Fig 2: NUAK1 phosphorylation targets in neuronal development and disease(a) Schematic shows substrates of NUAK1 identified by the chemical genetic screen which are binned by their roles in axonal, dendritic, and synaptic development, as well as splicing. Each substrate is color-coded according to its known function(s), with light green for endocytosis, teal for splicing, yellow for cytoskeleton, pink for axon, and blue for dendrites, with the legend in the lower left panel showing each individual color. Substrates with multiple functions are indicated with gradients of 2–4 colors. (b) Schematic representation of the role of PSD3 and NUAK1 in ARF6 activation. At the plasma membrane, PSD3 recruits inactive, GDP-bound ARF6. PSD3 exchanges GDP for GTP, generating active ARF6, which promotes the formation of PI(4,5)P2. This increase in PIP2 causes internalization of PSD3 and ARF6 on PIP2-enriched vesicles. In normal conditions, Rab5, ARF6 GAPs, and lipid phosphatases are recruited to these vesicles, leading to a cessation of ARF6 activity and depletion of PIP2. These vesicles can then continue in the endolysosomal pathway, or be recycled to the cell surface. When NUAK1 is not present, or when the PSD3 S476 phosphosite is mutated to alanine, ARF6 remains active and bound to GTP, accumulating in large, PIP2-enriched vesicles. (c) Schematic of the role of PSD3 in dendritic spine maturation. PSD3 activation of ARF6 leads to an increased proportion of mature, mushroom-shaped dendritic spines and a corresponding loss of immature filopodia. In normal conditions, PSD3 is phosphorylated by NUAK1, keeping ARF6 activation in check. When NUAK1 is inactive or when the PSD3 phosphosite is mutated to alanine, PSD3 phosphorylation is lost, leading to an aberrant increase in the proportion of active ARF6.

Fig 3: PSD3 is a bona fide direct phosphorylation target of NUAK1 kinase.(a) Two-dimensional schematic representation of the function domains of PSD3 isoform 1 (Uniprot # Q9NYI0), isoform 3 (Uniprot # Q9NYI0–3), and the shortened PSD3 construct used for biochemical experiments, PSD3 CC tail. PSD3 isoforms 1 and 3 share the same 3 conserved functional domains, including a Sec7 GEF domain (green) which activates ARF6, Pleckstrin Homology (PH) domain with specificity for PI(4,5)P2, and a Coiled Coil (CC) domain which binds to actin. The MS-identified phosphosite on PSD3, S1011 in isoform 1 or S476 in isoform 3, is indicated with a lollipop. The shortened CC tail domain used for protein purification contains an N-terminal GST tag followed by the C-terminal region of PSD3, from the CC domain to the end of the protein containing the S476/S1011 phosphosite. (b) ConSurf evolutionary conservation analysis of the MS-identified phosphosite of PSD3, S1011 in isoform 1 or S476 in isoform 3. The S1011 phosphoacceptor residue is boxed in violet. Residues are colored according to their evolutionary conservation, with warmer colors indicating conserved residues and cool colors indicating variable residues. (c) V5-tagged PSD3 isoform 3 WT or S476A was immunoprecipitated from HEK293T cells, followed by incubation with purified HM-NUAK1 WT and ATP. Immunoblots show V5 as a loading control for PSD3 and His as a control for NUAK1, and phospho-serine/threonine as a readout of phosphorylation for both NUAK1 and PSD3 at their respective molecular weights. (d) Quantification of phosphorylation of PSD3 by NUAK1, as measured by phospho-serine/threonine divided by V5 signal. Each experiment is normalized to WT PSD3, with S>A PSD3 as a ratio of WT signal (n=3 independent experiments; *** denotes p<0.001, student’s t-test). Error bars represent standard error of the mean. (e) HEK293T cells were co-transfected with mCherry-V5-PSD3 isoform 3 and either HA-empty vector or HA-NUAK1 WT. Cells were then lysed, and protein was immunoprecipitated using HA beads. Immunoblots show HA signal for HA-empty vector or HA-NUAK1, and V5 signal for mCherry-V5-PSD3. (f) Quantification of V5 signal for input and HA IP for n=3 independent experiments. Error bars represent standard error of the mean; ** denotes p<0.005, one way ANOVA with Tukey’s multiple comparisons. (g) HEK293T cells were co-transfected with mCherry-V5-PSD3 isoform 3 and either HA-empty vector or HA-NUAK1 WT. Cells were then lysed, and protein was immunoprecipitated using V5 beads. Immunoblots show HA signal for HA-empty vector or HA-NUAK1, and V5 signal for mCherry-V5-PSD3. (h) Quantification of HA signal for input and V5 IP, for n=3 independent experiments. Error bars represent standard error of the mean; * denotes p<0.05, one way ANOVA with Tukey’s multiple comparisons. (i) HM-NUAK1 WT or KD was incubated with GST-PSD3-CC tail WT or S>A in the presence of Mg2+ and ATPɣS, followed by alkylation with PNBM to generate thiophosphate esters as a readout of thiophosphorylation by NUAK1. (j) Quantification of phosphorylation of GST-PSD3-CC tail WT or S>A by HM-NUAK1 WT or KD. Data are normalized to WT PSD3 + WT HM-NUAK1 for each experiment, with all other samples as a percentage of WT/WT (n=3 independent experiments; **** indicates p<0.0001, one way ANOVA with Tukey’s multiple comparisons). Error bars represent standard error of the mean (k) Alphafold-predicted structure of NUAK1 WT (Uniprot # O60285) with the previously identified and validated substrate MYPT1 (Uniprot ID #O14974), along with adaptor protein Protein Phosphatase 1 beta (Uniprot ID #P62140). NUAK1 kinase domain is displayed as an electrostatic surface representation, with red-toned colors representing a negative charge, and blue-toned colors representing a positive charge. MYPT1 is colored in light blue, and amino acids within the NUAK1 substrate binding groove are color coded according to their position and side chain, with Leu at −5 and +4 in limon, Lys/His at −4, −3, and −1 in blue, and central phospho-acceptor Ser in green. Amino acid side chains are shown for positions −5 to +4 surrounding the MYPT1 S445 phosphosite. (l) Alphafold3 predicted structure of NUAK1 WT (Uniprot # O60285) with PSD3 isoform 1 WT (Uniprot # Q9NYI0), following the same color scheme as figure 4k. Amino acid side chains are shown for positions −5 to +4 surrounding the PSD3 S476 phosphosite.

Fig 4: NUAK1 mediated PSD3 phosphorylation at S476 restricts maturation of filopodia into dendritic spines(a) Representative images of DIV13 rat embryonic hippocampal neurons co-expressing membrane marker myristoylated-tdTomato (red) along with either sfGFP-tagged wildtype (WT) PSD3 or phosphomutant S476A (cyan). Scale bar is 10mm. (b) Magnified insets are shown to highlight the morphology of dendritic spines and the enrichment for PSD3 within spines. Grayscale and pseudocolored images show intensity of PSD3 distribution in WT and S476 conditions. Scale bar is 1μm. (c) Quantification of percent spines enriched in PSD3 in hippocampal neurons expressing WT-PSD3 and S476A phosphomutant (n>15 neurons for each condition from three experiments, **** denotes p<0.0001, t-test with Welch’s correction). (d) Protrusion density per micron dendrite length is plotted for neurons expressing WT-PSD3 and S476A phosphomutant (n>55 neurons for each condition from three experiments, ns denotes p>0.05, t-test with Welch’s correction). (e) Percent protrusions classified into stubby, thin, filopodia and mushrooms spines are plotted neurons expressing WT-PSD3 and S476A phosphomutant (n>55 neurons for each condition from three experiments, **** denotes p<0.0001 and * denotes p<0.05, 2-way ANOVA with multiple comparisons). All error bars are standard error of the mean. (f) Schematic depicts working model based on the results showing role of NUAK1 phosphorylation of PSD3 in regulation of ARF6 mediated membrane internalization and PI(4,5)P2 production. Role of PSD3 phosphorylation in maturation of dendritic spines through ARF6 is shown.