Role of PDXP in hippocampal PLP homeostasis.

(a) Age-dependent expression of PDXK and PDXP in murine hippocampi. Left panels, representative Western blots of three hippocampi for each genotype. The same blots were reprobed with α-actin antibodies as a loading control. The age of the investigated mice is indicated above the blots. Right panel, densitometric quantification of hippocampal PDXP and PDXK Western blot signals, corrected by the corresponding actin signals. Young mice were 18-42 days old, older mice were 252-351 days old; n=7 hippocampi were analyzed per group. Data are mean values ± S.D. Statistical analysis was performed with unpaired, two-sided t-tests; p-values are indicated. (b) Age-dependent, total PLP-concentrations in isolated hippocampi of PDXP-WT and PDXP-KO mice. PLP was derivatized with semicarbazide and analyzed by HPLC. Each symbol represents the result of the PLP determination in an individual hippocampus. Data were fitted by Gaussian least-squares analyses. (c) Determination of protein-bound and protein-depleted PLP in PDXP-WT and PDXP-KO hippocampal lysates of young (18-42 days old) and older mice (252-352 days old). The number of analyzed hippocampi is indicated in the bars. Data are mean values ± S.D. Statistical analysis was performed with two-way ANOVA and Tukey’s multiple comparisons test. Significant differences (adjusted P-values) in protein-depleted PLP levels are indicated. The exact age of analyzed mice is listed in Figure 1 – supplementary figure 1. Source data are available for this Figure.

Characterization of the 7,8-DHF/PDXP interaction.

(a) Determination of half-maximal inhibitory constants (IC50) of 7,8-DHF (2D-structure shown on top) for purified murine or human PDXP, using pyridoxal 5’-phosphate (PLP) as a substrate. Phosphatase activities in the presence of 7,8-DHF were normalized to the respective enzyme activities measured in the presence of the DMSO solvent control. Data are mean values ± S.D. of n=3 (human PDXP) and n=4 (murine PDXP) independent experiments. (b) IC50 values of different flavones for purified murine PDXP with PLP as a substrate. Phosphatase activities in the presence of flavones were normalized to the respective enzyme activities in the presence of the DMSO solvent control. All data are mean values ± S.D. The inhibition of PDXP by 3,7,8-trihydroxyflavone-4’-hydroxyphenyl (2D-structure shown on top) was assessed in n=6 independent experiments. All other data are from n=3 biologically independent experiments. Apparently missing error bars are hidden by the symbols. (c) Biolayer interferometry (BLI) measurements of the interaction of 7,8-DHF with purified murine PDXP. Left panel, example sensorgram overlayed with the global 1:1 binding model (red) and the negative control (gray). The dashed line indicates the start of the dissociation phase. Right panel, steady-state dose-response analysis for 7,8-DHF based on n=4 measurements. (d) Sensitivity of the indicated phosphatases to 7,8-DHF. Phosphatase activities in the presence of 7,8-DHF were normalized to the respective enzyme activities measured in the presence of the DMSO solvent control. Data are mean values ± S.D. of n=4 (PGP) or n=3 independent experiments (all other phosphatases). Phosphatase substrates and HAD phosphatase cap types are indicated in parentheses. PDXP, pyridoxal-5’-phosphate phosphatase (pyridoxal 5’-phosphate, C2); PGP, phosphoglycolate phosphatase (2-phosphoglycolate; C2); LHPP, phospholysine phosphohistidine inorganic pyrophosphate phosphatase (imidodiphosphate; C2); NT5C1A, soluble cytosolic 5’-nucleotidase 1A (AMP; C1); NANP, N-acetylneuraminate 9-phosphate phosphatase (6-phosphogluconate; C1); PHOP2, phosphatase orphan 2 (pyridoxal 5’-phosphate; C1); PSPH, phosphoserine phosphatase (O-phospho-L-serine; C1); PNKP, polynucleotide kinase phosphatase (3-phospho-oligonucleotide; C0); MDP1, magnesium-dependent phosphatase-1 (D-ribose-5-phosphate; C0); PTP1B (protein tyrosine phosphatase 1B; EGFR phospho-peptide); PP2B, protein phosphatase 2B/calcineurin (PKA regulatory subunit type II phospho-peptide); CIP, calf intestinal phosphatase (pNPP). Source data are available for this Figure.

Kinetic constants of PDXP-catalyzed PLP hydrolysis in the presence of 7,8-DHF.

X-ray crystal structures of human PDXP in complex with 7,8-DHF.

(a) The models were refined to a resolution of 1.5 Å for full-length human 7,8-DHF-PDXP with phosphate (PDB code 9EM1, colored in wheat yellow, left panel) and 1.5 Å for full-length human 7,8-DHF-PDXP without phosphate (PDB code 8S8A, colored in light blue, right panel). One protomer of each homodimeric PDXP is shown in cartoon representation and the other protomer in surface representation. 7,8-DHF is displayed in sphere representation with its C-atoms in cyan. Mg2+ ions are shown as deep purple spheres and phosphate ions are shown in sphere representation with the phosphorous atom in orange. (b) Orientation of 7,8-DHF in the active sites of human 7,8-DHF-PDXP in the presence or absence of phosphate. Structural details of bound 7,8-DHF and adjacent residues of the active sites are shown. Left, phosphate-containing 7,8-DHF-PDXP (wheat yellow, cartoon representation). Right, phosphate-free 7,8-DHF-PDXP (light blue, cartoon representation). 7,8-DHF is shown in stick representation (cyan C-atoms). The corresponding amino acids in murine PDXP are given in parentheses (see also Fig. 3 – figure supplement 1e, f). (c) Comparison of the Mg2+ coordination spheres. From left to right: human apo-PDXP (PDB: 2P27), human PDXP in complex with PLP (PDB: 2CFT), human PDXP in complex with 7,8-DHF in the presence of phosphate (PDB: 9EM1), human PDXP in complex with 7,8-DHF in the absence of phosphate (PDB: 8A8S). The catalytically essential Mg2+ is shown as a deep purple sphere. In 2CFT, Mg2+ was exchanged for Ca2+, which is shown here as a light brown-colored sphere. Water molecules are shown as blue spheres. (d) Verification of 7,8-DHF - PDXP interactions. Left panel, phosphatase activity of purified PDXP or the indicated PDXP variants. Data are mean values ± S.D. of n=3 independent experiments. Right panel, determination of the IC50 values of 7,8-DHF for purified PDXP or the indicated PDXP variants. Data are mean values ± S.D. of n=3 independent experiments. Apparently missing error bars are hidden by the symbols. Source data are available for this Figure.

Data Collection and Refinement Statistics.

Alignment of murine and human PDXP structures.

Effect of 7,8-DHF on the PLP/PL ratio in cultured hippocampal neurons from WT or PDXP-KO mice.

(a) Effect of long-term PDXP deficiency on total PLP levels in hippocampal neurons. Data are mean values ± S.E. of n=4 independent experiments. Statistical significance was assessed with a two-tailed, unpaired t-test. A representative image of primary hippocampal neurons stained for the neuronal marker protein MAP2 is shown in the insert (pixel intensities were color-inverted for better visualization). Scale bar, 100 µm. (b) Western blot analysis of PDXP and PDXK expression in hippocampal neuron samples shown in (a). The same blots were reprobed with α-actin antibodies as a loading control. The densitometric quantification of PDXK signals is shown on the right; data are mean values ± S.E. of n=4 independent experiments. (c) Effect of 7,8-DHF (20 µM, 45 min) or the DMSO solvent control (0.02% v/v, 45 min) on the PLP/PL ratio in hippocampal neurons of PDXP-WT or PDXP-KO mice. Source data are available for this Figure.

Analysis of total hippocampal PLP levels in PDXP-WT and PDXP-KO mice.

Each value represents the result of the PLP determination in an individual hippocampus. Analysis for statistically significant differences between PLP levels in PDXP-WT and PDXP-KO hippocampi (all ages combined; two-tailed, unpaired t-test) p<0.0001. Bold table entries indicate those hippocampal extracts that were further separated for an analysis of protein-depleted and protein-bound PLP (see Fig. 1c). Source data are available for this Table.

Identification of PDXP inhibitors.

A primary screen was conducted using 6,8-difluoro-4-methylumbelliferyl phosphate (DiFMUP) as an artificial substrate. Out of 41,182 screened compounds, 256 compounds were discarded that showed very high autofluorescence (as recognized by elevated fluorescence at the start of the kinetic curve); 26 compounds showed statistically significant PDXP activation, and 255 compounds showed PDXP inhibition (as recognized by an elevated or decreased slope of the kinetic curve, respectively). The average Z’-factor of the screen was 0.75 ± 0.112. These 281 compounds were selected for DiFMUP-based concentration-dependent validation, and the 46 most potent compounds were selected. A counter-screening was conducted in parallel, also in a concentration-dependent fashion, against the PDXP paralog and closest relative phosphoglycolate phosphatase (PGP). The 11 compounds that were inactive against PGP were validated in a secondary assay, using the PDXP substrate pyridoxal 5’-phosphate (PLP). One PDXP inhibitor hit blocked PLP dephosphorylation by ≥50%. Source data are available for this Figure.

PDXP inhibitor hits.

Determination of half-maximal inhibitory constants (IC50) of 11 PDXP inhibitory compounds (see InChI Key for chemical substance identification) using purified murine PDXP and pyridoxal 5’-phosphate (PLP) as a substrate. Data marked with an asterisk (*) are results of n=3 independent experiments. Because of the limited quantity of most compounds available for these assays, all other data are results of n=1 determinations. 7,8-DHF, 7,8-dihydroxyflavone; ∼, approximate IC50 value. Source data are available for this Table.

BLI measurements of the interaction of 7,8-DHF with purified murine PDXP.

Sensorgrams of three additional experiments overlayed with the global 1:1 binding model (red) and the negative control (gray). The dashed line indicates the start of the dissociation phase. Source data are available for this Figure.

X-ray crystal structures of murine PDXP in complex with 7,8-DHF.

(a) X-ray crystal structure of the assembled homodimeric, full-length murine PDXP in gray cartoon (protomer A) and surface (protomer B) representation. The model was refined to a resolution of 2.0 Å (PDB code 8QFW). 7,8-DHF is shown in sphere representation with its C-atoms in cyan. The Mg2+ ion is shown as deep purple sphere. The phosphate ion is shown as spheres with the phosphorous atom in orange. (b) The 2Fo − Fc electron density map of the depicted amino acids is contoured at an RMSD of 1.0 in blue mesh and superimposed with the refined model. The Fo − Fc polder electron density map of 7,8-DHF is contoured at an RMSD of 3.0 in green mesh. (c) A salt bridge between Arg62 in the B-protomer (in black) and Asp14 of a symmetry-related A-protomer (in gray) blocks the 7,8-DHF binding site in the crystal lattice of murine PDXP. 7,8-DHF (in stick representation with cyan C-atoms) is modeled based on the A-protomer. (d) In vitro phosphatase activity of the purified PDXP-D14A variant in the presence of 7,8-DHF. Data are mean values ± S.D. of n=3 independent experiments. Apparently missing error bars are hidden by the symbols. The purity of PDXP-D14A is shown in the Coomassie Blue-stained gel on the right. (e) Structural details of bound 7,8-DHF and adjacent residues of the active site of mPDXP (in gray cartoon representation). (f) Superimposition of the 7,8-DHF binding sites in the phosphate-containing murine PDXP (shown in gray) and human PDXP (in wheat yellow) structures. The corresponding amino acids are shown as gray or wheat yellow-colored sticks. 7,8-DHF bound to murine or human PDXP is shown as sticks colored in gray or cyan, respectively. The position of the Mg2+ and phosphate ions is based on human PDXP. Source data are available for this Figure.

7,8-DHF coordination in PDXP.

(a) The 2Fo − Fc electron density map of the depicted amino acids is contoured at an RMSD of 1.0 in blue mesh and superimposed with the refined model. The Fo − Fc polder electron density map of 7,8-DHF is contoured at an RMSD of 3.0 in green mesh. The Mg2+ ion is shown as a deep purple sphere. The phosphate ion is shown as a sphere with the phosphorous atom in orange. Left panel, human PDXP with phosphate (cartoon representation in wheat yellow); right panel, human PDXP without phosphate (cartoon representation in light blue). (b) Comparison of the 7,8-DHF and PLP binding sites in human PDXP with phosphate (wheat yellow, left panel), or human PDXP without phosphate (light blue, right panel). 7,8-DHF is shown in stick representation with cyan C-atoms. PLP (in stick representation with green C-atoms) was modelled based on a superposition of the human PDXP-PLP complex (PDB code 2CFT).

Alignment of human and murine PDXP.

Protein sequences of human PDXP (UniProtKB Q96GD0) and murine PDXP (UniProtKB P60487) were aligned with the EMBL-EBI multiple sequence alignment tool Clustal Omega version 1.2.4. PDXP residues found to engage in 7,8-DHF interactions (highlighted in red color) are identical in human and murine PDXP.

Salt bridge formation between Glu152 (Glu148) and Arg62 gates the active site entrance in PDXP.

Shown are views of the active site entrance in (a) hPDXP + 7,8-DHF with PO43-, (b) hPDXP + 7,8-DHF without PO43-, (c) apo-hPDXP, (d) hPDXP + PLP, (e) mPDXP + 7,8-DHF with PO43-, chain A, (f) mPDXP + 7,8-DHF with PO43-, chain B (inhibitor-free), (g) apo-mPDXP, chain A, (h) apo-mPDXP, chain B. The cap domain residue Glu152 in hPDXP (corresponding to Glu148 in mPDXP) is shown in blue, and the core domain residue Arg62 in hPDXP and mPDXP is shown in orange. 7,8-DHF is shown in stick representation with C-atoms in cyan. PLP is shown in stick representation with C-atoms in light green. The phosphate is shown in sphere representation with the phosphorous atom in orange. Mg2+ is shown as a deep purple sphere. hPDXP, human PDXP; mPDXP, murine PDXP. The respective PDB entries are indicated.

Purity of the employed PDXP and PDXP variants.

A Coomassie Blue-stained gel is shown.