Phosphoproteomic screening identifies physiological substrates of the CDKL5 kinase

Abstract

Mutations in the gene encoding the protein kinase CDKL5 cause a debilitating neurodevelopmental disease termed CDKL5 disorder. The impact of these mutations on CDKL5 function is poorly understood because the substrates and cellular processes controlled by CDKL5 are unclear. Here, we describe a quantitative phosphoproteomic screening which identified MAP1S, CEP131 and DLG5-regulators of microtubule and centrosome function-as cellular substrates of CDKL5. Antibodies against MAP1S phospho-Ser⁹⁰⁰ and CEP131 phospho-Ser³⁵ confirmed CDKL5-dependent phosphorylation of these targets in human cells. The phospho-acceptor serine residues in MAP1S, CEP131 and DLG5 lie in the motif RPXSA, although CDKL5 can tolerate residues other than Ala immediately C-terminal to the phospho-acceptor serine. We provide insight into the control of CDKL5 activity and show that pathogenic mutations in CDKL5 cause a major reduction in CDKL5 activity in vitro and in cells. These data reveal the first cellular substrates of CDKL5, which may represent important biomarkers in the diagnosis and treatment of CDKL5 disorder, and illuminate the functions of this poorly characterized kinase.

Keywords: CDKL5 disorder; centrosome; cilia; kinase; microtubule.

Figure 1. A quantitative phosphoproteomic approach for the identification of cellular CDKL5 targets

1. Pathogenic and non‐pathogenic CDKL5 variants. Schematic diagram shows modular domains in CDKL5 and the position of amino acid substitutions in humans that are either pathogenic, non‐pathogenic or of unknown consequence according to RettBASE (http://mecp2.chw.edu.au/cdkl5/cdkl5_variant_list_copy.php; Krishnaraj et al, 2017). NES: nuclear export signal; NLS: nuclear localization signal.

2. Generation of CDKL5 knockout human cells. U2OS cells modified with the Flp‐In™ T‐REx™ system were subjected to genome editing to disrupt CDKL5. Extracts of cells from two different knockout (KO) clones (7 and 13) were subjected to SDS–PAGE on the gel types indicated followed by immunoblotting with in‐house anti‐CDKL5 antibodies. “Hi” higher exposure; “lo” lower exposure. Asterisk: non‐specific band.

3. Stable expression of CDKL5 in knockout clone 13. The CDKL5 open reading frame (1,030 amino acid/115‐kDa isoform) was inserted at the FRT sites in CDKL5 knockout clone 13 from (B). Cells transfected with empty vector were used as control. Cells were incubated with the indicated concentrations of tetracycline (Tet), and extracts were immunoblotted with anti‐CDKL5 antibodies.

4. Phosphoproteomics workflow. CDKL5 knockout clone 13 and the same cells re‐expressing CDKL5 were lysed and protein extracts were digested using trypsin. After phosphopeptide enrichment by TiO₂ chromatography, peptides were isotopically labelled by TMT and combined. Combined peptides were fractionated by high‐pH reversed‐phase chromatography. Fractions were separated on a nano‐HPLC and analysed by quantitative mass spectrometry on an Orbitrap Fusion mass spectrometer. Data were analysed using MaxQuant software.

Source data are available online for this figure.

Figure 2. Identification of CDKL5 substrates in human cells

1. A

   Volcano plot showing potential CDKL5 substrates. The horizontal cut‐off line represents a P‐value of 0.05, and the vertical cut‐off lines represent a log₂ ratio of 0.58 (˜1.5‐fold) above which peptides were considered to differ significantly in abundance between CDKL5 KO cells expressing empty vector and CDKL5 KO cells expressing CDKL5. The mass spectrometry proteomics data for this figure have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Jarnuczak & Vizcaino, 2017) with the dataset identifier PXD009374.

2. B, C

   Sequence alignment of the Ser⁹⁰⁰ phosphorylation site in MAP1S (B) and the Ser³⁵ phosphorylation site in CEP131 (C), in the species indicated.

3. D

   Gene Ontology (GO) term enrichment of phosphopeptides that are more abundant in CDKL5‐expressing cells compared to knockout cells.

4. E

   String database network analysis of proteins which harbour phosphosites up‐regulated in CDKL5‐expressing cells compared to knockout cells. These are part of protein complexes involved in a wide range of biological functions.

Figure EV1. Dot blot analysis of phospho‐specific antibodies

1. Dot blot analysis of the MAP1S pSer⁹⁰⁰ antibodies. Increasing amounts of MAP1S pSer⁹⁰⁰ phosphopeptide immunogen and non‐phosphopeptide equivalent were spotted onto nitrocellulose, and membrane was subjected to dot blotting with two different concentrations of affinity‐purified MAP1S pS⁹⁰⁰ antibodies (0.2 μg/ml for 18 h or 1 μg/ml for 1 h). Two different antibody bleeds were tested.

2. Dot blot analysis of CEP131 pSer³⁵ antibodies. Increasing amounts of CEP131 pSer³⁵ phosphopeptide immunogen and non‐phosphopeptide equivalent was spotted onto nitrocellulose. After cross‐linking with glutaraldehyde (1% (v/v) in H₂O for 10 min), the membrane was subjected to dot blotting with two different concentrations of affinity‐purified CEP131 pSer³⁵ antibodies (0.2 μg/ml for 18 h or 1 μg/ml for 1 h). Two different antibody bleeds were tested.

Source data are available online for this figure.

Figure 3. CDKL5‐dependent phosphorylation of MAP1S and CEP131 in cells

1. CDKL5 phosphorylates MAP1S at Ser⁹⁰⁰ in human cells. HEK293 cells were co‐transfected with CDKL5 (wild type “WT” or kinase‐dead K⁴²R mutant) and/or FLAG‐MAP1S (wild type “WT” or a S⁹⁰⁰A mutant “SA”). Anti‐FLAG precipitates were subjected to Western blotting with the antibodies indicated. “Hi” higher exposure; “lo” lower exposure. The input extracts were also subjected to immunoblotting (lower panels). Three independent experiments were done, and one representative experiment is shown.

2. CDKL5 phosphorylates CEP131 at Ser³⁵ in human cells. HEK293 cells were co‐transfected with CDKL5 (wild type “WT” or kinase‐dead K⁴²R mutant) and/or FLAG‐CEP131 (wild type WT or a S³⁵A mutant). Anti‐FLAG precipitates were subjected to Western blotting with the antibodies indicated. “Hi” higher exposure; “lo” lower exposure. The input extracts were also subjected to immunoblotting (lower panels). Three independent experiments were done, and one representative experiment is shown.

Source data are available online for this figure.

Figure EV2. CDKLs 1–4 cannot phosphorylate MAP1S or CEP131 in cells

1. Schematic representation of CDKLs 1–5. The kinase catalytic domain is highlighted in dark blue. Amino acid numbers at the N‐ and C‐termini are indicated; the positions of conserved residues in the ATP binding sites that were mutated to render these protein “kinase‐dead” are also indicated.

2. HEK293 cells were co‐transfected with C‐terminally tagged FLAG‐tagged CDKLs 1,2,3,4 or 5 (wild type “WT” or the relevant kinase‐dead mutant) and HA‐MAP1S. Anti‐HA precipitates were subjected to Western blotting with the antibodies indicated. “Hi” higher exposure; “lo” lower exposure. The input extracts were also subjected to immunoblotting (lower panels). Three independent experiments were done, and one representative experiment is shown.

3. Same as (B) except that HEK293 cells were co‐transfected with C‐terminally tagged FLAG‐tagged CDKLs 1,2,3,4 or 5 (wild type “WT” or the relevant kinase‐dead mutant) and HA‐CEP131. Three independent experiments were done, and one representative experiment is shown.

4. HEK293 cells were transfected with C‐terminally tagged FLAG‐tagged CDKLs 1,2,3,4 or 5 (wild type “WT” or the relevant kinase‐dead mutant). Anti‐FLAG precipitates were incubated with the MAP1S S⁹⁰⁰ synthetic peptide in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺, and peptide phosphorylation was measured by Cerenkov counting. Data are represented as mean ± SEM from three independent experiments.

Source data are available online for this figure.

Figure 4. CDKL5 sequence determinants favouring phosphorylation by CDKL5

1. Sequence of synthetic peptides surrounding the sites of phosphorylation in putative CDKL5 substrates that were identified in the phosphoproteomic screening. The amino acid number of the phosphorylated residue in each peptide (highlighted in red) is listed.

2. Peptide kinase assays to investigate CDKL5 sequence specificity. Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or a K⁴²R kinase‐dead “KD” mutant) were incubated with the synthetic peptides from the proteins indicated (sequences shown in A) in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺, and peptide phosphorylation was measured by Cerenkov counting.

3. Same as (B), except that the peptides used were designed specifically to investigate the effect of amino acid substitutions at R⁸⁹⁷, P⁸⁹⁸, L⁸⁹⁹ and A⁹⁰¹ on the phosphorylation of MAP1S Ser⁹⁰⁰. The RPXSA motif is shaded in blue, and amino acid substitutions compared with the wild‐type MAP1S Ser⁹⁰⁰ peptide are shown in red.

4. Same as (C), except that phosphorylation of the indicated peptides from AMPH1 and MAP1S was compared.

Data information: In (B–D), phosphorylation of the control wild‐type MAP1S peptide is taken as 100%. The data are represented as mean ± SEM from three independent experiments.Source data are available online for this figure.

Figure 5. Mass spectrometric validation of DLG5 as a CDKL5 substrate

1. A–F

   Testing CDKL5‐dependent DLG5 phosphorylation in cells. HEK293 cells were co‐transfected with FLAG‐DLG5 (A, B), FLAG‐MAP1S (C, D) or FLAG‐CEP131 (E, F) together with either empty vector (−), wild‐type CDKL5 (WT) or kinase‐dead CDKL5 (K⁴²R). Cells were harvested after 24 h and lysed, and extracts were subjected to immunoprecipitation with anti‐FLAG‐agarose beads. Precipitates were resolved by SDS–PAGE and stained with Coomassie Brilliant Blue (COO, top panel). A small fraction of the precipitates was subjected to Western blotting with anti‐FLAG antibodies (A, C). The input extracts were also subjected to immunoblotting with the antibodies indicated in the lower panels. Three independent experiments were done, and one representative experiment is shown. Coomassie‐stained bands as shown in top panels of (A, C and E) were excised, destained and proteins digested. Peptides were extracted, TMT labelled and analysed using mass spectrometry. VSN‐calibrated and transformed, isotopically corrected reporter ion intensities of the phosphopeptides of interest are plotted in (B, D and F); the outline of the box plot indicates minimum and maximum values, and the middle line indicates the median. Higher intensity corresponds to higher phosphopeptide abundance in the relevant sample. Statistical testing was carried out using a t‐test (Computer Code EV1; Appendix Fig S4); to account for multiple testing, the significance threshold was adjusted from α = 0.05 to α = 0.00833 (six t‐tests) by Bonferroni correction (Computer Code EV1). All t‐tests resulted in a P‐value below the adjusted significance threshold. **P < 0.01; ***P < 0.001. Individual P‐values are shown in Table EV3.

2. G

   Sequence alignment of the Ser¹¹¹⁵ phosphorylation site in DLG5 in the species indicated.

Source data are available online for this figure.

Figure EV3. CDKL5 phosphorylates DLG5 Ser¹¹¹⁵ in vitro

Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or a K⁴²R kinase‐dead “KD” mutant) were incubated with the synthetic peptides indicated in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺, and peptide phosphorylation was measured by Cerenkov counting. Data are represented as mean ± SEM from three independent experiments.Source data are available online for this figure.

Figure EV4. Sequence conservation around the CDKL5 T‐loop

Schematic diagram showing alignment of catalytic subdomains VII and VIII from CDKL5 in species indicated. The position of the conserved ¹⁶⁸Y‐TEY¹⁷¹ (human numbering) motif is indicated.

Figure 6. T‐loop autophosphorylation is critical for CDKL5 activity

1. A, B

   Tyr‐autophosphorylation of CDKL5. Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or mutants K⁴²R kinase‐dead “KD”, Y¹⁶⁸A,Y¹⁷¹A or both) were subjected to immunoblotting with the antibodies indicated before or after incubation of precipitates with Mg²⁺‐ATP for 30 min at 30°C. Each experiment was done three times, and a representative experiment is shown.

2. C

   CDKL5 cannot phosphorylate Tyr‐containing synthetic peptides. Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or a K⁴²R kinase‐dead “KD” mutant) were incubated with the synthetic peptides indicated in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺, and peptide phosphorylation was measured by Cerenkov counting. Data are represented as mean ± SEM from three independent experiments.

3. H

   HEK293 cells were co‐transfected with untagged CDKL5 (wild type “WT” or the mutants indicated) and FLAG‐tagged MAP1S [wild type (WT), a Ser⁹⁰⁰Ala mutant (SA)] or empty vector (−). Anti‐FLAG precipitates were subjected to Western blotting with the antibodies indicated. The input extracts were also subjected to immunoblotting (lower panels). Each experiment was done three times, and a representative example is shown.

4. I

   Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or the mutants indicated) were incubated with a synthetic peptide corresponding to the sequence around the MAP1S Ser⁹⁰⁰ phosphorylation site, in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺. Peptide phosphorylation was quantitated in a scintillation counter. Data are represented as mean ± SEM from three independent experiments.

Source data are available online for this figure.

Figure EV5. T‐loop autophosphorylation is critical for CDKL5 activity towards CEP131

1. HEK293 cells were co‐transfected with untagged CDKL5 (wild type “WT” or the mutants indicated) and FLAG‐tagged CEP131 [wild type (WT), a Ser³⁵Ala mutant (SA)] or empty vector (−). Anti‐FLAG precipitates were subjected to Western blotting with the antibodies indicated. The input extracts were also subjected to immunoblotting (lower panels). Three independent experiments were done, and one representative experiment is shown.

2. Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or the mutants indicated) were incubated with a synthetic peptide corresponding to the sequence around the CEP131 Ser³⁵ phosphorylation site, in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺. Peptide phosphorylation was quantitated in a scintillation counter. Data are represented as mean ± SEM from three independent experiments.

Source data are available online for this figure.

Figure 7. Effect of pathogenic CDKL5 mutations on CDKL5 activity

1. HEK293 cells were co‐transfected with untagged CDKL5 (wild type “WT” or the mutants indicated) and HA‐tagged MAP1S. Anti‐HA precipitates were subjected to Western blotting with the antibodies indicated. The input extracts were also subjected to immunoblotting (lower panels). Three independent experiments were done, and one representative experiment is shown. Hi, high exposure; lo, low exposure.

2. Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or the mutants indicated) were incubated with a synthetic peptide corresponding to the sequence around the MAP1S Ser⁹⁰⁰ phosphorylation site, in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺. Peptide phosphorylation was quantitated in a scintillation counter. Data are represented as mean ± SEM from three independent experiments.

3. HEK293 cells were co‐transfected with untagged CDKL5 (wild type “WT” or the mutants indicated) and FLAG‐tagged CEP131. Anti‐FLAG precipitates were subjected to Western blotting with the antibodies indicated. The input extracts were also subjected to immunoblotting (lower panels). Three independent experiments were done, and one representative experiment is shown. Hi, high exposure; lo, low exposure.

4. Anti‐FLAG precipitates from HEK293 cells transiently expressing FLAG‐tagged CDKL5 (wild type “WT” or the mutants indicated) were incubated with a synthetic peptide corresponding to the sequence around the CEP131 Ser³⁵ phosphorylation site, in the presence of [γ‐³²P]‐labelled ATP‐Mg²⁺. Peptide phosphorylation was quantitated in a scintillation counter. Data are represented as mean ± SEM from three independent experiments.

Source data are available online for this figure.

Figure EV6. Effect of pathological mutations on CDKL5 tyrosine phosphorylation

HEK293 cells were transfected with FLAG‐tagged CDKL5 (wild type “WT” or a K⁴²R kinase‐dead “KD” mutant, or the mutants indicated). Anti‐FLAG precipitates were subjected to Western blotting with the antibodies indicated.Source data are available online for this figure.