Gene-Dosage-Dependent Association of DYRK1A with the Cytoskeleton in the Brain and Lymphocytes of Down Syndrome Patients

Abstract

The triplication of the DYRK1A gene encoding proline-directed serine/threonine kinase and located in the critical region of Down syndrome (DS) has been implicated in cognitive deficits and intellectual disability of individuals with DS. We investigated the effect of abnormal levels of this kinase on the cytoskeleton in brain and peripheral tissues of DS subjects. In DS tissues, the predictable ≈1.5-fold enhancement of the levels of DYRK1A protein was demonstrated. An association of DYRK1A with all 3 major cytoskeleton networks was identified using immunoprecipitation. We concentrated on the actin cytoskeleton because its association with DYRK1A was the most affected by the enzyme levels. As measured by co-immunoprecipitation in DS tissues, but not in fragile X lymphocytes, actin association with DYRK1A was reduced. This reduced association was dependent on the state of phosphorylation of cytoskeletal proteins and was present only in cells overproducing DYRK1A kinase; therefore, the effect was attributable to the DYRK1A gene dosage. Alterations of DYRK1A-actin assemblies were detected in newborn and infant groups, thereby linking DYRK1A overexpression with abnormal brain development of DS children. The identification of the actin cytoskeleton as one of cellular targets of DYRK1A action provides new insights into a gene-dosage-sensitive mechanism by which DYRK1A could contribute to the pathogenesis of DS. In addition, the presence of this DS-specific cytoskeleton anomaly in lymphocytes attests to the systemic nature of some features of DS. To our knowledge this is the first study conducted in human tissue that shows DYRK1A association with the cytoskeleton.

INTRODUCTION

DYRK1A gene, the homolog of Drosophila minibrain gene, encodes a proline-directed serine/threonine kinase (1, 2). The gene is located on chromosome 21 in the Down syndrome (DS) critical region containing a subset of genes implicated in DS (3). It is assumed that many of the phenotypic features of DS stem from enhanced expression of these genes (4, 5). Accordingly, DYRK1A levels in the brain tissue of DS subjects are increased in a gene-dosage-dependent manner (6), and its abnormal expression has been linked to brain anomalies and deficiencies in complex neuronal networks, both in the developing (7–10) and the mature (11, 12) CNS. This expression may lead to the formation of defective neuronal circuits and contribute to the pathogenesis of mental retardation in DS (13). Both DYRK1A triallelic (14–17) and monoallelic (10) mice showed morphogenesis defects of the brain and learning and memory deficits. Importantly, in humans, microcephaly and mental retardation were found to be associated with deletion (18) and truncation of DYRK1A (19). Moreover, similar phenotypes have been described for monosomy 21-associated mental retardation (20–23). These observations point to the crucial role of maintaining a dosage balance of this gene for normal development and CNS function. Changes in phenotypes of neuronal cells with abnormal expression of DYRK1A, such as reduced size, impeded neurite outgrowth, and lowered density of dendritic branching and spines, have also been reported (16, 24–27). These results further support the postulated role of DYRK1A in mental retardation and cognitive deficits in DS individuals.

Because of its ability to bind and phosphorylate numerous proteins (28, 29), multiple functions have been attributed to DYRK1A. DYRK1A contains the nuclear localization sequence (30); however, the majority of the endogenous kinase is found in the Triton-insoluble post-nuclear fraction (31), presumably a fraction containing cytoskeletal matrices. Protein-protein interactions have been widely employed to position a kinase in the proximity of its targets, and ultimately to determine its function in a defined cellular context. In this study, we use the immunoprecipitation (IP) technique with anti-DYRK1A antibodies to investigate the interaction of DYRK1A with cytoskeletal proteins in brain tissue and in immortalized B-lymphocytes of control and DS subjects. Our findings show that DYRK1A is associated with the cytoskeleton and this association, particularly with actin, is abnormally affected by the gene-dosage-sensitive mechanism, both in the brain and peripheral tissue of DS individuals.

MATERIALS AND METHODS

Brain Tissue

Brain tissue was obtained from 2 sources: the Brain and Tissue Bank for Developmental Disorders of the NICHD, University of Maryland at Baltimore and the Brain and Tissue Bank for Developmental Disabilities and Aging of the Institute for Basic Research (IBR Brain Bank). All groups represented subjects of both sexes and all ethnicities because more stringent selection of the cases was limited by the availability of the tissue. The cases used are listed in Table 1. All procedures involving human tissue were performed in accordance with the Declaration of Helsinki. Experimental protocols were approved by the Institutional Review Board of the NYS Institute for Basic Research in Developmental Disabilities. All specimens were kept frozen at −70°C until used.

Table 1.

Cases Studied by Age

Age group	Control (Age range)	Down Syndrome (Age range)
Newborns	n = 4 (14–35 days)	n = 3 21–30 days
Infants	n = 9 (3–40 months)	n = 8 (6–36 months)
Adults	n = 7 (31–85 years)	n = 14 (31–65 years)

Lymphoblastoid Cell Lines

Lymphoblastoid cell lines (LCLs) of control donors were obtained from the Corriell Cell Repositories (Camden, NJ). All DS and 2 fragile-X cell lines were established at the Department of Human Genetics of our Institute. The full list of LCLs used is presented in Table 2. Lymphocytes cultures were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum and a mixture of penicillin and streptomycin. All culture media and supplements were purchased from Mediatech, Inc. (Manassas, VA). Cells after harvest were washed 3 times with phosphate-buffered saline (PBS) and stored at −20°C until used. In some experiments freshly harvested cells without prior freezing were used for analysis.

Table 2.

Lymphoblastoid Cell Lines

Controls (n = 6)		Down Syndrome (n = 6)		Fragile X (n = 2)
	Age (years)		Age (years)		Age (years)
* AG08702	30	** DS405	48	** FraX3966	19
* GM07041	38	** DS68	55	** FraX3537	18
* GM07045	39	** DS206	57
* AG09387	43	** DS971	59
* AG08378	60	** DS211	60
* GM03657	68	** DS433	60

^*Cell lines obtained from Coriell Cell Repositories (Camden, NJ);

^**cell lines established at Institute for Brain Research in Developmental Disabilities, Staten Island, NY.

Antibodies

Anti-DYRK1A antibody 8D9 is an in-house produced monoclonal antibody raised against peptides containing the first 160 residues of rat DYRK1A (31), with the epitope located between residues 142–147 (isoform 1 numbering) containing non-phosphorylated Tyr-145 (32). Mouse monoclonal anti-DYRK1A antibody 7F3, with the epitope located between residues 74–78, was raised similarly as 8D9 (33). R420 rabbit polyclonal antibody aimed at the detection of isoform 2 of DYRK1A was produced by immunization with the peptide, the sequence of which matched the alternative junction of exon 5 and 6b (34). All antibodies used in the study are listed in Table 3.

Table 3.

Antibodies

Antibody	Antigen	Host	Source
7F3	DYRK1A, residues 1–160	mouse	(33)
8D9	DYRK1A, residues 1–160	mouse	(31, 32)
R420	DYRK1A, isoform 2	rabbit	this study
H143	DRYK1A, residues 621–763	rabbit	Santa Cruz Biotechnology, Santa Cruz, CA
G19	DRYK1A, N-terminus	goat	Santa Cruz Biotechnology
Z-5	glutathione S-transferase	rabbit	Santa Cruz Biotechnology
AC-15	β-actin, residues 3–16	mouse	Sigma-Aldrich, St. Louis, MO
B-5-1-2	β-tubulin	mouse	Sigma-Aldrich
N52	neurofilament 200, C-terminus	mouse	Sigma-Aldrich

Preparation of Brain and Lymphocytes Lysates, Immunoprecipitation and Immunoblot Analysis

Frozen dissected frontal cortex tissue was mortar-ground in liquid N₂. Pools were made by mixing 0.1 g of powdered tissue from each subject assigned to a given group solely based on age; 100 mg of pooled tissue was lysed in 1 ml of ice-cold RIPA buffer (1x PBS, pH 7.4, 1% NP-40, 0.1% sodium dodecyl sulfate (SDS), and 0.5% sodium deoxycholate) containing Complete protease inhibitor tablets (Roche Diagnostic, Indianapolis, IN) and, when necessary, phosphatase inhibitors (2 mM NaF, 1 mM Na₃VO₄, 4 μM cyclosporine, and 200 nM okadaic acid), followed by a short-pulse sonication. Lysates were clarified by microfuge centrifugation at 14 k rpm for 10 minutes; protein concentrations were determined by bicinchoninic acid assay (Thermo Scientific, Rockford, IL). To assure the specificity of binding, lysates were pre-cleared by incubation with Dynabeads Protein G (Invitrogen Dynal AS, Oslo, Norway). For immunoprecipitation, 1 mg of total protein of each brain lysate was mixed with Dynabeads – primary antibody conjugate, which was prepared by binding 2 μg of the antibody to 40 μl suspensions of prewashed beads. The volume of each sample was made 0.5 ml with complete RIPA buffer and the mixture was incubated with tilting and rotation for 1 hour at 4°C. The complexes were washed 3 times in 1 ml of PBST using a magnet and eluted by boiling for 5 minutes in 50 μl of 1x tricine sample buffer (Bio-Rad Laboratories, Hercules, CA). Ten μl (20% of total) of each eluate were separated electrophoretically on 8% tricine SDS-polyacrylamide gel and electro-transferred onto PVDF membranes (Bio-Rad). The membranes were blocked in 5% non-fat milk and incubated overnight at 4°C with the appropriate dilution of primary antibody. The immunoreactive bands were visualized by using species-specific anti-IgG alkaline phosphatase-conjugated secondary antibody (Thermo Scientific) and CDP-Star chemiluminescence reagent (New England Biolabs, Ipswich, MA). After chemiluminescence detection, blots were further developed by color reaction using BCIP/NBT substrate (Sigma-Aldrich, St. Louis, MO). Preparation of lymphocytes lysates was essentially the same as brain lysates; 15 × 10⁶ cells were lysed in 1 ml of ice-cold complete RIPA buffer. Lysates were clarified by centrifugation and the equivalent of 400 μg of total protein was taken for immunoprecipitation and immunoblot analysis, as described for the brain tissue. Cytochalasin D (CytD) and nocodazole (Noc) treatments were performed by incubating cell lysate at 30°C for 45 minutes in the presence of 25 μM CytD or 10 μM Noc (both from Sigma-Aldrich), and then processed for immunoprecipitation and immunoblotting.

Immunocytochemistry and Confocal Microscopy

Paraffin-embedded sections of white matter from DS fetal (20-week gestation) were stained by immunohistochemistry with monoclonal 7F3 anti-DYRK1A antibody, as described (33). For immunofluorescence staining, mouse Neuro2a (N2a) neuroblastoma cells grown in MEM with 10% fetal bovine serum on 8-well glass chamber slides (Nalge Nunc, Naperville, IL) were rinsed in PBS and fixed in 80% methanol for 20 minutes. After blocking for 1 hour in 10% fetal bovine serum in PBS, cells were incubated overnight at 4°C with primary antibodies diluted in blocking buffer containing 0.1% saponin, which was included in all subsequent steps of the procedure. Slides were developed by 1-hour incubation with FITC- or/and Cy 3-conjugated species-specific secondary antibodies (Jackson ImmunoResearch, West Grove, PA), mounted in the Vectashield mounting medium (Vector Laboratories, Burlingame, CA) and examined with a Nikon C1 three-laser confocal system mounted on a Nikon Eclipse 90i microscope (Nikon, Melville, NY).

Transfection of NIH3T3 Cells

NIH3T3 mouse fibroblasts were obtained from American Type Culture Collection (Manassas, VA) and propagated in DMEM medium supplemented with 10% fetal bovine serum. pCMV-Script-based DYRK1A expression vectors used for transfection were as previously described (35). For transfections, NIH3T3 cells were pre-plated for 24 hours onto a 6-well plate with cell density of 5 × 10⁵ cells per well in antibiotic-free DMEM medium and cultured under normal growth conditions. Next, cells were transfected with 2 μg of plasmid DNA (pCMV-Script, pCMV-MnbWT, and pCMV-MnbK188R) and Lipofectamine 2000 (Life Technologies, Carlsbad, CA), according to the manufacturer’s protocol followed by media change 5 hours afterwards. All experimental procedures were performed 48 hours post-transfection, according to standard protocols, as described for brain tissue and LCL cultures.

In Vitro Phosphorylation

After immunoprecipitation with anti-DYRK1A antibody-coated Dynabeads, 40 μl of the beads suspension were washed twice and resuspended in 100 μl of kinase buffer (25 mM Tris-HCl, pH 7.4, 100 mM NaCl, and 5 mM MgCl₂ supplemented with protease and phosphatase inhibitors). Phosphorylation was initiated by adding 100 μM ATP alone or with 1 μg of C-terminal truncated GST-DYRK1A 497 (35), and allowed to proceed at 30°C for 20 minutes with occasional gentle shaking. Immunoprecipitated complexes were then washed twice with PBST, eluted by boiling in the sample buffer and subjected to immunoblotting under standard conditions of this work.

Data Analysis

The densitometric quantification of immunoreactive bands was carried out with the 1D Scan EX 3.1 software program (Scanalytic Corp., Rockville, MD). The results were evaluated for statistical significance with the two-tailed Student t-test for independent samples using GraphPad Prism version 4.0 software program (GraphPad Software Inc. San Diego, CA), and considered significant at p < 0.05. Figures are representative of at least 3 independent experiments with similar results.

RESULTS

In our studies of DYRK1A protein expression in the brain of control and DS subjects co-fractionation of actin with DYRK1A was consistently observed (6). This prompted us to investigate the association of DYRK1A with actin and other cytoskeletal proteins. Detergent-soluble fractions of human brain tissue were immunoprecipitated with commercial (H143 and G19) and in-house produced (R420) anti-DYRK1A antibodies, and analyzed by immunoblotting with antibodies against major cytoskeleton networks. All 3 anti-DYRK1A antibodies produced very similar immunoprecipitation profiles and substantial enrichment of DYRK1A was evident in immunoprecipitates. Data for H143 antibody are presented in Figure 1. In the frontal cortex of control brain, neurofilament heavy subunit (NF-H), α-tubulin and β-actin were found to co-IP with DYRK1A. The association of DYRK1A with cytoskeleton was further confirmed by immunocytochemical staining of brain tissue and confocal microscopy of Neuro-2A (N2a) cells. There was strong staining of structures resembling the omnipresent cytoskeleton filamentous network in axons in fetal white matter (Fig. 1). In an earlier study of DYRK1A in adult brains, it was also found to be present in neuronal processes (33). In cultured mouse Neuro-2A cells doubly labeled with anti-DYRK1A and either NF-H, actin or tubulin antibodies, DYRK1A colocalized with all 3 main cytoskeletal networks, which is in accordance with the co-IP data. NF-H and tubulin colocalization was restricted to neurites and sites of neurite formation on the cell periphery. The strongest colocalization was with actin. Although concentrated in neurites, particularly in their endings, it was also present in the cell body.

Figure 1.

Association of DYRK1A with the cytoskeleton as revealed by immunoprecipitation and immunostaining of brain tissue and cultured mouse neuroblastoma neuro-2a (N2a) cells. Left panels: Frozen frontal cortex tissue of a control subject was homogenized in RIPA buffer and subjected to immunoprecipitation (IP) with anti-DYRK1A (H143) antibody. Immunoprecipitates were analyzed by Western blotting (WB) with antibodies against the indicated cytoskeletal proteins. All major structural cytoskeletal proteins were detected in the DYRK1A IP. Right panels: High magnification image of fetal (20-week gestation) Down syndrome (DS) white matter immunostained with anti-DYRK1A antibody (7F3). Merge images are of double immunofluorescent labeling of N2a cells with DYRK1A (H143) and either NF-H (N52), α-tubulin (B-5-1-2) or β-actin (AC-15) antibodies.

We next investigated whether the overexpression of DYRK1A in DS subjects could affect the association of DYRK1A with the neuronal cytoskeleton. The study included 6 controls and 12 DS subjects (Table 1). To reduce the individual variation among the cases (6), we created pools consisting of frontal cortex of control and DS subjects. Although DYRK1A was found to be part of the larger complex consisting of 3 main cytoskeleton networks, we focused on the actin cytoskeleton because its association with DYRK1A was the most prominent. Accordingly, using IP and quantification by immunoblotting, levels of DYRK1A and β-actin in immunoprecipitated complexes of control and DS subjects were measured and compared (Fig. 2). The specificity of the IP assay was tested using Dynabeads alone or beads coated with unrelated rabbit antibody (Fig. 2A). No signal with DYRK1A and only very weak signal with actin were detected in those samples, ruling out the possibility of non-specific binding. Figure 2B depicts a representative Western blot analysis of DYRK1A and actin immunoprecipitates together with plotted results of densitometric quantification. In accordance with our previous report (6), levels of DYRK1A in crude DS lysate were increased by 50% (149 ± 8.4%, p < 0.05). The levels of actin were similar in both pools asserting that the levels of actin were not altered in DS. In immunoprecipitates, DYRK1A was elevated by 37% (137% ± 6.5%, p < 0.05) in DS; however, the levels of actin were reduced by 43% (57% ± 4.6%, p < 0.05). Thus, overproduction of DYRK1A protein in DS can be observed at the level of its association with the actin cytoskeleton resulting in lowering the yield of actin, as measured in the IP assay.

Figure 2.

Quantification of DYRK1A and β-actin in immunoprecipitates of pooled frontal cortex tissue of control and Down syndrome (DS) subjects. (A) Specificity of DYRK1A and actin co-immunoprecipitation (IP). NIH3T3 cell lysate was used for IP with uncoated (Beads), anti-glutathione S-transferase (GST) antibody –coated, and anti-DYRK1A R420 antibody (DYRK1A) – coated Dynabeads. The levels of DYRK1A and β-actin in crude lysate and precipitates were detected by immunoblotting. (B) Analysis of DYRK1A and β-actin co-IP. Pools were made by mixing 0.1 g of tissue powdered in liquid N₂ (n = 6 for control and n = 12 for DS) of each brain. Parallel samples of brain lysates were immunoprecipitated with the rabbit polyclonal anti-DYRK1A antibody directed against the C-terminus (H143) of DYRK1A polypeptide and monoclonal β-actin antibody. IP and Western blot (WB) analysis of lysates and immunoprecipitates were conducted as described in Materials and Methods. 20 μg of each lysate and 10 μl of IP sample were taken for analysis. Blots were probed with 1:5000 dilutions of DYRK1A 8D9 and β-actin monoclonal antibodies. In the DS sample, IP with actin antibody was also done with lysate denatured by boiling. Representative Western blots and the results of quantification are shown. The results are expressed as percent of values recorded for the control pool. Each bar represents the mean value of 4 IP experiments conducted on 2 independently prepared pools. Error bars indicate SEM. CTR: control lysate; DS: Down syndrome lysate; DS 100°: boiled DS lysate. *Denotes the differences with statistical significance at the level of p < 0.05. (C) Patterns of DYRK1A-immunoreactive bands that co-IP with actin. DYRK1A and β–actin was independently IP with each respective antibody from control lysate and then compared by Western blotting. Running time for SDS-PAGE was extended to permit the separation of multiple DYRK1A bands approximately 90 kDa.

The association of DYRK1A with the actin cytoskeleton was confirmed by reverse immunoprecipitation with the actin antibody. In this case, the DS group, compared to control, showed both less actin and DYRK1A in precipitated materials. Actin was reduced by 70% and DYRK1A by 30% (Fig. 2B). However, the ratio of DYRK1A to actin in actin IP were quite similar to that observed in DYRK1A IP. In the DS pool there was 1.7-fold dominance of DYRK1A over actin compared to 1.6-fold in DYRK1A IP. Denaturing of DS lysates by boiling before IP (DS 100o in Fig. 2B) restored the yield of both DYRK1A and actin to the levels of the control pool; this indicates the importance of preserving the native conformation of actin-DYRK1A assemblies for displaying characteristic for DS patterns in the IP assay. As shown in Figure 2C, only the middle band of the normally seen triplet of DYRK1A-immunoreactive bands was recovered from the actin IP. Apparently, the middle band represents the only species with specific affinity toward the actin cytoskeleton.

These results establish the association of DYRK1A with the neuronal cytoskeleton and reveal altered DYRK1A complexes with the cytoskeleton in pathological brains of DS subjects. Because DS is a developmental disease it is important to examine it at different stages of brain development. This was tested in the pools grouping brain tissue of newborn and infants. A group of adults was included for comparison. The number of cases in each group and their age is listed in Table 1 and the results are presented in Figure 3. In lysates of newborn and infants groups there was no elevation of DYRK1A, as is observed in the brains of adult DS subjects (6). In the immunoprecipitates, however, DS subjects in all 3 groups showed an increase in DYRK1A: in the newborn group by 1.3-fold, infant group by 1.7-fold and adults by 1.5-fold in contrast to the DYRK1A measurements in the crude lysates. Presumably, this apparent discrepancy could be the reflection of overall lower numbers of cells expressing DYRK1A in DS during the early stages of brain development creating the dilution effect in the lysates. In DS pools of all age groups the actin yields were reduced by 50% (Fig. 3), resembling the results obtained for adult brains (Fig. 2). The overall yield of actin in the developmental groups, particularly in infants, was much above the levels recorded for adults (Fig. 3B). This could be the reflection of gross differences in the intensity of neuronal plasticity between the developing and mature brains. This also indicates DYRK1A involvement in the build-up of neuronal connectivity and its negative gene-dosage effect during a critical time of development.

Figure 3.

Immunoblotting profiles of DYRK1A immunoprecipitates in developing brains of control and Down syndrome (DS) subjects. Pools of frontal cortex tissue of newborns (n = 4 for control and n = 3 for DS), infants (n = 9 for control and n = 8 for DS), and adults (n = 7 for control and n = 14 for DS) were subjected to immunoprecipitation with R420 rabbit anti-DYRK1A antibody, and analyzed by immunoblotting with DYRK1A (8D9) and β-actin antibodies. Others details are as described in the legend to Figure 2. (A, B) Representative immunoblots of crude lysates (A) and immunoprecipitates (B) are shown. Results of densitometric measurements are expressed as a percent of immunoreactivity recorded for controls in each group. Each bar represents mean value ± SEM of 4 experiments conducted on 2 independently prepared pools. The differences between control and DS for all age groups were significant at *p < 0.05.

The availability of LCLs established from DS subjects prompted us to investigate whether observations from the brain tissue could be reproduced in peripheral cells. Figure 4A shows the results of immunoblot testing of 3 control and 5 DS LCLs lines. As in the brain tissue, all DS lines displayed elevated levels of DYRK1A and also showed some variability. This variability was particularly evident in the control group in which line from the 68 year old was much higher than the other 2 of this group and almost as high as the lowest line in DS group. Because of this variability, the study was conducted on pooled cases from each group (Fig. 4B). DYRK1A levels in DS crude lysates were enhanced by 39% (139% ± 8.5%, p < 0.05), thus close to the values obtained for the brain tissue (Fig. 2). No enhancement in DYRK1A expression was observed in fragile X cases. In DYRK1A immunoprecipitates, these proportions were retained with 37% (137% ± 22%, p < 0.05) enhancement over control and; however, the actin levels were reduced by 40% (6%0 ± 19%, *p < 0.05), as compared to the control and FraX pools. Thus, the changes of DYRK1A-actin-immunoprecipitable complexes found in the DS brain tissue were fully reproduced in DS LCLs, indicating the systemic nature of the underlying mechanism. Those changes appear to be specific for DS LCLs as they were not observed in 2 LCLs from donors with fragile X.

Figure 4.

Quantification of DYRK1A and β-actin in immunoprecipitates of pooled lymphoblastoid cell lines (LCLs) of healthy (CTR), Down syndrome (DS) and fragile X (FraX) subjects. (A) Levels of DYRK1A and β–actin in crude lysates. RIPA lysates were prepared from LCLs established from 3 control and 4 DS donors Samples of each lysate were immunoblotted with DYRK1A 8D9 and s- actin antibodies. (B) Immunoblot quantification of DYRK1A and β-actin in DYRK1A IP (R420). Preparation of pooled control (5 cases), DS (6 cases) and FraX (2 cases) LCLs, DYRK1A IP, and analysis of immunoprecipitates were as described in Materials and Methods. Samples were immunoblotted with DYRK1A 8D9 and β-actin antibodies. Plots of densitometric measurements are expressed as a percent of immunoreactivity recorded for the control pool. Each bar represents mean value ± SEM of 4 experiments conducted on 2 independently prepared pools. *Denotes the differences with statistical significance at the level of p < 0.05

Findings in LCLs create an opportunity to study a gene-dosage effect of DYRK1A in easily obtainable blood-derived cells. We further explored this opportunity to test the effects of low temperature, microfilament (CytD) and microtubule (Noc) disrupting agents, and conditions favoring phosphorylation on DYRK1A-actin complexes. All of those conditions are known to affect the integrity of cytoskeleton filamentous networks. We first tested effect of freezing cells before lysis. In the IP of lysates of freshly harvested cells, the yields of actin in control and DS cells were comparable (Fig. 5A). This was not the case in the IP of frozen cells lysates in which the DS sample had less actin. Although the 1.5 ratio of DS to control for DYRK1A was preserved in both conditions, the overall yield in IP samples of frozen cells was significantly lower. Thus, freezing may affect the stability of DYRK1A-actin complexes by disrupting actin filaments. That the actin polymerization process is essential for formation of complexes can be deduced from experiments with CytD treatment, a potent inhibitor of actin polymerization. CytD treatment of lysates before IP, markedly prevented formation of the complex, as judged from 80% reduction of the actin yield in DYRK1A co-IP (Fig. 5B). Importantly, Noc, which causes depolymerization of microtubules, had no effect. The polymerization and depolymerization of actin filaments is tightly regulated by phosphorylation and the elevated activity of DYRK1A kinase in DS may change the dynamic of those processes, thus affecting DYRK1A association with the actin cytoskeleton. If this were the case, lysates prepared without phosphatase inhibitors would be expected to affect control and DS samples to different degrees. Originally we used a phosphatase inhibitors cocktail, but later found that okadaic acid alone, a potent inhibitor of PP1 and PP2A Ser/Thr phosphatases, was enough for exerting the effect. Accordingly, lysates of lymphocytes were prepared with or without okadaic acid and processed for IP with DYRK1A antibody. By omitting okadaic acid in the lysis buffer, the markedly reduced actin yield that co-IP with DYRK1A in DS lymphocytes, was restored to the levels of control (Fig. 5C). This effect of okadaic acid on DYRK1A-actin complexes in DS suggests that their assembly and disassembly depends on the balanced phosphorylation of cytoskeletal proteins, possibly by DYRK1A kinase. To investigate phosphorylation dependency of the process more directly, we tested the effect of conditions favoring phosphorylation on already formed complexes. DYRK1A-actin complexes bound to Dynabeads coated with DYRK1A antibody were incubated in kinase buffer with or without ATP and then analyzed for their stability by immunoblot quantification. For comparison between control and DS, the experiments were conducted on freshly harvested cells without prior freezing to ensure that we started with comparable levels of actin (Fig. 5A). In both control and DS samples, incubation with ATP promoted disassociation of actin from the complexes, as judged from the decrease of actin, but not antibody-bound DYRK1A (Fig. 5D). The effect was much more pronounced in the DS sample in which the reduction of actin in samples incubated with ATP was by 78% vs. 31% in the control. These results reaffirm that the association of DYRK1A with the actin cytoskeleton is regulated by phosphorylation and the higher rate of actin dissociation in DS may stem from trisomy-driven overexpression of the DYRK1A gene.

Figure 5.

Conditions affecting DYRK1A and β–actin co-immunoprecipitation (IP) in control and Down syndrome (DS) lymphoblastoid cell lines (LCLs). (A) Effect of freezing. DYRK1A IP was performed with lysates prepared from freshly harvested cells of control AG09387 and DS405 and cells after freezing for 2 hours at -20°C. DYRK1A and β–actin complexes in IP were measured by immunoblotting. (B) Effect of cytochalasin D (CytD) or nocodazole (Noc) treatment on DYRK1A and β–actin co-IP. Equal aliquots of control (AG09387) cells lysate were incubated at 30°C with CytD or Noc before IP with DYRK1A antibody followed by immunoblot analysis for DYRK1A and β-actin. Load. crude lysate,(–) IP from lysate without treatment, CytD: IP from lysate treated with CytD, and Noc: IP from lysate treated with Noc. The results were quantified and shown as a bar graph. (C) Effect of phosphatase inhibitor okadaic acid (OKA). Lysates of control (GM07045) and DS (DS206) LCLs were prepared with or without 200 nM OKA; and then IP with anti-DYRK1A (R420) antibody. The representative immunoblot is shown together with the plot of densitometric measurements where each bar represents values for DS expressed as percent of corresponding untreated (- OKA) or OKA-treated (+ OKA) control. (D) Effect of incubating with ATP. DYRK1A IP was performed with lysates of control and DS LCLs as described. DYRK1A-actin complexes bound to Dynabeads were subsequently incubated for 30 min at 30°C with or without ATP in kinase buffer. After extensive washing, DYRK1A and β–actin retained on Dynabeads were then analyzed by immunoblotting. KB: incubated only with kinase buffer, ATP: incubated with ATP in kinase buffer.

Frozen brain tissue does not allow direct studies of gene-dosage effect; however, the use of transfected cell lines abrogates this limitation. Originally, we used mouse neuroblastoma N2a cells for these experiments, but due to the low and variable efficiency of transfection we switched to NIH3T3 fibroblasts, which can be easily transfected and also share the same developmental origin as LCLs. As shown in panels Figure 6A and B, the transfection of DYRK1A clone resulted in an average of 1.68-fold enhancements over the vector-transfected controls in lysates, and importantly, a similar increase was also recorded in the IP samples. The level of total s- actin was not affected by DYRK1A transfection; therefore, the level of actin can be used as readout for the analysis without complications. Similarly to what was observed in lymphocytes (Fig. 5), the yield of actin complexed with DYRK1A in the wild-type (WT) DYRK1A and vector-transfected cells was differentially affected by phosphatase inhibitors. Again, the levels of actin co-IP were reduced (0.61-fold decrease, p < 0.05) only when lysates of DYRK1A transfectants were prepared in the presence of phosphatase inhibitors. Tubulin was also found to co-IP with DYRK1A, but its yield was the same in all samples (Fig. 6A, B), indicating that microtubules association with DYRK1A does not depend on phosphorylation. That the kinase activity of DYRK1A is required for observing the effect can be deduced from experiments where cells were transfected with kinase-deficient K188R mutant and subjected to immunoprecipitation with DYRK1A antibody. Despite a similar elevation of DYRK1A in the WT and mutant transfected cells, the actin levels were decreased (0.61-fold, *p < 0.05) only in the WT transfectants (Fig. 6C). Because the only variable in this particular experimental system is kinase activity of DYRK1A, we conclude that the reduction in actin co-IP is a direct consequence of DYRK1A elevation. We subsequently analyzed the effects of DYRK1A on dissociating pre-formed DYRK1A-actin complexes. The experimental approach was the same as the one applied to lymphocytes (Fig. 5D), except the inclusion of exogenous recombinant WT DYRK1A in the sample during incubation. Accordingly, DYRK1A co-IP complexes were similarly prepared, incubated with ATP or ATP plus recombinant DYRK1A in kinase buffer, and then DYRK1A and actin retained on Dynabeads were eluted and analyzed by immunoblotting. Figure 6D shows that incubating complexes along with ATP can significantly reduce the level of bound actin and it can be further reduced if DYRK1A is present. These results reaffirm the conclusion that the assembly of DYRK1A-actin complexes depends upon phosphorylation, driven in part by DYRK1A kinase itself. Thus, its abnormal levels in DS may lead to derangement of this association. They further demonstrate that this effect of DYRK1A is independent of the cell type.

Figure 6.

Effect of exogenous DYRK1A on the formation and stability of DYRK1A-actin complexes in NIH 3T3 fibroblasts. (A) Effect of phosphatase inhibitors (PPIs) in DYRK1A over-expressing cells. NIH 3T3 fibroblasts were transfected with the wild-type (WT) DYRK1A or the cloning vector (VEC) for 48 hours before harvesting. Cells were frozen at -20°C for 2 hours, lysed in RIPA with (+) or without (-) PPIs, and processed for IP. Aliquots of cell lysates containing 400 μg of total protein were taken for IP with R420 DYRK1A antibody and immunoprecipitates were probed with DYRK1A (8D9), s-tubulin, and β-actin antibodies. As a control, crude lysates without IP (lysates) were also probed with DYRK1A (8D9) and β-actin antibodies. The intensity of immunoreactive bands was quantified and reported as independent values in arbitrary units. Representative Western blots and plots of densitometric quantification of DYRK1A and actin are shown. Each bar represents mean value ± SEM of 3 independent experiments. *Denotes the differences with statistical significance at the level of p < 0.05. (B) Effect of over-expression of kinase-deficient mutant. Cells were transfected with the cloning vector (VEC), wild-type (WT) or mutant (K188R) DYRK1A and processed with PPIs for IP and subsequent Western blotting as in (A). Quantification and analysis of the intensity of immunoreactive bands were also performed as described above. (C) Effect of exogenous recombinant DYRK1A on Dynabeads immobilized DYRK1A-actin complexes. Complexes bound to DYRK1A antibody-coated Dynabeads were prepared from untransfected NIH 3T3 lysate as described earlier. Beads were then incubated in kinase buffer for 30 minutes at 30°C, with or without ATP or recombinant truncated wild-type DYRK1A. The coated beads were then extensively washed with PBST and DYRK1A and actin that remained on the beads were quantified by immunoblotting. KB: incubated with kinase buffer only, ATP: incubated with ATP, and WT: incubated with ATP and DYRK1A.

DISCUSSION

Using immunoprecipitation with anti-DYRK1A antibodies, we have identified an association of DYRK1A with the cytoskeleton. In detergent-soluble fractions of human brain and cultured lymphocyte homogenates, DYRK1A was found to co-IP with actin, tubulin and NF-H proteins, the structural components of 3 main cytoskeleton networks (Figs. 1, 3, 6). The subcellular distribution of DYRK1A and cytoskeletal proteins was subsequently examined by confocal microscopy (Fig. 1) supporting results of IP experiments. Because its association with DYRK1A was the most affected by the enzyme levels, the actin cytoskeleton seemed to be the key target for DYRK1A kinase activity. Thus, we focused on the interaction of DYRK1A and actins, primarily β-actin in this study; however, the extension of this study to other cytoskeletal systems is currently under way.

The tissue and cell lysates for IP were prepared in RIPA buffer containing SDS and deoxycholate, with the intention of improving protein extraction and minimizing non-specific binding background. Co-IP of various cytoskeletal proteins with DYRK1A under such harsh conditions not only argues for the strength of the interaction but also allowed detection of subtle but consistent alterations in DYRK1A-actin complexes in samples prepared from DS subjects (Figs. 2, 4), in which the level of DYRK1A is elevated due to triplication of the gene. This phenomenon was also seen in mouse fibroblasts overexpressing the wild-type but not kinase-deficient mutant of DYRK1A (Fig. 6). Importantly, conditions favoring phosphorylation, i.e. the presence of phosphatase inhibitors (Fig. 5), are required for a display of altered properties of DYRK1A-actin complexes in DS samples. The importance of phosphorylation is also evident from experiments in which precipitated DYRK1A-actin complexes were exposed to ATP or active recombinant kinase (Figs. 5D, 6C). All those conditions promoted actin disassociation from the complex. Taken together, our data support a model that DYRK1A-actin complexes are maintained by a delicate balance of the kinase-phosphatase equilibrium. An ~50% increase in DYRK1A level due to gene-dosage elevation in DS subjects or transfection in NIH3T3 cells is sufficient to disturb this balance resulting in derangement of those complexes. To the best of our knowledge this is the first documentation of formation and gene-dosage regulation of the DYRK1A-actin complexes.

Actin depolymerization by CytD treatment and freezing cells before lysis could drastically reduce the yield of actin that co-IP with DYRK1A. It is conceivable that the effect of DYRK1A elevation may also be mediated through controlling the stability of actin filaments. Alterations in actin dynamic through increased stability of actin filaments was also observed in DYRK1A trisomic TgDyrk1A mice where DYRK1A is overexpressed to the levels observed in DS (27). The reduction in actin co-IP was already present in DS brains of newborn and infants (Fig. 3), linking the DYRK1A interaction with the cytoskeleton to well-documented deficiencies and delays in brain maturation associated with DS (36). The presence of this anomaly in newborn subjects contrasts with earlier studies on DS brain morphometry at the time of birth (37–40). It should also be stressed that alterations of the association of DYRK1A and actin in DS are systemic, as they were seen in both brain tissue and peripheral cells. The elevation of DYRK1A expression as well as the cytoskeleton abnormality seen in DS were not observed in fragile X cases, further implicating DYRK1A imbalance as a factor contributing to the altered stability of DYRK1A-actin complexes.

Actin filament dynamics are involved in shaping dendritic branching complexity and modulating spine/synapse densities. Altered DYRK1A gene dosage, as in DS brains (13) and triallelic (16, 27) or monoallelic (10, 24) mice, results in marked changes in spine/synapse densities and abnormalities in the dendritic arbors. DYRK1A expression has been associated with neurite formation in cultured neurons (26, 27, 41, 42), and its Drosophila orthologue-minibrain kinase (MNB) was also found to regulate actin-based protrusions, specifically in CNS-derived cell lines (43). Administration of tea extract containing DYRK1A inhibitor, epigallocatechin gallate, to transgenic hYACtgDYRK1A mice overexpressing DYRK1A nearly completely rescued brain defects in the mice (44). The downregulation of DYRK1A due to intragenic deletion (18) and truncation (19) of DYRK1A gene or partial monosomy of 21 containing locus of DYRK1A produced disease phenotypes very similar to those of DS patients (20–23). These studies point to a DYRK1A gene-dosage-sensitive mechanism as the chief contributor to mental retardation and cognitive impairment of DS individuals, but the key mechanism underlying its involvement is yet to be revealed. Our study identifying the actin cytoskeleton as a target of DYRK1A action provides new insight for better understanding of how this function of DYRK1A can be executed.

It is important to emphasize that DYRK1A is only a component of a common machinery of phosphorylation that involves many kinases and cytoskeletal proteins that modulates cytoskeleton plasticity. The kinases involved act in a highly regulated interplay inducing small variations in the phosphorylated states of cytoskeleton proteins, which are important for dynamic properties and integrated function of the entire cytoskeleton network. Abnormal activity of DYRK1A kinase may disturb this fine-tuning, causing subtle perturbations in the smooth operation of this complex machinery. The enormous complexity of this machinery makes identification of the putative target(s) of DYRK1A enzymatic action a daunting task. Purified actin itself is not phosphorylated to an appreciable level by recombinant DYRK1A, nor does actin bind directly to the kinase (Adayev, unpublished observation). This result suggests that DYRK1A complexing with actin needs adapter molecule(s). Actin interacts with a plethora of binding proteins (45). Likewise, more than 2 dozen putative substrates and interacting proteins of DYRK1A have been identified (28, 29). Unfortunately, a cross check of both lists did not immediately reveal any potential adapter candidate(s). DYRK1A has recently been found to phosphorylate N-WASP protein (46), which is involved in regulating actin polymerization and dynamics. Although a protein such as N-WASP could be a potential adaptor molecule, the co-IP and colocalization of tubulin and neurofilament proteins with DYRK1A (Figs. 1, 3) suggest that the adaptor molecules should also function as linkers of actin filaments with other cytoskeleton networks. The association of actin with DYRK1A has been found in every cell type examined so far; therefore, the distribution of adapter(s) should be ubiquitous. Other potential adapters could be heat shock protein 90, which we have found in DYRK1A and actin immunoprecipitates (Dowjat, unpublished observation). This molecular chaperone protein is known to be associated with all 3 cytoskeleton networks, apparently protecting their filamentous structure, and its biological function is regulated by numerous kinases (47). A systemic approach to characterize the adapter(s) is currently under way.

Of special interest are our findings in DS lymphocytes attesting to the systemic nature of some features of DS phenotype. One of those features could be Alzheimer disease (AD)-type dementia, a highly prevailing condition in aging DS patients. Thus, it will be interesting to explore whether a similar abnormality in the association of DYRK1A with actin exists in AD and whether it is required for the onset of clinical AD. Such a connection has been proposed in an previous publication from this laboratory (48), and fits the well-documented disruption of cytoskeleton in AD. If this is the case, then lymphocytes from DS and AD donors may serve as unique cellular model of the disease, and possibly, as diagnostic tool for predicting or confirming AD.