Effect of tyrosine autophosphorylation on catalytic activity and subcellular localisation of homeodomain-interacting protein kinases (HIPK)

Abstract

Background: Homeodomain interacting protein kinases (HIPKs) function as modulators of cellular stress responses and regulate cell differentiation, proliferation and apoptosis. The HIPK family includes HIPK1, HIPK2 and HIPK3, which share a similar domain structure, and the more distantly related HIPK4. Although HIPKs phosphorylate their substrates on serine or threonine residues, it was recently reported that HIPK2 depends on the autophosphorylation of a conserved tyrosine in the activation loop to acquire full catalytic activity and correct subcellular localization. In this study we addressed the question whether tyrosine autophosphorylation in the activation loop has a similar function in the other members of the HIPK family.

Results: All HIPKs contained phosphotyrosine when expressed in HeLa cells. Catalytically inactive point mutants were not tyrosine-phosphorylated, indicating that HIPKs are dual-specificity protein kinases that autophosphorylate on tyrosine residues. HIPK point mutants lacking the conserved tyrosine residue in the activation loop showed reduced catalytic activity towards peptide and protein substrates. Analysis of these mutants revealed that HIPK1, HIPK2 and HIPK3 but not HIPK4 are capable of autophosphorylating on other tyrosines. Inhibition of tyrosine phosphatase activity by treatment with vanadate enhanced global phosphotyrosine content of HIPK1, HIPK2 and HIPK3 but did not affect tyrosine phosphorylation in the activation loop. Mutation of the activation-loop tyrosines resulted in a redistribution of HIPK1 and HIPK2 from a speckle-like subnuclear compartment to the cytoplasm, whereas catalytically inactive point mutants showed the same pattern of cellular distribution as the wild type proteins. In contrast, mutation of the activating tyrosine did not increase the low percentage of cells with extranuclear HIPK3. HIPK4 was excluded from the nucleus with no difference between the wild type kinase and the point mutants.

Conclusions: These results show that HIPKs share the mechanism of activation by tyrosine autophosphorylation with the closely related DYRK family (dual-specificity tyrosine phosphorylation regulated kinase). However, members of the HIPK family differ regarding the subcellular localization and its dependence on tyrosine autophosphorylation.

Figure 1

Comparison of HIPK structures and tyrosine phosphorylation. A, Schematic illustration of HIPK1-4. The definition of the domains in HIPK1-3 was adapted from Kim et al. [26]. Nuclear localisation sequences (NLS) and the tyrosine residue in the activation loop (Y) are indicated. B, The tree illustrates the sequence identity of the catalytic domains of HIPK1-4 and DYRK1A and panel C shows the alignment of the activation loop sequences. Tyrosine residues known to be autophosphorylated in HIPK2 and DYRK1A are highlighted (red) [28,29,34]. D, GFP fusion proteins of HIPK1-4 and DYRK1A were immunoprecipitated from transiently transfected HeLa cells. Tyrosine phosphorylation was analysed by immunoblotting with a phosphotyrosine-specific antibody (pTyr). Total amounts of the recombinant proteins were detected with GFP antibody. The figure is representative of two experiments.

Figure 2

Tyrosine phosphorylation of HIPK mutants. HeLa cells were transiently transfected with the indicated expression constructs for HIPK1 (A), HIPK2 (B), HIPK3 (C) or HIPK4 (D). Sodium orthovanadate (Na₃VO₄) was added to every second sample for 1 h before lysis. GFP fusion proteins were immunoprecipitated and analysed by immunodetection with antibodies for pTyr and GFP. The panels are representative of 2–3 independent experiments.

Figure 3

Tyrosine phosphorylation in the activation loop of HIPK1, HIPK2 and HIPK3. HeLa cells were transfected with expression plasmids for wild type GFP-HIPK fusion proteins or the Tyr→Phe mutants thereof (YF). If indicated, cells were treated with sodium orthovanadate (Na₃VO₄) for 1 h or 2 h. Western blots of total cell lysates were detected with antibodies directed against pTyr361 in HIPK2, a general antibody for phosphotyrosine independent of the sequence context (PY99) and a GFP antibody. A, The pTyr361(HIPK2) antibody detects wild type HIPK1-3 but not the Tyr→Phe mutants. B, Effect of vanadate treatment. The column diagrams show the quantitative evaluation of 3 experiments. The background signal in the untransfected control samples was subtracted from the signal intensities obtained with the phosphospecific antibodies. Relative phosphorylation after vanadate treatment was calculated by normalization to the signal measured in untreated cells. Means + SEM, * p < 0.05, analysed by one-sample t test.

Figure 4

Maximal activity of HIPKs depends on the activation loop tyrosine. Wild type GFP-HIPK fusion proteins and the respective Tyr→Phe mutants were immunoprecipitated from HeLa cells and subjected to kinase assays with recombinant GST-p27^Kip1 (A), myelin basic protein (B) or DYRKtide (C). GFP served as background control. A, Phosphorylation of p27^Kip1 at Ser10 was detected by immunoblotting with a phosphorylation-specific antibody. For quantitative evaluation, pSer10 immunoreactivity was normalised to GFP immunoreactivity, which reflects the amount of kinase in the reaction. The blots illustrate a representative experiment, and the relative catalytic activities as determined from 3–4 assays are shown below the panels (means ± SD). One-sample t test: *, p < 0.05; **, p < 0.01. B and C, Phosphorylation of MBP and DYRKtide was measured in triplicate as incorporation of ³²P. Background values from the GFP control samples were subtracted and activities were normalised to the amount of kinase in the reaction as determined by GFP immunoreactivity. Column diagrams illustrate catalytic activities relative to HIPK2 (WT). The results were replicated in independent experiments, except for a missing value of HIPK1.

Figure 5

Subcellular distribution of HIPK mutants. COS-7 cells were transfected with expression vectors for GFP-HIPK1 (A), GFP-HIPK2 (B), GFP-HIPK3 (C) or GFP-HIPK4 (D). The cellular localisation of the HIPK constructs was evaluated by imaging GFP autofluorescence in relation to bisbenzimide-stained nuclei. For each kinase, cells were classified into 3 major patterns that are illustrated by representative images and icons. The cells shown were transfected with (top to bottom) HIPK1 WT, DN, YF; HIPK2 WT, WT, YF; HIPK3 WT, WT, DN and HIPK4 DN, DN. A series of images of the different patterns for each kinase and its mutants is provided in the additional material (Additional file 1: Figure S3). Here the graphs show the percentages of cells classified into the indicated patterns (means ± SD, n = 3). In each of 3 experiments, at least 90 cells were evaluated for all HIPK constructs. Scale bars, 10 μm.