Expression profile of Wdr45 protein in mouse tissues
When the expression profile was analyzed in a variety of mouse tissues by western blotting, Wdr45 with a molecular mass of ~40 kDa was detected in all tissues except small intestine (Fig.1A). While this ~40 kDa band is a major molecule in the brain tissues, liver, spleen, pancreas, ovary and uterus, it was weakly detected in heart, adrenal gland and skeletal muscle. An ~50 kDa band was detected in brain, heart and skeletal muscle, and an ~85 kDa protein was relatively well expressed in liver and kidney. Meanwhile, an ~110 kDa band was strongly and moderately expressed in pancreas and liver, respectively. In addition, tissue-specific bands with ~60 kDa, ~70 kDa and ~ 28 kDa were observed in the brain tissues, pancreas and skeletal muscle, respectively.
To gain some insight into the involvement of Wdr45 in neuronal development, we looked into its expression level in the whole brain extracts prepared from various developmental stages (Fig. 1B). Wdr45 with ~40 kDa was strongly detected during embryonic stages and then gradually decreased after postnatal day (P)8 to the lowest level at P30 (Fig. 1B). The ~110 kDa band also showed an expression pattern similar to that of the ~40 kDa protein, whereas weak expression of the ~50 kDa and ~60 kDa bands appeared to be weakly visualized after birth.
Although the expression level was not high in adult brain, Wdr45 is likely to be enriched at specific intracellular compartment such as synapses from the fact that functional defects in WDR45 is responsible for neurodevelopmental defects in BPAN. Biochemical fractionation of adult mouse brain was thus conducted, and the quality of each fraction was confirmed by detection of synaptophysin and PSD95, markers for pre- and post-synapses, respectively (Fig. 1C). Consequently, the ~40 kDa and ~80 kDa proteins, but not the ~50 kDa one, were detected in highly purified post synaptic density (PSD) (PSD-II) fraction where synaptophysin was little detected (Fig. 1C). The ~110 kDa and ~50 kDa proteins were only faintly observed at the PSD fraction (Fig. 1C). These results suggest that certain Wdr45 isoforms are distributed at PSD. Given that the synaptosome fraction contains both pre- and post-synaptic components and PSD-I fraction contains contaminated presynaptic proteins, Wdr45 also might be present in the pre-synaptic compartment. Taken together, we assume that Wdr45 is involved in synapse structure and/or function in differentiated neurons.
Immunohistochemical analysis of WDR45 in mouse brain
To determine the localization of Wdr45, we performed immunohistochemical analyses of cerebral cortex at embryonic day (E) 15, E17, P0, P7 and P30. At E15 and E17, Wdr45 was broadly detected in neurons in the marginal zone (MZ), cortical plate (CP), subplate (SP) and intermediate zone (IMZ) as well as neuronal progenitors in the ventricular zone (VZ) and subventricular zone (SVZ) (Fig. 2A, a and b, B). Although the broad distribution pattern was observed until P30, subcellular distribution altered after P7; relative nuclear enrichment of Wdr45 was frequently observed from E15 to P7 while cytoplasmic distribution came to be obvious at P30 (Fig. 2A,e, C). Meanwhile, it is noteworthy that Wdr45 was detected at the ventricular surface of neural progenitors from E15 to P0 (Fig. 2B). In addition, distinct immunoreactivity was observed in the neuropil of cerebral cortex at P30 (Fig. 2D), with a strong suggestion of enrichment at excitatory synapses. Collectively, Wdr45 was expressed in the nucleus of cortical neurons and the ventricular surface during corticogenesis and its subcellular localization changed from the nucleus to the cytoplasm and then neuropils in developed cerebral cortex.
We then carried out immunohistochemical analyses in hippocampus at E17, P0, P7 and P30 and in cerebellum at P30 (Supplementary Fig. 1). As in the case of cerebral cortex, Wdr45 appeared to be expressed in the nucleus and cytoplasm at E17 and P0 in the CA regions and dentate gyrus of hippocampus (Supplementary Fig. 1A and B). Then, the cytoplasmic distribution in the neurons of CA regions as well as dentate gyrus came to be detected at P7 and became clearer at P30 (Supplementary Fig. 1C and D). In cerebellum, almost all regions were positive for Wdr45, with relatively strong expression in Purkinje cells (Supplementary Fig. 1E). These results indicate possible roles of Wdr45 in hippocampus and cerebellum.
Characterization of RNAi- and expression-vectors
Given that WDR45 gene abnormalities are responsible for the early onset of ID in BPAN, Wdr45 is most likely to participate in the formation of cortical architecture. BPAN-causative mutants of WDR45 are supposed to be structurally unstable and undergo degradation 4. Indeed, a BPAN-causative mutant, c.700C > T/p.(Arg234*) (Wdr45-Rstop), as well as another mutant, c.439G>T (GVmut), with single amino acid substitution in the linker region between third and fourth b-propeller domains were hardly detected when expressed in primary cultured mouse cortical neurons (Fig. 3A). We thus conducted acute knockdown experiments to recapitulate the pathophysiological conditions in BPAN. To this end, 2 RNAi vectors were designed against mouse Wdr45, pSuper-Wdr45#1 and #2, both of which effectively knocked down Myc-Wdr45 expressed in COS7 cells (Fig. 3B). In addition, pSuper-Wdr45#1 and #2 silenced endogenous Wdr45 at least partially in cortical neurons in vitro (Fig. 3C). An RNAi-resistant version of Wdr45, Wdr45R, was shown to be resistant against pSuper-Wdr45#1 (Fig. 3D).
Role of Wdr45 in cortical neuron migration
We asked whether Wdr45 is involved in the migration of newly generated cortical neurons by in utero electroporation-based acute knockdown of Wdr45. pSuper-H1.shLuc (control), pSuper-Wdr45#1 or #2 was co-electroporated with pCAG-Turbo RFP into VZ progenitor cells, and electroporated brains were fixed and analyzed at P2. When compared to the control experiments, Wdr45-silencing had little effects on the migration and the cells were considered to normally make it to the layer II-IV (Supplementary Fig. 2).
Role of Wdr45 in axon extension and dendritic arbor development during brain development
Since aberrant synaptic network formation is tightly associated with defective brain development and function, Wdr45 may take part in axon pathfinding and/or dendrite growth during brain development. We thus first examined the interhemispheric axon projection of Wdr45-deficient cortical neurons in vivo. When Wdr45 was silenced in the VZ cells at E14.5, axon bundles from the hemisphere containing Wdr45-deficient cells normally reached the contralateral hemisphere at P2 (Fig. 4A) and this tendency was still observed at P30 (Fig. 4B). On the other hand, it is notable that axons of the deficient cells did not penetrate efficiently into the layer structure of contralateral cortex at P60 when compared to the control axons (Fig. 4C). Taken together, Wdr45 appeared to be important for callosal axon extension into the cortical structure after normal pathfinding from the ipsilateral to contralateral cortex.
We then examined the knockdown effects on dendritic arbor development of cortical neurons in vivo. Electroporation of pSuper-Wdr45#1 or #2 into the VZ cells at E14.5 gave rise to highly suppressed dendritic arborization at P10 (Fig. 5A). Sholl analyses revealed a significant decrease in the branch point number in the deficient neurons at P10 (Fig. 5B). Rescue experiments were then performed to exclude off-target effects. When an RNAi-resistant version of Wdr45, Wdr45R, was co-electroporated into the VZ progenitor cells with pSuper-Wdr45#1, morphological defects were rescued, indicating specific effects by Wdr45-silencing (Fig. 5A and B). We then measured the knockdown effects on the apical and basal dendrites, separately. As shown in Fig. 5C, apical oriented dendritic length was abnormally shortened in pSuper-Wdr45#1- or #2-transfected cells at P10, and the phenotype by pSuper-Wdr45#1 was rescued at least partially (Fig. 5C). Likewise, Wdr45-knockdown by pSuper-Wdr45#1 or #2 suppressed the basal dendrite development (Fig. 5D). In this analysis, while the phenotype tended to be rescued by Wdr45R, the effects were not statistically significant (Fig. 5D). Collectively, we concluded that Wdr45-deficiency impairs neuronal connectivity through defective axon elongation as well as aberrant dendrite arborization.
Role of Wdr45 in spine morphology regulation in vivo
As a next set of experiments, we examined the physiological role of Wdr45 in the spine density and morphogenesis in vivo. To this end, pSuper-H1.shLuc (control), pSuper-Wdr45#1 or #2 was co-electroporated at E14.5 with pCAG-loxP-GFP plus pCAG-M-Cre for sparse GFP-labeling. Brains were then fixed and analyzed at P10. Consequently, the total spine density was found to be significantly decreased when Wdr45 was silenced (Fig. 6A and B, left). The abnormal phenotype by pSuper-Wdr45#1 was rescued by co-expression of Wdr45R (Fig. 6A and B, left). In addition, silencing of Wdr45 caused reduction of mature mushroom-type spine number (Fig. 6B, middle and right). The phenotype by pSuper-Wdr45#1 was rescued by Wdr45R (Fig. 6B, middle and right). These results strongly suggest that Wdr45 plays a pivotal role in the synapse formation and/or maintenance.
Long term effects of Wdr45-knockdown on neuronal morphology in vivo
Although individuals with BPAN develop sudden-onset progressive dystonia, parkinsonism and dementia in adolescence, pathophysiological mechanism(s) underlying these clinical manifestations is yet to be clarified. To gain some insight into this aspect, we examined long-term effects of Wdr45-knockdown on the neuronal morphology by focusing on dendritic arbor development at P30, P60 and P90. To this end, we measured the intersection number of dendritic arbor as well as the total length of apical and basal dendrites. As shown in Fig. 7A - E, dendritic arbor complexity of Wdr45-deficient neurons remained poor when compared to the control cells at the abovementioned 3 time points. Likewise, both apical and basal dendritic length of the deficient neurons were shorter than those of the control cells at the time points (Fig. 7F and G). These results strongly suggest that Wdr45-deficiency prevents neuronal network formation even in the adult stage and causes sustained impairment of the synaptic function.
We then looked into the long-term effects of Wdr45-knockdown on spine density and morphology of layer II-III cortical neurons in vivo. While the reduced total spine density was still observed at P30, it was comparable to the control neurons at P60 (Fig. 8A and B, left). In contrast, the number of mature mushroom-type spine remained to be decreased at P60 (Fig. 8A and B, middle).
WDR45-deficiency has been reported to downregulate autophagy in adult mice 9,10 and LCLs from BPAN patients 4. We asked whether silencing of Wdr45 affects the autophagy pathway during cortical development. Since LC3 (microtubule-associated light chain 3)-II is a standard marker for autophagosomes and relatively specifically associated with nascent autophagosomes, quantification of LC3-positive puncta is crucial for assessing the numbers of autophagosomes in cells. Thus, the effect of Wdr45-silencing on puncta formation of LC3 was examined in dendrites of primary cultured cortical neurons, since the main site of autophagosome formation in neurons is dendrites 11. Consequently, the staining pattern of LC3 in the deficient neurons was comparable to the control cells (Supplementary Fig. 3).