Proteomic investigation of neural stem cell to oligodendrocyte precursor cell differentiation reveals phosphorylation-dependent Dclk1 processing

Oligodendrocytes are generated via a two-step mechanism from pluripotent neural stem cells (NSCs): after differentiation of NSCs to oligodendrocyte precursor/NG2 cells (OPCs), they further develop into mature oligodendrocytes. The first step of this differentiation process is only incompletely understood. In this study, we utilized the neurosphere assay to investigate NSC to OPC differentiation in a time course-dependent manner by mass spectrometry-based (phospho-) proteomics. We identify doublecortin-like kinase 1 (Dclk1) as one of the most prominently regulated proteins in both datasets, and show that it undergoes a gradual transition between its short/long isoform during NSC to OPC differentiation. This is regulated by phosphorylation of its SP-rich region, resulting in inhibition of proteolytic Dclk1 long cleavage, and therefore Dclk1 short generation. Through interactome analyses of different Dclk1 isoforms by proximity biotinylation, we characterize their individual putative interaction partners and substrates. All data are available via ProteomeXchange with identifier PXD040652. Supplementary Information The online version contains supplementary material available at 10.1007/s00018-023-04892-8.


Supplementary
Samples were taken at days 0, 3, 6, and 9, spheres dissociated and grown in adherent cultures on cover slips.Cover slips were stained with the respective antibodies and analyzed for the presence of astrocytes (glial fibrillary acidic protein, Gfap), neurons (β3-tubulin), microglia (cell surface glycoprotein F4/80), and oligodendrocytes (myelin basic protein, Mbp) by fluorescence microscopy.A-D) For all individual clusters (Fig. 2B), GO analyses were performed.Fold-enrichment and FDR p-values of all results for the GO categories "Biological Process" (left) and "Cellular Component" (right) are shown.Only categories with an FDR p-value < 0.01 and a fold enrichment > 4 are indicated.Data are only shown for clusters 1-4, as analysis of clusters 5 to 7 resulted in few or no significantly enriched GO terms (for all results, see Table S1).E) Western blot analysis of PurA and alpha-Tubulin expression in untransfected and PurA-transfected HeLa cells.Increase in PurAsignal intensity confirms lack of cross-reactivity of PurA and alpha-Tubulin antibodies.
Uncropped Western Blots and Agarose Gels

Figure S2 :
Figure S2: Post-processing of TMT data from neurosphere to oligosphere differentiation.A) Abundances (log2 values) of individual reporter ion channels of TMT data before normalization.B) Abundances (log2 values) of individual reporter ion channels of TMT data after LOESS and subsequent batch normalization.Batch normalization was based on a linear regression model to reduce batch effects between replicates.C) Pearson correlation heatmap for individual replicates and samples.D) Row-wise hierarchical clustering heatmap (based on Spearman correlation distance and complete linkage) for individual replicates and samples.Shown are zscored (row) normalized log2 abundances.

Figure S3 :
Figure S3: Results of paired t-tests between day 0 and individual subsequent time points.Shown are individual volcano plots for each comparison, -log10 p-values and log2 fold-change ratios are shown.Vertical lines indicate a fold-change of > 2 or < 0.5, horizontal lines an adjusted p-value of 0.05 (back-transformed to the scale of the original p-values).Numbers of significantly up-or down-regulated proteins are indicated (cutoff: adjusted p-value < 0.05 and fold change > 2 or < 0.5).A) Day 3 vs.day 0. B) Day 6 vs. day 0. C) Day 9 vs. day 0. D) Day 12 vs.day 0. E) Day 15 vs. day 0.

Figure S4 :
Figure S4: Gene Ontology (GO) analyses for individual clusters of the TMT dataset and investigation of PurA/alpha-Tubulin-antibody cross-reactivity.A-D) For all individual clusters (Fig.2B), GO analyses were performed.Fold-enrichment and FDR p-values of all results for the GO categories "Biological Process" (left) and "Cellular Component" (right) are shown.Only categories with an FDR p-value < 0.01 and a fold enrichment > 4 are indicated.Data are only shown for clusters 1-4, as analysis of clusters 5 to 7 resulted in few or no significantly enriched GO terms (for all results, see TableS1).E) Western blot analysis of PurA and alpha-Tubulin expression in untransfected and PurA-transfected HeLa cells.Increase in PurAsignal intensity confirms lack of cross-reactivity of PurA and alpha-Tubulin antibodies.

Figure S5 :
Figure S5: Quality control plots for the dimethyl labeling based phosphoproteomic analysis of neurosphere to oligosphere differentiation.A/B) Density plots showing the distribution of normalized log2 ratios of quantified protein groups (A) and class I phosphosites (B) for individual replicates of the different time point comparisons.C/D) Boxplots showing the distribution of normalized log2 ratios of quantified protein groups (C) and class I phosphosites (D) for individual replicates of the different time point comparisons.

Figure S6 :
Figure S6: Sequence pattern analysis of regulated phosphorylation sites.IceLogo 38 plots for up-and downregulated class I phosphosites in comparisons of different time points.

Figure S8 :
Figure S8: Proteomic analysis of Dclk1 isoform interactomes.A) Principal component analysis (PCA) of individual samples.B) Clustered heatmap analysis of individual samples (Manhattan distance, complete linkage clustering).Analyses shown in (A/B) are based on normalized log2 intensities after missing value imputation.

Figure S9 :
Figure S9: Differential expression analysis for individual Dclk1 isoform interactomes and overlap determination for proteins which are differentially interacting with individual Dclk1 isoforms.Binary comparisons for protein abundances of individual interactomes for different BirA*-Dclk1 constructs.Significantly differentially abundant proteins were determined based on a fold-change (log2 ≥ |0.58|) and adjusted p-value (q-value) cutoff (< 0.05).A) SK vs. K; B) DSK vs. DS; C) DSK vs. SK; D) DSK vs. K; E) DS vs. SK; F) DS vs. K.D: Dcx domain; K: kinase domain; S: SP-rich domain.Colored dots indicate regulated proteins (up = red, down = blue), numbers in the same color represent the number of proteins belonging to the respective category.G/H) Comparison of all proteins which were significantly upregulated for constructs containing a Dcx domain, therefore localizing to microtubules (G), or such without this domain, localizing to the cytosol or the nucleus (H).Venn diagrams were generated with the R package VennDiagram (version 1.7.3).

Figure S10 :
Figure S10: Analysis of Dclk1 isoform interactome phosphosites.A) Principal component analysis (PCA) of individual samples.B) Clustered heatmap of individual samples (Manhattan distance, complete linkage clustering).Analyses shown in (A/B) are based on normalized log2 intensities after missing value imputation.C) Count of regulated class I phosphosites per construct.Shown are mean values +/-SD (n = 3).

Figure S11 :
Figure S11: Differential expression analysis of phosphorylation sites identified in individual Dclk1 isoform interactomes.Binary abundance comparisons for individual phosphosites identified in the interactomes of different BirA*-Dclk1 constructs.Significantly regulated phosphosites were determined based on a fold-change (log2 ≥ |0.58|) and adjusted p-value (q-value) cutoff (< 0.05).Only fusion proteins containing the Dclk1 kinase domain were considered.A) DSK vs. K; B) DSK vs. SK; C) SK vs. K.D: Dcx domain; K: kinase domain; S: SP-rich domain.Colored dots indicate regulated proteins (up = red, down = blue), numbers in the same color represent the number of proteins belonging to the respective category.D/E/F) Comparison of significantly regulated phosphosites with the Figure 2