Senp1 expression is down regulated in response to membrane depolarization
To better understand the molecular mechanisms that mediate transcriptional repression in response to membrane depolarization in neurons, we evaluated the expression levels of several chromatin and epigenetic modulators in mouse primary cortical neurons depolarized with potassium chloride (KCl). In this model system, KCl-mediated depolarization leads to an influx of calcium through the L-type voltage-sensitive channels that potentiates changes in gene expression patterns [24]. KCl-depolarized neurons increase the expression of Bdnf and CREB phosphorylation, which is consistent with other studies (Additional file 1: Figure S1a and S1b). By comparing their relative mRNA levels of KCl-stimulated neurons versus un-stimulated neurons, we found that only the Senp1 and Tet1 genes were significantly reduced by membrane depolarization among several tested genes (Additional file 1: Figure S1c and S1d). In addition, using two different doses of KCl (25 mM and 60 mM) at different times (2 hrs and 24 hrs), we found that both treatments reduced the expression of Senp1 (Fig. 1a). Western blot analysis confirmed the silencing of Senp1 at the protein level following KCl treatment (Fig. 1b). In contrast, TTX treatment increased the expression of Senp1 (Additional file 1: Figure S1e). Recent work has shown that Tet1, an enzyme that converts the repressive 5′-methyl cytosine to the active 5′-hydroxymethylation DNA modification, is repressed after neuronal stimulation and/or fear conditioning [26], consistent with our results. In the current study, however, we have focused only on Senp1 expression. Taken together, these results show that the expression of SUMO-specific protease gene Senp1 is regulated by neuronal activity, with membrane depolarization reducing its expression.
Yy1 activates Senp1 transcription
To further study the underlying mechanisms, we first performed an in silico analysis of the Senp1 promoter using Jasper (http://jaspar.binf.ku.dk/) to identify potential DNA-specific binding transcriptional regulators. Our search results revealed numerous Yy1 binding sites within a 2.5-kilobase (kb) region upstream of the Senp1 transcriptional start site (TSS) (Fig. 2a). We deduced a consensus Yy1 motif for the Senp1 region (Fig. 2b). To determine whether Yy1 associates with the promoter region of Senp1 in neurons, quantitative chromatin immunoprecipitation (qChIP) was performed on chromatin prepared from adult whole mouse brain at two different Senp1 amplicons within the promoter region (situated at 80- and 1894-basepair (bp) upstream of the TSS, respectively) (Fig. 2a). Our results reveal Yy1 is enriched at the Senp1 promoter in neurons as compared with control at both the − 1894 and − 80 regions (Fig. 2c).
Yy1 can function as either a transcription repressor or activator [31]. To determine its transcription regulatory function on the Senp1 promoter, we co-transfected resting neurons with full-length myc-tagged Yy1 expression constructs and a plasmid harboring a 2541-bp fragment upstream of the mouse Senp1 TSS region driving luciferase expression. The co-expression of Yy1 stimulated luciferase activity as compared to control (Fig. 2d), indicating that Yy1 is a transcriptional activator for the Senp1 promoter. To establish that Yy1 is necessary for Senp1 transcription, we constructed short hairpin Yy1 RNAs (shRNAs) targeting three different regions within Yy1 and transfected these vectors into Neuro2A cells. As shown in Fig. 2e, shYy1–1, shYy1–2, and shYy1–3 deplete the Yy1 protein. The knockdown of Yy1 compromises the levels of Senp1 mRNA (Fig. 2f). Taken together, these results show that Yy1 activates Senp1 transcription in non-depolarized cells by directly binding the Senp1 promoter.
Yy1 forms a complex with Brd4 on the Senp1 promoter
The protein interaction partners of Yy1 are important determinants of its transcriptional activity [31]. A recent report showed that Yy1 interacts with Brd4 [32], a member of the BET family that contain two bromodomains and recognizes acetylated histones. Brd4 recruits Positive Elongation Factor, pTEF-b, phosphorylating RNA polymerase II (RNAP Pol II) Serine 2, which results in a release from gene pausing and facilitates productive gene transcriptional elongation [33, 34]. RNAP Pol II pausing is essential for the rapid induction of immediate early genes in response to stimuli in neurons [35]. To address whether Brd4 also participates in the depolarization-induced Senp1 downregulation, we first confirmed the physical interaction between Yy1 and Brd4 by co-immunoprecipitation (Co-IP) in neurons (Fig. 3a). With qChIP, we then established that Brd4 occupancy at the Senp1 promoter was altered following Yy1 knockdown (Fig. 3b, c). Consistent with a role for Brd4 in activating gene transcription, we found that Senp1 expression was compromised following Brd4 knockdown (Fig. 3d, e). When neuronal Brd4 was inhibited by the small molecule JQ1 which prohibits the binding of Brd4 bromodomain to acetylated histones [28, 36], KCl treatment did not further reduce the Senp1 levels suggesting that the depletion of Brd4 could be a major factor for neuronal depolarization induced Senp1 repression (Fig. 3f). Collectively our results demonstrate that Yy1 targets the BET family member Brd4 to the Senp1 promoter activating its transcription.
Membrane depolarization evicts the Yy1/Brd4 complex from the Senp1 promoter
To determine the role for Yy1 in Senp1 repression during neuronal depolarization, we asked whether membrane depolarization could directly alter Yy1 transcription. Yy1 mRNA levels increase 2 hour following depolarization (Fig. 3g). In contrast, we did not observe a change in the Yy1 protein levels after 2 hr of KCl treatment (data not shown). To resolve the paradox between transcriptional up-regulation of Yy1 and Senp1 down-regulation upon neuronal activity, we examined Yy1 occupancy on the Senp1 promoter in vivo using qChIP after membrane depolarization of primary mouse cortical neurons. We interrogated two different regions (− 1894 and − 80) upstream of the Senp1 TSS for Yy1 occupancy. When compared to the un-stimulated neurons, the enrichment of Yy1 on both Senp1 chromatin sites is greatly depleted after depolarization (Fig. 3h).
Because Yy1 recruits Brd4 to the Senp1 promoter, we examined whether membrane depolarization might also affect Brd4’s occupancy at these regions. qChIP was performed using Brd4 antibodies on chromatin prepared from un-stimulated and depolarized neurons. Our qChIP analysis showed that the in vivo binding of Brd4 to the Senp1 chromatin was depleted in response to neuronal activity (Fig. 3i). We also observed a dramatic reduction of Histone 4 acetylation (H4Ac) on these regions after neuronal depolarization (Fig. 3j). Our findings agree with previous studies suggesting that histone acetylation is dynamic and regulated by neuronal activity [37]. Taken together, we show that both Yy1 and Brd4 proteins are depleted from the Senp1 promoter upon membrane depolarization, and the histone PTM, H4 acetylation is reduced.
Membrane depolarization decreases Yy1 phosphorylation status resulting in its removal from the Senp1 promoter
One way that neurons employ for a rapid modulation of their transcription program in response to neuronal activity is by regulating the post-translational modifications (PTMs) of epistatic transcription factors. Yy1 can be modified by multiple PTMs, such as phosphorylation, acetylation, SUMOylation and ubiquitination [29, 37,38,39,40,41,42]. These modifications modulate either Yy1’s binding ability to DNA or interaction with protein co-factors [29, 37, 42, 43]. Previous studies have shown that Yy1 phosphorylation modulates its binding to the Egr2 and Talin2 promoters, and murine leukaemia virus long terminal repeat [38, 44, 45]. We asked whether the loss of Yy1 binding to Senp1 promoter upon neuronal depolarization might result from alterations in Yy1 phosphorylation levels induced by neuronal activity. To address this, we immunoprecipitated endogenous Yy1 from resting and depolarized primary neurons and checked its phosphorylation status. Western blot of the immunoprecipitants with anti-phospho-Serine antibodies suggests that neuronal activity reduces Yy1 phosphorylation (Fig. 4a).
During Schwann cell differentiation, Yy1 can be phosphorylated by MEK kinase on serine 184 (S184) and 247 (S247), regulating Egr2 expression in peripheral nerve myelination [44]. These Yy1 serine residues are conserved in vertebrate species, which underscores their importance. We next tested whether these Yy1 phosphorylation sites could regulate its binding to the Senp1 promoter. First, these two phosphorylation acceptor sites (S184 and S247) were mutated to alanine, creating the Yy1 mutant (Myc-Yy1-S184, 247A) (Fig. 4b). Yy1 phosphorylation was greatly compromised in the mutated form (Fig. 4b). Both wild type and mutated Myc-tagged Yy1 were immunopurified from Neuro2A cells following transfection and immobilized on the agarose beads (Fig. 4c, top). Their DNA binding abilities to Senp1 promoter region were tested in vitro [30]. Yy1 occupancy at the Senp1 promoter was abolished when the phosphorylation acceptor sites were mutated (Fig. 4c, bottom), arguing that Yy1 phosphorylation stabilizes its binding to the DNA of Senp1 promoter. Consistent with this, the wild type Yy1 activates Senp1 promoter in a luciferase reporter assay, whereas, the Yy1 mutant lost its ability to activate this reporter (Fig. 4d). Altogether, our results show that upon neuronal activity, the phosphorylation levels of Yy1 are reduced, leading to its deprivation from the Senp1 promoter and reduced Senp1 transcription. We identify two serine residues, S184 and S247, within the Yy1 protein that are crucial for the phosphorylation-dependent alterations in Senp1 transcription.
The PP1/PP2A phosphatases modulate Yy1 in depolarized neurons
Since Yy1 phosphorylation is critical for binding to the Senp1 promoter, we asked how membrane depolarization regulated Yy1’s phosphorylation status. Because Yy1 phosphorylation is inhibited in primary cortical neurons upon stimulation by KCl, we speculated that this change might result from the up-regulation of protein phosphatase activities. Neuronal activity enhances protein phosphatases 1 and 2A (PP1/PP2A) and protein phosphatase 2B (PP2B; also known as Calcineurin) [14], so we tested whether the depolarization-induced Senp1 repression was dependent on these enzymes. Okadaic acid (OA) and cyclosporin-A (CsA) can inhibit the PP1/PP2A and PP2B phophatases, respectively [14]. Primary neurons were treated with these pharmacologic inhibitors before depolarization and their effect on the transcription of Senp1 was measured by RT-qPCR. Okadaic acid treatment reversed the decrease in Senp1 levels after depolarization (compare the DMSO vehicle versus OA treatment) (Fig. 4e). In contrast, CsA treatment did not attenuate the repression of Senp1 caused by neuronal depolarization (Fig. 4e).
We then examined whether inhibition of PP1/PP2A phosphatases enhance Yy1 phosphorylation. Endogenous Yy1 proteins were immunoprecipiated from primary cortical neurons either with or without Okadaic acid treatment. We observed that inhibition of PP1/PP2A phosphatase activity enhances levels of phosphorylated Yy1 (Fig. 4a). Collectively, our results implicate that PP1/PP2A dephosphorylate Yy1 and that their phosphatase activity is indispensable for the neuronal activity-induced Senp1 deactivation.
Depletion of Yy1 reduces GluR1 protein expression levels
Senp1 is important for homeostatic synaptic scaling, controlling the trafficking of AMPA receptors (AMPARs), particularly the GluR1 subunit [20, 21]. The over-expression of Senp1 prevents the increased GluR1 surface levels following glycine-induced AMPAR expression [21]. We asked whether Yy1 could alter the total cellular levels of GluR1. Knockdown of Yy1 depleted the total GluR1 protein levels in primary cortical neurons (Fig. 5a). Interestingly, while Senp1 transcription is reduced, depletion of Yy1 did not significantly affect the mRNA levels of the GluR1 (also known as Gria1) gene (Fig. 5b), suggesting that Yy1 does not directly regulate GluR1 transcription. To further support that Yy1 regulates the expression of GluR1 at the protein level, primary cortical neurons were transfected with two different Yy1 shRNAs that have a coupled GFP expression enabling the identification of individual transfected neurons to knockdown Yy1. GluR1 immunostaining was performed on these neurons. Following Yy1 depletion we see a diminution of GluR1 immunostaining (Fig. 5c). The intensity of GluR1 staining was quantified and a statistically significant decrease in GluR1 signals was observed in Yy1 knockdown neurons (Fig. 5d). In addition, surface GluR1 was detected in neurons under non-permeant conditions and we found that Yy1 depletion also reduced the surface expression of GluR1 (Additional file 2: Figure S2). These findings suggest that Yy1 controls the expression of the GluR1 protein in neurons.
Senp1 overexpression rescues GluR1 levels following Yy1 depletion
Because Yy1 does not regulate GluR1 mRNA, we queried how Yy1 contributes to GluR1 expression in neurons. Previous studies have suggested that Senp1 plays a role in the localization of GluR1. As our above results indicate that Yy1 can regulate Senp1 expression, we questioned whether Senp1 also modulates the total amount of GluR1 in addition to its localization. We first transfected primary cortical neurons with a control or Flag-Senp1 expression plasmids under depolarizing or non-depolarizing conditions. Consistent with previous studies [10,11,12,13], neuronal activity reduces GluR1 expression (Fig. 6a). Interestingly, the over-expression of wild type Senp1 under depolarizing conditions rescued the expression of GluR1 (Fig. 6a). In contrast, depletion of Senp1 by siRNA reduced GluR1 expression (Fig. 6b).
Next, we asked whether Senp1 mediated Yy1’s regulation of GluR1 expression. To address this, we depleted Yy1 by siRNA and over-expressed a Flag-tagged full-length Senp1 protein in resting neurons (Fig. 6c). Yy1 knockdown reduced the levels of GluR1 protein while the overexpression of Flag-Senp1 indeed rescued the GluR1 levels (Fig. 6c). The over-expression of Senp1 protein did not affect the total level of endogenous Yy1 (Fig. 6c). Altogether, we determined that Yy1 controls GluR1 protein levels through Senp1 expression.