Abstract

Self-avoidance is a mechanism by which dendrites from the same neuron repel one another in order to establish uniform coverage of the dendritic field. The importance of self-avoidance for the development of complex arborization patterns has been highlighted by studies of Drosophila sensory and mouse retinal neurons. However, it is unclear whether branch patterning in the mammalian central nervous system is also governed by this strategy. We reduced Satb2 expression in a population of layer II/III pyramidal neurons in vivo by RNA interference and found that the somas of Satb2-deficient neurons clumped together, and their dendrites failed to expand laterally but instead formed fascicles. Furthermore, experiments showed that reducing Satb2 caused the adhesion of not only neighboring Satb2-deficient neurons but also neighboring wild-type neurons. Our results indicate a cell autonomous and non-cell autonomous role for Satb2 in regulating the adhesive and/or repulsive properties of cerebral pyramidal neurons.

Introduction

Different types of neurons in the nervous system develop distinct forms and shapes of dendritic arbors to fulfill their particular physiological functions. In order to achieve these shapes, growing dendrites respond to cell autonomous and non-cell autonomous cues that determine overall territories and the density of branches within the covered area. In Drosophila, dendrites of dendritic arborization (da) neurons exhibit a “self-avoidance” behavior in which branches from the same neuron seldom cross each other but those from different classes of da neurons frequently cross (Grueber et al. 2002; Hattori et al. 2008; Millard and Zipursky 2008; Schmucker and Chen 2009). Recent findings show that a cell surface molecule, Dscam (Down syndrome cell adhesion molecule), is required for the dendritic self-avoidance in both Drosophila da neurons and mouse retinal cells (Hughes et al. 2007; Matthews et al. 2007; Soba et al. 2007; Fuerst et al. 2008; Hattori et al. 2008; Millard and Zipursky 2008; Fuerst et al. 2009). Although mouse cortical neurons exhibit a similar contact-dependent inhibition behavior in vitro (Sestan et al. 1999), definitive in vivo evidence of avoidance-associated dendritic arborization and soma spacing in the mammalian nervous system outside the retina has yet to be reported.

The transcription factor Satb2 (specific AT-rich DNA-binding protein 2) was first identified as a gene mutated in human patients with cleft palate (FitzPatrick et al. 2003). Inactivation of Satb2 by homologous recombination leads to perinatal lethality, most likely as a result of the multiple craniofacial abnormalities (Britanova et al. 2006; Dobreva et al. 2006). Recently, Satb2 has been shown to be essential for the establishment of callosal projection neurons in the developing mouse cerebral cortex, as evidenced by the data from Satb2−/− mice that layer II/III pyramidal neurons take on layer V neuron fate and send their axons to subcortical regions instead of contralateral cortex (Alcamo et al. 2008; Britanova et al. 2008).

In this study, we examined the role of Satb2 in cortical dendritic development in vivo by knocking down its expression in cerebral pyramidal neurons of mice. We found that the somas of Satb2-deficient neurons clumped together, and their dendrites failed to expand within the upper layers of the cortex but instead formed fascicles. Furthermore, experiments showed that reducing Satb2 caused not only the adhesion of neighboring Satb2-deficient neurons but also of neighboring wild-type neurons. Together our data suggest that reducing Satb2 causes a gross imbalance of the adhesive and/or repulsive properties, which determine boundaries within and between dendritic arbors. Thus, we propose that Satb2-mediated regulation of gene transcription has a profound impact on processes that determine the form and size of dendritic arbors as well as soma spacing in the developing mammalian cortex.

Materials and Methods

DNA Constructs

The sequence of the Satb2-short hairpin RNA (shRNA) construct was targeted against nucleotides 843–861 of mouse Satb2 messenger RNA (NM_139146). An additional shRNA construct (Satb2-shRNA2) targeted against nucleotides 1324–1342 was also prepared. The complementary oligonucleotide was annealed and inserted into the pSUPER vector (OligoEngine). CAG-Satb2 was generated by cloning mouse Satb2 cDNA into the pCAGGS vector. The Satb2-shRNA resistant form of Satb2, CAG-Satb2R, was generated by introducing 10 nucleotide mutations within the Satb2-shRNA target site with no change in amino acid sequences.

HEK293 Cell Cultures

HEK293 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco) containing 10% fetal bovine serum (Hyclone). At 80% confluence, cells were transfected (for details, see Results) using lipofectamine 2000 (Invitrogen). Twenty-four hours later, cells were lysed in ice-cold radioimmunoprecipitation (RIPA) buffer containing 150 mM NaCl, 30 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 100 mM NaF, 1% Triton-X 100, and 0.01% sodium dodecyl sulfate (SDS). Samples were then loaded on sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), transferred, probed with mouse anti-Satb2 antibody (1:1000, Santa Cruz) or rabbit anti-GAPDH antibody (1:1000, Proteintech) and developed with species-specific horseradish peroxidase (HRP)-conjugated secondary antibodies (Proteintech), and visualized with enhanced chemiluminescence (Pierce).

Cortical Neuron Cultures

Cerebral cortex of newborn mouse was dissected and placed immediately in Hanks’ balanced salt solution (20 mM HEPES, 2 mM CaCl2, 5.4 mM KCl, 1 mM MgCl2, 136 mM NaCl, 1 mM Na2HPO4, and 5.6 mM glucose; pH 7.3). The tissue was then sliced and treated for 7 min at 37 °C with 0.05% (w/v) trypsin and 0.02% ethylenediaminetetraacetic acid (Gibco) in 2 mL phosphate-buffered saline (PBS) (137 mM NaCl, 6.5 mM Na2HPO4, 2.7 mM KCl, and 1.5 mM KH2PO4; pH 7.1). The digestion was stopped by adding 5 mL plating medium consisting of DMEM (Gibco) and 20% fetal bovine serum (Hyclone). Samples were then triturated with a fire-polished Pasteur pipette to single cells. Cells were then pelleted by centrifuging at 1000 rpm for 5 min and seeded in 12-well plates at a density of 3 × 105 cells/well in plating medium. One hour later, the medium was replaced by culture medium consisting of Neurobasal medium, B27 supplements, and 2 mM Glutamax-I (Gibco). For examination of Satb2-shRNA–induced effects in vitro, calcium phosphate transfection of pSUPER or Satb2-shRNA with CAG-Enhanced Green Fluorescent Protein (EGFP) was carried out at 6 days in vitro (DIV). At DIV9, cells were fixed in 4% paraformaldehyde, washed in PBS, and incubated with rabbit anti-GFP antibody (1:2000; Invitrogen) at 4 °C overnight, followed by incubation with Alexa Fluor 488–conjugated donkey anti-rabbit antibody (1:500; Invitrogen) for 3 h at room temperature.

For examination of Satb2-shRNA efficacy in cortical neurons, dissociated cortical neurons prepared as mentioned above were mixed with pSUPER or Satb2-shRNA together with CAG-EGFP, electroporated with the Nucleofector device (Amaxa), and then seeded in 12-well plates. Transfected neurons were harvested if percentages of transfected (i.e., EGFP-positive) neurons in the total of cells were above 30% at DIV3. Cells were lysed in ice-cold RIPA buffer. Protein samples were loaded on SDS-PAGE, transferred, probed with mouse anti-Satb2 antibody (1:1000; Santa Cruz) or goat anti-β-actin (1:1000; Proteintech), species-specific HRP-conjugated secondary antibodies (Proteintech), and visualized with enhanced chemiluminescence (Pierce).

In Utero Electroporation

Pregnant mice were deeply anaesthetized, and embryos were surgically manipulated, as described previously (Wang et al. 2007). Plasmids (2 μg/μL each in 0.5 μL) were injected directly into the lateral ventricles of the embryonic brain (n = 5 for each developmental stages). CAG-EGFP was also coinjected with other constructs at a concentration of 2 μg/μL. To label neurons relatively isolated from other transfected cells, the concentration of CAG-EGFP was lowered to 0.3 μg/μL. Five square electric pulses (30 V) were delivered through the uterus at 1-s interval with forceps-type electrodes while the uterus was kept wet with saline. For sequential electroporation, CAG-mCherry was transfected first, and 1 day later, a second round of electroporation was performed (pSUPER or Satb2-shRNA) or vice versa (n = 3 for each). All experimental manipulations have been reviewed and approved by the Animal Committee of Tongji University School of Medicine, Shanghai, China.

Immunohistochemistry, in situ hybridization, and TUNEL staining

Coronal brain sections (50-μm thick) were prepared from electroporated brains at different postnatal stages (n = 6 for each) and processed for immunostaining as described previously (Wang et al. 2007). To obtain 50-μm thick tangential sections, cortices were taken from P10 brains, flattened out, and cut from the superficial side. The following primary antibodies were used: rabbit anti-GFP (1:2000; Invitrogen), mouse anti-Satb2 (1:200; Santa Cruz), mouse anti-NeuN (1:500; Millipore), rabbit anti-glial fibrillary acidic protein (GFAP; 1:500; Dako), rabbit anti-S100 calcium-binding protein β (S100β; 1:600; Sigma), rabbit anti-γ-aminobutyric acid (GABA) (1:500; Sigma), and rabbit anti-Caspase-3 (1:1000; Cell Signaling) antibodies. Species-specific Alexa Fluor 488– or Cy3-conjugated secondary antibodies (1:500; Jackson ImmunoResearch) were used to detect primary antibodies.

For Satb2 and NeuN double immunostaining (both are mouse monoclonal antibodies), we employed the tyramide signal amplification system (TSA Plus Biotin Kit, PerkinElmer) (Shindler and Roth 1996). Brain sections were first incubated overnight at 4 °C with mouse anti-Satb2 antibody (1:50 000), a dilution at which no immunoreactivity for Satb2 could be detected by the conventional immunostaining procedure. Sections were then incubated with HRP-labeled goat anti-mouse IgG (1:100; KangChen) for 3 h and TSA Biotin Amplification Reagent (1:50; PerkinElmer) for 10 min at room temperature. Signals were visualized with Cy3-conjugated streptavidin (1:1000; Vector Labs). After being washed in PBS, sections were incubated with the second primary antibody, mouse anti-NeuN antibody (1:500) overnight at 4 °C, and Alexa Fluor 488–conjugated donkey anti-mouse IgG (1:500; Invitrogen) for 3 h at room temperature.

Antisense digoxigenin–labeled RNA probes were synthesized, and in situ hybridization was performed as described previously (Dai et al. 2008). The following probes were used: Cux2, Lmo4, Satb1, Satb2, Sox5 (Gray et al. 2004), ER81, Reelin, RORβ (Hanashima et al. 2004), and Ctip2 (nt2085–2921, NM_001079883). Terminal deoxynucleotidyl transferase mediated dUTP nick end labeling (TUNEL) staining was performed as described previously (Wang et al. 2007).

Imaging and Data Analysis

Confocal images were acquired on a Zeiss LSM510 or Leica TCS SP5 laser scanning confocal microscope and Z-stacks of 5 optical sections at 1-μm interval were collected. Images for callosal projections, in situ hybridizations, and neuronal migration were collected on a Nikon 80i microscope. To quantify dendritic branching in vitro, we performed Sholl analysis in which times of the dendrites of a neuron intersecting with the circumference of circles around the neuronal soma at every 10-μm interval were counted using ImageJ software (n = 30 for each). For quantification of knockdown efficacy in vivo, the number of EGFP/Satb2-double positive and EGFP-positive cells in pSUPER- or Satb2-shRNA–transfected brains (n = 5 for each) were counted in 3 different cortical regions (motor, somatosensory, and visual cortices), and the ratio of EGFP/Satb2-double labeled to the total of EGFP-labeled neurons was calculated. For migration analysis, the numbers of EGFP-labeled neurons in each cortical layers were counted in P4 and P7 mouse brains (n = 4 for each), and ratios of labeled neurons in each layer to the total population of the neurons in all layers were calculated. For quantitation of self-crossing, the number of crossing among sister dendrites in a relatively isolated neuron was counted (n = 23 for each). For quantitation of interneuronal fasciculation, the length of fasciculated dendrites of transfected neurons was measured (n = 20 for each). Dendritic fasciculation was defined by the following criteria: 2 or more dendritic branches from distinct neurons were located in an approximately parallel fashion and the distance between the branches was less than 1 μm. Student’s t-test was used for all comparisons in the present study.

Results

Satb2 Is Expressed in the Postnatal Developing Cerebral Cortex

We examined Satb2 expression in the mouse cerebral cortex at different postnatal stages. Consistent with previous reports (Britanova et al. 2005; Alcamo et al. 2008; Britanova et al. 2008), Satb2 transcripts were present throughout the cerebral cortex at postnatal day (P) 4 and P7 with highest levels in the superficial layers (Fig. 1A,B). The level of Satb2 was diminished by P14 and remained low at P21 (Fig. 1C,D). Double immunostaining of Satb2 with NeuN, a neuronal marker, and with glial cell markers (GFAP and S100β) showed that Satb2 was exclusively expressed in cortical neurons (Fig. 1E–G and Supplementary Fig. S1). As the first 2 weeks of postnatal life correspond to a period of extensive dendritic growth and arborization, the spatiotemporal pattern of Satb2 expression suggests that it is involved in cortical dendritic development.

Figure 1.

The expression of Satb2 in mouse cerebral cortex at various postnatal stages. (A) At P4, Satb2 is highly expressed in all cortical layers. (B) At P7, Satb2 is expressed at higher levels in the superficial layers relative to the deep layers. (C, D) At P14 and P21, Satb2 expression is reduced relative to P7 and presents at a homogeneous level in all cortical layers. (EG) Double immunostaining of Satb2 (red) and NeuN (green) in the cerebral cortex shows that all Satb2-positive neurons are NeuN-positive. I–VI indicate cortical layers. Scale bars: 100 μm (AD) and 50 μm (EG).

Figure 1.

The expression of Satb2 in mouse cerebral cortex at various postnatal stages. (A) At P4, Satb2 is highly expressed in all cortical layers. (B) At P7, Satb2 is expressed at higher levels in the superficial layers relative to the deep layers. (C, D) At P14 and P21, Satb2 expression is reduced relative to P7 and presents at a homogeneous level in all cortical layers. (EG) Double immunostaining of Satb2 (red) and NeuN (green) in the cerebral cortex shows that all Satb2-positive neurons are NeuN-positive. I–VI indicate cortical layers. Scale bars: 100 μm (AD) and 50 μm (EG).

Knocking Down Satb2 Leads to Soma Clumping and Dendritic Fasciculation

We investigated the role of Satb2 in the postnatal developing cortex by RNA interference (RNAi). We generated a shRNA expression vector that specifically targets mouse Satb2 (hereafter referred to as Satb2-shRNA). We tested its efficiency by measuring levels of exogenous Satb2 protein in HEK293 cells cotransfected with Satb2 expression vector and endogenous Satb2 protein in primary mouse cortical neurons. In both cell types, transfecting Satb2-shRNA caused a great reduction of Satb2 protein levels relative to cells transfected with empty pSUPER vector (Fig. 2A,B). The effectiveness of our RNAi was verified by determining that Satb2-shRNA was unable to reduce Satb2 protein levels in HEK293 cells cotransfected with an expression vector containing a form of Satb2 resistant to Satb2-shRNA (Satb2R, for details, see Materials and Methods; Fig. 2A).

Figure 2.

Satb2 knockdown increases dendritic branching in vitro. (A, B) Satb2 western blot. (A) Western blot shows that transfection of Satb2-shRNA into HEK293 cells greatly reduces expression of exogenous wild-type Satb2 but not of the shRNA-resistant form of Satb2 (Satb2R). (B) Satb2-shRNA also reduces endogenous Satb2 in cultured mouse cortical neurons. (C, D) Representative images of cultured cortical neurons cotransfected with pSUPER/EGFP or Satb2-shRNA/EGFP at DIV6 and examined at DIV9. Scale bar: 40 μm. (E) Sholl analysis of dendritic branching of cultured cortical neurons cotransfected with pSUPER/EGFP or Satb2-shRNA/EGFP. Error bars represent standard error of the mean, and asterisks indicate significant differences (**P < 0.01 and ***P < 0.001).

Figure 2.

Satb2 knockdown increases dendritic branching in vitro. (A, B) Satb2 western blot. (A) Western blot shows that transfection of Satb2-shRNA into HEK293 cells greatly reduces expression of exogenous wild-type Satb2 but not of the shRNA-resistant form of Satb2 (Satb2R). (B) Satb2-shRNA also reduces endogenous Satb2 in cultured mouse cortical neurons. (C, D) Representative images of cultured cortical neurons cotransfected with pSUPER/EGFP or Satb2-shRNA/EGFP at DIV6 and examined at DIV9. Scale bar: 40 μm. (E) Sholl analysis of dendritic branching of cultured cortical neurons cotransfected with pSUPER/EGFP or Satb2-shRNA/EGFP. Error bars represent standard error of the mean, and asterisks indicate significant differences (**P < 0.01 and ***P < 0.001).

Having established the efficacy and effectiveness of Satb2-shRNA, we tested its effect in cultured cortical neurons. Transfection of pSUPER or Satb2-shRNA was performed at DIV6, and dendrite morphology was examined at DIV9. We found that dendrite of Satb2-shRNA–transfected neurons was more complex relative to pSUPER-transfected neurons (Fig. 2C,D). Sholl analysis showed a dramatic increase of dendritic branching particularly in proximal dendrites (<60-μm distance from soma) in Satb2-shRNA–transfected neurons compared with control neurons (Fig. 2E), indicating that the dendritic arborization is abnormal in Satb2-knockdown cortical neurons.

We next introduced Satb2-shRNA into the dorsolateral ventricular zone of the embryonic forebrain via in utero electroporation at 15.5 days postcoitum (dpc) in order to transfect progenitors that give rise to layer II/III neurons in the cerebral cortex (Wang et al. 2007). CAG-EGFP plasmid (Ding et al. 2004) was simultaneously delivered to reveal the morphology of Satb2-deficient and control neurons. In control, mice electroporated with empty pSUPER, most EGFP-expressing neurons migrated to the superficial region of the cortex by P1 and completed their migration into layers II–III by P3 (data not shown). Upon reaching the superficial layers of the cortex, labeled pyramidal neurons established their final position by P3 and began a dramatic reorganization of their dendrites, particularly apical dendrites, over the next few days (Fig. 3A,C). These observations agree with those reported previously (Mizuno et al. 2007; Wang et al. 2007).

Figure 3.

Knockdown of Satb2 results in dendritic fasciculation and soma clumping in the visual cortex. (AF) At postnatal stages P4, P7, and P14, layer II/III pyramidal neurons transfected with Satb2-shRNA display a progressively worsening dendritic fasciculation and soma clumping phenotype (B, D, F). In contrast, dendrites of layer II/III pyramidal neurons in control pSUPER-transfected brains become progressively more complex and expand from P4 to P14 (A, C, E). (G, H) Tangential brain sections of Satb2-shRNA–transfected cortex reveal bundles of apical dendrites in layer I (arrows) and clusters of cell bodies (triangles) in layers II–III (H), whereas pyramidal neurons show the even distribution of somas in control brain (G). (I, J) High magnification of dendritic arborization in control and Satb2-shRNA–transfected brain sections at P7. (K, L) Expression of Satb2R rescues the Satb2-shRNA–induced dendrite arborization and soma spacing phenotypes. Scale bars: 100 μm (A–H, K, L) and 40 μm (I, J).

Figure 3.

Knockdown of Satb2 results in dendritic fasciculation and soma clumping in the visual cortex. (AF) At postnatal stages P4, P7, and P14, layer II/III pyramidal neurons transfected with Satb2-shRNA display a progressively worsening dendritic fasciculation and soma clumping phenotype (B, D, F). In contrast, dendrites of layer II/III pyramidal neurons in control pSUPER-transfected brains become progressively more complex and expand from P4 to P14 (A, C, E). (G, H) Tangential brain sections of Satb2-shRNA–transfected cortex reveal bundles of apical dendrites in layer I (arrows) and clusters of cell bodies (triangles) in layers II–III (H), whereas pyramidal neurons show the even distribution of somas in control brain (G). (I, J) High magnification of dendritic arborization in control and Satb2-shRNA–transfected brain sections at P7. (K, L) Expression of Satb2R rescues the Satb2-shRNA–induced dendrite arborization and soma spacing phenotypes. Scale bars: 100 μm (A–H, K, L) and 40 μm (I, J).

The migratory behavior of most Satb2-shRNA–expressing neurons did not differ from control neurons, as shown by similar ratios of transfected neurons in each cortical layer (Fig. 3A,B and Supplementary Fig. S2). Unlike controls, however, whose apical dendrites were radially arborized toward the pial surface at P4, the apical dendrites of Satb2-shRNA–transfected neurons appeared to adhere with dendrites of adjacent transfected neurons (Fig. 3A,B). This phenotype became more evident at P7, when apical dendrite fascicles, oriented toward the superficial surface, were observed in the cortex (Fig. 3C,D). In addition, the somas of Satb2-deficient neurons were clumped together in layers II–III, in contrast to the even distribution of control neuron somas (Fig. 3C,D,I,J). At P14, the bundles of dendrites of Satb2-shRNA–transfected neurons continued to grow and extend into layer I but did not expand radially like those of control neurons, which spread evenly in layer I (Fig. 3E,F). Immunostaining of GABA indicated the clumping neurons in Satb2-shRNA–transfected brains are not GABAergic inhibitory neurons (Supplementary Fig. S3). The fasciculation phenotype was most apparent in tangential sections. In layer I, dendrite bundles were separated from each other (arrows in Fig. 3H), and in layers II–III, clusters of neuronal cell bodies were similarly grouped (triangles in Fig. 3H). A similar phenotype was observed by transfection with another shRNA against Satb2 (Satb2-shRNA2), confirming the function of Satb2 in dendritic arborization and soma spacing (Supplementary Fig. S4).

Note that dendritic arborization and soma spacing phenotypes were more severe in posterior cerebral areas, such as the visual cortex (Fig. 4F), than in anterior cortical areas, such as the somatosensory and motor cortices (Fig. 4D,E). Satb2 immunostaining of these brain sections further confirmed the effectiveness of Satb2-shRNA, as indicated by a great reduction of Satb2 immunofluorescence in Satb2-shRNA–transfected neurons (Fig. 4D′,E′,F′). Interestingly, the ratios of EGFP/Satb2-double labeled neurons to the total number of EGFP-labeled neurons were decreased along the anterioposterior axis in Satb2-shRNA–transfected brains (Fig. 4G–I). These results indicated that the actual efficiency of Satb2-shRNA is different among these cortical regions (“low to high” gradient along the anterioposterior axis), and this would lead to less severe phenotype in the anterior cortex.

Figure 4.

Severity of the Satb2-shRNA phenotype varies along the anterior–posterior axis of the cortex. (AC) Control transfected (empty pSUPER) cortical neurons display similar morphologies in different cortical regions. (DF) The phenotypes resulting from Satb2-shRNA knockdown in the anterior cortex (motor and somatosensory cortices) are less severe than those in the posterior cortex (visual cortex). (A′–F′) Immunolabeling of Satb2 (red) in cortical sections from brains transfected with empty pSUPER or Satb2-shRNA vector in the relevant cerebral regions. (GI) Quantification of the ratio of Satb2/EGFP-double labeled neurons to the total number of EGFP-labeled neurons in Satb2-shRNA– and empty pSUPER-transfected (control) brains. Error bars represent standard error of the mean, and asterisks indicate significant differences (**P < 0.01 and ***P < 0.001). Scale bars: 200 μm (AF) and 50 μm (A′–F′).

Figure 4.

Severity of the Satb2-shRNA phenotype varies along the anterior–posterior axis of the cortex. (AC) Control transfected (empty pSUPER) cortical neurons display similar morphologies in different cortical regions. (DF) The phenotypes resulting from Satb2-shRNA knockdown in the anterior cortex (motor and somatosensory cortices) are less severe than those in the posterior cortex (visual cortex). (A′–F′) Immunolabeling of Satb2 (red) in cortical sections from brains transfected with empty pSUPER or Satb2-shRNA vector in the relevant cerebral regions. (GI) Quantification of the ratio of Satb2/EGFP-double labeled neurons to the total number of EGFP-labeled neurons in Satb2-shRNA– and empty pSUPER-transfected (control) brains. Error bars represent standard error of the mean, and asterisks indicate significant differences (**P < 0.01 and ***P < 0.001). Scale bars: 200 μm (AF) and 50 μm (A′–F′).

To precisely determine the nature of dendritic fasciculation, we lowered the concentration of cotransfected CAG-EGFP in in utero electroporation in order to label smaller number of neurons in a relatively scattered pattern (Fig. 5). Dendritic fasciculation was frequently observed in Satb2-shRNA–transfected neurons but rarely found in control neurons (Fig. 5A,B). Fasciculated dendrites were often bundled with dendrites of neighboring Satb2-shRNA–transfected neurons (Fig. 5E). We counted the number of dendritic crossings to quantify self-fasciculation (dendrites from the same neurons), measured the length of fasciculated dendrites (dendrites from distinct neurons), and found that both of them were dramatically increased in the Satb2-shRNA–transfected cortices compared with those in control cortices (Fig. 5C,F). These results showed that the dendritic fasciculation caused by Satb2 knockdown occurs among sister dendrites of the same neuron as well as dendrites of neighboring neurons.

Figure 5.

Dendritic fasciculation in Satb2-shRNA–transfected neurons in vivo. (A, B) An isolated control neuron shows nearly nonoverlapping pattern of dendritic branching, whereas dendrites of Satb2-shRNA–transfected neuron show self-crossing/fasciculation (arrows in B). (C) The number of self-crossing is significantly increased in Satb2-shRNA–transfected neurons compared with control neurons. (D, E) Dendritic fasciculation occurs among neighboring Satb2-shRNA–transfected neurons (triangles in E), but this is hardly present in pSUPER-transfected neurons (D). (F) The length of fasciculated dendrites between 2 neighboring neurons is dramatically increased in Satb2-shRNA–transfected neurons compared with control neurons. Error bars represent standard error of the mean, and asterisks indicate significant differences (***P < 0.001). Scale bar: 50 μm.

Figure 5.

Dendritic fasciculation in Satb2-shRNA–transfected neurons in vivo. (A, B) An isolated control neuron shows nearly nonoverlapping pattern of dendritic branching, whereas dendrites of Satb2-shRNA–transfected neuron show self-crossing/fasciculation (arrows in B). (C) The number of self-crossing is significantly increased in Satb2-shRNA–transfected neurons compared with control neurons. (D, E) Dendritic fasciculation occurs among neighboring Satb2-shRNA–transfected neurons (triangles in E), but this is hardly present in pSUPER-transfected neurons (D). (F) The length of fasciculated dendrites between 2 neighboring neurons is dramatically increased in Satb2-shRNA–transfected neurons compared with control neurons. Error bars represent standard error of the mean, and asterisks indicate significant differences (***P < 0.001). Scale bar: 50 μm.

To further verify whether the phenotype is caused by knockdown of Satb2, we performed cotransfection of Satb2-shRNA and Satb2R. The dendritic arborization and soma spacing defects were not observed (Fig. 3K,L). On the other hand, overexpressing wild-type Satb2 alone did not affect dendrite arborization and soma spacing (Supplementary Fig. S5). In addition, because Satb2-shRNA–transfected neurons were densely packed in layers II–III, the number of transfected neurons appeared to be reduced compared with that of control neurons (Fig. 3E,F), but we did not find any difference in TUNEL staining or Caspase-3 immunostaining between Satb2-shRNA–transfected and control brains (Supplementary Fig. S6). Taken together, we conclude that reducing Satb2 expression impairs dendritic arborization and soma spacing in the cerebral cortex.

Satb2 Exhibits Non-cell Autonomous Functions

To determine whether or not the role of Satb2 in somatic and dendritic adhesion is strictly cell autonomous, we used sequential in utero electroporation (Bai et al. 2003) to independently label 2 pools of pyramidal neurons while knocking down Satb2 in one of the pools. Embryos were electroporated with CAG-mCherry at 14.5 dpc and with Satb2-shRNA 1 day later or vice versa. mCherry itself did not affect dendrite morphology (Fig. 6A–C,G–I), but in sequentially electroporated cortices, mCherry-labeled neurons in layers II–III displayed a similar phenotype of dendritic fasciculation as Satb2-deficient neurons regardless of whether mCherry was transfected before or after Satb2-shRNA (Fig. 6D–F,J–L). These results suggest that reducing Satb2 has non-cell autonomous effects on dendritic arborization and spreading of cortical pyramidal neurons.

Figure 6.

Sequential in utero electroporation reveals a non-cell autonomous component of the Satb2-shRNA–induced phenotype. (AF) At P7, neurons transfected with mCherry on 14.5 dpc display similar dendritic arborization and soma spacing as neurons transfected with empty pSUPER on 15.5 dpc in the same cortex (AC). Note that mCherry-labeled neurons are located deeper in the cortex than control neurons. Neurons transfected with mCherry on 14.5 dpc exhibit impaired dendritic arborization, similar to neurons in the same cortical area transfected with Satb2-shRNA on 15.5 dpc (DF). (GL) At P7, neurons transfected with mCherry on 15.5 dpc display dendritic arborization and soma spacing as neurons transfected with empty pSUPER on 14.5 dpc in the same cortical region (GI). Neurons transfected with mCherry on 15.5 dpc exhibit impaired dendritic arborization, similar to neurons transfected with Satb2-shRNA on 14.5 dpc in the same cortical area (JL). Arrows indicate the dendritic defects in mCherry-expressing cells, which are located adjacent to Satb2-shRNA–transfected neurons. Scale bar: 100 μm.

Figure 6.

Sequential in utero electroporation reveals a non-cell autonomous component of the Satb2-shRNA–induced phenotype. (AF) At P7, neurons transfected with mCherry on 14.5 dpc display similar dendritic arborization and soma spacing as neurons transfected with empty pSUPER on 15.5 dpc in the same cortex (AC). Note that mCherry-labeled neurons are located deeper in the cortex than control neurons. Neurons transfected with mCherry on 14.5 dpc exhibit impaired dendritic arborization, similar to neurons in the same cortical area transfected with Satb2-shRNA on 15.5 dpc (DF). (GL) At P7, neurons transfected with mCherry on 15.5 dpc display dendritic arborization and soma spacing as neurons transfected with empty pSUPER on 14.5 dpc in the same cortical region (GI). Neurons transfected with mCherry on 15.5 dpc exhibit impaired dendritic arborization, similar to neurons transfected with Satb2-shRNA on 14.5 dpc in the same cortical area (JL). Arrows indicate the dendritic defects in mCherry-expressing cells, which are located adjacent to Satb2-shRNA–transfected neurons. Scale bar: 100 μm.

Knocking Down Satb2 Does Not Alter the Fate of Layer II/III Neurons

Satb2 is a determinant of layer II/III pyramidal neuron fates in the mouse cerebral cortex (Alcamo et al. 2008; Britanova et al. 2008). Satb2-expressing layer II/III neurons are callosal projection neurons, but these neurons take on deep-layer neuronal fate in Satb2-/- mice, as shown by the abnormal redirection of axons into subcortical regions and expression of deep layer–specific genes (Alcamo et al. 2008; Britanova et al. 2008). We thus explored whether layer II/III cortical neuronal fates are altered after knocking down Satb2 expression in the embryonic cortex from 15.5 dpc. To this end, we examined the expression of several cortical cell type– and layer-specific genes by in situ hybridization but did not detect any significant changes in terms of distribution and signal intensity (Fig. 7). For example, Ctip2 is a transcription factor necessary and sufficient for the extension of subcortical projections by deep cortical neurons (Arlotta et al. 2005); its expression is upregulated in layers II/III cortical neurons in Satb2-/- mice (Alcamo et al. 2008; Britanova et al. 2008) but remained unchanged in Satb2-shRNA–transfected brains compared with that in controls (Fig. 7I,J). Consistently, Satb2-shRNA–transfected cortical neurons showed normal development of callosal projections during the postnatal period examined (Supplementary Fig. S7). In addition, the expression of Satb1, the closest homolog for Satb2 and a factor essential for chromatin remodeling and gene expression (Han et al. 2008), was not altered either after Satb2 knockdown from 15.5 dpc (Fig. 7M,N).

Figure 7.

In situ hybridization shows cortical layer–specific gene expression in cortices transfected with Satb2-shRNA from 15.5 dpc. (AP) Expression patterns of Reelin (layers I and V), Cux2 (layers II–IV), RORβ (layer IV), ER81 (layer V), Ctip2 (layer V), Sox5 (layers V–VI), Satb1 (layers III–V), and Lmo4 (layers II–VI) are not significantly different between Satb2-shRNA–transfected cortices and that of controls. I–VI indicate cortical layers. Scale bar: 100 μm.

Figure 7.

In situ hybridization shows cortical layer–specific gene expression in cortices transfected with Satb2-shRNA from 15.5 dpc. (AP) Expression patterns of Reelin (layers I and V), Cux2 (layers II–IV), RORβ (layer IV), ER81 (layer V), Ctip2 (layer V), Sox5 (layers V–VI), Satb1 (layers III–V), and Lmo4 (layers II–VI) are not significantly different between Satb2-shRNA–transfected cortices and that of controls. I–VI indicate cortical layers. Scale bar: 100 μm.

Discussion

Conventional Satb2 knockout mice die shortly after birth (Britanova et al. 2006; Dobreva et al. 2006), and thus, these mice cannot reveal details concerning the role of Satb2 during postnatal stages of cerebral cortical development. In this study, we took advantage of in utero electroporation and RNAi techniques to knock down the Satb2 expression in mouse cortical neurons in vivo. We demonstrated that reducing Satb2 expression in a subset of layer II/III pyramidal neurons leads to cell autonomous and non-cell autonomous dendritic fasciculation and soma clumping of Satb2-deficient neurons and neighboring wild-type neurons. These phenotypes are consistent with the hypothesis that Satb2 regulates the expression of genes that are required for adhesion and/or repulsion of cortical pyramidal neurons.

Although it has not yet been established whether the self-avoidance and tiling contribute to dendritic arborization and soma spacing in the mammalian central nervous system (Emoto et al. 2004), dissociated cortical neurons show a contact-dependent inhibition of neurite growth (Sestan et al. 1999). Our in vitro data showed that Satb2 knockdown significantly increases the dendritic complexity, especially in proximal dendrites, leading to increased overlapping of dendritic branches (Fig. 2C,D). Consistent with this finding, we observed dendritic self-fasciculation in Satb2-shRNA–transfected neurons in vivo (Fig. 5). In addition, our in vivo data also revealed dendritic fasciculation among neighboring Satb2-knockdown neurons as well as those non–Satb2-shRNA-transfected neurons (Figs 4–6). We speculate that adhesion and/or repulsion mechanisms in cortical neurons are impaired by knockdown of Satb2, and further studies are needed to explore this important question.

We found that the severity of Satb2-knockdown phenotype varied across the cortex. Along the anterioposterior axis, reducing Satb2 had a progressively more profound impact on dendritic arborization and soma spacing from the rostral portion of the cortex to the caudal portion (Figs 3,4). Satb2 is homogeneously expressed in both anterior and posterior cortical regions with no significant difference in terms of pattern and level of expression (data not shown). Immunostaining of Satb2 in brain sections showed that the Satb2-knockdown efficiency was lower in the anterior cortical region than that in posterior cortex (Fig. 4), and this may be caused by unknown technical limitations of our electroporation. Alternatively, it is also likely that the transfected progenitors in the anterior cortex give rise to more neurons, which lead to dilution of Satb2-shRNA plasmids and finally results in decreased knockdown efficiency in individual neurons. Nevertheless, less severe phenotypes in the anterior cortex could be explained by reduced Satb2-knockdown efficiency.

Recently, much work has been focused on how intrinsic transcriptional programs control dendritic growth and branching in characteristic morphogenesis, especially in the Drosophila peripheral nervous system (Corty et al. 2009). The transcription factor Cut is undetected in class I da neurons but expressed in class II–IV da neurons and required for the more complex dendritic branching patterns in class II–IV da neurons (Grueber et al. 2003). In addition, the transcription factor Knot is also required for the development of highly branching arborization of class IV da neurons, and combinatorial expression code of Cut and Knot is essential for specific dendritic patterns in class III and IV da neurons in the Drosophila (Jinushi-Nakao et al. 2007). Recently, Cux1 and Cux2, the murine homologs of cut, are shown to be intrinsic regulators of dendritic branching in upper layer neurons of the cortex, as evidence by the findings that reducing Cux1 or Cux2 by shRNA leads to simplification of dendritic branching in mouse cerebral cortex (Cubelos et al. 2010). In the present study, we found that knockdown of Satb2 results in dendritic fasciculation and soma clumping. Thus, multiple transcription factors are involved in dendritic arborization and soma spacing, and it is of interesting to examine the relationship among these transcription factors in the dendritic arborization of cortical neurons.

In addition, Dscam and Dscam-like-1 (Dscaml1) mutant mice exhibit excessive dendritic fasciculation and soma clumping in several types of retinal neurons (Fuerst et al. 2008, 2009). The dendritic fasciculation and soma clumping seen in Satb2-deficient cortical pyramidal neurons are to some extent similar to the phenotype observed in Dscam or Dscaml1 mutant retinae (Fuerst et al. 2008, 2009). Given this similarity, it is possible that Dscam and/or Dscaml1 is also involved in dendritic arborization and soma pacing in the cerebral cortex, and their transcription is regulated by Satb2. In support of this hypothesis, reports have shown that Dscam and Dscaml1 are expressed in the mouse cerebral cortex (Agarwala 2001; Barlow et al. 2002). It is also interesting to explore whether the Dscam or Dscaml1 loci possess matrix-associated region (MAR) sites within their genomic DNA or whether these genes are regulated by MAR-binding transcription factors, such as Satb2.

Although layer II/III cortical neurons take on layer V cell fate in Sabt2-/- mice (Alcamo et al. 2008; Britanova et al. 2008), knocking down Satb2 expression from 15.5 dpc did not appear to alter the fate of transfected layer II/III neurons. We did not observe any discernable changes in layer II/III-specific gene expression (Fig. 7), and Satb2-shRNA–transfected neurons did not display changes in the timing or projection pattern of axon outgrowth (Supplementary Fig. S7). In Satb2 knockout mice, layer II–III neurons cannot be generated because their fate is altered in the beginning, and thus, the arborization and clumping phenotypes could not be observed. In addition, it may take a day or longer time for Satb2-shRNA constructs to generate enough siRNAs to knock down Satb2 expression, but Satb2-related cell fate determination events occur earlier during the development. This may also contribute to different phenotypes in Satb2-knockout and Satb2-knockdown cerebral cortex. Furthermore, Satb2 expression is reduced but not eliminated, and thus, these phenotypes observed in the present study are very likely caused by a partial loss of Satb2 rather than complete loss of this protein. Nevertheless, our results provide further insights into the role of Satb2 in cortical development by showing that it is also required for the dendrite arborization and soma spacing of layer II/III cortical neurons during the postnatal stages.

Supplementary Material

Supplementary material can be found at: http://www.cercor.oxfordjournals.org/

Funding

Key State Research Program from the Ministry of Science and Technology of China (2009ZX09501-030 to Y.-Q.D.); National Natural Science Foundation of China (30430260 to H.L. and 31030034 to Y.-Q.D.); and Tongji University’s Program for Young Excellent Talents (to L.Z.).

We thank Qiufu Ma and Gordon Fishell for providing in situ probes and Roger Y. Tsien for providing the mCherry plasmid. Conflict of Interest : None declared.

References

Agarwala
K
DSCAM, a highly conserved gene in mammals, expressed in differentiating mouse brain
Biochem Biophys Res Commun
 , 
2001
, vol. 
281
 (pg. 
697
-
705
)
Alcamo
EA
Chirivella
L
Dautzenberg
M
Dobreva
G
Farinas
I
Grosschedl
R
McConnell
SK
Satb2 regulates callosal projection neuron identity in the developing cerebral cortex
Neuron
 , 
2008
, vol. 
57
 (pg. 
364
-
377
)
Arlotta
P
Molyneaux
BJ
Chen
J
Inoue
J
Kominami
R
Macklis
JD
Neuronal subtype-specific genes that control corticospinal motor neuron development in vivo
Neuron
 , 
2005
, vol. 
45
 (pg. 
207
-
221
)
Bai
J
Ramos
RL
Ackman
JB
Thomas
AM
Lee
RV
LoTurco
JJ
RNAi reveals doublecortin is required for radial migration in rat neocortex
Nat Neurosci
 , 
2003
, vol. 
6
 (pg. 
1277
-
1283
)
Barlow
GM
Micales
B
Chen
XN
Lyons
GE
Korenberg
JR
Mammalian DSCAMs: roles in the development of the spinal cord, cortex, and cerebellum?
Biochem Biophys Res Commun
 , 
2002
, vol. 
293
 (pg. 
881
-
891
)
Britanova
O
Akopov
S
Lukyanov
S
Gruss
P
Tarabykin
V
Novel transcription factor Satb2 interacts with matrix attachment region DNA elements in a tissue-specific manner and demonstrates cell-type-dependent expression in the developing mouse CNS
Eur J Neurosci
 , 
2005
, vol. 
21
 (pg. 
658
-
668
)
Britanova
O
de Juan Romero
C
Cheung
A
Kwan
KY
Schwark
M
Gyorgy
A
Vogel
T
Akopov
S
Mitkovski
M
Agoston
D
, et al.  . 
Satb2 is a postmitotic determinant for upper-layer neuron specification in the neocortex
Neuron
 , 
2008
, vol. 
57
 (pg. 
378
-
392
)
Britanova
O
Depew
MJ
Schwark
M
Thomas
BL
Miletich
I
Sharpe
P
Tarabykin
V
Satb2 haploinsufficiency phenocopies 2q32-q33 deletions, whereas loss suggests a fundamental role in the coordination of jaw development
Am J Hum Genet
 , 
2006
, vol. 
79
 (pg. 
668
-
678
)
Corty
MM
Matthews
BJ
Grueber
WB
Molecules and mechanisms of dendrite development in Drosophila
Development
 , 
2009
, vol. 
136
 (pg. 
1049
-
1061
)
Cubelos
B
Sebastian-Serrano
A
Beccari
L
Calcagnotto
ME
Cisneros
E
Kim
S
Dopazo
A
Alvarez-Dolado
M
Redondo
JM
Bovolenta
P
, et al.  . 
Cux1 and Cux2 regulate dendritic branching, spine morphology, and synapses of the upper layer neurons of the cortex
Neuron
 , 
2010
, vol. 
66
 (pg. 
523
-
535
)
Dai
J-X
Hu
Z-L
Shi
M
Guo
C
Ding
Y-Q
Postnatal ontogeny of the transcription factor Lmx1b in the mouse central nervous system
J Comp Neurol
 , 
2008
, vol. 
509
 (pg. 
341
-
355
)
Ding
YQ
Yin
J
Kania
A
Zhao
ZQ
Johnson
RL
Chen
ZF
Lmx1b controls the differentiation and migration of the superficial dorsal horn neurons of the spinal cord
Development
 , 
2004
, vol. 
131
 (pg. 
3693
-
3703
)
Dobreva
G
Chahrour
M
Dautzenberg
M
Chirivella
L
Kanzler
B
Farinas
I
Karsenty
G
Grosschedl
R
SATB2 is a multifunctional determinant of craniofacial patterning and osteoblast differentiation
Cell
 , 
2006
, vol. 
125
 (pg. 
971
-
986
)
Emoto
K
He
Y
Ye
B
Grueber
WB
Adler
PN
Jan
LY
Jan
YN
Control of dendritic branching and tiling by the Tricornered-kinase/Furry signaling pathway in Drosophila sensory neurons
Cell
 , 
2004
, vol. 
119
 (pg. 
245
-
256
)
FitzPatrick
DR
Carr
IM
McLaren
L
Leek
JP
Wightman
P
Williamson
K
Gautier
P
McGill
N
Hayward
C
Firth
H
, et al.  . 
Identification of SATB2 as the cleft palate gene on 2q32-q33
Hum Mol Genet
 , 
2003
, vol. 
12
 (pg. 
2491
-
2501
)
Fuerst
PG
Bruce
F
Tian
M
Wei
W
Elstrott
J
Feller
MB
Erskine
L
Singer
JH
Burgess
RW
DSCAM and DSCAML1 function in self-avoidance in multiple cell types in the developing mouse retina
Neuron
 , 
2009
, vol. 
64
 (pg. 
484
-
497
)
Fuerst
PG
Koizumi
A
Masland
RH
Burgess
RW
Neurite arborization and mosaic spacing in the mouse retina require DSCAM
Nature
 , 
2008
, vol. 
451
 (pg. 
470
-
474
)
Gray
PA
Fu
H
Luo
P
Zhao
Q
Yu
J
Ferrari
A
Tenzen
T
Yuk
DI
Tsung
EF
Cai
Z
, et al.  . 
Mouse brain organization revealed through direct genome-scale TF expression analysis
Science
 , 
2004
, vol. 
306
 (pg. 
2255
-
2257
)
Grueber
WB
Jan
LY
Jan
YN
Tiling of the Drosophila epidermis by multidendritic sensory neurons
Development
 , 
2002
, vol. 
129
 (pg. 
2867
-
2878
)
Grueber
WB
Jan
LY
Jan
YN
Different levels of the homeodomain protein cut regulate distinct dendrite branching patterns of Drosophila multidendritic neurons
Cell
 , 
2003
, vol. 
112
 (pg. 
805
-
818
)
Han
H-J
Russo
J
Kohwi
Y
Kohwi-Shigematsu
T
SATB1 reprogrammes gene expression to promote breast tumour growth and metastasis
Nature
 , 
2008
, vol. 
452
 (pg. 
187
-
193
)
Hanashima
C
Li
SC
Shen
L
Lai
E
Fishell
G
Foxg1 suppresses early cortical cell fate
Science
 , 
2004
, vol. 
303
 (pg. 
56
-
59
)
Hattori
D
Millard
SS
Wojtowicz
WM
Zipursky
SL
Dscam-mediated cell recognition regulates neural circuit formation
Annu Rev Cell Dev Biol
 , 
2008
, vol. 
24
 (pg. 
597
-
620
)
Hughes
ME
Bortnick
R
Tsubouchi
A
Bäumer
P
Kondo
M
Uemura
T
Schmucker
D
Homophilic Dscam interactions control complex dendrite morphogenesis
Neuron
 , 
2007
, vol. 
54
 (pg. 
417
-
427
)
Jinushi-Nakao
S
Arvind
R
Amikura
R
Kinameri
E
Liu
AW
Moore
AW
Knot/Collier and cut control different aspects of dendrite cytoskeleton and synergize to define final arbor shape
Neuron
 , 
2007
, vol. 
56
 (pg. 
963
-
978
)
Matthews
B
Kim
M
Flanagan
J
Hattori
D
Clemens
J
Zipursky
S
Grueber
W
Dendrite self-avoidance is controlled by Dscam
Cell
 , 
2007
, vol. 
129
 (pg. 
593
-
604
)
Millard
SS
Zipursky
SL
Dscam-mediated repulsion controls tiling and self-avoidance
Curr Opin Neurobiol
 , 
2008
, vol. 
18
 (pg. 
84
-
89
)
Mizuno
H
Hirano
T
Tagawa
Y
Evidence for activity-dependent cortical wiring: formation of interhemispheric connections in neonatal mouse visual cortex requires projection neuron activity
J Neurosci
 , 
2007
, vol. 
27
 (pg. 
6760
-
6770
)
Schmucker
D
Chen
B
Dscam and DSCAM: complex genes in simple animals, complex animals yet simple genes
Genes Dev
 , 
2009
, vol. 
23
 (pg. 
147
-
156
)
Sestan
N
Artavanis-Tsakonas
S
Rakic
P
Contact-dependent inhibition of cortical neurite growth mediated by notch signaling
Science
 , 
1999
, vol. 
286
 (pg. 
741
-
746
)
Shindler
KS
Roth
KA
Double immunofluorescent staining using two unconjugated primary antisera raised in the same species
J Histochem Cytochem
 , 
1996
, vol. 
44
 (pg. 
1331
-
1335
)
Soba
P
Zhu
S
Emoto
K
Younger
S
Yang
SJ
Yu
HH
Lee
T
Jan
LY
Jan
YN
Drosophila sensory neurons require Dscam for dendritic self-avoidance and proper dendritic field organization
Neuron
 , 
2007
, vol. 
54
 (pg. 
403
-
416
)
Wang
CL
Zhang
L
Zhou
Y
Zhou
J
Yang
XJ
Duan
SM
Xiong
ZQ
Ding
YQ
Activity-dependent development of callosal projections in the somatosensory cortex
J Neurosci
 , 
2007
, vol. 
27
 (pg. 
11334
-
11342
)
Wang
C-Z
Shi
M
Yang
L-L
Yang
R-Q
Luo
Z-G
Jacquin
MF
Chen
Z-F
Ding
Y-Q
Development of the mesencephalic trigeminal nucleus requires a paired homeodomain transcription factor, Drg11
Mol Cell Neurosci
 , 
2007
, vol. 
35
 (pg. 
368
-
376
)