A novel de novo heterozygous DYRK1A mutation causes complete loss of DYRK1A function and developmental delay

Dual-specificity tyrosine phosphorylation-regulated kinase 1 A (DYRK1A) is essential for human development, and DYRK1A haploinsufficiency is associated with a recognizable developmental syndrome and variable clinical features. Here, we present a patient with DYRK1A haploinsufficiency syndrome, including facial dysmorphism, delayed motor development, cardiovascular system defects, and brain atrophy. Exome sequencing identified a novel de novo heterozygous mutation of the human DYRK1A gene (c.1185dup), which generated a translational termination codon and resulted in a C-terminally truncated protein (DYRK1A-E396ter). To study the molecular effect of this truncation, we generated mammalian cell and Drosophila models that recapitulated the DYRK1A protein truncation. Analysis of the structure and deformation energy of the mutant protein predicted a reduction in protein stability. Experimentally, the mutant protein was efficiently degraded by the ubiquitin-dependent proteasome pathway and was barely detectable in mammalian cells. More importantly, the mutant kinase was intrinsically inactive and had little negative impact on the wild-type protein. Similarly, the mutant protein had a minimal effect on Drosophila phenotypes, confirming its loss-of-function in vivo. Together, our results suggest that the novel heterozygous mutation of DYRK1A resulted in loss-of-function of the kinase activity of DYRK1A and may contribute to the developmental delay observed in the patient.


Cell culture and transfection. Human embryonic kidney 293T cells were cultured in Dulbecco's Modified
Eagle's Medium containing 10% foetal bovine serum (Welgene, Gyeongsan-si, Gyeongsangbuk-do, Republic of Korea) supplemented with 1% streptomycin and penicillin. The cells were seeded at approximately 50% confluency into cell culture plates and were maintained overnight at 37 °C under 5% CO 2 . When the cells reached 60-80% confluency, they were transfected with plasmids using the XtremeGene Transfection Reagent (Roche, Basel, Switzerland), according to the manufacturer's instructions. Transfected cells were incubated at 37 °C for 24 h prior to harvest or analysis.
Chemicals. We used the proteasome inhibitor MG132 (Calbiochem, San Diego, CA, USA), the lysosomal inhibitor NH 4 Cl (Sigma-Aldrich, St. Louis, MO, USA), the calpain inhibitor calpeptin (Calbiochem), and the autophagy inhibitor 3-methyladenine (Sigma-Aldrich) for protein degradation pathway analyses. All chemicals were dissolved in dimethyl sulfoxide (DMSO) prior to treatment, and the cells were treated with 10 μM of each compound for 15 h after plasmid transfection.
Polyubiquitination assay. Polyubiquitination assays were performed as described previously 26 . Briefly, the 293 T cells were co-transfected with FLAG-DYRK1A-E396ter and HA-ubiquitin plasmids using the XtremeGene Transfection Reagent (Roche) for 24 h. Cells were then treated with 10 μM MG132 for 11 h, harvested, and lysed in an immunoprecipitation (IP) buffer (Tris/HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA and 1% NP-40). The cell lysates were mixed with anti-FLAG M2 antibody-conjugated protein A/G-agarose beads (Santa Cruz Biotechnology) in 0.2% NP-40-containing IP buffer using a tubing-rotator at 4 °C for 2 h. The bound beads were washed three times with 0.2% NP-40-containing IP buffer and were solubilized with 2X sample buffer for western blot analysis.
Drosophila culture. Drosophila melanogaster were cultured at 25 °C on standard cornmeal media. Tau (human wild-type Tau, #51362) 27 , and all other stocks and balancers were obtained from the Bloomington Stock Centre (Bloomington, IN, USA). UAS-mnb transgenic flies were described in our previous report 28 . Based on amino acid sequence alignment, the human DYRK1A E396 residue corresponded to the Drosophila mnb D401 residue. Both wild-type full-length mnb and truncated mutant (mnb-D401ter) cDNAs were generated by PCR using Drosophila S2 cell cDNA as a template. Phusion Pfu PCR polymerase (Invitrogen) was used with the following primer pairs: 5ʹ-GAATTCATGTATAGATTAGAGGATACGA-3ʹ (forward, common for mnb-WT and mnb-D401ter), 5ʹ-TCTAGACTAATGTATAACTACAGGATTC-3ʹ (reverse for mnb-WT), and 5ʹ-TCTAGATCAGAAGAACTTGCGGGTCTTG-3ʹ (reverse for mnb-D401ter). The cDNAs of the full-length and truncated forms of mnb were ligated to the pUAST-FLAG vector digested with EcoRI/XbaI restriction enzymes to place the transgenes under the control of the UAS promoter. UAS-mnb-WT and UAS-mnb-D401ter transgenic flies were obtained by P-element-mediated germline transformation 29 . To analyse the eye phenotype, newly eclosed flies were collected and allowed to mate for 2-3 days. Wings from adult flies were dissected in 100% ethanol and mounted in Canada Balsam mounting medium (Gary's magic mountant). The eye and wing phenotypes were photographed using a SZ60 binocular microscope equipped with an eXcope K5 CCD system (Olympus, Tokyo, Japan). Fly tissue sizes were measured on multiple samples (n > 10) from each genotype using NIH ImageJ software. The average eye size was presented as a normalized percentage of the control eye size. For adult lethality assessments, at least 100 flies were included in each individual experiment. For statistical analyses, two-tailed unpaired Student's t-tests were used.
Immunohistochemistry. Immunostaining of larval samples was performed as previously described 30 .
Briefly, fixed larval preparations were washed in phosphate buffered saline with Tween-20 (PBST) three times (10 min each), blocked in PBST with 5% normal goat serum for 30 min, and incubated with primary antibody at room temperature for 2 h or at 4 °C overnight. Fluorescein isothiocyanate-labelled goat anti-horseradish peroxidase (1:50 dilution, The Jackson Laboratory, Bar Harbour, ME, USA) was used to detect the synaptic bouton in larval tissue. All images were collected using a FV1000 confocal microscope (Olympus) and processed using NIH ImageJ software. Neuromuscular junction (NMJ) quantifications were conducted based on published procedures 31 . The filopodial NMJ structure was processed from complete z-stacks using the entire NMJ of the A4 abdominal segment. All statistical comparisons were performed using Prism software (GraphPad, San Diego, CA, USA). P-values were calculated using two-tailed Student's t-tests.

Results
Case presentation. A 5-month-old female was referred to our hospital for evaluation of delayed development and facial dysmorphism. The patient was born after 39 weeks and 4 days of gestation by vaginal delivery after an uneventful pregnancy and weighed 2.89 kg. She was the third child of healthy Korean parents ( Supplementary Fig. S1), and her siblings showed normal development. Her phenotypic features included an epicanthal fold, tented mouth, short and deep philtrum, deep-set eyes, bi-temporal narrowing, micrognathia, wide nasal bone, sparse scalp hair, and prominent ears with underdeveloped ear lobes ( Supplementary Fig. S1). She also displayed rigidity in both her upper and lower extremities as well as thumb folding and deep tendon reflexes. An echocardiogram showed that she had a secondary atrial septal defect and peripheral pulmonary stenosis. Her chromosomal microarray examination revealed normal results. According to the Bayley Test (BSID-III), which measures early development in children, the patient had a language delay (composite score: 56; percentile score: 0.2), motor delay (composite score: 46; percentile score: <0.1), and poor adaptive behaviour skills (composite score: 66; percentile score: 1). At the time of her visit to the hospital, approximately 5 months after birth, her height was 63.6 cm (50 th percentile), her weight was 5.8 kg (10 th percentile), and her head circumference was 48.5 cm (25 th percentile). Based on these clinical manifestations, the patient was initially suspected to have either Ohdo or Hallermann-Streiff syndrome. However, we discovered a novel heterozygous DYRK1A variant www.nature.com/scientificreports www.nature.com/scientificreports/ (c.1185dup, p.E396ter, [NM_001396.3]) on exon 8 by exome sequencing. The novel variant was only detected in the patient sample and was not present in the control population (variant databases from 1000 Genomes Project, ESP 6500, and ExAC). We confirmed the variant by Sanger sequencing (Fig. 1a). Parental segregation of the variant was negative, indicating that the mutation occurred in a sporadic form. Moreover, the patient's brain MRI exhibited mild brain atrophy of both frontal lobes, thinning of the brainstem, hypoplasticity of the pituitary stalk and corpus callosum, and subcortical white matter hypomyelination, consistent with previously reported features of DYRK1A haploinsufficiency syndrome ( Supplementary Fig. S1) 21 . After discovering that the patient's disorder may be related to DYRK1A, rehabilitation therapy was initiated. At her 8-month follow-up (age: 13 months), the patient demonstrated delayed global developmental milestones and growth parameters. At 21 months of age, the patient developed seizures with generalized stiffness.

Analysis of the structure and molecular dynamics of the DYRK1A-E396ter protein. The novel
DYRK1A variant (c.1185dup) identified in this study was generated by the insertion of a single T at nucleotide position 1,185 of NM_001396.3. This insertion altered the translational frame, generated a premature translation termination codon (TGAG) at the E396 (GAG) position, and produced a truncated protein of 395 amino acids (DYRK1A-E396ter) (Fig. 1b). To gain a functional insight into the DYRK1A-E396ter protein, we primarily predicted its structure based on the crystal structure of the kinase domain in complex with its ATP-competitive inhibitor, DJM2005 32 (Protein Data Bank ID: 4MQ2) and substrate peptide (Fig. 2a). Structurally, the nonsense mutation occurred in the β-sheet of the CMGC insert, which only exists in CMGC kinases, thereby producing a protein lacking most of CMGC insert, α-helix H, and α-helix I at the C-terminal end of the catalytic domain. For the analysis of the molecular dynamics, we next performed energy minimization of the truncated structure by using the MODELLER software, and then the local molecular dynamics of the wild-type and E396ter DYRK1A proteins were assessed by using the Dynamut webserver with normal mode analysis function (Fig. 2b). As a result, we found that the DYRK1A-E396ter protein had elevated deformation energy throughout the protein compared to the wild-type one. Relatively high deformation energy was shown in a catalytic loop, an activation segment, and the loop between α8 and α9, suggesting a potential reduction of protein stability (Fig. 2c).

DYRK1A-E396ter is efficiently degraded by the proteasome.
To experimentally examine the protein stability of DYRK1A-E396ter in mammalian cells, we constructed a FLAG-tagged wild-type DYRK1A (FLAG-DYRK1A) expression clone and then introduced the mutation identified in the patient to produce a truncated protein of 395 amino acids (FLAG-DYRK1A-E396ter). In contrast to FLAG-DYRK1A-WT protein, FLAG-DYRK1A-E396ter protein was barely detectable by western blotting with an anti-FLAG antibody when overexpressed in 293 T cells (Fig. 3a). Since quantitative RT-PCR demonstrated that the transcripts were and substrate peptide is shown on the left. The C-helix, activation segment, catalytic loop, and CMGC insert are coloured in red, blue, orange, and hot pink, respectively. The N-and C-termini of the kinase domain of DYRK1A are represented by blue and red circles, respectively. The substrate peptide and inhibitor, DJM2005, are represented by a ball and stick model and coloured in dark and light grey, respectively. The predicted structure of DYRK1A-E396ter in complex with DJM2005 is shown on the right. The peptide sequences, which were not expressed in the mutant, are coloured dark grey and the location of the mutation (E396ter) is indicated by a red arrow. (b) Energy minimization of the wild-type and E396ter DYRK1A proteins was performed by using the MODELLER software, and the local molecular dynamics were assessed by using the Dynamut webserver with normal mode analysis function. Resulting molecular dynamics of the wild-type and E396ter DYRK1A proteins are presented in a tube style, which was generated by using PyMol software (version 1.3). The deformation energy is represented by thin to thick tubes coloured in blue (low), white (moderate), and red (high). (c) The deformation energy of the wild-type and E396ter DYRK1A proteins are coloured in blue and red, respectively. www.nature.com/scientificreports www.nature.com/scientificreports/ expressed at similar levels ( Supplementary Fig. S2), we then examined the involvement of protein degradation. Protein degradation occurs in a specific and coordinated manner through major proteolytic systems, such as the proteasome, lysosome, calpains, and autophagy. To determine which systems contributed to the degradation of DYRK1A-E396ter, 293T cells expressing FLAG-DYRK1A-E396ter were treated separately with four selective inhibitors of each proteolytic degradation pathway. MG132, NH 4 Cl, calpeptin, and 3-methyladenine were used to inhibit proteasome-, lysosome-, calpain-, and autophagy-mediated protein degradation pathways, respectively. Western blotting with an anti-FLAG antibody revealed a prominent FLAG-DYRK1A-E396ter protein (~45 kDa) band only after treatment with MG132 (Fig. 3b) and not after treatment with any of the other inhibitors, suggesting that FLAG-DYRK1A-E396ter was degraded by the proteasome. Most proteins degraded by the proteasome are marked with (poly)ubiquitin chains. Thus, we analysed the ubiquitination of DYRK1A-E396ter by co-expressing FLAG-DYRK1A-E396ter and HA-ubiquitin in 293 T cells. Polyubiquitination of FLAG-DYRK1A-E396ter was observed at high levels when co-expressed with HA-ubiquitin and at even higher levels after additional treatment with MG132 (Fig. 3c). Together, these results indicate that DYRK1A-E396ter is degraded by the ubiquitin-dependent proteasomal degradation pathway.
Loss-of-function of the DYRK1A patient mutant due to intrinsic inactivity. As the DYRK1A-E396ter protein was truncated at the catalytic domain, we sought to examine its kinase activity. We selected the Tau, a microtubule-associated protein, for the evaluation, because it is one of the most well-studied substrates of DYRK1A, and aberrant Tau phosphorylation is associated with the formation of neurofibrillary tangles in DS and Alzheimer's disease 10,33 . Phosphorylation of Tau at the T212 residue is highly dependent on DYRK1A and was chosen as a marker of DYRK1A kinase activity 34 . Tau was transiently co-expressed with each FLAG-DYRK1A-WT, FLAG-DYRK1A-E396ter, and FLAG-DYRK1A-K188R 1 (a catalytically inactive mutant of human DYRK1A) in 293 T cells. Phosphorylation of Tau at T212 was detected by western blotting with a phosphorylation-specific antibody. As previously reported, the co-expression of Tau and wild-type DYRK1A induced remarkable phosphorylation of Tau (Fig. 3a) 27 . In contrast, the DYRK1A-E396ter protein had little effect on Tau phosphorylation, which was similar to the DYRK1A-K188R protein.
We went on to examine whether FLAG-DYRK1A-E396ter possessed any intrinsic kinase activity when stabilized in the presence of MG132 (Fig. 3b). Despite a dramatic increase of FLAG-DYRK1A-E396ter protein level following the inhibition of proteasomal degradation, Tau phosphorylation was not rescued to an appreciable degree. These results indicate that the DYRK1A-E396ter mutant is catalytically inactive due to the truncation of the C-terminal region of the kinase domain. Additionally, we suggest that the intrinsic inactivity of DYRK1A-E396ter is the direct cause of its loss-of-function.

DYRK1A-E396ter does not have a dominant-negative effect on wild-type DYRK1A.
Because both wild-type and mutant DYRK1A alleles are present in the patient's cells, we investigated a potential dominant-negative effect of DYRK1A-E396ter on wild-type DYRK1A. The wild-type DYRK1A and DYRK1A-E396ter plasmids, alone or in a 1:1 combination, were co-transfected with the Tau plasmid, and the transfected cells were treated with MG132. As previously shown (Fig. 3a), Tau phosphorylation was strongly induced by the expression of wild-type DYRK1A. However, the additional co-expression of DYRK1A-E396ter did not impact Tau phosphorylation (Fig. 4). Similar results were observed even when the DYRK1A-E396ter protein was stabilized by MG132 treatment. Together, our findings indicate that DYRK1A-E396ter has little negative effect on wild-type DYRK1A with respect to Tau phosphorylation.
Loss-of-function of the DYRK1A patient mutant in a Drosophila model. Because basic biological and neurological properties are highly conserved between human and Drosophila, Drosophila models are widely used to understand the molecular pathology of human diseases. Indeed, the Drosophila genome is nearly 75% homologous with human disease genes 35 , and the Drosophila minibrain (mnb) gene is 82% identical to its human homolog, DYRK1A 36,37 . Loss-of-function mutations in mnb result in reduced brain size with abnormal visual and olfactory behaviour due to defects in neurogenesis and brain development 38 , which recapitulates the  , 15 h). The subsequent procedure was identical to that described in the legend of Fig. 3a. Uncropped full-sized blots are presented in Supplementary Fig. S4. www.nature.com/scientificreports www.nature.com/scientificreports/ microcephaly phenotype of DYRK1A haploinsufficiency in human patients. In addition, the tissue-specific overexpression of mnb also induces various phenotypic and neurological defects in central nervous system structure, consistent with various phenotypes of DS patients 39 .
To evaluate the functionality of the DYRK1A-E396ter protein in vivo, we generated transgenic Drosophila harbouring wild-type mnb (UAS-mnb-WT) or truncated mnb (UAS-mnb-D401ter). Based on sequence comparison, Drosophila D401 was equivalent to human E396 (Supplementary Fig. S3). For tissue-specific overexpression, transgenic mnb flies were crossed with flies expressing tissue-specific Gal4 drivers. The resulting expression of mnb-WT and mnb-D401ter proteins was analysed by western blotting of total protein extracts from transgenic flies (HS-Gal4 > UAS-mnb-WT or -D401ter) (Fig. 5a). Similar to our results in mammalian cells (Fig. 3a), the D401ter protein was expressed at much lower levels than the wild-type protein.
Next, the function of the D401ter protein was examined by comparing the phenotypes of mnb-D401ter flies with those of mnb-WT flies. Overexpression of mnb-WT in wing tissue using the MS1096-Gal4 driver showed shortening of the L5 vein in the adult wing, which we previously identified as one of the most prominent phenotypic defects 27 . In contrast, there was no recognizable phenotypic defect when the D401ter protein was overexpressed in the wing (Fig. 5b). Overexpression of human Tau in Drosophila is known to cause a severe eye degeneration phenotype 40 , which was worsened by co-expression with mnb-WT (Fig. 5c,d). Exacerbation of the eye degeneration phenotype is likely due to Tau hyperphosphorylation. In contrast, co-expression with mnb-D401ter had no additional effect.
As neuromuscular junction (NMJ) morphology of Drosophila larvae is used as a tool to assess neuronal synapse formation and integrity 41 , we analysed the effect of the D401ter mutation of mnb on NMJ morphology in transgenic flies. Overexpression of functional mnb-WT in the postsynaptic muscle tissue using mhc-Gal4 (g) Each mnb protein was ubiquitously or neuro-specifically overexpressed using the Actin5-or elav-gal driver, respectively, and embryonic lethality was examined. Viability was presented as a percentage. Two-tailed Student's t-tests were used to calculate P-values, which are depicted with an asterisk.

Discussion
In this study, we presented a patient with multiple congenital anomalies, including facial dysmorphism, developmental delay, and abnormalities in the cardiovascular system and brain structure (Supplementary Fig. S1). Exome sequencing and segregation analyses revealed that these phenotypic manifestations may be explained by a novel de novo heterozygous mutation of the DYRK1A gene. This novel mutation generated a translational termination codon and produced a C-terminally truncated protein (DYRK1A-E396ter) (Fig. 1a). Structurally, the DYRK1A-E396ter protein lacked the C-terminal end of the kinase domain, including the CMGC insert, α-helix H, and α-helix I. These regions, especially α-helices H and I, have been previously demonstrated to be required for the catalytic activity of DYRK1A. For instance, Arranz et al. showed that an R467ter DYRK1A mutant is completely inactive, and an F478 DYRK1A frame-shift mutant has less than 15% of the activity of wild-type DYRK1A. Their results indicate that even α-helix I, which is located at the C-terminal end of the kinase domain, is critically required for full kinase activity 42 . Consistent with this observation, we demonstrated that the DYRK1A-E396ter mutant was catalytically inactive. As this mutant protein was efficiently degraded by the proteasome and was barely detectable in mammalian cells, we evaluated its intrinsic kinase activity by inhibiting protein degradation with MG132. We found that even after protein stabilization, the DYRK1A-E396ter protein had little Tau-phosphorylating activity (Fig. 3b). According to an extensive analysis of DYRK1A missense mutants, lack of substrate phosphorylation activity is highly correlated with lack of tyrosine autophosphorylation 42 , which may be the case for the DYRK1A-E396ter protein.
More importantly, we further revealed that the mutant protein did not have a dominant-negative effect on the wild-type protein. Tau phosphorylation induced by the expression of the wild-type protein was barely affected by the additional expression of the mutant protein, even though the level of the mutant protein was recovered to the level of the wild-type protein by treatment with MG132 (Fig. 4). In the case of the two alleles are expressed in the patient's heterozygous cells, this result indicates a lack of dominant-negative effect of the mutant protein, which would be the result of the combined contribution of two effects: reduced stability of the mutant protein and no interference of the mutant protein on the activities of the wild-type one. We further confirmed the lack of dominant-negative function of the mutant protein in Drosophila as an in vivo model. Transgenic mnb-D401ter flies, which have the Drosophila mutation equivalent to human DYRK1A-E396ter, had no recognizable phenotypic defects in the wings, eyes, and NMJs, and no severe embryonic lethality, whereas mnb-WT transgenic flies had severe phenotypic defects and lethality (Fig. 5). Collectively, these results clearly show that the DYRK1A-E396ter protein is not only catalytically inactive but is also completely non-functional in mammalian cells and fly models.
According to previous reports, the C-terminal end of the DYRK1A kinase domain is required for protein stability as well as kinase activity [42][43][44] . Thus, loss of this region in the DYRK1A-E396ter mutant may directly affect protein stability. Indeed, we observed that the mutant protein was efficiently degraded by the ubiquitin-mediated proteasomal pathway and was consequently undetected in mammalian cells (Fig. 3b,c). It has been previously reported that DYRK1A degradation is mediated by binding of the E3 ubiquitin ligase SCF βTrCP to the N-terminus of DYRK1A 45 . In addition, the CDC37/HSP90 chaperone has been suggested to regulate DYRK1A protein stability through interaction with its N-terminal lobe 46 . Whether these two mechanisms are associated with the degradation of the DYRK1A-E396ter protein needs further experimental validation.
In conclusion, we have identified a novel de novo DYRK1A nonsense mutation in a patient with DYRK1A haploinsufficiency syndrome. The mutation generates a C-terminally truncated protein at the β-sheet of the CMGC insert within the kinase domain that behaves as a loss-of-function mutant in both mammalian cell and Drosophila models.