WD40-Repeat 47 Is Essential for Brain Connectivity Via Key Microtubule-Mediated Processes

WD40-repeat 47 is essential for brain connectivity via key microtubule-mediated processes

Meghna Kannan1,2,*, Christel Wagner1,*, Marna Roos3,*, Bruno Rinaldi4,*, Perrine Kretz1,

Lara McGillewie5, Séverine Bär4, Shilpi Minocha2, Claire Chevalier1, Alexandru Parlog1,

Chrystelle Po6, Jamel Chelly1, Jean-Louis Mandel1, Romina Romaniello7, Renato

Borgatti7, Amélie Piton1, Stephan C. Collins1,8, Craig Kinnear5, Yann Hérault1, Sylvie

Friant4, Ben Loos3 and Binnaz Yalcin1,2

1 Institute of Genetics and Molecular and Cellular Biology; National Centre for Scientific Research, UMR7104; National Institute of Health and Medical Research, U964; University of Strasbourg, France

2 Center for Integrative Genomics, Genopode, University of Lausanne, Switzerland

3 Department of Physiological Sciences, Stellenbosch University, South Africa

4 Université de Strasbourg, CNRS, GMGM UMR7156, F-67000 Strasbourg, France

5 SAMRC Centre for Tuberculosis Research, Department of Biomedical Sciences, Stellenbosch University, South Africa

6 ICube, UMR 7357, FMTS, University of Strasbourg

7 Neuropsychiatry and Neurorehabilitation Unit, Scientific Institute, IRCCS Eugenio Medea, Bosisio Parini, Lecco, Italy

8 Université de Bourgogne-Franche Comté, Esplanade Erasme, 21078 Dijon France

* Joint first authors

Correspondence should be addressed to: Dr. Binnaz Yalcin ([email protected])

ABSTRACT

The family of WD40-repeat (WDR) proteins is one of the largest in eukaryotes but little is known about their function in brain development. Among 26 WDR genes assessed, we found seven displaying a major impact in neuronal morphology when inactivated in mice. Remarkably, all seven genes showed corpus callosum defects, from thicker

(Atg16l1, Coro1c, Dmxl2 and Herc1), thinner (Kif21b and Wdr89) to absent (Wdr47), revealing a common implication of WDR genes in brain connectivity. We focused on the novel WDR47 protein sharing structural homology with LIS1, which causes lissencephaly. Mice lacking Wdr47 showed lethality, extensive fibre defects, microcephaly and sensory-motor gating abnormalities reminiscent of a patient harbouring mutation in WDR47. We demonstrated that WDR47 indeed shares functional characteristics with LIS1, as it participates in key microtubule-mediated processes. Interestingly, WDR47 specific CTLH domain was associated with functions in autophagy for the first time in mammals. Silencing WDR47 in hypothalamic GT1-7 neuronal cells and yeast models independently recapitulated these findings, demonstrating highly conserved mechanisms. Finally, our data identified two important

WDR47-interacting partners: SCG10 and Reelin. Taken together, these results make

WDR proteins an important family for studying neuronal regulation of microtubules and autophagy, bringing a new perspective to the biology of brain connectivity.

SIGNIFICANCE STATEMENT

We present the first identification of the relevance of WDR genes in brain connectivity, highlighting the power of unbiased mouse studies in the field of neuroscience. We focus

2 on the novel WDR47 protein sharing structural homology with LIS1, which causes lissencephaly. WDR47 plays a role in the regulation of microtubules dynamics, neuronal navigation, and fibre tract projections, in a similar fashion to LIS1 but with the distinctive particularity that WDR47 inhibits autophagic flux. This provides a functional link between autophagy biology and the CTLH domain for the first time in mammals.

Importantly, WDR47 uncovers a new aspect of corpus callosum biology pointing towards a link between the regulation of microtubule dynamics and autophagic flux for axonal outgrowth and guidance.

INTRODUCTION

The function of WD40-repeat (WDR) containing proteins, one of the largest eukaryotic protein families, is largely unknown. Their importance is however evident based on their highly-conserved repeating units from bacteria to mammals (1), commonly made of seven repetitive blades of 40 amino acids that end in tryptophan-aspartic acid (WD) dipeptide at the C terminus.

As demonstrated by crystallography studies including the crystal structure of the beta subunit of the G protein (2), a classical WDR protein, all WDR proteins are predicted to fold into a circularized beta-propeller structure, serving as a rigid platform

(or scaffold) for protein-protein interactions by providing many stable and symmetrical surfaces (3, 4). One reason why WD40-repeat domains may have been less studied than other common domains such as kinases, PDZ or SH3 domains (3), is that no WDR domain has yet been found with a catalytic activity of an enzyme (3), but this does not

3 mean scaffold domains are less important, to the contrary, their serving as a platform for multiple enzymatic reactions and signaling events is highly significant.

In recent years, human genetic studies have also begun to recognize the importance of WDR genes. Amongst 286 WDR genes annotated across both human and mouse genomes, mutations of 26 WDR genes (that is 9%) have so far been implicated in brain disorders, notably in intellectual disability associated with malformations pertaining to anomalies of the corpus callosum (Table S1). Among these, PAFAH1B1

(also known as LIS1, a WDR protein identified 20 years ago that regulates dynein activity and neuronal migration (5)) is linked with lissencephaly type 1, a severe malformation where the brain develops without convolutions (OMIM 607432) and the corpus callosum is thinner (6). Mutations in WDR62 cause autosomal-recessive primary microcephaly and hypoplasia of the corpus callosum (7) and WDR73 is implicated in

Galloway-Mowat syndrome characterised by microcephaly and thin corpus callosum

(8). Another example includes WDR81 linked to quadrupedal locomotion and significant reductions in volume of cerebellum and corpus callosum (9). Understanding the underlying pathophysiological mechanisms of callosal disorders is critical in establishing patient stratification and exploring novel therapeutic avenues.

Made of approximately 190 million axonal projections, the human corpus callosum is the largest inter hemispheric white matter tract in the brain (10), with neurons located mainly in neocortical layers II/III giving rise to callosal axons (11). The genetics of corpus callosum biology is however highly heterogeneous and, despite technological advances in next generation sequencing, 75% of callosal disorders have no identified genetic cause (12). Recent studies have suggested that a smaller corpus callosum is associated with a higher risk of autisms (13), bipolar disorder (14) and

4 schizophrenia (15). Corpus callosum abnormalities are often seen in conjunction with other defects such as smaller or larger brain sizes (16), and malformations of cortical development (MCD) (17). The formation of the corpus callosum is a process relying on axonal guidance cues and their receptors, such as Netrin/DCC, ROBO and Slit (18). This developmental process also relies on microtubule polymers that localize to the tip of the axon, known as the growth cone (19). However, much less is known about microtubules at the growth cone, yet they are the primary effectors of axonal movement and guidance

(20).

Less than 3% of WDR proteins have been functionally defined in the central nervous system, while for the remaining 97%, the function remains completely unknown (Table S1). Interestingly, several WDR proteins have been linked to microtubules in knockout mouse studies (for example LIS1 (21) and WDR62 (22)).

Composed of alpha and beta tubulin heterodimers, microtubules are critical components of the cytoskeleton, and their dynamics refers to the continuous remodelling between assembly (rescue) and disassembly (catastrophe) at their tip (23).

Proper regulation of this dynamic is essential and achieved through microtubule- associated proteins (MAPs). MAPs are of particular importance as they mediate the dynamic processes of microtubules by stabilizing and destabilizing tubulin.

In this study, we asked whether tubulin processes might be affected by mutations of WDR genes that result in corpus callosum anomalies, and what the underlying molecular mechanisms are, and ultimately how these underlie corpus callosum biology. To address these questions, we searched for brain defects in 26 WDR knockout mice including 17 parameters related to corpus callosum anatomy, and identified a third of WDR gene KO linked with major anomalies. We chose to focus on

5 the novel WDR47 gene, which shares structural similarities with LIS1 and results in agenesis of the corpus callosum. The agenesis of the corpus callosum (AgCC) is defined as a failure to develop the large bundle of fibres that connect the cerebral hemispheres

(12). WDR47 uncovers a new aspect of corpus callosum biology pointing towards a link between the regulation of microtubule dynamics and autophagic flux for axonal outgrowth and guidance.

RESULTS

Unravelling the importance of mouse WDR proteins in corpus callosum biology

26 WD40-repeat containing proteins were randomly selected amongst a manually curated list of 286 family members (Table S1), for neuroanatomical studies in 16-week old male knockout (KO) mice obtained through collaboration with the Wellcome Trust

Sanger Institute Mouse Genetics Program (24). We first carried out gene ontology term enrichment analysis in both the 286 and 26-gene sets and consistently found the same three most significant terms (P<0.001): protein complex binding, actin filament binding and histone binding (Fig. S1A).

Using a quantification of 66 morphological and 115 cellular measurements across 19 different brain regions in two coronal histological sections at Bregma 0.98 and -1.34mm (Table S2), in conjunction with 14 quality control parameters (Table S3), we found that mutations of seven WDR genes (Atg16l1+/-, Coro1c+/-, Dmxl2+/-, Herc1-/-,

Kif21b-/-, Wdr47-/- and Wdr89-/-) were associated with major neuroanatomical phenotypes (Fig. 1A, also see Tables S4,5 for a complete list of p-values and effect sizes). Neuroanatomical defects are described using MP (Mammalian Phenotype) terms

(http://www.informatics.jax.org/searches/MP_form.shtml). At the morphological level,

Dmxl2+/- and Herc1-/- KO mice displayed macrocephaly (MP:0000434), with an increase size of 37% (P=0.0005) and 20% (P=0.002), respectively, compared to wild-type (wt).

By contrast, Wdr47-/- and Kif21b-/- KO mice revealed microcephaly (MP:0000433), with a decrease size of 25% (P=0.03) and 20% (P=0.05), respectively, relative to wt. Loss of

Atg16l1 was associated with increased height (+11%) of the motor cortex

(MP:0012449, P=0.003), and increased height (+38%) of the radiatum layer of the hippocampus (MP:0011154, P=0.004), whereas loss of Wdr89 and Coro1c were associated respectively with ventricular atrophy (-13%) of the dorsal third ventricle

(MP:0008536, P=0.0009) and with an enlargement (+13%) of the lateral ventricles

(MP:0008535, P=0.0002). At the cellular level, Herc1-/- KO mice displayed a 37% increase in cell numbers in the mammillothalamic tract (P=0.003) and a 20% increase in the granular cortex (P=0.001), whilst Kif21b-/- KO mice showed a decreased number of cells (-17%) in the cingulate cortex (P=0.03).

Strikingly, all seven WDR gene KOs displayed corpus callosum anomalies (Fig.

1A, Tables S4, 5). Developmental mechanisms regulating dorso-ventral axes of the corpus callosum being distinctive, with on the one hand, pioneering axons projecting from the cingulate cortex crossing the dorsal region, and on the other hand, neurons from the neocortex regulating formation of the ventral region (25), we quantified several regions of the corpus callosum (the genu, soma and splenium). Depletion of

Dmxl2 and Herc1 had a stronger impact on the genu, whilst depletion of Atg16l1, Coro1c,

Kif21b and Wdr89 affected the soma only, indicating spatial and developmental specificity of their impact. In addition, Atg16l1+/-, Coro1c+/-, Wdr37-/- and Wdr89-/- KO mice exhibited cell count anomalies in the corpus callosum, the directionalities of which were in line with those of the morphological phenotypes. Wdr47 KO stood out as the

7 most severely affected, with complete agenesis of the corpus callosum (MP:0002196) at

Bregma -1.34mm (Fig. 1B).

WDR47 shares structural homology with LIS1 (Fig. 1C) and contains a LISH

(lissencephaly type 1-like homology) domain, the second most common neighbour of

WD40-repeat domain, known for its implication in dimerization (3). Unlike LIS1,

WDR47 is composed of a C-terminal to LisH domain (CTLH) whose function remains completely unknown in mammals.

Wdr47 is highly expressed in the brain and essential for survival in mice

To further investigate the function of WDR47, we developed two mouse models (tm1a and tm1b) using the International Mouse Phenotyping Consortium targeting mutation

(tm) strategy (26). This strategy relies on the identification of an exon common to all transcript variants, upstream of which a LacZ cassette was inserted to make a KO

(named tm1a), whereas tm1b creates a frame-shift mutation upon deletion of the selected exon (Fig. S1B).

We validated both tm1a and tm1b mouse models using qRT-PCR and determined that tm1a is a hypomorph allele in a series of tissues (cortex, hippocampus, spinal cord, cerebellum, striatum, thalamus, medulla, olfactory bulb, lung, liver and heart) suggesting that Wdr47 is skipping over the LacZ cassette restoring gene expression, whilst tm1b is a complete KO (Fig. 2A). Based on average relative expression to the wt, tm1a heterozygous (het) mice thereafter referred as Wdr47+/tm1a expressed 70% and tm1b het Wdr47+/tm1b expressed 50%. In homozygous (hom) animals, tm1a

(Wdr47tm1a/tm1a) and tm1b (Wdr47tm1b/tm1b) expressed 30 and 0%, respectively (Table

S6). This offers the opportunity to study the impact of gene dosage (70%, 50% and

30%) and complete gene KO. Using Western blot analysis, WDR47 protein analysis also confirmed minimal expression in the cortex of Wdr47tm1a/tm1a mice (Fig. 2B).

LacZ spatial and temporal expression from embryonic to adult stages throughout the whole brain and in peripheral tissues reveals Wdr47 expression mainly in the cortex, hippocampus, spinal cord, ventromedial hypothalamus and arcuate nucleus, with a cell-specific pattern in layers II/III of the cortex and pyramidal cells of the hippocampus (Fig. 2C, Fig. S1C). Wdr47 was less expressed in lung, liver and heart.

During development, Wdr47 expression gradually increased, from E9.5 through E18.5, reaching a peak at E18.5 (Fig. S1D), suggesting a role in late embryogenesis.

Adult mouse survival was assessed from 1,085 successfully genotyped mice derived from a “het-by-het” breeding scheme (Table S7). 5.7% hom, 54.2% het and

40.1% wt were obtained in Wdr47tm1a mice, and 0% hom, 55% het and 45% wt in

Wdr47tm1b mice, indicating severe lethality in KO mice. Wdr47 expression levels and lethality (expressed as percentage) exhibited a high negative correlation with an r2 coefficient of -0.9 (Fig. 2D) with males and females equally affected. To determine the window of death, we then tested mouse viability from E9.5 to P153 (Fig. 2E). Death rate was unaffected during embryogenesis with C-section derived E9.5 and E18.5 embryos alive at time of inspection. No abnormality in number of somites, limb morphology, heart beat and blood circulation was observed in E9.5 embryos (n=34) and E18.5

Wdr47tm1a/tm1a embryos (n=132) were observed to breath, turn pink, and move actively.

However, the percentage of Wdr47tm1a/tm1a mice decreased exponentially from birth to

55 days of age (P55), with a reduction of 36% by P16 and a further reduction of 64% by

P55. Mice that survived until P55, survived until adulthood. The cause of lethality

9 remains unknown, however histological assessment at E18.5 excluded lung anatomical defects.

It has been recently reported that lipid-enriched diet rescues lethality in a mouse model of frontotemporal dementia and amyotrophic lateral sclerosis (27). We thus maintained a separate colony of mice on a fortified diet (Mouse Breeder Diet 5021) with extra lipids (10.8% as opposed to 3% in a chow normal diet) and folic acid (3mg versus

0.7mg), and examined its effects on a total of 591 mice in both Wdr47tm1a and Wdr47tm1b mice (Fig. 2F). Remarkably, we found an almost complete trans-generational rescue of the lethal phenotype at the second generation. These results indicate that diet enrichment counterbalances the lethality effect, possibly by altering nutrient levels necessary in key processes for survival. There was no rescue in Wdr47tm1b/tm1b mice, suggesting that residual Wdr47 expression is necessary for diet induced survival reversal.

Wdr47 deficiency results in major fiber tract anomalies in mice

Next, we asked whether adult Wdr47 KO mice bred on an enriched diet would show reversal of neuroanatomical defects, since folic acid diet supplementation can promote neuronal proliferation and reduce apoptosis (28). For this purpose, we used the same set of 181 parameters than our previous analysis (Table S2), and assessed 16-week old

Wdr47tm1a/tm1a male mice derived from the second generation of mice bred with fortified diet as well as from the chow diet colony, and compared them to their respective matched wt.

Similar neuroanatomical defects were identified in fortified mice when compared to chow diet mice (Fig. S2A), suggesting no amelioration aside from viability tests. Accordingly, further tests on cortical layers revealed similar decrease in the height of layers II/III (-19.4%, P=6.98x10-5) and VI (-18.5%, P=3.6x10-4) at Bregma 0.98mm

(Fig. 3A) and reduction of all layers at Bregma -1.34mm (Fig. S2B) in both diets.

Focusing on layers II/III where neurons giving rise to callosal axons originate(11), we found a 22.2% reduction in cell number (P=5.7x10-4) in both groups, indicating that the microcephaly phenotype is likely due to a lesser number of cortical cells (Fig. 3B). This was replicated and validated by Cux-2 immunostaining, a postmitotic upper layer specific stain (Fig. S2C). Given that we did not observe an impact of fortified diet on neuroanatomy, subsequent analyses were done using the chow diet colony only.

To discriminate primary microcephaly (defined as reduction in brain size at birth) from progressive microcephaly, we studied brain morphology at E18.5 using a quantification of 67 measurements of size and surface (Tables S8, 9), across three coronal planes at stereotactic position Bregma 2.19, 3.51 and 6.75mm (Fig. S3A-C). In

Wdr47tm1a/tm1a mice, 14 phenotypes emerged as decreased, for example corpus callosum area at 2.19mm (-23.3%, P=5.6x10-3) and height of the motor cortex at 3.51mm (-10%,

P=0.02) (Fig. S4A), whereas the glial wedge and indusium griseum glia were not affected. We also measured cortical layers and found a significant reduction in cortical plate, sub- and ventricular zones (Fig. 3C). Additionally, we measured 63 parameters in mice aged eight days (P8) and found the same regions being affected (Fig. S4B, Table

S9). We also aged mice up to 56-week of age, and found that brain size phenotypes did not worsen (Fig. S4C-D). Taken together these results point towards primary microcephaly.

Sexual dimorphism in brain morphology was assessed through two additional analyses that varied in resolution. We first used micro computed tomography (µCT) that generated 512 coronal images per brain of which three were selected and quantified using a total of 78 brain measurements (Fig. S5, Table S9). We found microcephaly

(MP:0000433) and corpus callosum hypoplasia (MP:0000781) in Wdr47+/tm1a female mice, and smaller motor cortex (MP:0012450) in both Wdr47+/tm1a male and female mice. Second, we used a newly designed sagittal analysis of 95 variables for 22 unique brain regions down to cell-level resolution across three selected sections (Lateral 0.72,

1.32 and 2.52mm) having the advantage of adding new brain regions (such as the substantia nigra) while maintaining existing ones in Wdr47+/tm1a and Wdr47tm1a/tm1a mice (Fig. S6A-D, Tables S8, 9). We found a strong correlation between Wdr47 relative expression and severity of brain structural anomalies such as in the corpus callosum and anterior commissure, both in male and female mice, demonstrating that the function of Wdr47 in brain structure is highly sensitive to dosage (Fig. 3D, Fig. S7A).

Male and female mice showed overall a similar set of neuroanatomical anomalies, for example a reduction in the area of the corpus callosum of -23.6% (P=0.0071) for female and -24.7% (P=0.0079) for male Wdr47tm1a/tm1a mice, reduction in the total brain area, anterior commissure area (MP:0008226), internal capsule (MP:0012462) and fimbria of the hippocampus (MP:0012466) (Fig. 3E-F, Fig. 7B).

WDR proteins (such as WDR45) have recently been categorised as beta- propeller protein-associated neurodegeneration disorders with accumulation of iron in the brain (29). To test whether WDR47 could belong to this class of beta-propeller protein, we aged Wdr47 KO mice up to 56-week of age, but found no such accumulation

(data not shown).

Finally, the axonal (SMI-312R) and L1CAM markers were used to visualize and investigate neurofilaments and callosal neurons, respectively, during development. We found a major reduction in axonal processes both at E14.5 and E16.5 with thalamo- cortical projections unable to cross the diencephalon-telencephalon boundary (Fig. 3G).

Using magnetic resonance imaging (MRI), reduction in the amount of fiber projections across the whole brain in adult Wdr47tm1a/tm1a mice at 16-week of age corroborates these findings (Fig. 3H). The corpus callosum was severely affected ranging from the genu, with complete agenesis at the midline and the splenium, and the anterior commissure was almost completely missing (Fig. S7C).

Mouse Wdr47 depletion impairs growth cone dynamics ex vivo

The spatial distribution of WDR47 at sub-cellular level using conventional fluorescence microscopy in cortical and hippocampal primary neuronal cultures at E17.5 showed that endogenous WDR47 is homogeneously distributed throughout the cytoplasm, neurites, axons and growth cone, but not in the nucleus (Fig. S8A).

The impact of Wdr47 depletion on sub-neuronal architecture was assessed on similar preparation derived from Wdr47tm1a mice and littermate controls. Using double staining immunocytochemistry, we visualized the axon and growth cone using a marker for neurofilaments and microtubule-associated protein 2 (MAP2), a MAP enriched in the dendrites of neurons, and measured the area of the cell body (n=215 neurons), axonal length (n=519) and area of the growth cone (n=775). Primary neuronal cultures derived from Wdr47tm1a/tm1a mice displayed a severe reduction of growth cone areas, of -41% for the cortex (P=2.69x10-5) and -42% for hippocampus (P=1.88x10-6) (Fig. 4A), resembling the physiological collapse or catastrophe state in growth cone behavior(23).

Neuronal cultures derived from Wdr47+/tm1a mice also showed smaller growth cone areas of -17% for the hippocampus (P=1.08x10-5), and -37% for the cortex (P=8.6x10-6), showing that this phenotype is also sensitive to Wdr47 dosage although this is more evident in the hippocampus (Fig. 4A). Furthermore, these structures displayed a blunt tip and reduced filopodia protrusions whilst the cell area and length of axon did not differ (Fig. 4A). In addition, time-lapse recordings from live neurons over 24hrs showed that mutant growth cones are much less dynamic (SI videos 1-2).

The analysis of microtubule distribution network at the growth cone of neurons derived from Wdr47tm1b/tm1b mice using acetylated tubulin, as a marker of stable microtubules, revealed unusual shapes with a remarkable ring-like structure at the soma of the growth cone (Fig. 4B). Super-resolution single molecule localization microscopy also showed these unusual shapes and further established that tubulin molecules were widely dispersed in mutant as opposed to uniform and denser distributed cells in wt (Fig. 4C). Tau protein level, a microtubule-associated protein

(MAP) known to modulate the stability of axonal microtubules, was increased by about two-fold in Wdr47tm1a/tm1a cortical tissue samples when compared to wt (Fig. 4D).

Given the role of Tau in microtubule dynamics, we hypothesized that WDR47 might participate in microtubule stabilization. We tested this by treating hippocampal primary neuronal cultures derived from Wdr47tm1b/tm1b mice, characterised by a reduction of growth cone areas of -75% (n=162, P=1.89x10-5) (Fig. 4E), with a microtubule stabilizer compound (Epothilone D or EpoD) at two concentrations (10nM and 100nM), for 1.5hrs as recommended elsewhere (30). Remarkably, EpoD was able to dose dependently rescue growth cone size, up to +69.7% relative to vehicular-control

(DMSO) cells (Fig. 4E). Treatment with 10nM EpoD increased the size of growth cones

14 by 2.1 times (n=102, P=1.2x10-4) and treatment with 100nM EpoD by 3.3 times (n=143,

P=1.9x10-9). No significant changes were observed between EpoD-treated and non- treated groups in the wt. Thus, we demonstrated that Wdr47 indeed plays a role in stabilizing microtubules facilitating proper growth cone dynamics.

Wdr47 knockdown alters growth cone structure and neuronal navigation in vitro

To determine whether structural anomalies detected at the growth cone might be causing migration defects in a similar manner as observed in LIS1 (31), we used an in vitro scratch assay as a robust and well-developed method to measure cell migration in vitro (32). The basis of this assay relies on creating a scratch (or wound) in cell culture dishes and quantifying the time required for the cells to close it.

Rat hypothalamic GT1-7 cells treated with control or WDR47 specific siRNA were used in this assay instead of primary neuronal cells that are not adequate for post scratch recovery. WDR47 protein levels had residual expression of 37.3±9.3% (P<0.05) compared to control when treated with the WDR47 siRNA (Fig. 5A). Scanning electron microscopy (SEM) was used to evaluate surface morphology of hypothalamic neurons and demonstrated that siRNA treated cells are incapable of forming extended processes and had poorly defined regions of surface adhesion. Again, the growth cone appeared shrunken and characterized by an absence of filopodia-like extensions (Fig. 5B). At

24hrs post scratch introduction, the migration distance was significantly reduced with

WDR47 siRNA treatment (151.3±4.2µm) compared to the various control groups, untreated (184.5±3.4µm), mitomycin (used for inhibiting proliferation) (178.5±3.7µm), and siRNA (189.6±3.1µm) (Fig. 5C). In addition, images of the wound captured every

6hrs and the percentage of wound closure calculated showed for example a 25%

15 significant reduction in the WDR47 siRNA treated group compared to siRNA control group at 36hrs post scratch introduction (Fig. 5D).

We next asked whether knocking down Wdr47 in cells might also have an impact on tubulin network dynamics. Using super-resolution structured illumination microscopy (SR-SIM), neuronal architecture was visualized using acetylated tubulin and showed highly convoluted structure in the perinuclear region, however, tubulin protein levels were not significantly altered (Fig. 5E). Confocal microscopy of cells transfected with WDR47-YFP and immunostained for tubulin revealed co-localization in the perinuclear region and extended processes (Fig. 5F). Finally, we used transmission electron microscopy (TEM) to determine whether WDR47 siRNA-treatment affected neuronal ultrastructure in particular mitochondria but detected no appreciable anomalies (Fig. S8B).

WDR47 plays a role in cell homeostasis and autophagy

Several important WDR proteins (CDC20 (33), Atg18 (34), PIK3R4 (35)) have been identified in yeast before transposition to a mammalian system. WDR47 is not normally expressed in yeast, however, this unicellular model allows to rapidly test for a wider spectrum of biological processes. Hence, we took advantage of this to study WDR47

CTLH domain specificity (Fig. 1C), by overexpression of human WDR47 and LIS1 in S. cerevisiae (Fig. 6A).

Remarkably, delayed growth was recorded exclusively in WDR47-GFP transformed cells (Fig. 6B), suggesting that WDR47 is hijacking important cellular functions. After staining of the vacuolar membrane with the lipophilic dye FM4-64 (Fig.

6C), we observed that LIS1-GFP was mainly cytosolic but upon overexpression, associated to a large punctate structure adjacent to the vacuole, whereas WDR47-GFP was associated to smaller structures in the cytoplasm (Fig. 6C), but not to endosomes or

Golgi complex (Fig. S8C). These results demonstrate that when WDR47 and LIS1 are overexpressed in yeast cells, they have different intracellular localization and only

WDR47 overexpression impairs growth.

Autophagy being essential for cell viability upon nutrient starvation in yeast (36) and autophagosomes being formed at a single site next to the vacuolar membrane(37), we asked whether WDR47 or LIS1 could be localized to autophagy sites. We first used expression of non-tagged WDR47 or LIS1 with mCherry-Atg8, a homologue of mammalian LC3 (38). This revealed that WDR47 impaired yeast autophagy, whereas

LIS1 did not. Indeed, mCherry-Atg8 did not reach the lumen of the vacuole and accumulated into the cytoplasm (Fig. 6D). Co-expression of WDR47- or LIS1-GFP with mCherry-Atg8 also impaired yeast autophagy and further showed that WDR47 was associated to punctate structures that did not co-localize with Atg8 (Fig. 6E). These results suggest that WDR47 might interact with some yeast autophagy effector thereby inhibiting this cellular process.

To investigate this hypothesis in mammalian cells, we tested autophagy in GT1-7

WDR47 siRNA treated cells compared to siRNA control group. Levels of LC3 and p62 (or

SQSTM1), two key proteins involved in autophagy, were quantified. We found that LC3-

II levels, but not LC3-I, were significantly reduced. Immunofluorescence and confocal microscopy of cells stained with LC3 was performed and consistently, LC3 positive punctae are reduced in WDR47 siRNA treated cells (Fig. S9A). Confocal microscopy of cells transfected with WDR47 and immunostained for LC3 reveals co-localization of

WDR47 and autophagosomes (Fig. S9B). To assess whether this is due to increased autophagic flux or decreased autophagosomal synthesis, we treated both WDR47 siRNA treated and siRNA control groups with bafilomycin A1, a potent inhibitor of the H+

ATPase that reduces lysosomes functionality, and found that both LC3-II and p62 levels were increased (Fig. 6F), indicating enhanced autophagy flux. We also tested the expression of p62 and mTOR (an upstream regulator of autophagy) in cortices from adult Wdr47tm1a/tm1a mice, showing a clear specificity of WDR47 function in p62- mediated autophagy (Fig. 6G). Taking together, these data demonstrate an additional role of WDR47 in cell homeostasis and protein clearance.

SCG10 and Reelin are novel interacting partners of WDR47

To understand the molecular mechanisms by which WDR47 might regulate microtubule stability and autophagy, we next searched for interacting partners by screening a human foetal cDNA library using a yeast two-hybrid system.

Using the N terminus of WDR47 as bait, the superior cervical ganglion-10 protein

(SCG10) was identified as a putative WDR47-interacting partner (Table S10). SCG10 is a well-established microtubule-destabilising protein (39) typically regulated by JNK1, a protein kinase of the MAPK family known to phosphorylate SCG10 rendering it inactive

(40). To gain insight into the mechanistic basis of this interaction, we first studied localization of WDR47, SCG10 and JNK1 in primary cortical neurons. WDR47 co- localized with SCG10 in the cytoplasm but not in the growth cone, whereas JNK1 showed co-localization with SCG10 in the cytoplasm as well as neurites (Fig. S10A).

SCG10 relative mRNA expression levels showed no difference between Wdr47tm1a/tm1a and wt mice preparations derived from the cortex, spinal cord, thalamus and the liver

(Fig. S10B). Co-localization of WDR47 and SCG10 was confirmed to occur in the cytoplasm of GT1-7 hypothalamic cells (Fig. 7A). Western blot analysis of endogenous

SCG10 normally gives rise to four bands that range from 20 to 25kDa, representing distinct phosphorylation states(41). We assessed this in WDR47 siRNA-treated cells compared to control siRNA cells, and quantified a 22kDa (unphosphorylated) and

25kDa (phosphorylated) band, but only detected a trend for decreased 22kDa SCG10 and increased 25kDa SCG10 (Fig. 7B).

Importantly, the WD40 domain of WDR47 was identified as a Reelin-interacting protein at its Reeler domain (Fig. S10C). The extracellular protein Reelin is an essential factor in the control of neuronal migration(42), also controlling microtubule stability and synaptic activity(43). The interaction between WDR47, Reelin and tubulin was confirmed in immortalized GT1-7 hypothalamic neuronal cells by co-localisation, with

Reelin and WDR47 co-localising in the cytoplasm (Fig. 7C), and both Reelin and WDR47 co-localised with tubulin (Fig. 7D). Coimmunoprecipitation assay confirmed the interaction between WDR47 and Reelin (Fig. S10D). Together these results show that

WDR47 physically interacts with two microtubule-regulating proteins: SCG10 and

Reelin.

DISCUSSION

WDR proteins have recently emerged in the field of neuroscience but their function and, ultimately, their participation in shaping the mammalian brain remains to be addressed.

Here, we report the analysis of 26 WD40-repeat (WDR) proteins in brain anatomy, and focus on the functional characterisation at the whole organism, cellular and molecular level using a combination of three model systems (KO mouse, siRNA and yeast) of a

19 novel member, WD40-repeat 47. WDR47 is also known as Nemitin (44), a protein sharing structural homology with LIS1, a WDR protein identified 20 years ago and associated with lissencephaly (45). Our findings highlight four important points.

First, 27% of assessed WDR genes gave rise to brain anomalies when inactivated in mice, Atg16l1+/-, Coro1c+/-, Dmxl2+/-, Herc1-/-, Kif21b-/-, Wdr47-/- and Wdr89-/-, demonstrating the functional importance of WDR genes in brain connectivity, notably in the genesis of the corpus callosum, a commissure that provides higher order neurological advantages in placental mammals. This is in line with the emerging role of

WDR genes in human brain pathologies associated with corpus callosum anomalies.

Examples include WDR73 (8), WDR81 (9), ERCC8 (46), and HERC1 (47). Herc1-/- knockout mice were coincidently processed in our study and showed macrocephaly and enlarged corpus callosum, reminiscent of the radiographic features of human patients with HERC1 mutations, demonstrating the pertinence of rodent screens for translating neuroanatomical disorders in humans.

Second, WDR47 has been shown to associate with microtubules in vivo (44), but for the first time, we report here the precise function of WD40-repeat 47 in brain connectivity and corpus callosum genesis. Our work provides new mechanistic insights demonstrating that WDR47 plays a regulatory role in key microtubule-mediated processes by stabilizing microtubules. We show that both callosal and corticofugal neurons have fibre tract defects, in conjunction with microcephaly and reduced number of cells in the cortex, in particular in layers II/III where the cell bodies interconnecting cortical neurons originate from and provide a molecular model underlying these neuropathological features. An increase in characteristic ring-like structures at the growth cone might be another underlying model since these have been recognized as

20 novel forms of structural organization of the cytoskeleton (48), with a potential role in microtubule nucleation (49), a process that has previously been linked to midline crossing in the fruit fly (50).

The interaction between the N-terminal of WDR47 and SCG10 (a very-well known microtubule destabilizer promoting catastrophe at the growth cone (51)), led us to think that WDR47 might also be a regulator of SCG10 activity in the JNK1 pathway.

This is further supported by KO mouse studies of Jnk1 (52) and Stmn2 mouse ortholog of SCG10 (Gabriele Grenningloh, personal communication), which, remarkably show high behavioural similarities (for example motor coordination defects) when compared to Wdr47 KO mice (see Dataset S1-2, SI Appendix). In addition, we saw an interaction between WD40 domain of WDR47 and Reelin (an important player in neuronal migration (43)), shedding light on the role of WDR47 in the Reelin-signaling pathway, in a similar way to LIS1(53). Our working model is summarised in Figure 8.

Third, altered levels of LC3-II and p62 proteins, concomitantly with increased autophagy flux, suggest that WDR47, unlike LIS1, is involved in the regulation of key effectors of autophagy in the brain. Autophagy is a protein clearance process of aberrant or obsolete cellular structures and organelles crucial in cell homeostasis (54).

Microtubules seem an important player in the autophagy process in mammals (55) however, their specific implication in autophagy is unclear (56). A potential molecular mechanism by which WDR47 regulates autophagy might be through its interaction with the light chain of MAP8 (also known as MAP1S) (44). In a similar way to MAP8 (57),

WDR47 might bridge the autophagy machinery, in particular LC3-II-bound autophagosomes, with microtubules.

Unlike other cells of the mammalian system, post-mitotic neurons are more vulnerable to damage from cellular debris such as damaged organelles and obsolete intracellular macromolecules, thus would require a well-regulated degradation pathway. The possible association of WDR47-related autophagy to brain wiring is supported by a recent report showing that the WDR autophagy scaffold protein (ALFY) is required for neuronal connectivity in the mouse brain (58). Strikingly, the lack of Alfy in mice led to perinatal lethality, microcephaly, absence of the corpus callosum, and hypoplasia of the internal capsule (58), reminiscent of Wdr47 phenotypes, suggesting a similar mode of action. It may therefore be that WDR47 regulates the cell ability to detect environmental cues via its microtubule stabilization role, which in turn regulates proteins degradation pathway necessary for neuronal shape and motility.

The final significant finding of this study is the essential role of WDR47 for survival. The scarcity of patients harbouring mutations in WDR47 supports this. The inability of S. cerevisiae to grow possibly due to the inhibition of autophagy upon overexpression of WDR47 further supports the essentiality of WDR47 for survival, but also suggests that mirrored protein levels (too little or too much) have identical effects on pathogenicity and underlying biological processes. This is a characteristic of scaffold proteins, indicating that WDR47 might be serving as a support for the interaction of other proteins such as SCG10 and Reelin. Although we were unable to identify the precise reason of death, cardiac and breathing failure were excluded. Autophagy being dramatically upregulated in the neonatal stage to overcome the starvation period (59), mice could be dying of lack of nutrients since an enriched lipid diet rescued lethality.

In conclusion, this study presents the first identification of the relevance of WDR genes in brain connectivity, highlighting the power of unbiased and high-throughput

22 mouse LoF studies in the field of neuroscience. These mouse models could explain some of the missing genetics in corpus callosum biology (12), and help pave the way for better stratification system of complex neurodevelopmental steps. WDR47 plays a role in the regulation of microtubules dynamics, neuronal navigation, and fibre tract projections, in a similar fashion to LIS1 (60-62), but with the distinctive particularity that WDR47 inhibits autophagic flux. This provides a functional link between autophagy biology and the CTLH domain (63) for the first time in mammals (this association was previously made with the vacuole import and degradation pathways in yeast (64)), whilst strengthening the emerging link between autophagy and microtubules assembly.

Although a definite association of WDR47 with human brain disorders has not been made yet (see Dataset S3 and SI Appendix), WDR47 should be considered as a novel candidate gene for corpus callosum abnormalities (isolated or associated with malformations of cortical development) and motor coordination deficiencies, possibly through compound heterozygosity or somatic mosaicism in the brain. Considering that the microtubule stabilizer drug restored cellular defects, it will be interesting to test whether the behavioral anomalies, as already shown in an Alzheimer mouse model (65), could also be restored. This might open up novel therapeutic perspectives in the clinic with the aim of relieving symptoms in patients suffering from this novel class of diseases we refer to as “WDRopathies”.

METHODS

Wdr47 knockout construction

The construction of alleles is based on the “Knockout-first allele” method (Fig. S1B)

(26). A “critical” exon (exon 5) present on all transcripts, was first identified. A promotor-less targeting cassette, embryonic stem-cell clone EPD0046_1_C04 from

KOMP (Knockout Mouse Program), was inserted in intron 4 in C57BL/6N blastocyst, which created a frame-shift mutation, thus generating the KO-first allele (tm1a allele).

Tm1a mice were bred into mice expressing Cre recombinase under the ROSA26 promoter to generate null alleles (tm1b). A heterozygous-by-heterozygous breeding scheme was used to propagate the various lines.

Wdr47 genotyping

Genomic DNA was extracted from mouse ear or tail clippings. Primers were designed specifically for each mouse model (see Fig. S1B) with the following sequences: Ef

5’AGGTTGTCATGCAGTCTGGG3’, Er 5’GGATGACTATAAAGCGGTGCAAG3’, Kr

5’CTCCTACATAGTTGGCAGTGTTTGGG3’, L3f 5’TCCTTTGCTAACTTCCACTATCC3’, L3r

5’TCAGCCTGGTCTACAGAGTTA3’, Lxr 5’ACTGATGGCGAGCTCAGACCATAAC3’ and Mq1f

5’GGGATCTCATGCTGGAGTTCTTCG3’. PCR reactions were carried out using the

FastStart PCR Master (Roche Life Science) under the following thermocycler conditions:

95°C for 4min (1 cycle), 94°C for 30sec followed by 62°C for 30sec and 72°C for 1min

(34 cycles), 72°C for 7min (1 cycle), and 20°C for 5min (1 cycle). Expected size of PCR product is shown in Fig. S1B.

LacZ expression

Brain samples from mice at E14.5, E16.5, E18.5 and 16 weeks of age and whole embryos at E9.5 were collected. Samples were dropped-fixed in fixative solution (10%

Formaldehyde, 0.4% Gluteraldehyde, 0.04% Np40, and 0.02% NaDC) for 2hrs and washed in 1xPBS. The samples were incubated in X-Gal solution (5mM Potassium

Ferricyanide, 5mM Potassium Ferrocyanide, 1mg/ml X-Gal, 2mM Magnesium Chloride in 1xPBS) at 4°C overnight. The samples were washed in 1x PBS and sectioned at 50µm thickness using a vibratome and post-fixed in 4% paraformaldehyde. The sections were counterstained in 0.1% Safranin solution and mounted on slides before imaging using the Hamamatsu slide scanner.

Histo-morphometric and statistical analyses

Methods associated to histology, micro computed tomography (µCT) and magnetic resonance imaging (MRI) are described in SI Appendix. Each image was QCed in order to assess whether: 1) the section is at the correct position, 2) the section is symmetrical,

3) the staining is of good quality, and 4) the image is good quality. Only images that fulfilled all the quality control checks were processed. To avoid experimenter bias, the same person measured the 78 parameters on adult coronal planes, 67 on embryonic planes, and 95 on sagittal planes (Table S8). Once data collection was completed, we checked for the presence of erroneous measures, typos and outliers, and corrected the measurement if necessary. We used a linear mixed model as implemented in Phenstat

(66) to assess significance and critically evaluated any results with p-values higher than

0.0001. We considered a phenotype hit if sections were undamaged and measurements were accurate.

DATA ACCESS

Data are available upon request from the corresponding author.

DECLARATION

The authors declare no conflicts of interest.

REFERENCES

1. Andrade MA, Perez-Iratxeta C, & Ponting CP (2001) Protein repeats: structures, functions, and evolution. Journal of structural biology 134(2-3):117-131. 2. Sondek J, Bohm A, Lambright DG, Hamm HE, & Sigler PB (1996) Crystal structure of a G- protein beta gamma dimer at 2.1A resolution. Nature 379(6563):369-374. 3. Stirnimann CU, Petsalaki E, Russell RB, & Muller CW (2010) WD40 proteins propel cellular networks. Trends in biochemical sciences 35(10):565-574. 4. Smith TF, Gaitatzes C, Saxena K, & Neer EJ (1999) The WD repeat: a common architecture for diverse functions. Trends in biochemical sciences 24(5):181-185. 5. Sasaki S, et al. (2000) A LIS1/NUDEL/cytoplasmic dynein heavy chain complex in the developing and adult nervous system. Neuron 28(3):681-696. 6. Saillour Y, et al. (2009) LIS1-related isolated lissencephaly: spectrum of mutations and relationships with malformation severity. Arch Neurol 66(8):1007-1015. 7. Bilguvar K, et al. (2010) Whole-exome sequencing identifies recessive WDR62 mutations in severe brain malformations. Nature 467(7312):207-210. 8. Colin E, et al. (2014) Loss-of-function mutations in WDR73 are responsible for microcephaly and steroid-resistant nephrotic syndrome: Galloway-Mowat syndrome. American journal of human genetics 95(6):637-648. 9. Gulsuner S, et al. (2011) Homozygosity mapping and targeted genomic sequencing reveal the gene responsible for cerebellar hypoplasia and quadrupedal locomotion in a consanguineous kindred. Genome Res 21(12):1995-2003. 10. Lindwall C, Fothergill T, & Richards LJ (2007) Commissure formation in the mammalian forebrain. Curr Opin Neurobiol 17(1):3-14. 11. Fame RM, MacDonald JL, & Macklis JD (2011) Development, specification, and diversity of callosal projection neurons. Trends Neurosci 34(1):41-50. 12. Paul LK, et al. (2007) Agenesis of the corpus callosum: genetic, developmental and functional aspects of connectivity. Nat Rev Neurosci 8(4):287-299.

13. Paul LK, Corsello C, Kennedy DP, & Adolphs R (2014) Agenesis of the corpus callosum and autism: a comprehensive comparison. Brain 137(Pt 6):1813-1829. 14. Li J, et al. (2014) A comparative diffusion tensor imaging study of corpus callosum subregion integrity in bipolar disorder and schizophrenia. Psychiatry Res 221(1):58-62. 15. Balevich EC, et al. (2015) Corpus callosum size and diffusion tensor anisotropy in adolescents and adults with schizophrenia. Psychiatry Res 231(3):244-251. 16. Edwards TJ, Sherr EH, Barkovich AJ, & Richards LJ (2014) Clinical, genetic and imaging findings identify new causes for corpus callosum development syndromes. Brain 137(Pt 6):1579-1613. 17. Romaniello R, et al. (2016) Clinical Characterization, Genetics, and Long-Term Follow-up of a Large Cohort of Patients With Agenesis of the Corpus Callosum. J Child Neurol. 18. Chedotal A (2011) Further tales of the midline. Curr Opin Neurobiol 21(1):68-75. 19. Dent EW, Gupton SL, & Gertler FB (2011) The growth cone cytoskeleton in axon outgrowth and guidance. Cold Spring Harb Perspect Biol 3(3). 20. Lowery LA & Van Vactor D (2009) The trip of the tip: understanding the growth cone machinery. Nat Rev Mol Cell Biol 10(5):332-343. 21. Sapir T, Elbaum M, & Reiner O (1997) Reduction of microtubule catastrophe events by LIS1, platelet-activating factor acetylhydrolase subunit. EMBO J 16(23):6977-6984. 22. Chen JF, et al. (2014) Microcephaly disease gene Wdr62 regulates mitotic progression of embryonic neural stem cells and brain size. Nat Commun 5:3885. 23. Kirschner MW & Mitchison T (1986) Microtubule dynamics. Nature 324(6098):621. 24. White JK, et al. (2013) Genome-wide generation and systematic phenotyping of knockout mice reveals new roles for many genes. Cell 154(2):452-464. 25. Richards LJ, Plachez C, & Ren T (2004) Mechanisms regulating the development of the corpus callosum and its agenesis in mouse and human. Clin Genet 66(4):276-289. 26. Skarnes WC, et al. (2011) A conditional knockout resource for the genome-wide study of mouse gene function. Nature 474(7351):337-342. 27. Llewellyn KJ, et al. (2014) Lipid-enriched diet rescues lethality and slows down progression in a murine model of VCP-associated disease. Hum Mol Genet 23(5):1333-1344. 28. Yuan Q, et al. (2013) Folic acid supplementation changes the fate of neural progenitors in mouse embryos of hyperglycemic and diabetic pregnancy. J Nutr Biochem 24(7):1202-1212. 29. Saitsu H, et al. (2013) De novo mutations in the autophagy gene WDR45 cause static encephalopathy of childhood with neurodegeneration in adulthood. Nature genetics 45(4):445-449, 449e441. 30. Brizuela M, et al. (2015) The microtubule-stabilizing drug Epothilone D increases axonal sprouting following transection injury in vitro. Mol Cell Neurosci 66(Pt B):129-140. 31. Grabham PW, Seale GE, Bennecib M, Goldberg DJ, & Vallee RB (2007) Cytoplasmic dynein and LIS1 are required for microtubule advance during growth cone remodeling and fast axonal outgrowth. J Neurosci 27(21):5823-5834. 32. Liang CC, Park AY, & Guan JL (2007) In vitro scratch assay: a convenient and inexpensive method for analysis of cell migration in vitro. Nat Protoc 2(2):329-333. 33. Yu H (2007) Cdc20: a WD40 activator for a cell cycle degradation machine. Mol Cell 27(1):3- 16. 34. Zhang C & Zhang F (2015) The Multifunctions of WD40 Proteins in Genome Integrity and Cell Cycle Progression. J Genomics 3:40-50. 35. Stoetzel C, et al. (2016) A mutation in VPS15 (PIK3R4) causes a ciliopathy and affects IFT20 release from the cis-Golgi. Nat Commun 7:13586. 36. Suzuki SW, Onodera J, & Ohsumi Y (2011) Starvation induced cell death in autophagy- defective yeast mutants is caused by mitochondria dysfunction. PloS one 6(2):e17412. 37. Shibutani ST & Yoshimori T (2014) A current perspective of autophagosome biogenesis. Cell Res 24(1):58-68.

38. Kabeya Y, et al. (2000) LC3, a mammalian homologue of yeast Apg8p, is localized in autophagosome membranes after processing. EMBO J 19(21):5720-5728. 39. Riederer BM, et al. (1997) Regulation of microtubule dynamics by the neuronal growth- associated protein SCG10. Proceedings of the National Academy of Sciences of the United States of America 94(2):741-745. 40. Tararuk T, et al. (2006) JNK1 phosphorylation of SCG10 determines microtubule dynamics and axodendritic length. J Cell Biol 173(2):265-277. 41. Bondallaz P, Barbier A, Soehrman S, Grenningloh G, & Riederer BM (2006) The control of microtubule stability in vitro and in transfected cells by MAP1B and SCG10. Cell Motil Cytoskeleton 63(11):681-695. 42. Falconer DS (1951) Two new mutants, 'trembler' and 'reeler', with neurological actions in the house mouse (Mus musculus L.). J Genet 50(2):192-201. 43. Lee GH & D'Arcangelo G (2016) New Insights into Reelin-Mediated Signaling Pathways. Front Cell Neurosci 10:122. 44. Wang W, et al. (2012) Nemitin, a novel Map8/Map1s interacting protein with Wd40 repeats. PloS one 7(4):e33094. 45. Lo Nigro C, et al. (1997) Point mutations and an intragenic deletion in LIS1, the lissencephaly causative gene in isolated lissencephaly sequence and Miller-Dieker syndrome. Hum Mol Genet 6(2):157-164. 46. Laugel V, et al. (2010) Mutation update for the CSB/ERCC6 and CSA/ERCC8 genes involved in Cockayne syndrome. Hum Mutat 31(2):113-126. 47. Nguyen LS, et al. (2016) A nonsense variant in HERC1 is associated with intellectual disability, megalencephaly, thick corpus callosum and cerebellar atrophy. Eur J Hum Genet 24(3):455- 458. 48. Gallo G (2013) More than one ring to bind them all: recent insights into the structure of the axon. Dev Neurobiol 73(11):799-805. 49. Petry S & Vale RD (2015) Microtubule nucleation at the centrosome and beyond. Nat Cell Biol 17(9):1089-1093. 50. Schenck A, et al. (2004) WAVE/SCAR, a multifunctional complex coordinating different aspects of neuronal connectivity. Dev Biol 274(2):260-270. 51. Grenningloh G, Soehrman S, Bondallaz P, Ruchti E, & Cadas H (2004) Role of the microtubule destabilizing proteins SCG10 and stathmin in neuronal growth. J Neurobiol 58(1):60-69. 52. Komulainen E, et al. (2014) JNK1 controls dendritic field size in L2/3 and L5 of the motor cortex, constrains soma size, and influences fine motor coordination. Front Cell Neurosci 8:272. 53. Assadi AH, et al. (2003) Interaction of reelin signaling and Lis1 in brain development. Nature genetics 35(3):270-276. 54. Abada A & Elazar Z (2014) Getting ready for building: signaling and autophagosome biogenesis. EMBO reports 15(8):839-852. 55. Mackeh R, Perdiz D, Lorin S, Codogno P, & Pous C (2013) Autophagy and microtubules - new story, old players. Journal of cell science 126(Pt 5):1071-1080. 56. Monastyrska I, Rieter E, Klionsky DJ, & Reggiori F (2009) Multiple roles of the cytoskeleton in autophagy. Biol Rev Camb Philos Soc 84(3):431-448. 57. Xie R, Nguyen S, McKeehan WL, & Liu L (2010) Acetylated microtubules are required for fusion of autophagosomes with lysosomes. BMC Cell Biol 11:89. 58. Dragich JM, et al. (2016) Autophagy linked FYVE (Alfy/WDFY3) is required for establishing neuronal connectivity in the mammalian brain. Elife 5. 59. Kuma A, et al. (2004) The role of autophagy during the early neonatal starvation period. Nature 432(7020):1032-1036.

60. Cahana A, et al. (2001) Targeted mutagenesis of Lis1 disrupts cortical development and LIS1 homodimerization. Proceedings of the National Academy of Sciences of the United States of America 98(11):6429-6434. 61. Dujardin DL, et al. (2003) A role for cytoplasmic dynein and LIS1 in directed cell movement. J Cell Biol 163(6):1205-1211. 62. Reiner O (2013) LIS1 and DCX: Implications for Brain Development and Human Disease in Relation to Microtubules. Scientifica (Cairo) 2013:393975. 63. Kobayashi N, et al. (2007) RanBPM, Muskelin, p48EMLP, p44CTLH, and the armadillo-repeat proteins ARMC8alpha and ARMC8beta are components of the CTLH complex. Gene 396(2):236-247. 64. Alibhoy AA, Giardina BJ, Dunton DD, & Chiang HL (2012) Vid30 is required for the association of Vid vesicles and actin patches in the vacuole import and degradation pathway. Autophagy 8(1):29-46. 65. Zhang B, et al. (2012) The microtubule-stabilizing agent, epothilone D, reduces axonal dysfunction, neurotoxicity, cognitive deficits, and Alzheimer-like pathology in an interventional study with aged tau transgenic mice. J Neurosci 32(11):3601-3611. 66. Karp NA, Melvin D, Sanger Mouse Genetics P, & Mott RF (2012) Robust and sensitive analysis of mouse knockout phenotypes. PloS one 7(12):e52410.

FIGURES AND FIGURE LEGENDS

Figure 1. Relevance of mouse WDR genes in brain structure (A) Top right:

Schematic representation of brain features plotted in two coronal planes according to p- values for Atg16l1+/-, Coro1c+/-, Dmxl2+/-, Herc1-/-, Kif21b-/-, Wdr89-/- and Wdr47-/- KO mice (n=3 in each group). The left half schematic image represents the striatum section

(Bregma 0.98mm) and the right one shows the hippocampus section (Bregma -

1.34mm). White indicates p-value higher than 0.05 (ns for not significant) and grey is no data (nd) for absence of data. Top left: Histograms of effect sizes relative to wild-type animals (100%) colored according to the significance level. Circles with crosses indicate cell count measurements in each brain region and are plotted in the histograms according to p-value and effect size. Bottom left: Numbers around the circle show assessed unique brain regions together with a list of acronyms (full description provided in Table S2). Statistical analyses were carried out using the linear mixed model framework within Phenstat (66). * denotes the absence of the assessed brain region (corpus callosum). (B) Brain coronal section at Bregma 1.34mm, double stained with cresyl violet for neuron and luxol blue for myelin, showing a representative image of the corpus callosum (cc) in wild-type animal (top), compared to agenesis of the corpus callosum (AgCC) in a Wdr47-/- mouse (bottom). (C) Structural similarities and dissimilarities between WDR47 and LIS1.

Figure 2. Characterization of Wdr47 mouse models (A) Wdr47 relative expression using qRT-PCR in Wdr47+/tm1a (n=3), Wdr47tm1a/tm1a (n=3) and matched wt (n=3), across eleven tissues (cortex, hippocampus, spinal cord, cerebellum, striatum, thalamus, medulla, olfactory bulb, lung, liver and heart), and in Wdr47+/tm1b (n=3) and matched wt

(n=3) across four tissues (cortex, hippocampus, spinal cord and heart). Normalization was done using the house keeping gene, GNAS. The average relative expression of

Wdr47 across all tissues is calculated in Table S6. (B) WDR47 protein profiling in cortex of wt (n=3) and Wdr47tm1a/tm1a (n=3). Normalization was done using beta-actin. (C) LacZ staining in E18.5 and adult Wdr47+/tm1a mice across layers II/IIII of the cortex, pyramidal cells (py) and dendate gyrus (DG) of the hippocampus, piriform cortex (pir), arcuate nucleus (Arc) and ventromedial part (VMH) of the hypothalamus. The experiment was replicated 3 times and all showed the same result. (D) Correlation plot between Wdr47 average expression (%) and percentage mouse lethality. A total of 843

Wdr47tm1a and 242 Wdr47tm1b mice were used. A linear regression model was fitted

(dashed red line) (r2=0.9). (E) Mouse survival outcome carried out at nine time points both in Wdr47tm1a male and female mice: E9.5 (n=34), E18.5 (n=132), P0 (n=36), P16

(n=28), P55 (n=97), P90 (n=130), P111 (n=89), P132 (n=75), and P153 (n=53).

Expected ratio indicates ratio of 25% for wt, 50% for Wdr47+/tm1a and 25% for

Wdr47tm1a/tm1a, accordingly to a ‘het-by-het’ breeding scheme. (F) Mouse survival outcome upon supplementation in fortified diet (Mouse Breeder Diet 5021) with extra lipids (10.8% as opposed to 3% in a chow normal diet) and folic acid (3mg versus

0.7mg) in Wdr47tm1a and Wdr47tm1b mice across three generations from F0 (n=621), F1

(n=112), F2 (n=82), and from F0 (n=118), F1 (n=53), F2 (n=37), in Wdr47tm1a and

Wdr47tm1b mice respectively. Expected ratio is based on the "het-by-het’ breeding scheme. All plots are represented as mean+s.e.m. ***P<0.001, **P<0.01, *P<0.05,

#P<0.07. Statistical analysis was done using Student’s t-test (two-tailed).

Figure 3. Major fiber tracts defects and microcephaly in mice (A) Height of cortical layers in adult Wdr47tm1a/tm1a mice (n=6) compared to matched wt (n=6) at Bregma

0.98mm, combining chow and fortified diet fed mice. Sections were stained with cresyl violet and luxol blue. (B) Number of cells and cell size in layers II/III in Wdr47tm1a/tm1a mice (n=6) compared to wt (n=6). (C) Height of neocortical layers at 2.19mm in sections stained with cresyl violet from wt (n=5) and Wdr47tm1a/tm1a (n=4) embryos at E18.5. MZ, marginal zone; CP, cortical plate; SP, subplate; IZ, intermediate zone; SVZ, sub- ventricular zone; VZ, ventricular zone. All plots are represented as mean+s.e.m.

***P<0.001, **P<0.01, *P<0.05. (Student’s t-test, two-tailed). (D) Sagittal sections stained with cresyl violet and luxol blue in mice with reducing relative expression of

Wdr47. cc, corpus callosum; HP, hippocampus; aca, anterior commissure anterior part.

(E) Left: heat-map of p-values of 22 brain regions quantified in section 7 (Lateral

0.72mm, Fig. S6) across Wdr47+/tm1a and Wdr47tm1a/tm1a mice, both male and female, versus respective wt, n=3 in each group. 1, TBA, TBA_width, TBA_Height1, TBA_Height2;

2, cc, cc_Height, cc_TOL; 3, HP, TILpy, DG, Mol, Rad, Or; 4, Ds; 5, hif; 6, fi; 7, f; 8, aca; 9, Cg,

Cg_Height; 10, M1, S1; 11, CPu; 12, ThNu; 13, sm; 14, Mnu; 15, SN; 16, InfC; 17, och; 18,

TCA, Folia, IGL, Med; 19, Pn, fp; 20, 7N; 21, LV; 22, 4V (abbreviations described in Table

S8). Right: histograms of percentage effect size in comparison to controls (100%). (F)

Summary heat-map of p-values for 25 and 31 sagittal brain regions quantified in critical section 8 (Lateral 1.32mm) and 9 (Lateral 2.52mm), respectively (Fig. S6, Table S8) across Wdr47+/tm1a and Wdr47tm1a/tm1a mice, both male and female. (G) Confocal microscopy images of projection patterns in the developing brain using axonal (SMI-

312R) and commissural (L1CAM) markers at E14.5 (n=2) and E16.5 (n=2) (scale is

1mm for first and second panels, and 100μm for third panel). Th, thalamus; DTB, diencephalon-telencephalon barrier; TRN, thalamic reticular nucleus; ic, internal

34 capsule; Ctx, cortex. (H) Brain morphology studied in Wdr47tm1a/tm1a male mouse using diffusion tensor MRI.

Figure 4. Microtubule-stabilizing role of WDR47 at the growth cone ex vivo (A)

Florescent microscopy images of primary neurons derived from wt and Wdr47tm1a/tm1a embryos at E17.5 stained with anti-MAP2 (red) and SMI-312R (green) (top panel scale

50μm, bottom panel 5μm) in hippocampal (HP) and cortical (Ctx) primary neuronal cultures. Area of cell body (nHP(Wdr47tm1a/tm1a)=95, nHP(wt)=120); length of axon

(nHP(Wdr47tm1a/tm1a)=231, nHP(wt)=167, nCtx(Wdr47tm1a/tm1a)=65, nCtx(wt)=56) and area of growth cones (nHP(Wdr47tm1a/tm1a)=115, nHP(Wdr47+/tm1a)=239, nHP(wt)=129, nCtx(Wdr47tm1a/tm1a)=78, nCtx(Wdr47+/tm1a)=126, nCtx(wt)=88) were quantified using

ImageJ and analysed using the Kruskal-Wallis test. (B) Microtubule architecture was studied in the growth cones of hippocampal primary neurons by staining for acetylated tubulin, revealing an odd ring-like arrangement (indicated using white arrows) in the

Wdr47tm1b/tm1b neurons using regular fluorescent microscopy. (C) Super-resolution single molecule localization microscopy of hippocampal growth cone stained with acetylated tubulin showing dispersed distribution in Wdr47tm1b/tm1b (scale 5μm). (D)

Western blot analysis of endogenous Tau levels in three Wdr47tm1a/tm1a mice as compared to matched wt mice (n=3). Quantification of relative protein expression is normalized against β-actin. (E) Images of primary hippocampal neurons derived from

Wdr47tm1b/tm1b and wt embryos at E17.5, treated with 10 and 100nM Epothilone-D

(EpoD), scale 5μm. Growth cone area was quantified using ImageJ after 1.5hrs of treatment. All plots are represented as mean±s.e.m. ***P<0.001, **P<0.01, *P<0.05

(Students t-test, two-tailed).

Figure 5. Involvement of WDR47 in neuronal migration in vitro (A) Western blot verification of WDR47 knock-down in rat GT1-7 hypothalamus-derived neuronal cells reveals a significant reduction of WDR47 protein levels (37.3±9.3%, *P<0.05) in WDR47 siRNA-treated cells compared to control siRNA cells (71±11.9 %) 24hrs post transfection (n=7). (B) Scanning electron micrographs of control siRNA and WDR47 siRNA neurons at the level of the axon and growth cone in the migration zone. Scale bar is 2μm. (C) The right panel shows transmission light micrographs of an in vitro 36hrs neuronal migration assay of rat GT1-7 neuronal cells treated with control or WDR47 siRNA. The dashed red lines show the edge of the wound. The bottom panel shows images at higher magnification. White arrows indicate neurons at the edge of the wound. The average migration distance is significantly reduced with WDR47 siRNA treatment (151.3±4.2μm) compared to the mitomycin control group (178.5±3.7μm) or control siRNA group (189.6±3.1μm) (n=4, *P<0.05), at 24hrs post-scratch introduction.

Similarly, the migration velocity (μm/min) is significantly reduced with WDR47 siRNA- treatment (0.11±0.003μm/min) compared to mitomycin control group

(0.12±0.003μm/min), and control siRNA group (0.13±0.002 μm/min) (n=4, *P<0.05).

(D) Percentage of wound closure is shown over time, indicating reduced closure in

WDR47 siRNA treated cells (n=4, *P<0.05). (E) Representative confocal (left) and SR-

SIM (right) fluorescent micrographs of acetylated tubulin networks in control siRNA

(top panel) and WDR47 siRNA-treated cells (bottom panel). Scale bars are 20μm and

5μm, respectively. Acetylated tubulin appears to increase and to robustly organize in the perinuclear region, forming a highly dense and convoluted network. Western blot analysis of acetylated tubulin is carried out in control siRNA compared to WDR47 siRNA-treated cells. Normalisation is done using housekeeping gene GAPDH. (F)

Representative confocal microscopy of cells transfected with WDR47-YFP (yellow) and

38 immuno-stained for acetylated tubulin (red) reveals pronounced colocalization (white) in the perinuclear region as well as extended processes. Bottom right panel shows co localization only. Nuclei are counterstained with Hoechst, scale bar 10μm.

Figure 6. WDR47 is a key effector of autophagy (A) Wild-type BY4742 yeast strain transformed with the empty plasmid pAG413-EGFP was used as control. (B) Drop test growth assays were done on wild-type yeast cells (BY4742) transformed with pAG413

(CEN, expression) or pAG423 (2μl, overexpression) plasmids bearing LIS1 or WDR47, as indicated. Mid-log phase cultures of the indicated yeast cells were serially diluted to the indicated OD600 and spotted onto SC-His medium. Growth was evaluated after 2 days of incubation at 30°C. (C) Living wild-type yeast cells (BY4742) expressing or overexpressing human LIS1-GFP or WDR47-GFP were observed by fluorescence microscopy after staining of the vacuoles with the FM4-64 lipid dye. The merge shows the merge between the GFP and DsRED images and the merge+DIC, the merge between the GFP, DsRED and DIC images. (D) Wild-type BY4742 (control) or vps15Δ (negative control) yeast cells transformed with mCherry-Atg8 plasmid and wild-type BY4742 cells co-transformed with expression plasmid (pAG413) bearing LIS1 or WDR47 cDNA were observed by fluorescent microscopy after incubation for 4hrs in SD-N medium to induce autophagy. (E) Living wild-type yeast cells (BY4742) expressing human LIS1-

GFP or WDR47-GFP and mCherry-Atg8 were observed by fluorescence microscopy after induction of autophagy by incubation in SD-N medium. (F) Western blot quantification of LC3 and p62 relative protein levels in the presence and absence of Bafilomycin A1 treatment for the assessment of autophagic flux in response to WDR47 siRNA treatment. GAPDH is used as a loading control. (G) Representative western blot images of protein expression of p62, mTOR and phospho-mTOR in wt and Wdr47tm1a/tm1a mice.

Quantification of relative protein expression, normalized against β-actin is plotted as mean±s.e.m. **P<0.01, *P<0.05 (n=6 Wdr47tm1a/tm1a and n=9 wt, male and female). The scale bars represent 5μm.

Figure 7. SCG10 and Reelin interacts with WDR47 (A) Co-localization of SCG10 and

WDR47 in GT1-7 hypothalamic neuronal cells. Top left panel shows expression in GT1-7 cells transfected with pEYFP-WDR47 (green). Top right panel shows expression of

SCG10 labelled with Texas red anti-SCG10 antibody (red). Bottom left panel is the overlay of the top panel images and bottom right panel is the three dimensional co- localization of WDR47and SCG10. (B) Western blot analysis of SCG10 relative protein levels in response to WDR47 siRNA treatment. GAPDH is used as a loading control.

*P<0.05. Statistical analysis was done using Student’s t-test (two tailed). (C) Co- localizatio of Reelin with WDR47 in GT1-7 cells. Top left panel shows expression of

Reelin labelled with Texas red anti-Reelin antibody. Top right panel shows expression in GT1-7 cells transfected with pEYFP-WDR47 (yellow fluorescence). Bottom left panel is the overlay of top panel images. Bottom right panel is the three dimensional co- localization of WDR47 and Reelin. (D) Co-localization of Reelin and WDR47 with tubulin in GT1-7 hypothalamic neuronal cells. The top left panel shows expression of reelin in

GT1-7 cells labelled with Texas red anti-Reelin antibody, followed by paclitaxel oregon green 488-labelled tubulin, overlay and three dimensional co-localization of Reelin and tubulin. The bottom panel shows expression in GT1-7 cells transfected with pEYFP-

WDR47 (yellow), followed by paclitaxel oregon green 488-labelled tubulin, overlay and three dimensional co-localisation of Reelin and tubulin.

Figure 8. Summary of WDR47 working model (A) Schematic representation of a neuron with normal expression of Wdr47. WDR47 shares striking structural homology with LIS1, a well-established microtubule interacting protein (MAP) that regulates dynein activity. This study reveals that WDR47 is also a MAP. More specifically, WDR47 regulates microtubule dynamics by acting as a stabilizer of microtubules. SCG10, a microtubule destabilizer is an important player in growth cone dynamics and is regulated by JNK1 through phosphorylation. In the cytoplasm, WDR47 acts a negative regulator of p62 and LC3-mediated autophagy. Besides, WDR47 interacts with Reelin and SCG10. (B) Lack of Wdr47 leads to destabilization of the microtubules, causing an excess of growth cone catastrophes along with an abnormal ring-like architecture.

Wdr47 also plays an important role in LC3-mediated autophagy resulting in an increased autophagic flux. Finally, absence of Wdr47 gives rise to a significant increased level of Tau. Together, these defects explain impairment in neuronal motility, which in turn, may result in the agenesis of the corpus callosum.