Neural Differentiation in HDAC1-Depleted Cells Is Accompanied by Coilin Downregulation and the Accumulation of Cajal Bodies in Nucleoli

Cajal bodies (CBs) are important compartments containing accumulated proteins that preferentially regulate RNA-related nuclear events, including splicing. Here, we studied the nuclear distribution pattern of CBs in neurogenesis. In adult brains, coilin was present at a high density, but CB formation was absent in the nuclei of the choroid plexus of the lateral ventricles. Cells of the adult hippocampus were characterized by a crescent-like morphology of coilin protein. We additionally observed a 70 kDa splice variant of coilin in adult mouse brains, which was different to embryonic brains and mouse pluripotent embryonic stem cells (mESCs), characterized by the 80 kDa standard variant of coilin. Here, we also showed that depletion of coilin is induced during neural differentiation and HDAC1 deficiency in mESCs caused coilin accumulation inside the fibrillarin-positive region of the nucleoli. A similar distribution pattern was observed in adult brain hippocampi, characterized by lower levels of both coilin and HDAC1. In summary, we observed that neural differentiation and HDAC1 deficiency lead to coilin depletion and coilin accumulation in body-like structures inside the nucleoli.


Introduction
Cajal bodies (CBs) are striking nuclear structures consisting of accumulated proteins that play various roles in nuclear processes. These structures were designated Cajal's accessory bodies (cuerpo accesorio) and were discovered for the first time in rat brain neurons [1]. A role of the CBs during neurogenesis was also significantly studied and summarized by Lafarga et al. [2] and Baltanás et al. [3]. At this moment, it is well known that the function of these structures is dynamic because CBs regulate RNA synthesis and the assembly of ribonucleoproteins (RNPs) [4]. Moreover, Tapia et al. [5] showed that the symmetrical dimethylation of arginines on coilin supports the formation of CBs, positive on survival motor neuron (SMN) proteins and small nuclear ribonucleoproteins (snRNPs). These regulatory factors probably determine the association of CBs with the spliceosome and a role for CBs in pre-mRNA splicing [6]. Conversely, coilin hypomethylation depreciates its function and causes the disintegration of canonical CBs into small microfoci. Unmethylated coilin does not support the formation of robust CBs but is located inside the dense fibrillar component of the nucleoli. In this form, there is no link between the coilin nuclear pattern and global transcription activity [5]. On the other hand, canonical CBs, which are nonmembrane nuclear components, are prominent structures in dividing cells with high transcriptional activity [4]. CBs have a diameter of 0.5-1.0 m and contain many proteins, including the abovementioned p80 coilin, which becomes increasingly more phosphorylated during mitosis and, particularly in human embryonic stem cells, is present at high levels in the nucleoplasmic pool [7][8][9][10][11][12]. However, coilin is not completely essential because knockout of coilin in mice is not lethal [13]. On the other hand, coilinpositive CBs play an important role in genome organization in terms of gene expression and pre-mRNA splicing via their association with many chromosomes. The periphery of 2 Stem Cells International these chromosomes represents a site of interaction for genes that are poised for transcription and thus associates with regulatory components. Human chromosome 1 is a key player in these processes, and its periphery is frequently occupied by CBs [14]. The use of chromosome conformation capture analysis (4C-seq), a novel molecular biology method, has revealed an association between highly expressed histone genes, sn/snoRNA coding loci, and CBs, which are involved in intraand interchromosomal clusters [14,15]. This interaction is of immense functional importance during transcription and especially splicing. CBs are also highly mobile structures, as revealed by single-particle tracking analysis and fluorescence recovery after photobleaching (FRAP) [10,[16][17][18][19]. For example, we recently demonstrated the constrained local motion of individual CBs after cell exposure to -radiation. Furthermore, in mouse embryonic stem cells (mESCs), the coilin dispersed in the nucleoli and accumulated in CBs was characterized by a reduced mobile fraction compared to the GFPtagged coilin in the nucleoplasm [10]. FRET (fluorescence resonance energy transfer) analysis additionally revealed a specific interaction between coilin and SMN protein in CBs as well as the appearance of coilin-coilin dimerization [17]. However, as regards to DNA repair machinery, our experiments did not show coilin-SMN interaction or coilin dimerization in UVA-induced DNA lesions, which are characterized by pronounced coilin recruitment [10]. Together, the abovementioned results illustrate the dynamic behavior of coilin and CBs, which is required not only for optimal pre-mRNA processing but also for DNA repair [15].
Interestingly, in some tumor cells, the functional properties of coilin are associated with both CBs and nucleoli. The nucleoli contain many different proteins that play a role during the transcription of ribosomal genes or during DNA repair [18][19][20]. In UVA-damaged chromatin, we observed the recruitment of the upstream binding factor UBF, a major transcription factor for ribosomal genes, and we noted a similar response for coilin [10,21]. As determined by Boulon et al. [22], UVA and UVC cause the disintegration of coilinpositive CBs, and ionizing irradiation has a similar, notable effect of CB disruption [23,24]. Thus, nucleolar proteins, including coilin that also appears in nucleoli of tumor cells, appear to be involved in the DNA repair machinery, which is, for example, also activated in Purkinje cells during neurodegeneration, characterized by the disintegration of nucleoli and CBs [3].
In this study, we focused especially on the nuclear distribution patterns of the CBs, and we studied coilin levels in embryonic and adult mouse brains and during neural differentiation of mESCs. Based on the initial observations of Raymond Cajal, who noted that the CBs are striking nuclear components of the rat brain and, more specifically, the pyramidal cells of the human cerebral cortex [1,25], we analyzed the nuclear distribution patterns and formation of the CBs in the hippocampus and olfactory bulbs (OBs) of adult mouse brains. We also investigated the distribution of coilin in the ventricular ependyma of e15.5 embryonic brains. Furthermore, to elucidate the CB dynamics in neurogenesis, we analyzed the formation of CBs during the neural differentiation of wild-type (wt) and HDAC1 double-knockout (dn) mESCs. From the view of neural differentiation, it was shown that embryonic neural progenitor stem cells are characterized by a high level of HDAC1, while HDAC2 is expressed during neural differentiation and pronouncedly in terminally differentiated neurons [26]. Differentiation processes in the brain are also regulated by HDAC3, as shown by Volmar and Wahlestedt [27]. Moreover, in neural progenitor stem cells, functional HDAC3 was found to be responsible for the balance between cell proliferation and differentiation [28]. Based on these data we addressed the following hypothesis: whether neural differentiation and HDAC1 depletion can affect the levels of coilin and the nuclear distribution of Cajal bodies because we expected that depletion of some HDAC induces chromatin relaxation; thus this nuclear event could change distribution pattern of CBs. We also analyzed HDAC1 depletion in order to show how changes in histone acetylation, a central epigenetic factor responsible for chromatin accessibility [29,30], can change the level of coilin, which is methylated when it accumulates in CBs [5].

The Nuclear Distribution Pattern of Cajal Bodies in the
Embryonic and Adult Mouse Brain. We inspected sections of embryonic and adult mouse brains and observed the formation of single, robust CBs at the cortex periphery in embryonic brains at stage e15.5 after fertilization (Figures 1(a)-1(c)). We additionally found that in approximately 90% of the cell nuclei at the cortex periphery, the Cajal bodies (CBs) were located away from clusters of centromeric heterochromatin called chromocenters (Figure 1(c)).
The cell nuclei in adult brains were highly positive for the coilin protein, particularly in the chondroid plexus of the lateral ventricle (Figure 2(a)). However, the cells in this region did not have easily discernable CBs. Next, we observed clustering of coilin inside the cell nuclei occupying the cortex periphery in adult mouse brains (Figure 2

Levels and Nuclear Distribution Pattern of Coilin, Fibrillarin, and SC35 in Mouse Brain and Pluripotent or Differentiated mESCs.
In comparison to nondifferentiated and differentiated wt mESCs, pan-acetylation of lysines was very high in HDAC1 dn mESCs and their differentiated counterpart (Figure 4(a)). In these experiments, we addressed a question if hyperacetylated surroundings of CBs in HDAC1 dn mESCs could change formation or maintenance of CBs, which is regulated by methylation-related processes [5].
Here, western blot analysis revealed reduced levels of coilin (80 kDa) during neural differentiation of wt mESCs (Figures 4(b) and 4(e)(A)). We also analyzed the levels of coilin in nondifferentiated and differentiated wt and HDAC1 dn mES cells. Our statistical analysis, using Student's -test, documented significant changes at * ≤ 0.05 when we compared nondifferentiated and differentiated wt mESCs (Figures 4(b) and 4(e)(A)). In HDAC1-depleted cells, the difference was even more pronounced: a significantly different result (at * * ≤ 0.01) was found when we compared nondifferentiated and differentiated HDAC1 dn cells (Figures 4(b) and 4(e)(A)). We also examined the coilin levels in mouse brains at various developmental stages. We studied the whole brains of e13.5, e15.5, and e18.5 embryonic stages and adult mice (Figure 4(c)). Compared to embryonic brains, which are characterized by the 80 kDa coilin variant, we observed a different splice variant of coilin (∼70 kDa) in adult brains. During mouse brain development, coilin levels were stable at the e13.5, e15.5, and e18.5 developmental stages. Interestingly, mouse ESCs were characterized by a very low level of 80 kDa coilin in comparison to embryonic brains (Figure 4(c)). In parallel with coilin, we analyzed fibrillarin Coilin distribution in the cortex periphery of an adult brain. ((c)(A), (c)(B)) Hippocampal regions visualized by hematoxylin-eosin staining, an image from the brain atlas (see [31]). (d-f) Accumulation of coilin in crescent-like foci in the hippocampal region of an adult brain. DAPI staining is used to visualize cell nuclei. Coilin (red) was labeled by a secondary antibody conjugated with Alexa 594. Nuclear distribution of coilin in cells 1 and 2 (e) is shown in graphs 1 and 2. Fluorescence intensity along white lines with arrows was measured using the Image J software (NIH freeware). (f) shows a high density of coilin in the hippocampus (hippocampal blade) of an adult mouse brain. levels in the mouse brains because individual CBs colocalize with fibrillarin in many cell types (Figures 4(c), 5(a), and 5(b)). By western blots, in mouse adult brains, we observed a very low level of fibrillarin (see two western blot expositions in Figure 4(c)), especially compared to mouse embryonic stem cells (mESCs). In our samples, shown in Figure 4(c), we found that when the level of coilin was high, the level of fibrillarin was low and vice versa.
Using western blot, we also detected the levels of 70 kDa coilin variant and 39 kDa fibrillarin in the OBs of the adult brain, the adult hippocampus, the brain cortex, and the whole adult brain (Figures 4(d) and 4(e)(B)). Compared to OBs, the hippocampus and the brain cortex were characterized by coilin depletion, which was accompanied by a decrease in HDAC1 level when it was normalized to total protein level and -tubulin (Figures 4(d), 4(e)(B), and 4(e)(C)). The fibrillarin levels were not substantially different in the brain regions analyzed (Figure 4(d)).
Here, we also compared the nuclear pattern of CBs in mESCs and the human cancer cells line HeLa, which has been used by many authors for CBs studies [17,33]. In HeLa cells, the CBs were always positive for both coilin and    Because CBs are nuclear regions associated with splicing processes, we additionally analyzed the spatial link between CBs and SC35-positive nuclear speckles (Figures 6(a)-6(f)). In mouse pluripotent ESC colonies, we observed high levels of coilin in the nuclear interior, and these regions were surrounded by the SC35 protein (Figures 6(a)-6(c)). We found that most of the CBs and SC35-positive nuclear speckles were spatially distinct, but some of them partially overlapped. This nuclear distribution pattern was identical in both wt and HDAC1 dn (Figures 6(a)-6(f)).

Discussion
CBs, which were first described by Cajal [1], consist of several proteins, including p80 coilin. The functional properties of coilin in CBs were characterized by Andrade et al. [34] and Raška et al. [12]. CBs are also the sites for various factors that play roles during pre-mRNA splicing, pre-ribosomal RNA processing, and histone pre-mRNA maturation [7,33,35]. Moreover, CBs are highly mobile structures, as demonstrated by photobleaching experiments [16,17].
Here, we addressed the morphology of CBs in embryonic and adult brains and during the in vitro induction of mESC neural differentiation. Previously, in certain human and mouse ESCs (particularly at the periphery of mESC colonies), we observed the accumulation of coilin into visible CBs [10]. Conversely, human and mouse pluripotent ESCs, particularly those at the center of the colony, are highly positive for diffusely dispersed coilin protein (Figures 6(a) and 6(b); [10]). Thus, our results indicate that the peripheries of ESC colonies are more prone to spontaneous differentiation, which is characterized by an appearance of CBs [36]. Here, the formation of robust CBs or coilin-positive microfoci was more pronounced after induced neural differentiation, especially in HDAC1 dn ES cells (compare Figures 5(b) and 5(c)(A)-5(c)(D)). Our analyses confirmed that embryonic stem cells, characterized by an immense differentiation potential, are a good tool for the studies of nuclear architecture. For example, Butler et al. [37] showed the formation of CBs as a consequence of the spontaneous differentiation that frequently appears at the periphery of human ESC colonies, and it is also the case documented here. A good experimental model in which the formation of CBs is studied are the cells of the embryonic brain (particularly the cells in the prominent neurogenic regions destined for pronounced differentiation). For our analysis, we selected the hippocampus and the OBs (Figures 2(c)-2(f) and 3(b)(A)-3(b)(E)). Our data fit well with the original observations of Cajal, who noted the appearance of CBs in primary cells, such as the pyramidal cells from the human cerebral cortex (see also [8]) and the cells undergoing terminal differentiation (Figures 2(d), 3(b)(C), and 5(c)(A)-5(c)(D)). Here, for the first time, we show the accumulation of coilin in a crescent-like structure that is specific to the hippocampal regions of the adult brain (Figures 2(d) and 2(e)). Furthermore, OBs were characterized by well-visible CBs surrounded by high levels of coilin (Figures 3(b) These results support the conclusions of other authors that noted cell-type specificity regarding the size, morphology, and numbers of CBs [15,[38][39][40][41][42][43][44].
Here, we additionally revealed a link among the focal accumulation of coilin in the nucleolus, decreased coilin levels and HDAC1 depletion. This connection was particularly observed during the neural differentiation of mESCs and in the hippocampus (Figures 2(d), 2(e), 4(b), 4(d), 4(e)(A)-4(e)(C), 5(c)(C), and 5(c)(D)). In these cases, coilin was depleted and accumulated into robust CBs or microfoci inside the nucleoli of cells with an HDAC1 deficiency or HDAC1 decreased level. Thus, changes in histone acetylation, mediated by HDAC1 function, likely affected the interaction between coilin and chromatin-related factors. Accumulation of coilin to the nucleoli was found to be linked to coilin hypomethylation [5]. Interestingly, both HDAC1 depletion and coilin hypomethylation likely caused the coilin transition to the fibrillarin-positive dense fibrillar component of the nucleoli (compare Figures 4(b) and 4(e)(A) with Figures  5(b), 5(c)(C), and 5(c)(D) and [5]). Moreover, it seems to be possible that coilin could be hypomethylated in hyperacetylated surroundings in the genome, which can be caused by HDAC1 depletion. This epigenetic nuclear event could also be a consequence of HDAC1-dependent changes in chromatin accessibility.

Conclusion
Nuclear bodies, including CBs, are functionally important nuclear compartments containing accumulated proteins that play roles in many nuclear processes, including transcription, splicing, and DNA repair. The morphology and nuclear distribution patterns of these nuclear bodies likely reflect their functional properties, which contribute to the molecular mechanisms that maintain the balance between cell physiology and pathophysiology. We showed here that coilin is highly expressed in brain tissue, especially in the embryonic brain. Cajal bodies, recognized by accumulated coilin, were found to be localized inside nucleoli, especially in HDAC1depleted cells, which was accompanied by coilin downregulation. These results show that epigenetic events, such as histone acetylation (or lysine pan-acetylation) affecting the accessibility of regulatory elements to chromatin, can stand behind changes in the nuclear distribution pattern of Cajal bodies.
HeLa-Fucci cells were purchased and cultivated as previously described [48].

Confocal Microscopy and Image Analysis.
We acquired images with a Leica TCS SP5 X confocal microscope (Leica Microsystems, Germany). Image acquisition was performed using a white light laser (WLL) with the following parameters: 1024 × 1024-pixel resolution, 400 Hz, bidirectional mode, and zoom 8-12. For 3D projections, we obtained 30-40 optical sections with axial steps of 0.3 m. 3D projection reconstruction was conducted using the Leica Application Suite (LAS) software. The scanning of larger biological objects, such as embryonic brain sections, was conducted in tile scanning mode with the Leica software, as previously described [49].

Statistical Analysis.
We used Excel software for data presentation. Florescence intensity and density of western blot fragments were calculated by ImageJ software (NIH freeware). Statistically significant results at ≤ 0.05 ( ≤ 0.01) are labeled by asterisks * ( * * ). Statistical analysis was performed by Student's -test, a tool of Sigma Plot 8.0 software.