Transcription factor Sp4 regulates dendritic patterning during cerebellar maturation

Knockdown of Sp4 Leads to Exuberant Dendritic Arborization in Cerebellar Granule Neurons.

To investigate the function of Sp4 in postmitotic neurons, we used RNAi to knock down Sp4 in cerebellar granule neurons. This approach allows us to examine the effects of Sp4 loss of function during a late developmental window in postmitotic cells and avoids potential complications due to Sp4 functions in progenitor or neighboring cells, as well as compensatory mechanisms that may act during ontogeny. Morphogenesis of cerebellar granule neurons has been faithfully recapitulated in vitro under depolarizing culture conditions, which support maximal health and survival of rat cerebellar granule neurons; we have used these culture conditions in our analysis (36–38). We generated plasmids encoding two different short hairpin RNAs that specifically reduced expression of cotransfected FlagSp4 in Neuro 2A cells (Fig. 1A) and endogenous nuclear Sp4 protein in cerebellar granule neurons [Fig. 1B and supporting information (SI) Fig. 7]. There was no increase in apoptosis in granule neurons transfected with either Sp4 RNAi or control as judged by the appearance of condensed chromatin 5 days after transfection (Fig. 1B and data not shown).

Strikingly, we found that the dendritic morphology of neurons transfected with either of the two Sp4 RNAis was highly aberrant compared with neurons transfected with a control hairpin or RNAi for the related transcription factor Sp1 (Fig. 1C and SI Fig. 8). Immunostaining of cerebellar granule neurons for the axonal tau-1 or dendritic Map-2 proteins revealed that the major alteration in granule cells depleted for Sp4 was in the dendrites (SI Fig. 7 B and C).

We quantified the changes in dendritic patterning by counting the number of primary, secondary, and tertiary dendrites per neuron. Both Sp4 RNAi constructs increased the number of dendrites at least 2-fold compared to control-transfected neurons (Fig. 1D). The frequency of branch points on the shaft of a primary dendrite for neurons depleted of Sp4 was 3- to 4-fold higher than control-transfected neurons, supporting the observation that Sp4 knockdown neurons have more complex dendritic arbors (Fig. 1E). Total dendritic length was not significantly affected after knocking down Sp4 (SI Fig. 8B). Similarly, total axonal growth was not reduced after depletion of Sp4 (data not shown). Thus, depletion of Sp4 led to altered dendritic patterning, but did not compromise the growth of neuronal processes.

To confirm that the observed exuberant dendritic arborization was due specifically to the depletion of Sp4 and not because of nonspecific off-target effects of the RNAi pathway, we generated a Flag-Sp4 “rescue” vector bearing two silent point mutations in the Sp4 cDNA that was resistant to RNAi-mediated knockdown (Fig. 2A). When the the Flag-Sp4 rescue vector was cotransfected with the Sp4RNAi into granule neurons, the numbers of dendrites were completely restored to wild-type levels (Fig. 2B). Taken together, these data show that the more complex dendritic arbors induced by Sp4 RNAi are the result of specific depletion of Sp4.

Sp4 Is Required for Dendritic Maturation of Granule Neurons in Cerebellar Cortex.

The finding that depletion of Sp4 leads to more complex dendritic trees in primary granule neurons led us to investigate Sp4 function in dendritic development in the context of the cerebellum cortex. We used sagittal cerebellar slices maintained in nondepolarizing concentrations of potassium, where tissue is preserved and dendrites develop their physiological morphology (20, 35). We introduced the control, Sp4, or Sp1 RNAi constructs into cerebellar slices using biolistic method of gene delivery. We confirmed that tissue structure was intact by staining with the DNA dye Hoechst (Fig. 3A). Control-transfected granule neurons displayed mature dendritic arbors with two or three robust primary dendrites with few branches (Fig. 3A). Neurons depleted of Sp4 exhibited exuberant dendritic trees, with more numerous and highly branched dendrites around the soma compared with control Scr or Sp1 RNAi-transfected neurons (Fig. 3B). The appearance of Sp4 knockdown neurons closely resembled classic descriptions of granule neurons at an intermediate stage of dendritic maturation, exhibiting numerous, highly branched, primary dendrites that are later remodeled (9, 31). Quantitative analysis confirmed that Sp4 knockdown led to an increase in the number of dendrites and the frequency of branching compared to control and Sp1 RNAi, consistent with the phenotype observed in dissociated primary neurons (Fig. 3 C and D). Taken together, these results suggest that Sp4 contributes to dendritic patterning by controlling remodeling and/or restricting branching associated with maturation of cerebellar granule neurons.

Sp4 Knockdown Impairs Dendritic Pruning and Leads to Persistent Branching.

Dendritic development involves highly dynamic and choreographed events, including the formation and elimination of primary dendrites and branches (4–6, 39). To determine the steps of dendritic development controlled by Sp4, we performed a time-course analysis. We transfected primary cerebellar granule neurons with control or Sp4 RNAi and monitored the number of primary and secondary dendrites over time. Consistent with the pruning that occurs during development in vivo, as described by Ramon y Cajal and others (9, 31), in control neurons, we observed a decrease in the number of dendrites from day 4 to day 6 in vitro. In contrast, pruning of dendrites was blocked in neurons depleted for Sp4 (Fig. 4 A–C and SI Fig. 9). In fact, we observed that the addition of branches was persistent in Sp4 RNAi-transfected neurons, thereby accumulating more than twice as many branches and resulting in a 4-fold increase of branching points compared to control neurons (Fig. 4 C and D). Total dendritic length was not affected on depletion of Sp4 (SI Fig. 9B). These studies reveal that Sp4 is required for pruning and limiting branching during dendritic maturation.

Sp4 Is Required for Dendritic Remodeling Regulated by Neuronal Activity.

It has been observed that maturation of cerebellar granule neurons in vivo is concomitant with the arrival of mossy fibers, raising the possibility that dendritic remodeling in these cells is regulated by neuronal activity, as shown in many other types of neurons (9, 40–44). We therefore investigated, first, whether depolarization regulates remodeling in cerebellar granule neurons, and, second, whether Sp4 contributes to activity-dependent remodeling in these cells. Neurons transfected with GFP were maintained in conditioned medium with 25 mM potassium up to 4 days in vitro (DIV) to allow dendritic growth and branching. Then granule neurons were switched to fresh medium containing depolarizing (25 mM) or nondepolarizing (5 mM) concentrations of KCl for 3 days. A vector expressing Bclx_L was included in these studies to prevent apoptosis induced by low potassium (38). At 4 DIV, we observed complex dendritic trees, which were remodeled to simpler arbors when neurons were maintained in high potassium concentrations for the next 3 days (Fig. 5A). When switched to nondepolarizing media, however, dendrites were not pruned, showing more complex trees (Fig. 5A and SI Fig. 10). Quantitation revealed a significant reduction of dendrites under depolarizing concentrations of potassium, whereas the number of dendrites was maintained in nondepolarizing conditions (Fig. 5B). Consistent with these findings, pharmacological inhibition of l-type voltage-sensitive calcium channels (l-VSCC) with nimodipine in cerebellar slices from P9 pups delayed dendritic pruning (SI Fig. 11). Thus, dendritic remodeling in cerebellar granule neurons is regulated by depolarization and voltage-sensitive calcium channels.

We next induced knockdown of Sp4 in neurons maintained in depolarizing or nondepolarizing conditions to investigate the contribution of Sp4 function to activity-dependent remodeling. In control RNAi-transfected neurons, the number of dendrites was reduced in neurons maintained in depolarizing compared to nondepolarizing conditions (Fig. 5C). In contrast, in neurons depleted of Sp4, there was not a significant difference in the number of dendrites on membrane depolarization (Fig. 5C). Thus, Sp4 knockdown blocked depolarization-induced remodeling. These findings support the conclusion that Sp4 is required for activity-dependent remodeling in cerebellar granule neurons.

Overexpression of Sp4 Promotes Dendritic Pruning in Nondepolarizing Conditions.

To investigate the contribution of Sp4 transcriptional activity to activity-dependent dendritic pruning, we generated a deletion mutant of Sp4 lacking the Zinc finger DNA-binding domain (Sp4 ΔZnF). Wild-type Sp4, but not Sp4ΔZnF, activated expression of a cotransfected Sp4-responsive reporter, p35-luciferase (45) (Fig. 6A). Transfection with wild-type Sp4 led to a significant decrease in the number of dendrites in granule neurons maintained in nondepolarizing conditions (Fig. 6B). In contrast, expression of transcriptionally inactive Sp4ΔZnF did not lead to a change in dendrite number (Fig. 6B). These results demonstrate that high levels of Sp4 are sufficient to promote dendritic pruning in these cells when calcium signaling is blocked, and, furthermore, transcriptional activity of Sp4 is required to promote elimination of dendrites during the development of cerebellar granule neurons.

Plasmids.

Flag-Sp4 and Flag-Sp1 were expressed from the CMV promoter in pRC-Flag-Sp4 and pRC-Flag-Sp1. Short hairpin RNAs were expressed from the U6 promoter in pBSSK (51). Sequences targeted by the short hairpin RNAs were: Scrambled Sp3 control (gggaattaatatgcacacaggcc), Sp4 #1 nucleotides 1551– 1571 (gggctccaactttaacacctt), Sp4 #2 nucleotides 1804–1824 (gggtgctgcgggtgttcaagt), and Sp1 nucleotides 881–901 (gggaacatcaccttgctacct) [accession nos. NM 003112 (Sp4) and XM 028606 (Sp1)]. Silent mutations resistant to Sp4 RNAi hairpin #1 were introduced into Flag-Sp4 by PCR, and nucleotides were mutated to gggctccaacACtaacacctt. The Sp4ΔZnF mutant was generated by an inframe deletion of Sp4 amino acids 625–775. The reporter plasmid containing p35 promoter sequence −467 upstream of the luciferase gene has been described (45).

Cell Culture and Transfection.

Cerebellar granule neurons were obtained from postnatal day 6 rat pups as described (52). Briefly, neurons were maintained in BMEM supplemented with 10% FCS, 25 mM KCl, penicillin (50 units per ml), streptomycin (50 μg/ml), and 2 mM glutamine at 4.2 × 10³ cells per mm² density. After 18 h, 10 μM cytosine arabinofuranoside was added. Cells were transfected by calcium phosphate precipitation at DIV 2 as described (37). In some experiments, cerebellar granule neurons were switched to BEM with 5 mM KCl or 25 mM KCl 2 days after transfection. Where indicated, a vector expressing the antiapoptotic protein Bcl-x_L was included.

Neuro2A cells were plated at 3.6 × 10⁴ cells per cm² 24 h before transfection with Lipofectamine (Invitrogen, Carlsbad, CA). The ratio of Flag-Sp4 vector and U6/RNAi hairpin vector was 1:5. Relative luciferase assays were performed as indicated by the manufacturer (Promega, Madison, WI).

Cerebellar Slice Cultures and Transfection.

Slice cultures and transfections were done as described (20). Briefly, cerebella from postnatal day 9 rats were cut sagitally into 400-μm sections, transferred onto a porous membrane (Millipore Corporation, Billerica, CA), and maintained in α-MEM medium (5 mM KCl) supplemented with 12.5 mM Hepes, 30 mM glucose, 0.25% HBSS, 25% Horse serum, 2 mM glutamine, penicillin (50 units per ml), and streptomycin (50 μg/ml). Individual neurons were transfected after 3 days by using biolistics (Helios gene gun; Bio-Rad, Hercules, CA) with 1 μm gold particle coated with the test plasmids together with GFP and Bcl-x_L expression plasmids.

Immunofluorescence and Immunoblotting.

At the indicated times, cells were fixed in 4% paraformaldehyde. Neurons were immunolabeled by using a rabbit antibody to GFP (Molecular Probes, Eugene, OR) or Sp4 (Santa Cruz Technology, Santa Cruz, CA) or monoclonal antibody against MAP-2 (Sigma–Aldrich, St. Louis, MO), Tau-1 (Chemicon International, Temecula, CA), or GFP (Molecular Probes), followed by goat anti-rabbit or anti-mouse conjugated to Cy2 or Cy3 (Amersham Biosciences, Piscataway, NJ) and stained with the DNA dye bisbenzimide (Hoechst 33258). Neuro2A extracts were prepared in 1% Nonidet P-40 lysis buffer for 30 min on ice, followed by sonication and centrifugation at 15,000 × g for 15 min. Lysates were resolved by SDS/PAGE and immunoblotted with monoclonal anti-flag antibody (Sigma–Aldrich) or polyclonal antibody against GFP (BD Biosciences, San Jose, CA).

Morphometric Analysis of Dendrites.

Images of individual transfected neurons with no overlapping processes from other transfected neurons were captured randomly in a blinded manner at ×400 magnification using a Nikon eclipse TE2000 epifluorescence microscope with a CCD camera (Diagnostic Instruments, Sterling Heights, MI). Digital zoom magnification and quantitation of the length of individual primary, secondary, and tertiary dendrites were performed by using SPOT imaging software. Total dendritic length corresponds to the sum of the length of all individual dendrites per neuron. The number of branching points in 100 μm primary dendrites was determined by counting total secondary dendrites and total length of all primary dendrites in a neuron. The following formula was applied:

where TS is total number of secondary dendrites and TPL is total primary dendritic length.

Statistical differences were determined by ANOVA and a post hoc Tukey test or a two-tailed t test using GraphPad Prism Software. For every condition, quantitation was performed in three independent experiments with similar results.