), and MAP2K3 (Hs00177127_m1) and, 2 unaltered control genes, NTRK1 (Hs01021011_m1) and DUSP2 (Hs00358879_m1).
Gene ontology and pathway analysis (KEGG pathways; ref. 26) was carried out using the DAVID functional annotation tool ( http://david.abcc.ncifcrf.gov/) and Pathway-Express as part of Onto-Tools ( http://vortex.cs.wayne.edu/ontoexpress/) scoring the pathways, with a significant P value/corrected gamma P value of less than 0.05 in both methods. This was performed individually for the HEK293/HeLa MYST4 siRNA expression data, the mouse Qkfgt/gt adult dorsal cortex versus control expression data, the mouse Qkfgt/gt E12.5 dorsal telencephalon versus control expression data, and the HEK293/HeLa MYST4 siRNA ChIP-CHIP data.
The patient cell line was transiently transfected with a human MYST4 wild-type construct (SC308942, Origene) using Lipofectamine 2000 (Invitrogen). Normalization of expression levels were measured by quantitative real-time PCR as mentioned above. Amounts of MEK1/2, ERK1/2, AKT, phospho-MEK1/2, phospho-ERK1/2, and phospho-AKT were determined by Western blotting of the cell lysates from patient and control lymphoblastoid cell lines as well as from the MYST4-transfected patient cell line. Cells were lysed (lysis buffer containing 50 mM Tris/HCl, pH 7.5, 100 mM NaCl, 2 mM MgCl2, 1% Igepal CA-630, 10% glycerol, 20 mM β-glycerolphosphate, 1 mM orthovanadate, EDTA-free inhibitor cocktail-Roche), and the lysates were cleared by centrifugation. For similar loading the protein amount was normalized by Bradford assay (Bio-Rad), and the proteins were separated by SDS-PAGE (10% polyacrylamide) and detected by Western blotting using antibodies against MEK1/2 (Cell Signaling Technology), ERK1/2 (Cell Signaling Technology), AKT (Cell Signaling Technology), phospho-MEK1/2 (Ser 217/221, Cell Signaling Technology), phospho-ERK1/2 (Thr202/Tyr204, Cell Signaling Technology), and phospho-AKT (Ser473, Cell Signaling Technology).
Quantitative real-time PCR statistical analysis was performed according to the ΔΔCt method. All experiments were performed in triplicate, and error bars represent SD unless otherwise stated. All statistical analysis of the expression and ChIP-CHIP results were generated with Partek Genomic Suite 6.5. These analyses were performed in triplicate, and the ANOVA calculation was applied to identify the fold changes of differentially expressed genes and the MAT score of ChIP-CHIP binding sites. Unless otherwise stated, P values of less than 0.05 were considered significant.
This study was approved by the Ethical Review Board of the Medical Faculty of the Friedrich-Alexander University Erlangen-Nuremberg. After written informed consent, peripheral blood samples for DNA and RNA extraction and cell culture were obtained from the family members.
We thank the family for giving their consent for this study. This study was supported by the IZKF grant E5 to A. Rauch, the BMBF network grant SKELNET (GFGM01141901) to A. Rauch and A. Reis, the DFG grant ZE524/4-1 to M. Zenker, the DFG grant TH896/3-1 to C.T. Thiel, and grants from the F.R.S.-FNRS and the Australian NHMRC. This work was performed as part of our research study addressing the genetics of mental retardation, which was approved by the research ethics committee of the Medical Faculty of the University of Erlangen-Nuremberg. M.R. Ahmadian was supported by the NGFNplus-program of the German Ministry of Science and Education [BMBF, grant 01GS08100] and the Research Committee of the Medical Faculty of the Heinrich-Heine University Düsseldorf.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2011;121(9):3479–3491. doi:10.1172/JCI43428.
Hamish Scott’s present address is: Department of Molecular Pathology, The Centre for Cancer Biology, The Institute of Medical and Veterinary Science and The Hanson Institute, SA Pathology, and The School of Medicine, The University of Adelaide, Australia.