Deletion of DLK in postmitotic glutamatergic neurons does not alter gross morphology of hippocampus.

(A) Western blot of DLK and β-actin from hippocampal tissue of Vglut1Cre/+;DLK(cKO)fl/fl and DLK(cKO)fl/fl littermate controls (age P60, each lane representing protein sample from individual mice, N=3 mice/genotype). (B) Quantification of DLK protein level normalized to β-actin. Statistics: Unpaired t-test, *** P<0.001. (C) Confocal z-stack (max projection) images of coronal sections of the dorsal hippocampus immunostained for NeuN. Mice are P15 and P60, respectively. Dashed boxes in CA1 pyramidal layers are enlarged below. Scale bar: 1,000 μm in hippocampi; 100 μm in CA1 layer. (D) Quantification of CA1 pyramidal layer thickness in P15 and P60 mice. Each dot represents averaged thickness from 3 sections per mouse, N≥4 mice/genotype per stage. Statistics: Two-way ANOVA with Holm-Sidak multiple comparison test; ns, not significant. (E) Confocal z-stack images of hippocampus CA1, CA3, and DG regions immunostained with Tuj1 (age P60). Dashed outlines mark ROI (region of interest) for fluorescence intensity quantification. Scale bar: 100 μm. (F,G,H) Tuj1 mean fluorescence intensity (MFI) after thresholding signals in dendritic regions in each hippocampal area. N=4 control, and 5 Vglut1Cre/+;DLK(cKO)fl/flmice. Statistics: Unpaired t-test. ns, not significant.

Induced DLK overexpression in hippocampal glutamatergic neurons causes degeneration of CA1 neurons.

(A) Coronal sections from dorsal hippocampus of P10, P15, and P60 mice, stained for NeuN (shown as z-stack of max intensity projection). Dashed boxes mark CA1 pyramidal layers enlarged below. Scale bar: 1,000μm in hippocampi; 100μm in CA1 layer. (B) Quantification of CA1 pyramidal layer thickness. Data points represent averaged measurement from 3 sections per mouse, N≥4 mice/genotype at each stage. Statistics: Two-way ANOVA with Holm-Sidak multiple comparison test. ns, not significant; *** P<0.001; **** P<0.0001.

A select set of genes in hippocampal glutamatergic neurons show translation dependency on the expression levels of DLK.

(A) Volcano plot showing RiboTag analysis of gene expression in Vglut1Cre/+;H11-DLKiOE/+;Rpl22HA/+vs Vglut1Cre/+;Rpl22HA/+ (age P15). 260 genes showing differential expression with adjusted p-values < 0.05 are shown in red. Names of genes with p<1E-10 are labeled. (B) Volcano plot showing RiboTag analysis of genes in Vglut1Cre/+;DLK(cKO)fl/fl;Rpl22HA/+vs Vglut1Cre/+;Rpl22HA/+ (age P15). 36 genes showing differential expression with adjusted p-values < 0.05 are shown in blue. Names of genes with p<1E-10 are labeled. (C) Rank-rank hypergeometric overlap comparison of gene expression in DLK(iOE) and DLK(cKO) datasets shows enrichment of similar genes when DLK is low or high. Color represents the -log transformed hypergeometric p-values (blue=weaker p-value, red=stronger p-value). (D,E) Gene ontology (GO) analysis of significantly upregulated or downregulated genes, respectively, when DLK expression is increased in hippocampal glutamatergic neurons. Colors correspond to P-values. Circle size represents fold enrichment for the GO term, with number on X position showing # of significant genes included in the GO term. (F) SynGO sunburst plot shows significant enrichment for differentially expressed genes when DLK expression is increased in hippocampal glutamatergic neurons. (G,H) Pie charts show distribution of synaptic genes whose expression exhibits dependency when DLK expression is increased in hippocampus, based on in situ data in CA1, CA3, and DG in dorsal hippocampus (P56) in the Allen Mouse Brain Atlas.

Stmn4 and microtubule homeostasis show dependency on the expression levels of DLK.

(A) RNAscope analysis of Stmn4 and Vglut1 mRNAs in hippocampal neurons. Dashed circle outlines single nuclei. Shown as single-slice confocal image; scale bar 10μm. (B,C) Quantification of the ratio of Stmn4 to Vglut1 RNAscope puncta in same nuclei of CA1 and CA3 neurons, respectively. N >50 cells per genotype, from 6,3,3 mice of respective genotype and 4 sections per mouse. Statistics: One way ANOVA with Dunnett’s multiple comparison test, ns, not significant; ** P<0.01; **** P<0.0001. (D-F) Confocal images of immunostained CA1 sections for Tuj1, tyrosinated tubulin, and acetylated tubulin, respectively, in P15 control and Vglut1Cre/+;H11-DLKiOE/+mice. SR denotes stratum radiatum region used for quantification. (G) Normalized tyrosinated tubulin mean fluorescence intensity (MFI) after thresholding signals in SR in CA1. N=9, 6 mice, 3 sections averaged per mouse. (H) Normalized acetylated tubulin MFI after thresholding signal in SR in CA1. N=6, 4 mice, 3 sections averaged per mouse. (I-K) Confocal images of immunostained CA1 sections for Tuj1, tyrosinated tubulin, and acetylated tubulin, respectively, in P60 control and Vglut1Cre/+;H11-DLKiOE/+mice. (L) Normalized tyrosinated tubulin MFI after thresholding signal in SR in CA1. N=9, 9 mice, 3 sections averaged per mouse. (M) Normalized acetylated tubulin MFI after thresholding signal in SR in CA1. N=6, 6 mice, 3 sections averaged per mouse. (N-P) Confocal images of immunostained CA1 sections for Tuj1, tyrosinated tubulin, and acetylated tubulin, respectively, in P60 control and Vglut1Cre/+;DLK(cKO)fl/fl mice. (Q) Normalized tyrosinated tubulin MFI after thresholding signal in SRin CA1. N=5, 7 mice, 3 sections averaged per mouse. (M) Normalized acetylated tubulin MFI after thresholding signal in SR in CA1. N=6, 7 mice, 3 sections averaged per mouse. All tubulin images shown as maximum projection of z-stack, scale bar 10μm. Dashed lines outline pyramidal layer. Arrows point to apical dendrites with elevated signal; arrowheads point to thin neurites with elevated signal. Statistics: (G,H,L,M,Q,R) Unpaired t-test. ns, not significant; * P<0.05.

Hippocampal dorsal CA1 glutamatergic neurons show altered synapses when DLK expression is increased.

(A) Confocal images of Bassoon and Homer1 immunostaining in CA1 stratum radiatum (SR) of control and Vglut1Cre/+;DLK(cKO)fl/flmice at P60. (B-C) Quantification of Bassoon and Homer1 puncta density, respectively. (D) Quantification of synapses displaying co-localization of Bassoon and Homer1. Data points represent individual ROIs quantified from 5-6 images per mouse across 3 sections. N=7 control, and 8 Vglut1Cre/+;DLK(cKO)fl/flmice. (E) Confocal images of Bassoon and Homer1 immunostaining in CA1 SR of control and Vglut1Cre/+;H11-DLKiOE/+ mice at P15. (F-G) Quantification of Bassoon and Homer1 puncta density, respectively. (H) Quantification of synapses displaying co-localization of Bassoon and Homer1. All images shown as single slice; scale bars, 5μm in panel images, and 1μm in enlarged images. Data points represent individual ROIs quantified from 5-6 images per mouse across 3 sections, N=9 control, and 6 Vglut1Cre/+;H11-DLKiOE/+ mice. Statistics: unpaired t-test or Mann-Whitney U test if not passing normality. ns, not significant; *** P<0.001; **** P<0.0001.

DLK promotes short neurite formation in primary hippocampal neurons.

(A) Confocal images of DIV2 primary hippocampal neurons from genotypes indicated and immunostained with DLK and STMN4. tdTomato labeling of neurons is from Cre-dependent Rosa26-tdTomato. Red arrows point to some of the thin neurites from neurons with increased DLK expression. Scale bar 10μm. (B) Quantification of STMN4 immunostaining integrated density (Area X MFI) in neuronal soma, relative to DLK integrated density. N≥3 cultures/genotype, ≥60 cells/genotype. Spearman correlation r=0.7454. (C) Quantification of neurons with no, one, or more than one axon in each genotype. Number of neurons: 47 from 3 Vglut1Cre (control) cultures, 49 from 3 DLK(cKO) cultures, 42 from 4 DLK(iOE) cultures. Statistics: Fisher’s exact test shows significance (P<0.0001) between the three genotypes and number of axons. Axon formation in control vs DLK(cKO): P=0.1857. Formation of multiple axons in control vs DLK(cKO): P>0.9999. Axon formation in control vs DLK(iOE): P=0.0042. Formation of multiple axons in control vs DLK(iOE): P=0.0001. (D) Quantification of number of primary neurites, which include both branches and filopodia, per neuron. Number of neurons: 55 from 4 Vglut1-cre (control) cultures, 70 from 4 DLK(cKO) cultures, 45 from 5 DLK(iOE) cultures. Statistics, Kruskal-Wallis test with Dunn’s multiple comparison test. **** P<0.0001. (E) Confocal z-stack images show that neurons with high expression of DLK have filopodia structures (arrows) around the soma and on axons with tyrosinated tubulin, scale bar 10μm. (F) Confocal z-stack images show that neurons with high expression of DLK have filopodia structures (arrows) around the soma and on axons without acetylated tubulin, scale bar 10μm.

Increasing DLK expression alters synapse formation in primary hippocampal neurons.

(A) Confocal images of DIV14 neurons of indicated genotype, co-stained with Bassoon and DLK. (B) Quantification of bassoon puncta density. (C) Quantification of average bassoon puncta size from individual ROIs. Number of neurons: 30 from 3 Vglut1-cre (control) cultures, 41 from 3 DLK(cKO) cultures, 46 from 4 DLK(iOE) cultures. Statistics: One way ANOVA with Dunnett’s multiple comparison test. ns, not significant; * P<0.05. (D) Confocal z-stack images of DIV14 neurons from Vglut1Cre/+, Vglut1Cre/+;DLK(cKO)fl/fl, and Vglut1Cre/+;H11-DLKiOE/+, labeled by Rosa26-tdTomato. Lower panels show zoomed view of dendritic spines. Scale bar 10μm top, 5μm bottom. (E) Measurement of dendritic spine density. (F) Mushroom spine density. (G) Distribution of spine types. (E-G) Number of neurons: 35 from 3 Vglut1-cre (control) cultures, 31 from 3 DLK(cKO) cultures, 31 from 3 DLK(iOE) cultures. Statistics: One way ANOVA with Dunnett’s multiple comparison test. * P<0.05; ** P<0.01.

Additional evidence for expression levels of DLK and effects on hippocampal morphology at 1 year of age

(A) Confocal images of Dlk and Vglut1 RNA expression in hippocampal glutamatergic neurons at P15. Scale bar 1000μm, 10μm zoomed ROI. (B) Long exposure western blot of DLK and β-actin from hippocampal tissue, arrow indicates lower band for DLK associated with conditional knockout (age P60, each lane representing protein sample from individual mice, N=3 mice/genotype). (C) IGV visual of representative Vglut1Cre/+ and Vglut1Cre/+;DLK(cKO)fl/fl reads for Dlk from P15 RiboTag data reveal reduced expression of floxed exon (∼1/3 level of control). Other exons remained at a similar level as control. Dlk exon structure shown along the bottom in dark blue, with start ATG noted. Red region indicates exon deleted in conditional knockout. Height of reads (y-axis) in gray or blue represents number of reads overlapping a given sequence. (D,E) Confocal z-stack images of coronal sections from ∼1 year old mice from CA1, CA3, and DG dorsal hippocampus immunostained for NeuN and DAPI in (D) control and Vglut1Cre/+;DLK(cKO)fl/flmice (1 year), (E) control and Vglut1Cre/+;H11-DLKiOE/+mice (44-46 weeks old). Outlines highlight cell somas of pyramidal layer or granule cell layer. Scale bar 100μm. (F,G,H) Quantification of cross-sectional area from CA1, CA3, or DG dorsal hippocampus, respectively. Data points represent individual mice, averages taken across 3 sections per mouse, N≥3 mice/genotype. Statistics: One way ANOVA with Dunnett’s multiple comparison test. ns, not significant; **** P<0.0001.

Evidence for induced DLK expression visualized at RNA and protein levels.

(A) Confocal z-stack images of Vglut1 and Dlk mRNA at P15 in control and Vglut1Cre/;H11-DLKiOE/+ mice. Scale bar 1000μm. Inset shows single slice CA1 neurons, with dashed line showing individual nuclei as used for quantification, scale bar 10μm. (B) Quantification of puncta shown as ratio of Dlk puncta to Vglut1 puncta overlapping with individual CA1 neuron nuclei. Statistics: Mann-Whitney U test. **** P<0.0001. (C) Confocal z-stack images of control and Vglut1Cre/+;H11-DLKiOE/+mice immunostained for DLK protein (P60). Scale bar 1000μm. Blue box, magnified view of pyramidal cell layer/stratum pyramidale (SP), scale bar 10μm. Yellow box, magnified view of stratum lacunosum-moleculare (SLM), scale bar 10μm.

Region dependent vulnerability observed with increased DLK expression.

(A) Confocal z-stack images of ventral hippocampus in control and Vglut1Cre/+;H11-DLKiOE/+ mice immunostained for NeuN protein (P60). Dashed line regions are enlarged to the right. Top zoomed view shows dorsal (posterior) CA1 neurons with some thinning of pyramidal layer in mice with increased DLK. Bottom zoomed view shows ventral CA1 neurons appearing similar between genotypes. Scale bar: 1,000μm in hippocampi; 100μm in CA1 layer. (B) Quantification of pyramidal layer thickness across CA1. CA1 dorsal (anterior) data from Figure 1B. Data points represent individual mice, averages taken across 2-3 sections per mouse. Statistics: Unpaired t-test. ns, not significant; **** P<0.0001. (C) Confocal z-stack images of P60 dorsal and ventral hippocampus immunostained for p-c-Jun. Dashed line regions are enlarged to the right. Scale bar: 1,000μm in hippocampi; 100μm in CA1 layer.

Additional evidence for DLK induced hippocampal neuron death.

(A) Confocal z-stack imaging of GFAP immunostaining at P15 and P120. Scale bar 500μm. Boxed areas represent areas for quantification. (B-D) Mean fluorescent intensity of GFAP at P15 in CA1, CA3, and DG regions, respectively. N= 5, 4 mice, quantified from 3 sections per mouse. Statistics: Unpaired t-test. ns, not significant; * P<0.05. (E) Single slice confocal image of TUNEL positive P15 CA1 pyramidal neuron in Vglut1Cre/+;H11-DLKiOE/+ overlapping with DLK transgene-dependent tdTomato expression and pyknotic nuclei. Scale bar 25μm. (F) Quantification of TUNEL positive neurons in CA1 pyramidal layer. Data points represent individual mice, averages taken across 3 sections per mouse, N=4 mice. Statistics: Unpaired t-test. ** P<0.01.

Evidence for RiboTag samples and additional analysis of genes showing differential dependence on DLK expression levels.

(A) Western blot validation of immunoprecipitation of HA-tagged Rpl22. (B) qRT-PCR expression of Vglut1 (glutamatergic neuron marker), Wfs1 (CA1 marker), Vgat (inhibitory neurons), Gfap (astrocytes) shows expression of transcripts expressed in glutamatergic neurons and depletion of non-glutamatergic neuron transcripts. N=3 biological replicates, data shown as fold change of expression of marker genes in immunoprecipitated glutamatergic neuron RNA relative to whole hippocampal RNA. Normalized to Gapdh expression. (C) Venn diagram showing overlap of genes with statistically significant differential translation in DLK(cKO) and DLK(OE). (D,E) Heatmaps of significant genes from DLK(cKO) and DLK(OE). Columns represent individual mice expression levels; rows represent individual genes with the right-hand labels showing which dataset the gene was found to be statistically significant in. Data were normalized by row, with color keys shown above the heatmap. (F,G,H,I) Pie charts show expression of differentially expressed genes based on adult, endogenous expression patterns in the Allen Mouse Brain Atlas in situ data. (F,H) Upregulated, or (G,I) downregulated genes when DLK expression is (F,G) increased or (H,I) conditionally knocked out are categorized based on expression patterns in CA1, CA3, and DG in dorsal hippocampus. (J) Sunburst plot shows significant enrichment for differentially expressed genes from DLK(cKO) relating to the synapse. (K) Pathway analysis of expression data for mice with increased DLK compared to control. Nodes represent sets of genes involved in pathways, with size dependent on the number of genes in the pathway. Nodes are clustered in shaded circles based on related pathways. All pathways shown have q-value (false discovery rate <0.05). Red represents pathways enriched in mice with increased DLK. Blue indicates pathways downregulated in mice with increased DLK.

Levels of c-Jun and p-c-Jun show dependency on expression levels of DLK.

(A) Confocal z-stack images of CA1, CA3, and DG in control and Vglut1Cre/+;DLK(cKO)fl/fl mice immunostained for c-Jun (P60). Scale bar 50μm. Corresponding DAPI staining is shown alongside for reference, dashed lines outline region used for quantification. Below, quantification of mean fluorescence intensity (MFI) of c-Jun in pyramidal or granule cell layer of each region. Data points represent MFI of individual mice normalized to average from control animals. 3 sections per mouse, N=4,4 mice. Statistics: Unpaired t-test. ns, not significant. (B) Confocal z-stack images of control and Vglut1Cre/+;DLK(cKO)fl/fl mice immunostained for p-c-Jun (P60). Scale bar 50μm. Below, quantification of MFI of p-c-Jun in pyramidal or granule cell layer of each region. Data points represent MFI of individual mice normalized to average from control animals. 3 sections per mouse, N=6, 7 mice. Statistics: Unpaired t-test. ns, not significant; * P<0.05. (C) Confocal z-stack images of CA1, CA3, and DG in control and Vglut1Cre/+;H11-DLKiOE/+ mice immunostained for c-Jun (P15). Scale bar 50μm. Data points represent MFI of individual mice normalized to average from control animals. 3 sections per mouse, N=7,8 mice. Statistics: Unpaired t-test. ** P<0.01; **** P<0.0001. (D) Confocal z-stack images of CA1, CA3, and DG in control and Vglut1Cre/+;H11-DLKiOE/+ mice immunostained for p-c-Jun. Scale bar 50μm. Data points represent average number of p-c-Jun positive nuclei in ROIs from individual mice normalized to average from control animals. 3-5 sections per mouse, N=4,5 mice. Statistics: Unpaired t-test. ** P<0.01; *** P<0.001.

Stathmin transcript abundance and western blot analysis of DLK, STMN4 in mice aged P10 to 1 yr.

(A) RiboTag transcript abundance of Stathmin family members in transcripts per million (TPM). Differential expression analysis significance shown (padj). ns, not significant; **** P<0.0001. (B,C) Western blot of hippocampal tissue from P1, P8, P15, P60, and ∼1yr old mice blotted for DLK, Flag, STMN4, STMN2, and actin in Vglut1Cre/+;DLK(cKO)fl/fl and Vglut1Cre/+;H11-DLKiOE/+, respectively. (D,E) Relative DLK protein level normalized to actin and P1 control. N=3 mice/genotype. (F,G) Relative STMN4 protein level normalized to actin and P1 control. N=3 mice/genotype. Statistics: Two-way ANOVA with Sidak multiple comparisons test. ns, not significant; * P<0.05.

Additional evidence for Stmn4 and microtubule expression in DLK(cKO) and DLK(iOE) in hippocampus.

(A,B,D,E) Confocal single slice images of Stmn4 and Vglut1 RNAscope stained sections from (A,B) CA3 and (C,D) DG in (A,C) overexpression mice and (B,D) conditional knockout mice. Scale bar 10μm. (E) Confocal z-stack images of Tuj1 immunostaining of neuron specific beta tubulin in CA1, CA3, and DG at P15. Scale bar 50μm. (F) Confocal z-stack images of MAP2 immunostaining of CA1 pyramidal layer at P60 from control and Vglut1Cre/+;H11-DLKiOE/+ mice. Arrows point to apical dendrites with elevated signal. Arrowheads point to thin neurites with elevated signal. Scale bar 10μm. (G) Normalized MAP2 MFI after thresholding signal in SR. N=9, 9 mice, 3 sections averaged per mouse. Statistics: Unpaired t-test. ns, not significant.

Increasing expression of DLK alters VGLUT1 pattern in dorsal CA1.

(A) Confocal single slice image of VGLUT1 immunostaining of control and Vglut1Cre/+;DLK(cKO)fl/flmice in CA1 SR at P60. Scale bar 5μm, inset scale bar 1μm. (B) VGLUT1 puncta density and (C) size. Data points represent individual ROIs quantified from 5-6 images per mouse across 3 sections. N=5 control, and 8 Vglut1Cre/+;DLK(cKO)fl/flmice. (D) Confocal single slice image of VGLUT1 immunostaining of control and Vglut1Cre/+;H11-DLKiOE/+ mice in CA1 SR at P15. Scale bar 5μm, inset scale bar 1μm. (E) VGLUT1 puncta density and (F) size. Data points represent individual ROIs quantified from 5-6 images per mouse across 3 sections, N=9 control, and 8 Vglut1Cre/+;H11-DLKiOE/+mice. (G-I) Quantification from control and Vglut1Cre/+;DLK(cKO)fl/flmice for (G) Bassoon puncta size, (H) Homer1 puncta size, and (I) size of Bassoon-Homer1 overlap. Data points represent individual ROIs quantified from 5-6 images per mouse across 3 sections. N=7 control, and 8 Vglut1Cre/+;DLK(cKO)fl/flmice. (J-L) Quantification from control and Vglut1Cre/+;H11-DLKiOE/+ mice for (J) Bassoon puncta size, (K) Homer1 puncta size, and (L) size of Bassoon-Homer1 overlap. Data points represent individual ROIs quantified from 5-6 images per mouse across 3 sections, N=9 control, and 6 Vglut1Cre/+;H11-DLKiOE/+ mice. Statistics: unpaired t-test or Mann-Whitney U test if not passing normality. ns, not significant; * P<0.05; ** P<0.01; **** P<0.0001.

Analysis of Stathmins in primary cultured hippocampal neurons from DLK(cKO) and DLK(iOE).

(A) Confocal z-stack image of Rosa26-tdTomato labeling of neuron morphology at DIV2. Red arrowheads point to long processes considered as axons. Scale bar 100μm. (B) Co-immunostaining of DLK and STMN4 in control growth cone show non-overlapping puncta. (C) Confocal z-stack images of DIV2 primary hippocampal neurons from genotypes indicated immunostained with DLK and STMN2. tdTomato labeling of neurons is from Rosa26-tdTomato. Scale bar 10μm. (D) Quantification of association between DLK level and STMN2 in cell soma. N=3 cultures/genotype, ≥45 cells/genotype. Spearman correlation r=0.4693.

Additional evidence for STMN2 and STMN4 antibodies

(A) Confocal single slice images of STMN2 and STMN4 antibody comparison in DIV3 control primary hippocampal neurons. Dashed box outlines enlarged region. Scale bar 10μm full cell, 1μm enlarged region. (B) Western blot of P15 hippocampal tissue for STMN2 and STMN4 antibodies. Lanes 1-3 control for Vglut1Cre/+;H11-DLKiOE/+, and lanes 7-9 control for Vglut1Cre/+;DLK(cKO)fl/fl genotype referred to as “+”. Lanes 4-6 Vglut1Cre/+;H11-DLKiOE/+ referred to as “Tg”. Lanes 10-12 Vglut1Cre/+;DLK(cKO)fl/flreferred to as “-”. (C) STMN2 or STMN4 protein level normalized to actin. N=3 mice/genotype. Statistics: One way ANOVA with Sidak’s multiple comparison test. ns, not significant; * P<0.05.

Primers