Microcephaly-Associated Protein WDR62 Regulates Neurogenesis Through JNK1 in the Developing Neocortex

Cell Reports Article

Microcephaly-Associated Protein WDR62 Regulates Neurogenesis through JNK1 in the Developing Neocortex

Dan Xu,1 Feng Zhang,1 Yaqing Wang,1 Yiming Sun,1 and Zhiheng Xu1,2,* 1State Key Laboratory of Molecular Developmental Biology, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, West Lincui Road, Chaoyang District, Beijing 100101, China 2Center of Parkinson’s Disease, Beijing Institute for Brain Disorders, Beijing 100101, China *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2013.12.016 This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

SUMMARY somes or mitotic spindle poles (Bilgu¨ var et al., 2010; Guernsey et al., 2010; Hussain et al., 2012; Manzini and Walsh, 2011; Nich- Mutations of WD40-repeat protein 62 (WDR62) have olas et al., 2010; Wollnik, 2010; Yu et al., 2010). Mutations been identified recently to cause human MCPH (auto- in ASPM (MCPH5) and WD40-repeat protein 62 (WDR62; somal-recessive primary microcephaly), a neurode- MCPH2) genes account for more than half of MCPH worldwide velopmental disorder characterized by decreased (Mahmood et al., 2011; Thornton and Woods, 2009). brain size. However, the underlying mechanism is During the initial stages of cortical development, self-renewing unclear. Here, we investigate the function of WDR62 symmetric cell division of progenitors generates sufficient numbers of cells to serve as a pool for ongoing neurogenesis in brain development and the pathological role of WDR62 (Dehay and Kennedy, 2007; Franco and Mu¨ ller, 2013; Huttner mutations. We find that WDR62 knockdown and Kosodo, 2005; Knoblich, 2008; Wang et al., 2011). The pro- leads to premature differentiation of neural proliferation and differentiation of NPCs mainly take place in the ven- genitor cells (NPCs). The defect can be rescued by tricular and subventricular zones (VZ and SVZ, respectively). wild-type human WDR62, but not by the five MCPH- Disturbance of the symmetric cell divisions can lead to reduction associated WDR62 mutants. We demonstrate that of the neural progenitor pool, a subsequent decrease in prolifer- WDR62 acts upstream of JNK signaling in the control ation, and decline of neuron production levels. The end result is a of neurogenesis. Depletion of JNK1 and WDR62 in- smaller than usual brain or microcephaly (Dehay and Kennedy, curs very similar defects including abnormal spindle 2007; Huttner and Kosodo, 2005; Manzini and Walsh, 2011; Noc- formation and mitotic division of NPCs as well as pre- tor et al., 2004; Thornton and Woods, 2009; Wollnik, 2010). The mature NPC differentiation during cortical develop- newborn neurons transiently become multipolar with multiple processes within the SVZ and lower intermediate zone (IZ) and ment. Thus, our findings indicate that WDR62 is then change to a bipolar morphology and migrate along radial required for proper neurogenesis via JNK1 and pro- fibers to the cortical plate (CP) where they further mature (Marıń vide an insight into the molecular mechanisms under- and Rubenstein, 2003; Noctor et al., 2001; Tsai et al., 2005). lying MCPH pathogenesis. Recent studies have identified WDR62 as a causative gene of autosomal recessive microcephaly (Bilgu¨ var et al., 2010; Nicho- INTRODUCTION las et al., 2010; Yu et al., 2010). Similar to our findings for POSH (plenty of SH3 domains) (Kukekov et al., 2006; Xu et al., 2003, Development of the mammalian cortex requires precise timing 2005), WDR62 was originally reported as a scaffold protein for of proliferation/self-renewal of neural progenitor cells (NPCs), the JNK pathway (Cohen-Katsenelson et al., 2011; Wasserman differentiation, neuronal migration, and maturation. Abnormality et al., 2010). WDR62 has 13 WD40 domain repeats in its N-termi- in any of these processes may lead to brain developmental dis- nal half and MKK7/JNK binding domains and six potential JNK orders, such as primary autosomal-recessive microcephaly phosphorylation sites in its C-terminal half. Although WDR62 (MCPH) and lissencephaly (Kaindl et al., 2010; Kang et al., has been proposed to play an important role in spindle formation 2011; Manzini and Walsh, 2011; Marıń and Rubenstein, 2003; and the proliferation of neuronal progenitor cells (Bilgu¨ var et al., Thornton and Woods, 2009; Tsai et al., 2005). MCPH is a neural 2010; Nicholas et al., 2010; Yu et al., 2010), the exact role of developmental disorder in which patients display significantly WDR62 during cortical development and the underlying mecha- reduced brain size and mental retardation (Mahmood et al., nism by which WDR62 regulates neurogenesis and thereby brain 2011). Interestingly, eight genes (MCPH1–MCPH8) underlying size are unknown. primary microcephaly identified so far are predicted to be asso- In this study, we explore the role of WDR62 in cortical develop- ciated with the function of mitotic apparatus, such as centro- ment. We find that knockdown of WDR62 expression in the rat

104 Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors A B

Figure 1. WDR62 Knockdown Results in Altered Cell Distribution and Loss of Radial Glial Fibers (A–C) Coronal sections from rat brains electroporated at E16.5 and inspected at E20.5. Slices were stained with Hoechst for nuclei. WDR62 shRNAs result in reduction of the fraction of cells in the CP and VZ/SVZ but an increase in the IZ. (A and C) Images of cortices electroporated with bicistronic constructs encoding both EGFP and shRNAs, including scrambled (Ctrl) or WDR62-directed small hairpin siRNAs (shW1, shW2, or shW3). Scale bars, 100 mm. (B) Quantification of cell distribution. Data are mean ± SEM. ***p < 0.0001, one-way ANOVA. Ctrl, n = 9; shW1, shW2, and shW3, n = 6. n, brain slice number from different brains. (D) Coronal sections from rat brains electroporated at E16.5 and inspected at E18.5. Slices were stained with GFP antibody in order to detect weak GFP signals. Yellow-boxed areas denote zoomed areas shown in the right two columns. Scale bar, 50 mm. See also Figures S1 and S2. embryonic cortex disturbs the redistribution of newborn neu- line) (Figures S1D–S1G). We used these shRNAs to deplete rons. Wild-type (WT) WDR62, but not the mutants found WDR62 expression in NPCs via in utero electroporation (IUE) in patients with MCPH, can rescue the defects incurred by at embryonic day 16.5 (E16.5) in rat. There were striking differ- WDR62 knockdown. We further demonstrate that WDR62 plays ences in the distribution of transfected cells between WDR62 an essential role in the proliferation and differentiation of NPCs shRNA- and control shRNA-transfected brains inspected through the regulation of JNK1 activity. 4 days (E20.5) or 8 days (postnatal day 2 [P2]) later (Figures 1A and S2A). In the knockdown brains (E20.5), very few cells RESULTS reached the CP, and significantly more cells resided in the IZ compared with controls (Figures 1A and 1B), which apparently WDR62 Knockdown Leads to Defects in NPC correlated well with the knockdown efficacy of shRNAs (Fig- Development that Can Be Rescued by WT WDR62 but ure S1). At P2, most of the WDR62 knockdown cells were Not Mutants Found in Patients with MCPH located in the white matter (WM) of the cerebrum instead (Fig- In order to examine the role of WDR62 in brain development, we ure S2A). More importantly, WDR62 depletion resulted in signif- first assessed the expression of WDR62. In agreement with a icant loss of cells remaining in the proliferative regions of the VZ/ previous report by Nicholas et al. (2010), WDR62 is expressed SVZ 4 days after IUE (Figures 1A–1C). We also did the IUE at E14 in the VZ and SVZ of developing rat forebrain and is located in and inspected the brains at E17.5 and observed similar defects the spindle poles of HEK293 cells in a cell-cycle-dependent as those electroporated at E16.5 and inspected at E20.5 manner (Figures S1A–S1C). We then screened bicistronic con- (Figure S2B). structs encoding both EGFP (or dsRed) and shRNA for WDR62 The decrease of cell number in the VZ/SVZ may be partially knockdown capacity. We found three different shRNAs (shW1, due to cell death because WDR62 knockdown led to a modest shW2, and shW3) capable of knocking down levels of WDR62 increase of cells positive for the activated form of caspase-3 efficiently in different cell types, including cortical neurons and (Figures S2C and S2D). Because neurons are derived from radial mouse neuroblastoma N2a cells (a neural progenitor-like cell glial cells that function as NPCs during brain development to

Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors 105 A Figure 2. WDR62 Knockdown-Incurred Defects Can Be Rescued by WT Human WDR62, but Not Mutants Found in Patients with MCPH (A) Diagram of human WDR62 and sites of different mutants used in (B). (B) WDR62 knockdown-incurred defects can be rescued by human WDR62. Plasmids expressing B either WT or different mutants of human WDR62 were coelectroporated with plasmids expressing control or WDR62 shRNA and analyzed as in Fig- ure 1A. Scale bar, 100 mm. (C) Conﬁrmation of coexpression of WT WDR62 (stained with WDR62 antibody and shown in red) with WDR62 shRNA (green). (D) Quantiﬁcation of cell distribution in (B). Data are mean ± SEM. ns, p > 0.05, *p < 0.05, **p < 0.01, and ***p < 0.0001, one-way ANOVA. Vector + Ctrl and Vector + shW2 n = 5; all others, n = 3. n, brain slice number from different brains. See also Figure S3. C

of the defects incurred by WDR62 knockdown (Figures 2B–2D). In addition, hWDR62 could rescue WDR62 knockdown-incurred defects in a dosage- dependent manner (Figure S3). This indicates that the function of WDR62 is conserved in human and rodents. In addition, these data show that the pheno- D type caused by the WDR62 shRNA was WDR62 dependent and was unlikely to be the result of off-target effects.

WDR62 Is Essential for the Coordinated Proliferation and Differentiation of Neural Progenitors Our above results suggest a role for WDR62 in NPC proliferation and differentiation. We went on to confirm this idea by examining the overlap of the GFP+-electroporated cell population with different progenitor cell markers including Sox2 establish radial units in neocortex (Noctor et al., 2001), we exam- and Pax6, markers for apical progenitor cells and radial glial ined the effect of WDR62 knockdown 2 days after IUE (E18.5) cells, and Tbr2, a marker for intermediate or basal progenitor before the depletion of cells in the VZ/SVZ. Knockdown of cells 2–4 days after IUE at either E14.5 or E16.5. WDR62 knock- WDR62 led to the disappearance of most radial glial fibers down led to a significant decrease of both Sox2+ and Pax6+ cells from the transfected cells (Figure 1D), but it did not have (Figures 3A, S4A, S4C, and S4D). Interestingly, depletion of apparent effects on cortical structure or on radial glial fibers of WDR62 led to increased proportion of Tbr2+ cells among those untransfected cells (Figure S2E). Therefore, WDR62 depletion cells in the VZ/SVZ 2 and 3 days after IUE (Figures 3B and is likely to cause the loss of NPC radial glial fibers cell S4E). However, the fraction of Tbr2+ cells decreased when all autonomously. cells in the whole cortex were counted 3 or 4 days after IUE Because mutations of WDR62 are linked to MCPH, we co- (Figures 3B and S4B). This suggests that WDR62 knockdown transfected WDR62 shRNAs with WT or five different mutated induces the premature differentiation of Sox2 and Pax6+ cells hWDR62s discovered in patients with MCPH (Nicholas et al., into Tbr2+ cells in the beginning and their depletion later. 2010)(Figure 2A). Mouse WDR62, which can be targeted by We went on to analyze the effect of WDR62 depletion on cell the WDR62 shRNAs, was used as a negative control. The WT cycle and observed a decrease of cells positive for Ki67, a hWDR62, but not the different mutants, provided strong rescue marker for proliferating cells, and a substantial decrease in

106 Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors A B

C D

(legend on next page) Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors 107 mitotic activity based on phosphor-histone H3 staining at E19.5, Katsenelson et al., 2011; Wasserman et al., 2010). We generated 3 days after IUE (Figures 3C and 3D). We also examined whether truncated mutants of WDR62 including WDR62 WD40 in which the decrease of NPCs would lead to their differentiation into neu- the MKK7/JNK binding domains and JNK phosphorylation sites rons. As shown in Figure S4F, there were significant increases of were deleted, and WDR62 WD40D in which the WD40 domains WDR62 knockdown cells positive for Tuj1, a marker for immature were deleted. WDR62 WD40 and WDR62 WD40D did not pro- neuron. In addition, we adopted a NeuroD promoter-reporter vide the apparent rescue of the defect caused by WDR62 knock- construct that drives GFP expression. Almost all the WDR62 down (Figure 5A). This indicates that the WD40 domains and knockdown cells expressed GFP, indicating their differentiation MKK7/JNK binding domains are essential for the normal function into neurons (Figure S4G). We also analyzed FoxP2, a marker of WDR62 during brain development. for deep-layer cortical neuron. WDR62 knockdown led to an in- WDR62 has been shown to be able to activate the JNK crease of FoxP2+ cells among those in the CP and a dramatic pathway (Bogoyevitch et al., 2012; Cohen-Katsenelson et al., decrease of FoxP2+ cells when all cells in the whole cortex 2011; Wasserman et al., 2010), and JNKs are important for brain were counted (Figure 3E). Finally, we examined the effect of development, although their role in NPC proliferation and differ- WDR62 depletion on cell cycle and observed a significant in- entiation in vivo remains unclear (Kuan et al., 1999; Sun et al., crease in cell-cycle exit, indicating an overall decrease in cell 2007; Westerlund et al., 2011). We therefore investigated the proliferation (Figure 4A). Taken together, these findings show relationship between the JNK pathway and WDR62 in brain that WDR62 depletion in NPCs inhibits their self-renewal and development. We mutated two amino acids in the JNK binding induces their premature differentiation. domain of WDR62 (PM6), which has been shown to abolish its ability to bind to JNKs (Bogoyevitch et al., 2012), and found WDR62 Knockdown Affects the Spindle Pole Structure that the mutant could not rescue the neurogenic defects incurred in the VZ/SVZ by knockdown of WDR62 (Figure 5A). In addition, this mutant Proper centrosome function and its inheritance are important for lost its ability to induce the phosphorylation of JNK, an indicator the maintenance of neural progenitors in neocortex (Feng and for JNK activation (Figure 5B). These results suggest that Walsh, 2004; Manzini and Walsh, 2011; Wang et al., 2009). WDR62-JNK interaction is necessary for WDR62 expression- Because WDR62 mutations have been proposed to affect cen- induced JNK activation and that JNK is likely to be involved in trosomal function (associated with spindle pole) (Bilgu¨ var et al., the normal function of WDR62 during brain development. 2010; Nicholas et al., 2010; Yu et al., 2010), we next tested We therefore screened bicistronic constructs encoding both whether WDR62 depletion affects spindle alignment. We first dsRed (or EGFP) and shRNA for JNK1 knockdown capacity arrested HeLa cells in G2/M phase with nocodazole and then and found two different shRNAs (shJ1-1 and shJ1-2) capable washed away the drug to release the arrest. We found a substan- of knocking down the expression of JNK1 (Figure S5A). Interest- tial increase of WDR62 knockdown cells with multipolar spindles ingly, JNK1 depletion with different shRNAs incurred very similar (Figure 4B), indicating that WDR62 depletion compromises spin- defects as knockdown of WDR62 during cortical development, dle organization during mitosis. We further analyzed this effect including the redistribution of most cells to the IZ (Figures 5C in vivo and inspected the NPCs in the VZ/SVZ of E18.5 brains and S5B). Similar to depletion of WDR62, JNK1 knockdown 2 days after IUE with WDR62 shRNA. We noticed the presence also led to a dramatic decrease of cells positive for Sox2 and a of multiple centrosomes in many NPCs and the decreased size significant increase of Tuj1+ cells (Figures 5D and 5E), suggest- of many centrosomes (Figure 4C). These results suggest that ing that JNK1 depletion may also inhibit the self-renewal and WDR62 leads to the appearance of abnormal spindle pole, which induce the premature differentiation of NPCs. In addition, knock- could be the result of abnormal centrosome biogenesis or aber- down of JNK1 also resulted in a significant increase of cells with rant mitosis. They also suggest that WDR62 may play a role in multipolar centrosomes after the nocodazole arrest and release cell cycle. To test this, we examined the role of WDR62 in neural procedure (Figure 5F). progenitor-like N2a cells. Knockdown of WDR62 led to a significant increase of cell population in the G1/G0 phase accompa- WDR62 Controls Neurogenesis through JNK1 Signaling nied by a decrease in both S and G2/M phases (Figure 4D). Because knockdown of WDR62 and JNK1 leads to very similar defects, we went on to explore the relationship between them. Knockdown of JNK1 Incurs Similar Defects as WDR62 In HeLa cells arrested in M phase with nocodazole, WDR62 Depletion during Cortical Development and phosphorylated JNK colocalized with each other mainly on WDR62 largely consists of the WD40 domain repeats and the a spindle pole-like structure (Figure 6A). To further confirm the MKK7/JNK binding domains (Bogoyevitch et al., 2012; Cohen- location of the two proteins, HeLa cells were lysed with detergent

Figure 3. WDR62 Depletion Decreases NPC Proliferation and Induces Its Differentiation (A–E) Coronal sections from rat brains electroporated at E14.5 and inspected at E17.5 (A and B) or electroporated at E16.5 and inspected at E19.5 (D) or E20.5 (C and E). (A–E) Images of cortices electroporated with control or shWDR62s and stained for SOX2 (A), TBR2 (B), Ki67 (C), P-H3 (D), or FoxP2 (E) are shown. Boxed areas denote zoomed areas shown in the upper columns (A, B, and C) or left columns (D). Arrows indicate examples of cells that are double positive for GFP and Ki67 (C), or P-H3 (D). Lower or right panels show quantiﬁcation of GFP+ cells that overlap with individual markers. The mitotic index of the GFP+ cell population was measured by P-H3 staining (D). All data in (A)–(E) are mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.0001, one-way ANOVA. All groups, n = 3. n, brain slice number from different brains. Scale bars, 100 mm (A–E). See also Figure S4.

108 Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors A

(legend on next page)

Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors 109 (the nucleus and centrosome structures were preserved) prior decrease of those between 60 and 90. The shift of the cleavage to fixation. WDR62 and phosphorylated JNK colocalized well plane of WDR62 and JNK1 knockdown NPCs indicates that on spots around the nuclei, implicating that both of them are those cells are likely to undergo asymmetric division, which will located on the centrosome (Figure 6B). We also inspected the lead to disturbance of cell proliferation and a subsequent deple- localization of activated JNK in the cortex and found that it colo- tion of the neural progenitor pool (Fish et al., 2006; Huttner calized well with g-tubulin in the VZ/SVZ (Figure 6C). This further and Kosodo, 2005; Knoblich, 2008; Wang et al., 2009). Taken supports the notion that the activated form of JNK is located on together, our results indicate that WDR62 controls the symmetric the centrosome. and asymmetric division of NPCs in the neocortex through the We have shown above that WDR62-JNK interaction is essen- regulation of JNK1 activity and probably by affecting its localiza- tial for WDR62-induced JNK activation. We postulated that tion on the spindle pole. WDR62 may also regulate the localization of phosphorylated JNK on the spindle. We depleted WDR62 and inspected DISCUSSION phosphorylated JNK in HeLa cells arrested in M phase. The phosphorylated JNK signals in WDR62 knockdown cells were Previous studies have shown the conserved centrosomal asso- reduced significantly, especially for those on the spindle struc- ciation of different MCPH proteins and the role of MCPH genes in ture (Figure 6D). Similarly, reduced P-JNK signal was also neurogenesis (Lizarraga et al., 2010; Manzini and Walsh, 2011; noticed in the VZ of cortex 2 days after IUE with shWDR62 at Thornton and Woods, 2009; Wollnik, 2010). Our study demon- E18.5 (Figure S6). This suggests that WDR62 may play a role in strates that knockdown of WDR62 results in redistribution of the regulation of JNK activity during brain development. most cells to the IZ, due to the depletion of NPCs and defects Because WDR62 plays a role in JNK activation and WDR62 in neuronal migration. We further show that WDR62 controls and JNK1 depletions incur very similar defects, it is therefore the proliferation and differentiation of NPCs within the context reasonable to predict that JNK1 may act downstream of of corticogenesis through the regulation of JNK activity. This WDR62 during brain development. To test this hypothesis, we may occur by affecting the location of activated JNK to orches- investigated whether the defects incurred by different WDR62 trate the proper formation of the spindle and by this means sym- shRNAs (shW1 and shW2) could be rescued by coexpression metric or asymmetric division of NPCs. of WT or constitutively active JNK1 (CA JNK1). As shown in Fig- The depletion of NPCs in the VZ/SVZ and the loss of radial ures 7A and S7, expression of CA JNK1 provided very significant glial fibers that take place after knockdown of WDR62 suggest rescue of WDR62 knockdown-induced NPC depletion in the VZ/ defects in the maintenance of cell proliferation and the occur- SVZ, whereas the WT JNK1 provided partial rescue. In addition, rence of premature neuronal differentiation. This is supported expression of CA JNK1 could also significantly suppress WDR62 by the findings of significantly decreased mitotic index and knockdown-incurred loss of apical NPCs (Sox2+, Figure 7B) and increased cell-cycle exit index in WDR62-depletion NPCs. A increase immature neurons (Tuj1+, Figure 7C). This suggests that substantial decrease of Pax6+, Sox2+, and Tbr2+ cells is accom- WDR62 plays an important role in cortical development through panied by an increase of Tuj1+ neurons and disturbance of the regulation of JNK signaling. Foxp2+ neurons after WDR62 knockdown. Thus, our findings Spindle orientation has been indicated in the control of cell fate indicate a role for WDR62 in the maintenance of both apical of NPCs in the cortex (Huttner and Kosodo, 2005; Manzini and and basal progenitor pools during neurogenesis. In addition, Walsh, 2011; Wang et al., 2011), and we have shown above WDR62 depletion may affect neuronal lineage formation during that both WDR62 and JNK1 play a role in spindle formation. cortical development. We therefore investigated the role of WDR62 and JNK1 in the Depletion or mutations of many MCPH proteins have been regulation of mitotic orientation of NPCs, an indicator for sym- shown to lead to the pulling of NPCs out of cell cycle prema- metric or asymmetric division. As shown in Figure 7D, depletion turely and consequently to premature neuronal differentia- of either WDR62 or JNK1 led to a dramatic increase of cells with tion or cell death during neurogenesis (Buchman et al., 2011; cleavage planes of small angles crossing the apical surface, Gruber et al., 2011). We explored the potential underlying particularly between 0 and 30 accompanied by a significant mechanism by which WDR62 affects the coordinated

Figure 4. WDR62 Regulates Spindle Formation in NPCs (A) Images of the VZ/SVZ cortices electroporated with Ctrl, shW1, or shW2 at E16.5 and stained for GFP, Ki67, and BrdU 24 hr after BrdU labeling at E19.5. Arrowheads mark GFP, BrdU-double positive but Ki67-negative cells. Arrows mark GFP, BrdU, Ki67 triple-positive cells. Right panel shows cell-cycle exit index measured at E20.5. Data are mean ± SEM. ***p < 0.0001, one-way ANOVA. For each group, three brain slices from different brains were analyzed. (B) Effect of WDR62 depletion on mitotic spindle morphology. HeLa cells transfected with Ctrl or shWDR62 were arrested in G2/M phase and then released as described in Experimental Procedures. Cells were fixed and stained with a-tubulin antiserum and Hoechst (left panels). Right panel shows quantification of cells with multipolar spindles in Ctrl (n = 65) and WDR62 shRNA (n = 80) transfected cells from two independent experiments. Data are mean ± SEM. *p = 0.012, t test. (C) WDR62 knockdown affects mitotic spindle formation. Coronal sections from brains electroporated at E16.5 and inspected at E18.5 are shown. Slices were stained with Hoechst, GFP antibody for transfected cells (green) and g-tubulin antibody for centrosomes (red). Examples of cells with three centrosomes are outlined in the shW1-transfected brain. (D) Cell-cycle profiles of N2a cells transfected with either Ctrl or shW1 (left panels). Right panel shows quantification of cells transfected with Ctrl, shW1, or shW2 in different cell-cycle phase as analyzed by FACS. Data are mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.0001, one-way ANOVA. ns, p > 0.05. Results are from three independent experiments. Scale bars, 100 mm (A), 10 mm (B), and 20 mm (C).

110 Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors A

B C

D E

Figure 5. JNK1 and WDR62 Knockdowns Have Similar Phenotypes, and Interaction with JNK1 Is Important for WDR62 Function (A) Interaction with JNK1 is important for WDR62 function. WDR62 mutants including WDR62 WD40 (aa 1–743), WDR62 WD40D (aa 743–1,523), or PM6 (WDR62 L1299A/L1301A) were electroporated together with control or WDR62 shRNA and analyzed as in Figure 2. Right panel shows quantiﬁcation of cell distribution. Data are mean ± SEM. ns, p > 0.05, *p = 0.036, **p = 0.007, and ***p = 0.0002, one-way ANOVA. Vector+Ctrl and Vector+shW2, n = 5; all others, n = 3. (B) Interaction with JNK1 is required for WDR62-induced JNK activation. HEK293 cells were transfected with pCMS.EGFP, pCMS.EGFP.WDR62, or pCMS.EGFP.WDR62 PM6. Thirty-six hours later, cells were collected, and JNK activity (P-JNK) was determined by western blotting. Similar results were obtained from three experiments. (legend continued on next page)

Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors 111 proliferation and differentiation of NPCs. The presence of mul- deletion of Jnk1, Jnk2,orJnk3 survives without overt structural tiple spindle poles and centrosomes and disturbed mitotic abnormalities (Kuan et al., 1999). Although functional redun- orientation (increase of asymmetric divisions) in WDR62 knock- dancy between JNK1 and JNK2 may exist (Kuan et al., 1999), down NPCs indicates the deregulation of mitosis. During the it will be helpful to investigate cortical neurogenesis especially revision of this manuscript, WDR62 depletion-incurred malfor- NPC differentiation more thoroughly in jnk1 KO mice via methods mation of spindle poles in a cultured cell line was also reported adopted recently. by Bogoyevitch et al. (2012). Although we have detected an Although our study uncovers a functional interaction between increase of cell death in WDR62-depleted neurons, it is not WDR62 and JNK1, further studies will be necessary to elucidate very significant. WDR62 is therefore more likely to play a crucial the molecular mechanisms by which WDR62 regulates the JNK role mainly in the timing of differentiation, and the other pathway. We have shown previously that POSH and JIPs directly observed WDR62 knockdown-induced defects may be a result associate with one another as well as components of the JNK of this mistiming. pathway to form a multiprotein complex (PJAC) and play an Several microcephaly-associated proteins such as MCPH1 important role in JNK activation (Kukekov et al., 2006; Xu et al., and ASPM have been shown to regulate the Chk1-Cdc25b 2005). We reported very recently that POSH controls the migra- and Wnt signaling pathways, respectively, during neurogenesis tion of neurons during cortical development (Yang et al., 2012). It in vivo (Buchman et al., 2011; Gruber et al., 2011). Because will therefore be intriguing to investigate in the future whether WDR62 serves as a scaffold for the JNK pathway (Bogoyevitch POSH can coordinate with WDR62 to form a complex to regulate et al., 2012; Cohen-Katsenelson et al., 2011), we investigated the JNK pathway and by this means influence different aspects the role of JNKs in the normal function of WDR62. Our results of cortical neurogenesis such as NPC proliferation, differentia- with deletion of the MKK7/JNK binding domains and JNK bind- tion, and neuronal migration. ing domain point mutations (PM6) indicate that interaction We investigated the effect of mutations in human WDR62 that with JNK is essential for WDR62’s ability to activate JNK and are found in patients with autosomal recessive microcephaly. for its normal function during brain development. Interestingly, The five mutants could not effectively rescue the defects induced a WDR62 mutant found in patients with MCPH (PM5) in which by WDR62 knockdown compared with WT WDR62. This indi- the JNK phosphorylation sites are deleted could not rescue cates that these mutations are very likely to result in loss of the defect incurred by WDR62 depletion. This implies a potential function and that transgenic mice and rats with knockin of these positive feedback loop between WDR62 and JNK activity, mutants will be valuable animal models of autosomal recessive similar to what we have reported previously between the JNK microcephaly. pathway scaffold proteins, POSH and JIPs (JNK-interacting proteins), and JNK signaling (Kukekov et al., 2006; Xu et al., EXPERIMENTAL PROCEDURES 2005). Through knockdown of different JNK isoforms, we found that Materials JNK1 and WDR62 knockdowns have very similar phenotypes, Antibodies used for immunohistochemistry were as follows: rabbit anti- WDR62 was from Bethyl Laboratories; rabbit anti-Ki67, rabbit anti-sox2, rabbit including defects in neuronal distribution and premature differ- anti-Pax6, rabbit anti-Caspase-3, mouse anti-Tuj1, rat anti-BrdU, rat anti- entiation of NPCs during cortical development, formation of Phosphor-H3, rabbit anti-Foxp2, mouse anti-a-tubulin and anti-g-tubulin, abnormal spindle poles, as well as the mitotic division mode of and chicken anti-GFP were from Abcam; rabbit anti-Tbr2 was from Millipore; NPCs. Because WDR62 knockdown-induced NPC depletion and rabbit antiactivated JNK was from Promega. The following antibodies and premature neuronal differentiation could be partially rescued were used for western blot: rabbit anti-WDR62 was made by MBL International by WT JNK1 and very significantly by a CA JNK1, JNK1 is very (B&M Biotech), rabbit anti-GAPDH and mouse anti-Phospho-SAPK/JNK were likely to act downstream of WDR62 during cortical development. from Cell Signaling Technology, rabbit anti-GFP was from Abcam, and mouse anti-a-tubulin was from Sigma-Aldrich. It is interesting to notice that WT JNK1 and CA JNK1 cannot rescue the migration defect incurred by shWDR62. There are DNA Constructs two possibilities for this. One is that the JNK1 effect could be in- Human and mouse WDR62 was created by cloning WDR62 cDNA sequence dependent of WDR62. Another very likely possibility is that JNK1 into the pCMS.EGFP vector. Point mutations of human WDR62 were plays a role in inhibiting NPC differentiation. Expression of JNK1 created with the QuikChange Site-Directed Mutagenesis Kit. The primers in WDR62 knockdown cell can rescue the depletion of NPCs used to amplify WD40 domain and WD40D domain were as follows: 0 0 0 0 when there is still residual WDR62 in the beginning of transfec- 5 -ATGGCGGCCGTAGGGT-3 and 5 -CAGTGCCAGATGAACACGCA-3 , and 50-ATGCTGGGCCCGGAGATCA-30 and 50-TCAGTGCCCCCGTGC-30, tion, but it cannot rescue the migration defect because WDR62 respectively. WT and CA JNK1 in pCDNA3.1 were gifts from Dr. Roger Davis. is depleted later during NPC development after the cells left Small hairpin constructs were cloned into either pSiren-RetroQ-DsRed or the VZ/SVZ. The Jnk1 and Jnk2 dual-deficient mouse embryos pLentiLox 3.7. Sequence for scrambled (control) shRNA was 50-TCATGCCTC exhibit exencephaly and die around E11–E12, whereas single TATCTACGTC-30. WDR62 small hairpin constructs containing the following

(C) Coronal sections from brains electroporated at E15 and inspected at E19. Quantiﬁcation of cell distribution is shown in the right panel. Data are mean ± SEM. ***p < 0.0001, one-way ANOVA. All groups, n = 4. (D and E) Control or JNK1 shRNAs were transfected and analyzed as in Figure 3. Data are mean ± SEM. *p < 0.05 and **p < 0.007, t test. All groups, n = 4. (F) Quantiﬁcation of cells with multipolar spindles in Ctrl (n = 144) and JNK1 shRNA (n = 107)-transfected cells (analyzed as in Figure 4A). Data are mean ± SEM. *p = 0.02, t test. Scale bars, 100 mm (A and C), 20 mm (D), 50 mm (E), and 10 mm (F). n, brain slice number from different brains (A and C–E). See also Figure S5.

112 Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors A Figure 6. WDR62 Regulates JNK Activity and Localization (A) Activated JNK colocalizes with WDR62. HeLa cells were arrested in G2/M phase as in Figure 4A. Fixed cells were stained with phospho JNK antibody (P-JNK; green), WDR62 antibody (red), and DAPI (blue). (B) Activated JNK colocalizes with WDR62 on a spindle pole-like structure. HeLa cells were lysed with 0.5% Triton X-100 for 2 min. The detergent- insoluble cellular structures including the nucleus and spindle pole were fixed and stained for WDR62 (red) and P-JNK (green). (C) P-JNK colocalizes with g-tubulin (Tub) in the B VZ/SVZ of E18 cerebral cortex. Zoom-in views of rectangles are shown in the right panels. (D) WDR62 regulates JNK activity. HeLa cells were transfected, arrested in M phase, and fixed as in Figure 4A. Cells were stained with P-JNK antibody (red) and DAPI (blue). Cells arrested in M phase are labeled with arrows (shRNA-transfected cells are in green; untransfected cells are in white). C Right panel shows quantification of relative P-JNK fluorescence signal intensity of cells in M phase (transfected cells/untransfected cells) (Ctrl, transfected cells [n = 12]; untransfected cells [n = 8]; shW1, transfected cells [n = 8]; untransfected cells [n = 10]). Data are mean ± SEM. **p = 0.0038, t test. All scale bars, 10 mm. See also Figure S6.

Cell Culture, Transfection, Synchronization, FACS Analysis, and Western Blotting Primary neuron cells were cultured as described previously with minor modification (Sun et al., 2010). Briefly, telencephalons from E16 Sprague- Dawley rats were mechanically dissected and cultured in B27 supplement (GIBCO) DMEM. All cell lines used were cultured in DMEM with 10% D FBS. Transfection and western blotting were performed as previously described (Xu et al., 2003). G2/M synchronization was performed as previously described by Whitfield et al. (2002). Mitotic HeLa cells were analyzed at the next synchronous M phase around 18–20 hr later. For FACS analysis, transfected cells were collected and washed with PBS. Hoechst 33342 was added to cells to the final concentration of 10 mg/ml and incubated at 37C for 45 min. A total of 20,000 GFP+ cells were collected and analyzed by flow cytometer (BD; FACSCanto II) per sample.

IUE and Cell-Cycle Exit Assay Pregnant Sprague-Dawley rats were provided by the animal center of the Institute of Genetics target sequences were used: shW1, 50-GAGAAGGTGCTTGGCATCA-30; and Developmental Biology (IGDB), Chinese Academy of Sciences. shW2, 50-GCTACCAACATGACATGG-30; and shW3, 50- GTGGAAATCTGAG All experimental procedures involved were performed according to proto- GATCCA-30. JNK1 small hairpin constructs containing the following target cols approved by the Institutional Animal Care and Use Committee at sequences were used: J1-1, 50- ATCATGAAAGAATGTCCTAC-30; and J1-2, IGDB. IUE was performed as described previously (Sun et al., 2010; Yang 50-GGTGCATTATGGGAGAAAT-30. qPCR primers for WDR62 were 50- ACCA et al., 2012). For rescue analysis, WDR62 shRNAs were electroporated TCCGCTTCTGGAATATG-30 and 50-CTGGATGTCGTTCTCCACATAG-30. with double the amount of rescue constructs. For cell-cycle exit experi- Primers for b-actin were 50-AGGGAAATCGTGCGTGAC-30 and 50-GATAGT ments, E16.5 electroporated rats were intraperitoneally injected with BrdU GATGACCTGACCGT-30. (50 mg/kg) at E19.5, sacriﬁced, and ﬁxed 24 hr later. Brain slices were

Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors 113 A

(legend on next page)

114 Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors treated with 1 N HCl for 10 min on ice and 30 min at 37C prior to Feng, Y.Y., and Walsh, C.A. (2004). Mitotic spindle regulation by Nde1 controls immunostaining. cerebral cortical size. Neuron 44, 279–293. Fish, J.L., Kosodo, Y., Enard, W., Pa¨ a¨ bo, S., and Huttner, W.B. (2006). Aspm Brain Sectioning, Immunohistochemistry, Confocal Imaging, and specifically maintains symmetric proliferative divisions of neuroepithelial cells. Quantification of Cells Proc. Natl. Acad. Sci. USA 103, 10438–10443. Embryonic brain sectioning and immunohistochemistry were performed using Franco, S.J., and Mu¨ ller, U. (2013). Shaping our minds: stem and progenitor a standard protocol as described before (Sun et al., 2010). Sections were cell diversity in the mammalian neocortex. Neuron 77, 19–34. imaged on an LSM 700 (Carl Zeiss) confocal microscope as described by Yang et al. (2012). Cell counts were analyzed with Imaris 364 or ImageJ. All Gruber, R., Zhou, Z., Sukchev, M., Joerss, T., Frappart, P.O., and Wang, Z.Q. data were analyzed using Prism software (GraphPad). Tests used were two- (2011). MCPH1 regulates the neuroprogenitor division mode by coupling the tailed t test and one-way ANOVA paired with Tukey posttest. centrosomal cycle with mitotic entry through the Chk1-Cdc25 pathway. Nat. Cell Biol. 13, 1325–1334. SUPPLEMENTAL INFORMATION Guernsey, D.L., Jiang, H., Hussin, J., Arnold, M., Bouyakdan, K., Perry, S., Babineau-Sturk, T., Beis, J., Dumas, N., Evans, S.C., et al. (2010). Mutations Supplemental Information includes seven figures and can be found with this in centrosomal protein CEP152 in primary microcephaly families linked to article online at http://dx.doi.org/10.1016/j.celrep.2013.12.016. MCPH4. Am. J. Hum. Genet. 87, 40–51. Hussain, M.S., Baig, S.M., Neumann, S., Nu¨ rnberg, G., Farooq, M., Ahmad, I., ACKNOWLEDGMENTS Alef, T., Hennies, H.C., Technau, M., Altmu¨ ller, J., et al. (2012). A truncating mutation of CEP135 causes primary microcephaly and disturbed centrosomal We would like to thank Drs. L.A. Greene, Z. Wang, and Y. Zhang for their function. Am. J. Hum. Genet. 90, 871–878. constructive advice and editing of the manuscript, Dr. D. Ng for providing Huttner, W.B., and Kosodo, Y. (2005). Symmetric versus asymmetric cell divi- human WDR62 cDNA, Dr. L.H. Tsai for pNeuroD-GFP plasmid, Dr. R. Davis sion during neurogenesis in the developing vertebrate central nervous system. for WT and CA JNK1 constructs, and Dr. N. Jing for neuroblastoma N2a cell Curr. Opin. Cell Biol. 17, 648–657. line. This work was supported by grants from the Chinese Academy of Sci- ences (XDA01010105), the MOST (China) (2014CB942801/2012CB517904/ Kaindl, A.M., Passemard, S., Kumar, P., Kraemer, N., Issa, L., Zwirner, A., 2012YQ03026006), the NSF (China) (31271156/31270826), and BIBD- Gerard, B., Verloes, A., Mani, S., and Gressens, P. (2010). Many roads lead PXM2013_014226_07_000084. to primary autosomal recessive microcephaly. Prog. Neurobiol. 90, 363–383. Kang, E., Burdick, K.E., Kim, J.Y., Duan, X., Guo, J.U., Sailor, K.A., Jung, D.E., Received: July 27, 2012 Ganesan, S., Choi, S., Pradhan, D., et al. (2011). Interaction between FEZ1 and Revised: October 23, 2013 DISC1 in regulation of neuronal development and risk for schizophrenia. Accepted: December 11, 2013 Neuron 72, 559–571. Published: January 2, 2014 Knoblich, J.A. (2008). Mechanisms of asymmetric stem cell division. Cell 132, 583–597. REFERENCES Kuan, C.Y., Yang, D.D., Samanta Roy, D.R., Davis, R.J., Rakic, P., and Flavell, R.A. (1999). The Jnk1 and Jnk2 protein kinases are required for regional Bilgu¨ var, K., Oztu¨ rk, A.K., Louvi, A., Kwan, K.Y., Choi, M., Tatli, B., Yalnizoglu, specific apoptosis during early brain development. Neuron 22, 667–676. D., Tu¨ ysu¨ z, B., Caglayan, A.O., Go¨ kben, S., et al. (2010). Whole-exome sequencing identifies recessive WDR62 mutations in severe brain malforma- Kukekov, N.V., Xu, Z., and Greene, L.A. (2006). Direct interaction of the tions. Nature 467, 207–210. molecular scaffolds POSH and JIP is required for apoptotic activation of Bogoyevitch, M.A., Yeap, Y.Y., Qu, Z., Ngoei, K.R., Yip, Y.Y., Zhao, T.T., Heng, JNKs. J. Biol. Chem. 281, 15517–15524. J.I., and Ng, D.C. (2012). WD40-repeat protein 62 is a JNK-phosphorylated Lizarraga, S.B., Margossian, S.P., Harris, M.H., Campagna, D.R., Han, A.P., spindle pole protein required for spindle maintenance and timely mitotic pro- Blevins, S., Mudbhary, R., Barker, J.E., Walsh, C.A., and Fleming, M.D. gression. J. Cell Sci. 125, 5096–5109. (2010). Cdk5rap2 regulates centrosome function and chromosome segrega- Buchman, J.J., Durak, O., and Tsai, L.H. (2011). ASPM regulates Wnt signaling tion in neuronal progenitors. Development 137, 1907–1917. pathway activity in the developing brain. Genes Dev. 25, 1909–1914. Mahmood, S., Ahmad, W., and Hassan, M.J. (2011). Autosomal Recessive Cohen-Katsenelson, K., Wasserman, T., Khateb, S., Whitmarsh, A.J., and Primary Microcephaly (MCPH): clinical manifestations, genetic heterogeneity Aronheim, A. (2011). Docking interactions of the JNK scaffold protein and mutation continuum. Orphanet J. Rare Dis. 6, 39. WDR62. Biochem. J. 439, 381–390. Manzini, M.C., and Walsh, C.A. (2011). What disorders of cortical development Dehay, C., and Kennedy, H. (2007). Cell-cycle control and cortical develop- tell us about the cortex: one plus one does not always make two. Curr. Opin. ment. Nat. Rev. Neurosci. 8, 438–450. Genet. Dev. 21, 333–339.

Figure 7. WDR62 Controls Neurogenesis through JNK1 Signaling (A) WDR62 knockdown-induced NPC developmental defects can be rescued by CA JNK1. Control or WDR62 shRNAs were cotransfected with vector or CA JNK1 and analyzed as in Figure 2. Quantiﬁcation of cell distribution is shown in the right panel. Data are mean ± SEM. *p < 0.05 and ***p < 0.0001, one-way ANOVA (n = 7 for all groups). ns, p > 0.05. (B and C) WDR62 knockdown-incurred NPC depletion (B) and premature differentiation (C) can be rescued by CA JNK1. Control or WDR62 shRNAs were cotransfected with vector or CA JNK1 and analyzed as in Figure 3. Data are mean ± SEM. ns, p > 0.05, *p < 0.05, and ***p < 0.0001, one-way ANOVA (B, n = 4; C, n = 3 for all groups). n, brain slice number from different brains. Scale bars, 100 mm. (D) WDR62 and JNK1 knockdown affects mitotic division. Coronal sections from brains electroporated at E16.5 and inspected at E18.5 are shown. Slices were stained with Hoechst for nucleus, GFP antibody for transfected cells (green). Examples of cells undergoing symmetric or asymmetric divisions are outlined. Lower panel shows quantiﬁcation of the orientation of transfected cells undergoing mitotic division. Data are mean ± SEM. ns, p > 0.05; **p < 0.01; and ***p < 0.001 one- way ANOVA. Ctrl n = 3, average of 24 transfected cells counted/animal; shW1 n = 3, average of 19 transfected cells counted/animal; shJ1-1 n = 3, average of17 transfected cells counted/animal. Results are from two independent experiments. See also Figure S7.

Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors 115 Marıń, O., and Rubenstein, J.L. (2003). Cell migration in the forebrain. Annu. Wasserman, T., Katsenelson, K., Daniliuc, S., Hasin, T., Choder, M., and Aron- Rev. Neurosci. 26, 441–483. heim, A. (2010). A novel c-Jun N-terminal kinase (JNK)-binding protein WDR62 Nicholas, A.K., Khurshid, M., De´ sir, J., Carvalho, O.P., Cox, J.J., Thornton, G., is recruited to stress granules and mediates a nonclassical JNK activation. Kausar, R., Ansar, M., Ahmad, W., Verloes, A., et al. (2010). WDR62 is associ- Mol. Biol. Cell 21, 117–130. ated with the spindle pole and is mutated in human microcephaly. Nat. Genet. Westerlund, N., Zdrojewska, J., Padzik, A., Komulainen, E., Bjo¨ rkblom, B., 42, 1010–1014. Rannikko, E., Tararuk, T., Garcia-Frigola, C., Sandholm, J., Nguyen, L., et al. Noctor, S.C., Flint, A.C., Weissman, T.A., Dammerman, R.S., and Kriegstein, (2011). Phosphorylation of SCG10/stathmin-2 determines multipolar stage A.R. (2001). Neurons derived from radial glial cells establish radial units in exit and neuronal migration rate. Nat. Neurosci. 14, 305–313. neocortex. Nature 409, 714–720. Whitfield, M.L., Sherlock, G., Saldanha, A.J., Murray, J.I., Ball, C.A., Alexander, Noctor, S.C., Martıńez-Cerden˜ o, V., Ivic, L., and Kriegstein, A.R. (2004). K.E., Matese, J.C., Perou, C.M., Hurt, M.M., Brown, P.O., and Botstein, D. Cortical neurons arise in symmetric and asymmetric division zones and (2002). Identification of genes periodically expressed in the human cell cycle migrate through specific phases. Nat. Neurosci. 7, 136–144. and their expression in tumors. Mol. Biol. Cell 13, 1977–2000. Sun, Y., Yang, T., and Xu, Z. (2007). The JNK pathway and neuronal migration. Wollnik, B. (2010). A common mechanism for microcephaly. Nat. Genet. 42, J. Genet. Genomics 34, 957–965. 923–924. Sun, Y.M., Fei, T., Yang, T., Zhang, F., Chen, Y.G., Li, H.S., and Xu, Z.H. (2010). Xu, Z.H., Kukekov, N.V., and Greene, L.A. (2003). POSH acts as a scaffold for The suppression of CRMP2 expression by bone morphogenetic protein a multiprotein complex that mediates JNK activation in apoptosis. EMBO J. (BMP)-SMAD gradient signaling controls multiple stages of neuronal develop- 22, 252–261. ment. J. Biol. Chem. 285, 39039–39050. Xu, Z., Kukekov, N.V., and Greene, L.A. (2005). Regulation of apoptotic c-Jun Thornton, G.K., and Woods, C.G. (2009). Primary microcephaly: do all roads N-terminal kinase signaling by a stabilization-based feed-forward loop. Mol. lead to Rome? Trends Genet. 25, 501–510. Cell. Biol. 25, 9949–9959. Tsai, J.W., Chen, Y., Kriegstein, A.R., and Vallee, R.B. (2005). LIS1 RNA inter- Yang, T., Sun, Y., Zhang, F., Zhu, Y., Shi, L., Li, H., and Xu, Z. (2012). POSH ference blocks neural stem cell division, morphogenesis, and motility at multi- localizes activated Rac1 to control the formation of cytoplasmic dilation of ple stages. J. Cell Biol. 170, 935–945. the leading process and neuronal migration. Cell Rep. 2, 640–651. Wang, X., Tsai, J.W., Imai, J.H., Lian, W.N., Vallee, R.B., and Shi, S.H. (2009). Yu, T.W., Mochida, G.H., Tischfield, D.J., Sgaier, S.K., Flores-Sarnat, L., Sergi, Asymmetric centrosome inheritance maintains neural progenitors in the C.M., Topç u, M., McDonald, M.T., Barry, B.J., Felie, J.M., et al. (2010). Muta- neocortex. Nature 461, 947–955. tions in WDR62, encoding a centrosome-associated protein, cause micro- Wang, X., Lui, J.H., and Kriegstein, A.R. (2011). Orienting fate: spatial regula- cephaly with simplified gyri and abnormal cortical architecture. Nat. Genet. tion of neurogenic divisions. Neuron 72, 191–193. 42, 1015–1020.

116 Cell Reports 6, 104–116, January 16, 2014 ª2014 The Authors