A FOXO-Pak1 transcriptional pathway controls neuronal polarity

Abstract

Neuronal polarity is essential for normal brain development and function. However, cell-intrinsic mechanisms that govern the establishment of neuronal polarity remain to be identified. Here, we report that knockdown of endogenous FOXO proteins in hippocampal and cerebellar granule neurons, including in the rat cerebellar cortex in vivo, reveals a requirement for the FOXO transcription factors in the establishment of neuronal polarity. The FOXO transcription factors, including the brain-enriched protein FOXO6, play a critical role in axo-dendritic polarization of undifferentiated neurites, and hence in a switch from unpolarized to polarized neuronal morphology. We also identify the gene encoding the protein kinase Pak1, which acts locally in neuronal processes to induce polarity, as a critical direct target gene of the FOXO transcription factors. Knockdown of endogenous Pak1 phenocopies the effect of FOXO knockdown on neuronal polarity. Importantly, exogenous expression of Pak1 in the background of FOXO knockdown in both primary neurons and postnatal rat pups in vivo restores the polarized morphology of neurons. These findings define the FOXO proteins and Pak1 as components of a cell-intrinsic transcriptional pathway that orchestrates neuronal polarity, thus identifying a novel function for the FOXO transcription factors in a unique aspect of neural development.

Figure 1.

FOXO transcription factors establish neuronal polarity in cerebellar granule neurons. (A) Granule neurons were electroporated before plating using the Amaxa nucleofection kit with the control U6 or U6/foxo RNAi plasmid. Four days after transfection, lysates were subjected to immunoblotting with a FOXO1, FOXO3, or FOXO6 antibody. FOXO RNAi substantially reduced levels of endogenous FOXO1, FOXO3, and FOXO6 in neurons. The asterisk indicates nonspecific band. (B) Cerebellar granule neurons transfected with the control U6 or U6/foxo RNAi plasmid and a GFP expression plasmid were subjected 4 d after transfection to immunocytochemistry with an antibody to GFP (see Supplemental Fig. 3 for additional lower-magnification panels). Arrows, arrowheads, and asterisks indicate dendrites, axons, and cell body, respectively. Bar, 50 μm. (C) Granule neurons transfected and analyzed as in B were scored as polarized or nonpolarized. FOXO knockdown significantly increased the number of neurons that fail to acquire a polarized morphology (P < 0.01; t-test, n = 3). (D–F) Granule neurons were transfected with the Amaxa electroporation device with the control U6 or U6/foxo RNAi plasmid and the GFP expression plasmid and grown at low density. Five days after transfection, neurons were subjected to immunocytochemistry with the GFP antibody and an antibody to the dendritic marker MAP2 (D) or the axonal marker Tau1 (E). Enrichment of Tau1 and MAP2 was quantified in F. Tau1 and MAP2 enrichment are defined as the intensity of Tau1 or MAP2 immunostaining in the longest neurite divided by the intensity in the second-longest neurite. FOXO knockdown neurons displayed significantly increased MAP2 enrichment (P < 0.001; t-test, n = 3) and significantly reduced Tau1 enrichment (P < 0.01; t-test, n = 3) when compared with control U6-transfected neurons. Arrowheads and arrows point to the longest process and other processes, respectively. Asterisks indicate cell bodies. (G) Morphometric analysis of granule neurons transfected as in B revealed that FOXO RNAi significantly reduced the length of the longest process (axon in control), and concomitantly increased the length of secondary processes (dendrites in control) (P < 0.001; t-test, 213 neurons measured). (H) Lysates of 293T cells transfected with the control U6 or U6/foxo RNAi plasmid together with an expression vector encoding GFP-tagged FOXO6 (FOXO6-WT) or the RNAi-resistant mutant FOXO6 (FOXO6-Res) were subjected to immunoblotting with the GFP antibody (top panel) or an antibody to ERK1/2 (bottom panel). (I–K) Granule neurons transfected with the control U6 or U6/foxo RNAi plasmid, together with the FOXO6-Res expression plasmid or its control vector and an expression plasmid encoding DsRed, were subjected 4 d after transfection to immunocytochemistry with an antibody to DsRed. FOXO6-Res significantly reduced the percentage of nonpolarized neurons in the background of FOXO RNAi (P < 0.01; ANOVA, n = 3). The length of the longest process (axon in control) was significantly reduced and the length of secondary processes (dendrites in control) was significantly increased upon FOXO RNAi (P < 0.001; ANOVA, 200 neurons measured), but not in FOXO6-Res-expressing neurons in the background of FOXO knockdown, when compared with control U6-transfected neurons. Arrows, arrowheads, and asterisks indicate dendrites, axons, and cell body, respectively. Bar, 50 μm.

Figure 2.

FOXO transcription factors play a critical role in the switch from nonpolarized to polarized morphology in neurons. (A) Granule neurons transfected with the control U6 or U6/foxo RNAi plasmid and the GFP expression plasmid were scored as polarized or nonpolarized at each time point. While a majority of control neurons exhibited a polarized morphology at 2 d in vitro (DIV2), FOXO RNAi-transfected neurons failed to polarize over time. (B) Granule neurons plated on etched coverslips were transfected 8 h later with the U6 control or U6/foxo RNAi plasmid together with the GFP expression plasmid. Twenty hours after plating, individual live neurons were imaged in 12-h intervals over the course of 86 h. Nonpolarized control neurons acquired a polarized morphology within the first 36 h of observation. In contrast, FOXO knockdown neurons failed to polarize in the same amount of time. Arrows, arrowheads, and asterisks indicate dendrites, axons, and cell body, respectively. Bar, 50 μm. (C,D) Quantification of the developmental stage of individual neurons transfected and analyzed as in B. Neurons were grouped into five different morphological developmental stages, as described by Powell et al. (1997), with some modification. Stages 1–2 represent nonpolarized neurons bearing no neurites (stage 1), or several unspecified processes (stage 2). Stages 3–5 designate polarized neurons, including bipolar neurons bearing two axon-like processes (stage 3), multipolar neurons with an axon and short dendrites (stage 4), and multipolar complex neurons with long axons and elaborate dendritic arbors (stage 5). The majority of control neurons (85%) starting at stages 1–2 reached the polarized stages 4–5 by 5 d in vitro (DIV5), while a large proportion of FOXO knockdown neurons (68%) remained in stage 2. Both control and FOXO knockdown neurons that had already acquired a polarized morphology at the beginning of the analysis remained polarized throughout the course of observation. (DIV) Days in vitro.

Figure 3.

FOXO transcription factors promote axo–dendritic polarization in hippocampal neurons. (A–E) Hippocampal neurons were transfected with the control U6 or U6/foxo RNAi plasmid and the GFP expression plasmid. Four days after transfection, neurons were subjected to immunocytochemistry with the GFP antibody and Tau1 (B) or MAP2 (C) antibody. The percentage of neurons that failed to acquire a polarized morphology is quantified in A. Enrichment of Tau1 and MAP2 was quantified in D and E, respectively. A significant proportion of FOXO knockdown neurons failed to acquire a polarized morphology (P < 0.01; t-test, n = 3), and displayed significantly reduced Tau1 enrichment (P < 0.0001; t-test, 40 neurons measured) and significantly increased MAP2 enrichment (P < 0.0001; t-test, 57 neurons measured) when compared with control U6-transfected neurons. Arrowheads and arrows point to longest process and secondary processes, respectively. The asterisks indicate cell bodies. The double dagger points to the Tau1-positive axons of untransfected neurons. (F,G) Hippocampal neurons were transfected with the control U6 or U6/foxo RNAi plasmid together with the FOXO6-Res and GFP expression plasmids and were analyzed as in A. FOXO6-Res significantly reversed the FOXO RNAi-induced neuronal polarity phenotype (P < 0.05; ANOVA, n = 3). Arrowheads and arrows point to the longest process and other processes, respectively. (H–J) Hippocampal neurons were transfected as in F and, 4 d later, were analyzed as in B–E. Enrichment of Tau1 and MAP2 was quantified in H and I, respectively. A significant proportion of FOXO knockdown neurons failed to acquire a polarized morphology (P < 0.05; ANOVA, n = 3), and displayed significantly reduced Tau1 enrichment (P < 0.0001; ANOVA, 51 neurons measured) and significantly increased MAP2 enrichment (P < 0.0001; ANOVA, 57 neurons measured) compared with control U6-transfected neurons. These phenotypes were significantly reversed by FOXO6-Res (P < 0.0001; ANOVA, 51 neurons measured; Tau1 and P < 0.0001; ANOVA, 57 neurons measured; MAP2). Representative images of MAP2 immunostaining are shown in J.

Figure 4.

FOXO knockdown disrupts the establishment of neuronal polarity in the cerebellar cortex in vivo. (A,B) P3 rat pups were injected in the cerebellum with a GFP expression plasmid and then subjected to electroporation. Five days later, at P8, pups were sacrificed and coronal sections of cerebella were subjected to immunohistochemistry with a monoclonal antibody to GFP (green) and a rabbit polyclonal antibody to calbindin (red), the latter to label Purkinje cells. Transfected GFP-positive cerebellar granule neurons bear dendrites and have associated parallel fibers (PF) along the ML. Bars, 50 μm. (C) Coronal sections of cerebella electroporated as in A with the control U6-cmvGFP or U6/foxo-cmvGFP RNAi plasmid were subjected to immunohistochemistry with the GFP antibody (green) and the calbindin antibody (red). The bottom panels show a higher magnification of the numbered cells. In control animals (U6), granule neurons in the IGL were typically associated with parallel fibers. In contrast, FOXO knockdown (U6/foxo) led to loss of associated parallel fibers. Concomitant with the decrease in parallel fiber abundance, the length of secondary processes in the IGL was increased in granule neurons in FOXO knockdown animals as compared with granule neurons in control transfected animals. Arrows and arrowheads indicate secondary processes in the IGL (dendrites in control animals) and parallel fibers, respectively. (D) Quantification of total length of secondary processes in the IGL of granule neurons in animals electroporated and analyzed as in C. FOXO knockdown significantly increased the length of secondary processes in the IGL in granule neurons (P < 0.001; t-test, 335 neurons measured). (E) Quantification of parallel fiber phenotype upon FOXO knockdown in vivo. The percentage of granule neuron somas in the IGL that were associated with parallel fibers was significantly reduced in FOXO knockdown animals as compared with control transfected animals (P < 0.001; t-test, n = 3, 811 neurons measured). (F,G) P8 rat pups electroporated at P3 with the control U6-cmvGFP or U6/foxo-cmvGFP RNAi plasmid together with the FOXO6-Res expression plasmid or its control vector were analyzed as in A–E. Expression of FOXO6-Res in the background of FOXO knockdown in vivo significantly reduced the length of secondary processes in the IGL (dendrites in control animals) (P < 0.01; ANOVA, n = 3, 216 neurons measured) and significantly increased the number of parallel fibers associated with IGL granule neurons (P < 0.05; ANOVA, n = 3, 2655 neurons measured) as compared with FOXO knockdown animals.

Figure 5.

FOXO knockdown-induced impaired neuronal polarity phenotype in vivo is sustained in later stages of development. (A) P3 rat pups were injected in the cerebellum with the control U6-cmvGFP or U6/foxo-cmvGFP RNAi plasmid and then subjected to electroporation. Nine days, later at P12, pups were sacrificed and coronal sections of cerebella were subjected to immunohistochemistry with the GFP (green) and calbindin (red) antibodies, the latter to label Purkinje cells. In control animals (U6), granule neurons in the IGL were typically associated with parallel fibers. The FOXO knockdown-induced loss of parallel fibers was sustained at this later stage of development (P12). In addition, secondary processes in the IGL (dendrites in control) appeared to be much longer in FOXO knockdown animals as compared with control animals. Arrows and arrowheads indicate dendrites and parallel fibers, respectively. Bar, 50 μm. (B,C) Higher magnification of granule neurons in the cerebellar cortex in animals electroporated and analyzed as in A. The numbered dendritic tips shown in B are magnified in C. Mature dendrites in control animals bear dendritic claws at their ends (indicated by brackets), which represent characteristic post-synaptic structures (Shalizi et al. 2006). In contrast, the aberrant long secondary processes in the IGL in FOXO knockdown animals have tapered ends lacking dendritic claws. Bars: B, 50 μm; C, 10 μm. (D) Quantification of total length of secondary processes in the IGL of granule neurons in animals electroporated and analyzed as in A. FOXO knockdown significantly increased total secondary process length in granule neurons (P < 0.001; t-test, n = 3 brains, 172 neurons measured). (E) Quantification of parallel fiber phenotype upon FOXO knockdown in vivo. The percentage of granule neurons in the IGL that were associated with parallel fibers was significantly reduced in FOXO knockdown animals as compared with control transfected animals (P < 0.01; t-test, n = 3 brains, 809 neurons measured). (F) Quantification of the number of dendritic claws in control and FOXO knockdown animals. FOXO knockdown significantly reduced the number of secondary processes in the IGL (dendrites in control) bearing claws (P < 0.005; t-test, n = 3, 141 neurons measured).

Figure 6.

Identification of Pak1 as a direct target of FOXO transcription factors in neurons. (A) Granule neurons were transfected at high efficiency with the control U6 or U6/foxo RNAi plasmid. Two days later, RNA was extracted and reverse-transcribed for use in quantitative PCR of genes encoding proteins implicated in the establishment of neuronal polarity. Knockdown of FOXO transcription factors significantly reduced expression of several polarity genes. Pak1 expression was the most robustly down-regulated of all of the genes tested. Arrows indicate genes that are significantly reduced in FOXO knockdown neurons as compared with U6 control transfected neurons (P < 0.05; t-test, n = 3). (B,C) Pak1 mRNA abundance was assessed by quantitative RT–PCR in cultured granule neurons (B) or in the cerebellum (C) at the indicated time points. Pak1 mRNA abundance increases preceding the onset of polarization. (D,E) Pak1 protein expression was analyzed by immunoblotting of lysates prepared from cultured granule neurons (D) or from cerebellar lysates (E) at the indicated time points. Pak1 expression increases during the period of polarization. (F) Granule neurons were transfected at high efficiency with the control U6 or the U6/foxo plasmid. Four days later, lysates were prepared and subjected to immunoblotting with the indicated antibodies. FOXO knockdown triggered the down-regulation of Pak1 protein levels in neurons. (G) The Pak1 promoter contains putative FOXO-binding sites. Sequence alignment of a fragment of rat, mouse, and human Pak1 promoters is shown along with the engineered mutations in the putative FOXO-binding sites. (H) Granule neurons were transfected with a luciferase reporter gene under the control of a 1.4-kb region of the rat Pak1 promoter containing conserved FOXO-binding sites (Pak1-Luc) and an expression plasmid encoding FOXO1, FOXO3, FOXO6, or the control plasmid, together with a Renilla reporter to serve as control for transfection efficiency. Expression of FOXO transcription factors significantly increased the activity of the Pak1-Luc reporter gene (P < 0.01; ANOVA, n = 3). (I) Granule neurons were transfected with a plasmid encoding FOXO6 or its control vector together with Pak1-Luc or the Pak1 promoter containing mutations within the putative FOXO-binding site (Pak1 Mut 1/2-Luc) and the tk-Renilla reporter. Expression of FOXO6 robustly induced the expression of the Pak1-Luc reporter gene (P < 0.001; ANOVA, n = 3), but failed to effectively induce the expression of the Pak1 Mut 1-Luc or the Pak1 Mut 2-Luc reporter gene. (J) FOXOs occupy the promoter of the endogenous Pak1 gene in granule neurons by ChIP analysis. Granule neuron chromatin was subjected to immunoprecipitation with a control IgG antibody or with antibodies to FOXO1, FOXO3, and FOXO6. Immunoprecipitates were analyzed by quantitative PCR using primers designed to amplify the promoter of the Pak1 gene encompassing the putative FOXO-binding sequence or the first exon of the GAPDH gene as control. Data are plotted as the relative FOXO/IgG immunoprecipitation efficiency. FOXO occupancy at the Pak1 gene is significant relative to the GAPDH gene (P < 0.005; t-test, n = 3).

Figure 7.

The polarity-associated protein kinase Pak1 mediates FOXO-dependent neuronal polarity. (A) Granule neurons transfected with the control U6, U6/foxo, U6/scr, or U6/pak1 RNAi plasmid and the GFP expression plasmid were subjected 4 d after transfection to immunocytochemistry with the GFP antibody. Knockdown of Pak1 significantly increased the number of nonpolarized neurons as compared with control U6/scr (P < 0.0001; ANOVA, n = 3) and phenocopied FOXO knockdown. Bar, 50 μm. (B) Granule neurons were transfected with the control U6, U6/foxo, or U6/pak1, or both the U6/foxo and U6/pak1 RNAi plasmids, together with the GFP expression plasmid and subjected to immunocytochemistry 4 d later. While individual Pak1 or FOXO knockdown increased the number of nonpolarized neurons (P < 0.0001; ANOVA, n = 3), simultaneous FOXO and Pak1 knockdown did not additively increase the number of nonpolarized neurons as compared with Pak1 knockdown. (C) Granule neurons transfected with the control U6 or U6/foxo RNAi plasmid together with a plasmid expressing Pak1 or its control vector and the GFP expression plasmid were subjected 4 d after transfection to immunocytochemistry with the GFP antibody. (D) Expression of Pak1 significantly reduced the percentage of nonpolarized neurons in the background of FOXO RNAi (P < 0.01; ANOVA, n = 3). (E) Morphometric analysis shows that the length of the longest process (axon in control) was significantly reduced and the length of secondary processes (dendrites in control) was significantly increased upon FOXO RNAi (P < 0.0001; ANOVA, n = 3). Pak1 expression in the background of FOXO RNAi significantly increased the length of the longest process and significantly reduced the length of secondary processes as compared with FOXO RNAi alone (P < 0.001; ANOVA, n = 3). A total of 636 neurons were measured. (F–H) Coronal sections of P8 rat pups electroporated at P3 with the control U6-cmvGFP or U6/foxo-cmvGFP RNAi plasmid together with the Pak1 expression plasmid or its control vector were subjected to immunohistochemistry with the GFP antibody. Expression of Pak1 in the background of FOXO knockdown in vivo significantly reduced the length of secondary processes in the IGL (P < 0.05; ANOVA, n = 3) and significantly increased the number of parallel fibers associated with IGL granule neurons (P < 0.01; ANOVA, n = 3) as compared with FOXO knockdown animals. Arrows and arrowheads indicate dendrites and parallel fibers, respectively. Bar, 50 μm.