BAF chromatin remodeling complex

Abstract

The multi-subunit chromatin remodeling BAF complex controls different developmental processes. Using cortex-specific conditional knockout and overexpression mouse models, we have recently reported that BAF170, a subunit of the vertebrate BAF chromatin remodeling complex, interacts with transcription factor (TF) Pax6 to control cortical size and volume. The mechanistic basis includes suppression of the expression of Pax6 target genes, which are required for genesis of cortical intermediate progenitors (IPs) and specification of late neuronal subtype identity. In addition, we showed that a dynamic competition between BAF170 and BAF155 subunits within the BAF complex during progression of neurogenesis is a primary event in modulating the size of the mammalian cortex. Here, we present additional insights into the interaction between the BAF complex and TF Pax6 in the genesis of IPs of the developing cortex. Furthermore, we show that such competition between BAF170 and BAF155 is involved as well in the determination of the size of the embryonic body. Our results add new insights into a cell-intrinsic mechanism, mediated by the chromatin remodeling BAF complex that controls vertebrate body shape and size.

Introduction

Organ and overall body size is determined by a combination of cell size and number.¹ In the last few years, progressive advances in understanding the regulation of the size and thickness of the mammalian cortex have been made.^2-5 In higher mammals, the cortex serves as an integrative center for perception of sensory inputs, cognitive function, and consciousness. Composed of multiple neuronal subtypes and diverse glial cells, the cortex underwent a pronounced expansion during evolution that is assumed to underlie the high intellectual capability of primates.^6-8

The cortex is a complex structure, which is organized both radially in 6 layers and tangentially in numerous functional domains. Cortical size is determined during embryonic development and reflects the balance between radial (thickness) and tangential (surface area) growth of the neuronal layers.⁹ Cortical thickness is determined by the number of neurons generated per radial unit of cortex, while the cortical surface depends on the number of such radial units.⁹ Radial units are established by ventricular radial glial cells (vRGs) that are located in the ventricular zone (VZ) and were long thought to be only a scaffold supporting the radial migration of newborn neurons.⁹ Recent evidence indicated, however, that vRGs are pluripotent progenitors generating both neuronal and glial cells.^10-12 Research in recent years has revealed that the main type of cortical neurons, the glutamatergic projection neurons, are generated during embryogenesis by vRGs.^10-12 During neurogenesis, a proliferating vRG cell at the apical surface produces one vRG (for self-renewal of the pool) and, via different modes of neurogenesis, either one neuron (direct mode) or one intermediate progenitor (IP) (indirect mode) (Fig. 1). IPs are located mostly in the second germinative zone, the subventricular zone (SVZ).^6,7,13,14 Before entering terminal neurogenic divisions, the IPs can undergo a limited number of proliferative divisions to amplify their pools, thereby expanding the neurogenic output^3,6,7 (Fig. 1A). Therefore, the indirect mode of neurogenesis is assumed to account for the increase in cortical thickness and surface during the evolution of primates.^3,6,7 Although the enlarged size of the cerebral cortex is a hallmark of mammalian brain evolution, the mechanisms underlying this key feature are still largely unknown.

Figure 1. Coupling of Pax6, BAF chromatin remodeling, and the REST-corepressor complex controls modes of cortical neurogenesis. (A) The schema illustrates how epigenetic control through BAF complex timely modulates the modes of cortical neurogenesis: direct mode (RG = RG renewal + neuron) and indirect mode (RG = RG renewal + IPs) followed by IP multiplication in SVZ and generation of neurons. Bars under the schema represent the dynamic expression of BAF170 and BAF155, respectively, during cortical neurogenesis. (B) Our findings indicate that during early corticogenesis (until E14.5), TF Pax6 recruits a BAF complex (incorporating Brm-ATPase, BAF155, BAF170), that is assembled to the REST-corepressor complex, thus mediating chromatin remodeling and negative control of Pax6 downstream target genes. At least 3 of them, as shown in our study, are involved in the specification and generation of IPs (Tbr2) and upper layer neuronal fates (Cux1, Tle1), which are predominantly expressed after stage E14.5. During early neurogenesis (before E14.5), the stoichiometric presence of both BAF170 and BAF155 in the complex prevents the euchromatin state of these genes, thus excluding their heterochronic expression and allowing for normal neurogenesis which predominantly follows the direct mode. After E14.5, the expression of BAF170 decreases, accompanied by the enhanced presence of BAF155 subunit(s), leading to the relaxed chromatin state of the promoters of such genes, thereby promoting late neurogenesis. Abbreviations used: ventricular zone, VZ; subventricular zone, SVZ; lower layer, LL; upper layer, UL; mantle zone, MZ; radial glial cells, RG; intermediate progenitor, IP; neuron, N. The curved arrows indicate self-renewal of the RG cells.

During corticogenesis, the decision of progenitors to either self-renew or differentiate is regulated by extrinsic factors and cell-intrinsic programs largely mediated by neurogenic transcription factors (TFs).^6,7,15 In particular, TF Pax6, which is an intrinsic determinant of the pluripotent RG cells, plays an essential role in cortical patterning and neurogenesis.^16-18 In Pax6 deficiency, as shown previously by us and other groups, the RG cells generated half of the normal neuronal number, showed defects in cell cycle progression and IP genesis, and almost completely failed to produce neuronal subtypes with upper cortical layer identity.^16,19-22 In fact, generation of lower layer (L-6,-5) and upper layer (L-4,-3,-2) neuronal subtypes is controlled by the timely expression of sets of developmental regulators at a particular developmental stage.^8,23 Such simultaneous expressions of batteries of genes involve epigenetic regulation of chromatin condensation.^15,24 Interestingly, a rising number of neurodevelopmental disorders and mental diseases (e.g., schizophrenia, autism, Coffin­–Siris syndrome, Nicolaides–Baraitser syndrome) have been recently shown to be causally related to mutations in genes encoding chromatin remodeling proteins, such as the subunits of the chromatin remodeling BAF (or mSWI/SNF) complex.^25,26 In vertebrates, BAF complexes contain 2 interchangeable core ATPase subunits (Brg1 or Brm), in combination with at least 15 different BAF (Brg1/Brm-associated factor) subunits.^26,27

Using mouse transgenic lines for cortex-specific conditional knockout (cKO) and overexpresssion (cOE) of the BAF170 subunit, we recently reported that BAF170 is an intrinsic factor of RG cells, which, through interactions with TF Pax6, controls cortical size and thickness. We also found that competition between the BAF170 and BAF155 subunits in the BAF complex affects chromatin modifications and binding of the Pax6/REST corepressor complex to Pax6 target genes, thereby playing an essential role in timely IP genesis and the maintenance of the proper expansion of the mammalian cortex²⁸ (Fig. 1A). Here we introduce our recent findings and provide additional insights on how the genetic interaction between TF Pax6 and BAF170 regulates the production of IPs. Furthermore, we provide novel evidence that BAF170 acts as a key factor to control the size of the embryonic body.

Results and Discussion

BAF170 expression controls the accessibility of TF Pax6 to the downstream target gene Tbr2, thereby regulating generation of Tbr2+ IPs

We recently reported that through its transitory expression in vRGs during a defined developmental window (E12.5-E14.5), BAF170 competes with the BAF155 subunit in the BAF complex during early cortical neurogenesis, when the direct mode predominates (Tuoc et al., 2013) (Fig. 1A). Using in vivo magnetic resonance imaging, we obtained accurate measurements showing that the thickness, volume, and surface area of cerebral cortex were greatly increased in BAF170cKO mice compared with the wild-type (WT) control. We were also able to show that the loss of BAF170 in BAF170cKO mutants led to the incorporation of additional BAF155 subunit(s) into the BAF complex, thereby promoting the euchromatin state and enhancing the binding efficiency of TF Pax6 to its targets, including Tbr2, Cux1, and Tle1, which are expressed in IPs, late RGs, and neurons with upper layer identities generated during late (indirect) neurogenesis. Moreover, the enlarged cortical size due to the conditional inactivation of BAF170 was almost fully rescued in Pax6-deficient background (Pax6cKO). Thus, the generated cortex-specific double conditional KO for Pax6 and BAF170 (dcKO) showed a similar number of Tbr2+ IPs as in the cortex of WT mice,²⁸ indicating that Pax6 and BAF170-dependent regulation of Tbr2 expression mediates cortical size alteration.

To further investigate whether the enhanced number of Tbr2+ IPs in BAF170cKO cortex is causally related to Pax6 function, we sought to examine whether Pax6 overexpression (Pax6OE) in the background of BAF170 knockdown (BAF170KD) in vivo might synergistically enhance IP genesis (Fig. 2A and B). To study this possibility, we electroporated silencing plasmids containing small hairpin RNAs (shRNAs) against BAF170 (shBAF170#1,#2) and Pax6 expression plasmids (CMV-Pax6) into E13.5 cortices via in utero electroporation (IUE) and examined the genesis of Tbr2+ IPs 1–2 days later (Fig. 2C and D). Consistent with published results from transgenic Pax6 gain-of-function (GOF)^20,29 and BAF170 loss-of-function (LOF) models,²⁸ we found that both the overexpression of Pax6 and the silencing of BAF170 led to an increase in the number of Tbr2+ IPs in the developing cortex. Strikingly, the overexpression of Pax6 in a BAF170KD background increased the number of Tbr2+ IPs nearly 2-fold compared with controls, producing a more pronounced effect than that observed in cortices either after activation of Pax6 (CMV-Pax6) or knockdown of BAF170 (BAF170KD) (Fig. 2C and D). Taken together, these new results further highlight the fact that the regulation of Tbr2 expression and IP genesis directly depend on genetic interactions between TF Pax6 and BAF170. In addition, these findings indicate that manipulating the endogenous expression level of chromatin remodeling factor BAF170 and TF Pax6 in cortical RG cells may provide a means for generating larger numbers of neuronal progenitors (IPs) and neurons, possibly offering a potential for therapeutic strategies.

Figure 2. Interaction between TF Pax6 and BAF170 controls the generation of Tbr2+ IPs. (A) Manipulation of endogenous expression level of Pax6 and BAF170 in RGs via in utero electroporation (IUE). The indicated plasmids were electroporated in utero into E13.5 cortices and analyzed 1 d later. After double immunostaining with GFP/Pax6 or GFP/BAF170 antibodies, the strength of the Pax6 or BAF170-fluorescent signal was quantified in confocal microscopy images using ImageJ software (NIH), as previously described.⁵⁰ Upon electroporation of CMV-Pax6 expression vector (left side), based on the detected strength of the fluorescent signal, 3 subpopulations of Pax6+ progenitors (Pax6 extreme high, Pax6 high, and Pax6 low as pointed by filled and empty arrows, and arrowheads, respectively) were detected. Two populations of BAF170+ progenitors (BAF170 high, BAF170 low as pointed by empty arrows, and arrowheads, respectively) were easily distinguishable upon electroporation of the hairpin BAF170 construct (shBAF170, right side). The filled and empty arrows point to GFP+ and GFP-cells, respectively. (B) Quantitative estimation of the effectiveness of BAF170 gain-and loss-of-function experiments. The diagram represents the relative percentage (compared with EV-only controls) of Pax6 extremely high+ cells in the Pax6-overexpression experiment (achieved via electroporation of CMV-Pax6 plasmid), and BAF170 high+ cells in the BAF170-knockdown experiment (via shBAF170/#1,#2 vectors) as shown in (A). The results indicate that under the experimental conditions a successful overexpression of PAX6 (by CMV-Pax6) and knockdown (KD), shBAF170#1) via IUE was achieved. (C) The indicated plasmids were electroporated into the cortices of E13.5 embryos, and 1 d later, tissues were harvested and processed for double immunostaining with GFP/Tbr2 antibodies. Note the significant increase of Tbr2 immunoreactivity (Tbr2+IPs) upon overexpression of Pax6 (CMV-Pax6) in the BAF170KD cortex (shBAF170/CMV-Pax6). (D) The diagrams present a statistic comparison of the number of GFP+/Tbr2+ cells in the above described experiment. Note that overexpression of Pax6 (CMV-Pax6) in the background of BAF170 knockdown (shBAF170#1) synergistically enhances the generation of Tbr2+ IPs. Values are presented as means ± SEMs (n = 4). Scale bars = 250 µm.

Using a comparative analysis of genes regulated by BAF170 and Pax6, we showed previously that most of the targets that are positively regulated by TF Pax6, are repressed upon BAF170 overexpression.^28,29 Among them, many are known to have important roles in neural development, genesis of IPs, and cortical layer formation such as Tbr2,³⁰ Cux1,³¹ Tle1,³² Er81,³³ Ngn2,³⁴ AP2ɤ.³⁵ In addition, Pax6 and BAF170-dependent expression of some genes, such as Trnp1, Cdk4, and cyclin D1, were recently identified as key factors in controlling the expansion of cortical size.^36,37 Further studies are needed to determine whether Pax6 directly recruits BAF170 to the promoters of these genes to regulate their expression. It is important to note, however, that despite the massive expansion of the cortical surface of BAF170cKO mice, we never detected signs of cortical folding,²⁸ which had been recently demonstrated in the cortex of Trnp1-deficient mice (Stahl et al., 2013). Thus, although dramatic expansion of IPs may contribute to an increase in cortical thickness and surface, this alone seems insufficient to cause gyrencephaly in naturally lissencephalic species like mouse.

BAF170 controls the body size of mouse embryo

The intriguing observation that the replacement of BAF170 with the BAF155 subunit in the BAF170-deficient cortex severely affected neurogenesis and cortical morphology prompted us to study the consequences of BAF170LOF in the body.

The in situ hybridization (ISH) analysis with a BAF170-specific probe revealed abundant accumulation of BAF170 transcripts in the entire central nervous system (CNS) of E14.5 embryos, while the other organs showed only scarce expression. Immunohistochemistry (IHC) with BAF170 antibody revealed that the BAF170 protein is predominantly expressed in the entire CNS, including the dorsal root ganglions of the spinal cord as well as the nasal cartilages, and at a much lower level in most other organs (e.g., tongue) (Fig. 3A, A', and A"). Based on the genepaint expression pattern database³⁸ at E14.5, the BAF155 subunit shows a homogenous expression in the whole body. Complimentary to the distribution of BAF170 protein in the embryo body, IHC with BAF155 antibody disclosed expression mostly confined to tissues outside the CNS (Fig. 3C–C").

Figure 3. Complimentary expression pattern of BAF170 and BAF155 in the body of developing embryo. Immunohistochemical analysis (IHC) with BAF170 and BAF155 antibodies was performed on E13.5 sagittal sections of control (CMV-Cre) (A and C) and BAF170 full KO mice (BAF170fl/fl;CMV-Cre)(B and D). Images on the right side show spinal cord and tongue tissues in higher magnification (A"–D"). The recombination through CMV-Cre leads to full loss of BAF170 expression in the embryos. Note that similar to upregulation of BAF155 in the cortex-specific BAF170cKO,²⁸ the global loss of BAF170 led to an obvious upregulation of BAF155 expression in the entire CNS. Scale bars = 500 µm.

We have presented evidence that during early cortical neurogenesis, competition between BAF170 and BAF155 subunits has a strong modulatory role on the mode of RG cell division and cortical expansion.²⁸ To address whether a similar subunit interaction may define the embryo size as well, we generated and analyzed a full BAF170KO line by crossing BAF170fl/fl mice with the ubiquitous deleter CMV-Cre line, driving Cre-recombinase activation in all body cells.³⁹ The mutants died shortly after birth (at postnatal stage P0-P3) with reduced breathing capability and severe developmental retardation. As expected, the global loss of BAF170 led to a greater upregulation of BAF155 expression in the CNS (Fig. 3A, A", B, B", C, C", D, and D") and to a milder extent in the tissues outside the CNS (e.g., tongue) (Fig. 3A, A', B, B', C, C', D, and D'). Our previous results indicated that although BAF170 negatively controlled the expression of BAF155 at the protein level, the manipulation of BAF170 expression in cKO and cOE mouse mutants did not affect the transcription and promoter activity of BAF155.²⁸ Together, these findings suggest that the complimentary expression of BAF170 and BAF155 (high in CNS and low in other organs, respectively) most likely is a result of competition between these 2 subunits within the BAF complex.

To evaluate the role of altered BAF155 expression levels in BAF170 deficiency, we examined the morphology of embryos (at stages of E13.5, E15.5, E17.5) and newborn mice. Interestingly, BAF170KO embryos have an obviously bigger body size and significantly greater body weight at all examined embryonic stages (E13.5, E15.5, E17.5, and P2) as compared with the WT controls (Fig. 4).

Figure 4. BAF170 controls size and weight of embryonic body. The morphology of body size (A) and statistical analysis of total weight (B) indicated that BAF170 regulates the size and weight of the embryonic body. Values are presented as means ± SEMs (n = 6). Scale bars = 1000 µm.

It is well known that body size is predominantly controlled by growth factors and cellular signaling.^40-42 However, increasing evidence suggests that organ-intrinsic mechanisms do play a significant role in determining organ and body size.^43-45 To further investigate whether BAF170 has an intrinsic role in controlling body size, we examined the effect of BAF170 overexpression in transgenic mice. We generated mice constitutively overexpressing BAF170 (referred thereafter as BAF170OE) by crossing BAF170cOE_fl/fl²⁸ with CMV-Cre mice.³⁹ BAF170OE mice died between 1–2 months after birth. We observed smaller body size in BAF170OE embryos and mice compared with that of control mice. Also, BAF170OE animals showed significantly lower weight as compared with controls at all examined stages (E13.5, E15.5, E17.5, and P2) (Fig. 4). However, apoptosis in BAF170 mutants was not detectable.²⁸ Despite the fact that the BAF170KO and BAF170OE mice did not survive beyond stage P1-P3, thereby denying us the possibility to examine the role of BAF170 in late development, these novel data strongly suggest that the distinct presence of BAF170 vs. BAF155 subunits within the BAF complex is involved in regulating the size and weight of the whole body.

The progressive transition from embryonic stem cells (ES) to neural progenitors is accompanied by subunit exchanges within the embryonic stem (esBAF) and neural progenitor (npBAF) chromatin remodeling complexes.^46-48 This process is accompanied by the induced expression of BAF170 replacing one of the BAF155 subunits and the recruitment of Brm as the ATP-ase subunit in the npBAF complex. In our previous study, we showed that during early neurogenesis (predominantly direct mode), TF Pax6 recruits BAF170 (that is only expressed in RGs, up to stage E14.5), and specifically Brm (not Brg1) to the promoters of the Pax6 target genes (e.g., Tbr2, Cux1, Tle1 involved mostly in late and indirect neurogenesis) in order to repress their early expression.²⁸ As shown here, we observed an overall increase in body weight in the BAF170KO embryos, highly reminiscent of the phenotype of the adult Brm-null KO mice.⁴⁹ Taken together, these results allow us to assume that, similar to developing cortex, BAF170 subunit and ATPase Brm are part of the same BAF complex and possibly play similar roles in controlling cell proliferation, organ size, and body shape. In future studies, it will be interesting to decipher the interplay between BAF chromatin remodeling complexes and transcription factors in the regulation of different organs and body size.

Conclusion

A wealth of evidence indicates that as a part of epigenetic control, chromatin remodeling exerts a transcriptional control on the expression of specific sets of genes in distinct cell types and certain developmental stages. Based on our recent findings, we outlined here one of the possible mechanisms through which a specific subunit assembly of the BAF complex modulates the generation of IPs during cortical neurogenesis, which ultimately affects cortical volume and size. The process includes the timely binding of the BAF170-, BAF155-subunits, and Brm-ATPase of the BAF complex to TF Pax6, and the recruitment of the REST-corepressor complex to promoters of Pax6 downstream target genes involved in specification of IPs and late progenitors, generating mostly upper layer neuronal fate. Additionally, we also found that in the embryo body, variation of BAF170 expression level causes a complimentary change in the BAF subunit composition, seemingly mostly affecting the expression level of BAF155, thus controlling the size of the whole body. Our novel findings suggest that competition between the 2 paralogs, BAF170 and BAF155, is a part of the molecular mechanism, which plays a significant role in organogenesis during embryonic development.

Materials and Methods

Animals and in utero electroporation

BAF170^fl/fl28, BAF17OE,²⁸ and CMV-Cre³⁹ mice were maintained in a C57BL6/J background. In utero electroporation was performed as described previously.⁵⁰ Plasmids that were used for the electroporation: pCAX-Pax6 (mouse Pax6 cDNA in pCAX, a gift from Dr Noriko Osumi, Tohoku University), shBAF170 (target sequences for mouse BAF170; SABiosciences). Animals were handled in accordance with the German Animal Protection Law and with the permission of the Bezirksregierung Braunschweig.

Immunohistochemistry (IHC)

IHC was performed as previously described.¹⁹ Primary polyclonal (pAb) and monoclonal (mAb) antibodies were used in this study (working dilution; sources): Pax6 mAb (1:200; DSHB), Pax6 rabbit pAb (1:200; BABCO), BAF170 rabbit pAb (1:100; Bethyl), BAF155 mouse mAb (Santa Cruz), Tbr2 rabbit pAb (1:300; Chemicon), GFP chick pAb (1:1000; Abcam). Alexa 488, Alexa 568 (1:400; Invitrogen).

Image acquisition and statistical analysis

All images were acquired with standard (Leica DM 6000) and confocal (Leica TCS SP5) fluorescence microscopes. Images were further analyzed with Adobe Photoshop. Statistical analyses were done using Student t-test and Mann-Whitney test. All graphs were plotted as mean ± SEM.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.