Modulating the expression of Chtop, a versatile regulator of gene-specific transcription and mRNA export

ABSTRACT

Chtop binds competitively to the arginine methyltransferases PRMT1 and PRMT5, thereby promoting the asymmetric or symmetric methylation of arginine residues, respectively. In cooperation with PRMT1, Chtop activates transcription of certain gene groups, such as the estrogen-inducible genes in breast cancer cells, the 5-hydroxymethylcytosine-modified genes involved in glioblastomagenesis, or the Zbp-89-dependent genes in erythroleukemia cells. Chtop also represses expression of the fetal γ-globin gene. In addition, Chtop is a component of the TREX complex that links transcription elongation to mRNA export. The regulation of Chtop expression is, therefore, a key process during the expression of certain gene groups and pathogenesis of certain diseases. Our recent study revealed that cellular levels of Chtop are strictly autoregulated by a mechanism involving intron retention and nonsense-mediated mRNA decay. Here, we summarize roles of Chtop in gene-specific expression and highlight our recent findings concerning the autoregulation of Chtop.

Introduction

The expression of specific genes is a key event in cell differentiation and maintaining a differentiated state. These processes are often controlled by chemical modification (such as acetylation, methylation etc.) of chromatin or transcription factors. Some modifications are involved in both gene activation and repression, whereas others are involved in only one of these processes. Among chemical modifications, methylation of arginine residues, which is found frequently within glycine-arginine-rich (GAR) domains, is involved in both; asymmetrical methylation of arginine (ω-N^G, N^G-dimethylated arginine) is involved only in transcriptional activation [1], whereas symmetrical methylation of arginine (ω-N^G, N^'G-dimethylated arginine) is specific to transcriptional repression [2–4]. The former is catalyzed by Type I arginine methyltransferases (PRMT1, 3, 4, 6 and 8 in human) and the latter by Type II PRMTs (PRMT5, 7 and 9) [3,5]. Both types use S-adenosyl-l-methionine as the methyl donor for transfer to the side-chain nitrogens of arginines. Although these methylation types do not alter the overall charge of arginine, they increase steric hindrance and hydrophobicity and also decrease hydrogen-bonding capacity, thereby altering the capacity of the arginine-methylated proteins to interact with other proteins/RNAs. For example, arginine residues in GAR domain–containing RNA-binding proteins are believed to be key contributors to RNA-protein interactions, and indeed arginine methylation regulates protein-protein interactions. Among the known PRMTs, PRMT1 and PRMT5 compete for binding to Chtop (chromatin target of PRMT1)—the main topic of this review—and catalyze the methylation of arginines in Chtop [6]. Therefore, Chtop can potentially regulate both transcriptional activation and repression. Notably, both PRMT1 and PRMT5 are able to methylate exactly the same arginine (R3) of histone H4 yet in different ways (asymmetric vs. symmetric) and thus have opposite effects on transcription [1,2,7]. Indeed, Chtop participates in activation of several gene groups in cooperation with PRMT1, and Chtop and PRMT5 contribute to suppression of the fetal globin gene (γ-globin) [8].

Although Chtop was previously known as “friend of PRMT1” (FOP) [6,9,10] or as “small protein rich in arginine and glycine” (SRAG) [11], here we use “Chtop” because the HUGO Gene Nomenclature Committee assigned this name for the corresponding human gene C1orf77. The full transcript from this gene contains six exons (1–6) and five introns (Int 1–5) that encode 248 amino acid residues including a GAR domain, and two additional isoforms are produced probably by alternative splicing—either by skipping exons 2–4 (SRAG-3) or exon 5 (SRAG-5) [11]. As expected from its structural features [12], Chtop associates with RNA, including its own mRNA, U5 and U6 snRNAs, and rRNA [10,11,13]. With respect to its association with rRNA, Chtop shares sequence similarities in its GAR region with those of fibrillarin and nucleolin, both of which are trans-acting factors involved in ribosome biogenesis and localize in the nucleolus during rRNA transcription [11]. The level of Chtop is highest in resting cells and is downregulated at G₂/M phase of the cell cycle [11]. Because overexpression of Chtop inhibits the cell cycle, causing cell death [11], it is believed that Chtop level is tightly regulated in normal cells to maintain its functional homeostasis. Although 25–33% of the Chtop population is found within the nucleolus, the localization of which is regulated by the N terminus of the protein, the rest localizes in the nucleus (mostly nuclear speckles) and tightly associates with facultative heterochromatin (co-localizes with trimethylated lysine-27 of histone H3, H3K27me3) in vertebrate interphase cells [6,11,14]. This coincides with the roles of Chtop in transcription of specific gene groups that respond to cellular transformation or intercellular stimuli such as estrogen or glioblastoma-related factors, as described below.

Roles of Chtop in transcriptional regulation

One gene group regulated by Chtop is that induced by estrogen (17β-estradiol, E2) [6]. Chtop binds the promoter region of the gene pS2 in an E2-dependent manner in MCF7 cells (human E2-responsive breast cancer cell line) and is required for the stable binding of estrogen receptor α (ERα) to the promoter upon induction with E2 (Fig. 1A). E2-induced transcription is a kinetically complex cyclical process involving iterative association of ∼50 transcription factors with the pS2 promoter, a concept called “transcriptional clock” that proceeds through several sequential cycles—an initial transcriptionally unproductive cycle followed by productive cycles [15]. The initial unproductive cycle involves nucleosomal reorganization at the pS2 promoter facilitated by the nucleosomal remodeling complex, followed by recruitment of histone acetyl transferases, histone methyltransferases, and basal transcription factors. This process is required for promoter commitment, which establishes a state ready for the binding of liganded ERα. Métivier et al. reported the molecular processes underlying the transcriptional clock [15], and readers may consult this paper for details. Briefly, in this cycle, unliganded ERα binds inefficiently to the pS2 promoter and is subsequently degraded by the proteasome unless E2 is available. Because deficiency of Chtop does not affect the binding of unliganded ERα to the promoter [6], Chtop is probably not involved in this unproductive cycle even though the methylation mark H4R3 is involved in this process. Then, the transcriptionally productive cycle starts via the binding of E2-liganded ERα to the pS2 promoter, which engages at least 30 different proteins on the promoter, including PRMT1 and histone acetyl transferases, the nucleosome-remodeling complex, basal transcription factors, RNA polymerase II, and transcriptional mediators and elongators to promote transcriptional cycling. During this process, PRMT1 associates with a specific remodeling SWI/SNF complex containing liganded ERα, Brg1 (ATPase of SWI/SNF complex), TATA-binding protein (TBP), and TF_IIA (basal transcription factor) or with a complex containing liganded ERα, TBP, TF_IIA, SRC, and Tip60 [15] (Fig. 1A). Given that Chtop associates with PRMT1 and is required for stable E2-dependent recruitment of ERα to the pS2 promoter, it is likely that Chtop functions as a component of one or both of these PRMT1-containing complexes during the transcriptionally productive cycles. However, the direct interaction between ERα (or the aforementioned PRMT1-binding partners) and Chtop and the requirement of PRMT1 for the binding of Chtop (or vice versa) to the pS2 promoter remain undetermined. In this regard, our previous finding that Chtop associates with Cip1-interacting zinc-finger protein (Ciz1) and scaffold attachment factor B2 (SAFB2) [10] is very intriguing because Ciz1 is a DNA-binding co-activator of ERα, which ensures hypersensitivity to estrogen [16], whereas SAFB2 suppresses ERα-induced transcription [17]. Chtop may control the association of those proteins with ERα during the transcription of ERα target genes.

Figure 1.

Schematic illustration of regulation of gene-specific transcription by Chtop. A. Activation of estrogen (E2)-induced genes by Chtop [6]. Chtop receives dimethylation of arginine asymmetrically (aDMA) and symmetrically (sDMA), catalyzed by PRMT1 and PRMT5, respectively, and binds to both PRMT enzymes in a mutually competitive manner. Upon induction with E2, Chtop is recruited to the pS2 promoter and induces the binding of ERα to the promoter. During this process, PRMT1 associates with a complex containing TBP, TF_IIA, ERα, and Brg1 or a complex containing TBP, TF_IIA, ERα, SRC, and Tip60. B. Activation of glioblastoma-related genes [18]. Chtop binds and stabilizes 5hmc (formed by the action of TET1) and recruits the methylosome (composed of PRMT5, MEP50, ERH, and PRMT1) to the promoter of the glioblastoma-related genes. The recruited PRMT1 methylates H4R3 and activates transcription. C. Activation of Zbp-89 target genes [19]. Chtop recruits 5FMC (composed of Wdr18, Pelp1, Tex10, Las1L, and Senp3) to sumoylated Zbp-89 (SUMO-Zbp-89). This recruitment is dependent on the aDMA of Chtop by PRMT1. Then, Senp3 removes the SUMO group from SUMO-Zbp-89 and increases the association of RNA polymerase II with a subset of Zbp-89 target genes. D. Hypothesis for repression of γ-globin by Chtop. Chtop recruits PRMT5 to the γ-globin locus, causing symmetrical dimethylation of H4R3 (H4R3me2s) and subsequent binding of DNA methyltransferase 3A (Dnmt3A) that methylates local DNA, resulting in γ-globin repression upon further methylation of H4K20 (H4K20me3) with SUV4–20h.

The second group of genes regulated by Chtop comprises genes involved in glioblastomagenesis, including EGFR, AKT3, CDK6, CCND2, and BRAF in that 5-hydroxymethylcytosine (5hmC) is over-presented in their promoters and intragenic regions [18]. Chtop is upregulated in human primary proneural glioblastoma cells compared with neural progenitor cell lines and can bind 5hmC-containing sequences dependent on its asymmetrical methylation by PRMT1 [18]. Primary proneural glioblastoma cells also upregulate TET1 [Fe(II) and 2-oxoglutarate-dependent DNA dioxygenase 1 belonging to the ten-eleven translocation family (TET1)], which converts 5-methylcytosine to 5hmC to maintain an appropriate level of 5hmC in glioblastoma cells. Therefore, high-level expression of both Chtop and TET1 is seemingly necessary for glioblastomagenesis. Despite the high expression of TET1, however, human neural progenitor cells produce low levels of 5hmc and do not transform into glioblastoma cells under normal conditions [18]. Thus, high-level expression of Chtop is critical during glioblastomagenesis. In glioblastoma cells, Chtop, when bound to 5hmC-containing sequences, recruits the methylosome [composed of PRMT1, PRMT5, MEP50 (methylosome protein 50) and ERH (enhancer of rudimentary homolog)] to chromatin in a TET1-dependent manner (Fig. 1B). This recruitment is required for asymmetrical or symmetrical dimethylation of H4R3. Consistent with this, Chtop and PRMT1 are necessary for the expression of EGFR, AKT3, CDK6, CCND2, and BRAF whereas PRMT5 is required for the repression of CCND2. Thus, Chtop plays central roles via it ability to recruit the methylosome to the genes containing 5hmC responsible for glioblastomagenesis.

The third gene group regulated by Chtop is targeted by Zbp-89 (Kruppel-type zinc-finger protein 89, BERF-1, or BECOL1) [19]. Zbp-89 is a GC-rich sequence–specific Kruppel-type zinc-finger transcription factor that either activates or represses a variety of genes that regulate many cell processes, including differentiation, myogenesis and hematopoiesis, growth and proliferation, and apoptosis [20]. For example, Zbp-89 is a transcriptional activator in the context of expression of the proto-oncogene β-catenin (CTNNB1) in colorectal cancer cells [21] or of the matrix metalloproteinases (MMP-1 and MMP-3) in cells that respond to inflammatory cytokines [22]. In contrast, Zbp-89 represses expression of ODC (encoding ornithine decarboxylase) for the biosynthesis of polyamines during the growth and differentiation of many cell types [23], of the vimentin gene in stage-specific cells during tissue development [24], and of the PU.1 gene during regulation of stress-induced hematopoiesis in adult mouse bone marrow [25]. Despite the involvement of Zbp-89 in transcription of a number of genes, the molecular mechanisms by which Zbp-89 functions as an activator or repressor have been reported for only a few cases; e.g., Zbp-89 activates transcription of p21^waf1 (encoding a cyclin-dependent kinase inhibitor) by bridging between acetyltransferase p300 and specificity factor 1 (Sp1) on the promoter [26], whereas it suppresses transcription of ODC by competing with Sp1 for the binding to its GC-rich promoter [23] or by recruiting HDAC1 (histone deacetylase) to the vimentin promoter [24,27]. Chtop provides an additional unique mechanism by which Zbp-89 regulates Zbp-89-bound genes. In mouse erythroleukemia cells, for example, Chtop (along with PRMT1) is required for the recruitment of Zbp-89 to a subset of Zbp-89 target genes, including Dusp6, Zbp-89, Atf5, and Tubb1 [19]. PRMT1-methylated Chtop interacts with a nuclear protein complex named “Five Friends of Methylated Chtop” (5FMC) that is composed of five proteins, namely Pelp1 (proline-glutamate and leucine-rich protein 1), Las1L (LAS1-like protein), Tex10 (testis-expressed 10 protein), Senp3 (SUMO1/sentrin/SMT3 specific peptidase 3), and Wdr18 (WD repeat domain 18 protein); this interaction recruits 5FMC to the promoter or coding region of Zbp-89 target genes in mouse erythroleukemia cells, thereby promoting recruitment of RNA polymerase II [19] (Fig. 1C). This recruitment enables one of the 5FMC components, Senp3, to remove a SUMO group from sumoylated Zbp-89 and abolishes the inhibitory effects of sumoylated Zbp-89 on transcription of Zbp-89 target genes [19] (Fig. 1C). Zbp-89 is sumoylated possibly at K115 and K356 by an E1 enzyme (SAE1/SAE2) and an E2 enzyme (Ubc9) during this process [28]. Chtop activates transcription by recruiting SUMO-specific proteases to Zbp-89 target genes.

Chtop reportedly represses the expression of γ-globin, which produces a γ-globin chain to form the fetal tetramer hemoglobin with the α-globin chain (α2γ2) [9]. This fetal hemoglobin is responsible for oxygen transport in fetal blood. In adulthood, γ-globin is silenced, and instead β-globin is produced to form adult hemoglobin (α2β2). Defects or mutation in this β-globin gene causes β-thalassemia or sickle cell disease, which presents various symptoms including anemia, hemolysis, etc. Severity of either disease may be fatal [29]. Because the severity of the associated anemia is associated with the concentration of fetal hemoglobin among patients with β-thalassemia or sickle cell disease, reactivation of suppressed γ-globin is the main focus for developing therapeutics [29]. As expected from known mechanisms underlying the silencing of other disease-related genes, γ-globin in adults is silenced via deacetylation of histones by histone deacetylase 1 (HDAC1) and HDAC2 [30,31] and assembly of the repressor complex (NuRD/Mi2β) by action of the transcription factor GATA-1 and its binding protein (FOG-1, friend of GATA-1) [32]. In addition, at least four transcription factors (BCL11A, MYB, KLF1, SOX6) are involved in the expression of γ-globin as a negative regulator [33–38]. Among these factors, BCL11A and SOX6 integrate the actions of many of the aforementioned factors described as part of the mechanism underlying γ-globin silencing; i.e., SOX6 bridges between γ-globin loci and repressor complex NuRD/Mi2β and HDACs via the binding to a complex comprising BCL11A-GATA-1-FOG-1 [38], whereas GATA-1 binds directly to the –556GATA silencer motif in γ-globin and recruits the GATA-1-FOG-1-NuRD/Mi2β repressor complex to the gene in mice [32].

Chtop is another protein involved in silencing of γ-globin in adulthood and contributes to the suppression of up to ∼30% of total expression of fetal γ-globin, as estimated from its depletion in erythroid progenitor cells [9]. Therefore, depletion of Chtop increases the level of γ-globin in human erythroid progenitor cells, and levels are comparable with those observed in individuals who are haploinsufficient for BCL11A [39,40]. However, the mechanism by which Chtop contributes to γ-globin silencing remains unknown. Because Chtop deficiency reduces the expression of SOX6 but not BCL11A [9], Chtop may control the amount of cellular SOX6. However, a plausible mechanism has been posited for a role of Chtop in regulating DNA methylation [2,41]. Notably, γ-globin is methylated by DNA methyltransferase 3A (Dnmt3A), which is recruited to symmetrically methylated arginines in histone H4R3 (H4R3me2s), as catalyzed with PRMT5; this methylation leads to the assembly of histone-modifying enzymes including SUV4–20h, resulting in induction of another repressive mark H4K20me3 [2,8]. Therefore, it is likely that Chtop recruits PRMT5 to γ-globin and allows subsequent DNA methylation by PRMT5. This hypothesis is illustrated in Fig. 1D.

Roles for Chtop in processes coupling pre-mRNA splicing with mRNA export

In addition to the roles for Chtop in regulating gene-specific transcription, it has another role in steps that are required for ensuring timing from the processing events of nascent pre-mRNAs to the export of mature mRNAs to the cytoplasm [42]. Chtop plays this role as a component of the TREX (TRanscription-EXport) complex that functions during a series of post-transcriptional events as such 5′-cap formation, co-transcriptional splicing, and 3′-end processing of mRNAs [43,44]. As a component of the TREX complex, Chtop is recruited with another RNA-binding protein, Alyref, in a mutually exclusive manner onto 5′-end region of mRNA by promoting the ATPase and RNA helicase activities of Uap56. Then, Chtop and Alyref are methylated by PRMT1 and play critical roles in the binding of the nuclear RNA export factor Nxf1 to mRNA; this recruitment eventually leads to the translocation of the TREX-mRNA-Nxf1 complex from speckles to the nucleoplasm [44].

Because Chtop is a component of the nuclear SMN (survival motor neuron protein) complex and is required for the association of this complex with heterogeneous nuclear ribonucleoproteins (hnRNPs) and for the dissociation of ATP-binding proteins from SMN complex [10], it may also determine the timing of the assembly of hnRNPs and/or ATP-binding proteins on nascent pre-mRNAs during the events coupling pre-mRNA splicing with mRNA export. In addition, analogous to Alyref as an adaptor for RNAs [44], Chtop can associate with many mRNAs including those encoding proteins related to Cajal body formation and function such as FLASH (FADD-like IL-1 β-converting enzyme (FLICE)-associated huge protein), NPAT (nuclear protein, ataxia-telangiectasia), and FMR1 (fragile X mental retardation protein 1) [45–48] (Fig. 2). Furthermore, Chtop also interacts physically with the nucleus-localized noncoding RNAs NEAT1 and NEAT2, which are involved in paraspeckle assembly and function [49,50] (Fig. 2). These data indicate that Chtop contributes substantively to the steps regulating events occurring from transcription to the nuclear export of mRNAs.

Figure 2.

Association of Chtop with coding and noncoding RNAs. Pulldown analysis using FLAG-tagged Chtop from a T-REx 293 cell extract. RNAs pulled down were amplified by RT-PCR with primer sets corresponding to each of the RNAs indicated. As controls, RNAs pulled down with IgG, FLAG, or FLAG-hnRNP A were also analyzed by RT-PCR with the same primer sets.

Autoregulation of Chtop expression

Because Chtop plays critical roles in regulating gene expression at multiple levels from transcription to the nuclear export of mRNAs, the expression of Chtop is controlled strictly in normal cells to maintain its functional homeostasis [11]. This is achieved by a negative feedback mechanism utilizing intron retention and nonsense-mediated mRNA decay (NMD). In this mechanism, Chtop interacts with exon 2 of its own mRNA via its arginine-glycine-rich (RG) domain and with the stem-loop region in intron (Int) 2 via its N-terminal (N1) domain; binding to either the exon or intron causes retention of Int2, which contains a stop codon near its 5′ end. This stop codon leads to NMD of Chtop mRNA. On the other hand, hnRNP H binds both Int2 of the Chtop mRNA and N1-RG of the Chtop protein, and this accelerates Int2 excision from the Chtop pre-mRNA in a manner dependent on Chtop level; thus, Chtop and hnRNP H regulate Int2 retention of Chtop mRNA antagonistically (Fig. 3) [13]. Because Chtop controls the association of the nuclear SMN complex with hnRNPs including hnRNP H as described above [10], Chtop may facilitate auto-regulation of the Chtop expression by sequestrating hnRNP H to SMN complex. This in turn implies that the ratio of Chtop to hnRNP H may help determine the cellular level of Chtop.

Figure 3.

Autoregulation of the cellular level of Chtop [11]. Schematic illustration for the proposed mechanism by which the cellular level of Chtop is regulated by intron retention and NMD. Increased Chtop level enhances its binding to the stem-loop region in intron 2 (Int2) and/or exon 2 (Ex2) of its own mRNA via its N1 domain and its RG region, respectively. Either type of binding inhibits the excision of Int2. Int2-retained Chtop mRNA contains a stop codon at the 5′ side of Int2 and is subsequently degraded by NMD. hnRNP H binds Int2 of Chtop RNA at two different regions and promotes excision of Int2. Chtop antagonizes the action of hnRNP H for Int2 splicing. Chtop binding to N1-RG may also reduce the binding of hnRNP H to Chtop mRNA.

A number of studies have reported mechanisms by which the level of a particular protein is regulated by modulating alternative pre-mRNA splicing followed by NMD [51–57]. Examples include RNA-binding proteins, such as SR proteins (splicing regulators), polypyrimidine tract–binding protein, and hnRNPs [58–60], the levels of which are often controlled by alternative use of a 5′ or 3′ splice site, exon skipping, or inclusion of mutually exclusive exons. In line with this, mRNAs encoding proteins involved in chromatin modification are currently part of the list of alternative splicing-NMD–regulated mRNAs as revealed by deep sequencing of RNAs of the brain cortex and many cancers [61,62]. These analyses also revealed that the alternative splicing and NMD mechanism occurs in certain tissues or in a cell type–specific manner and that intron retention coupled with NMD is a physiological mechanism of gene expression control in normal granulopoiesis [63]. Several recent reviews offer in-depth analyses of models for alternative splicing-NMD that is not involved in intron retention [64–68]. Intron retention is one mode of alternative splicing that creates proteomic versatility from most genes, and it is widely considered as a distinct mechanism that regulates gene expression. Besides the degradation via NMD, intron-retaining transcripts are remain in the nucleus until they become necessary or undergo degradation in the nucleus as other levels of mechanisms regulating gene expression [69–73]. Also, in some cases, they are transported to specific organelles within the cell [74]. Darya et al [69]. currently review the various roles of intron retention in gene expression, and readers may consult this review for details. Despite the importance of intron retention in gene expression, however, its regulatory mechanism remains mostly unknown. Recently, RNA sequencing technology identified hnRNP L as being responsible for 3′-terminal intron retention via the analysis of transcriptome changes upon depletion of hnRNP L in T cells, but the molecular mechanism was not determined [58]. Although the general concepts underlying the mechanisms causing intron retention were proposed during in the past 5–6 years [57,75,76], only a few proteins controlling intron retention have been identified. One such protein is the poly(A)-binding protein nuclear 1 (PABPN1), which binds to an adenosine (A)-rich region in the 3′ untranslated region of its mRNA and promotes retention of the 3′-terminal intron, leading to clearance of intron-retained mRNAs by the nuclear exosome [77]. One other example is BS69, which specifically recognizes K36me3-modified histone variant H3.3. BS69 controls intron retention dependent upon its association with H3.3K36me3-bound chromatin during ongoing transcription by antagonizing the binding of pre-RNA splicing regulators such as elongation factor Tu GTP-binding domain containing 2 (a component of the spliceosomal U5 snRNP) [78]. One more protein, namely methyl-CpG binding protein 2 (MeCP2) [79] is added to the list of proteins that regulate intron retention. The insufficient binding of MeCP2 near the splice junction of a pre-mRNA decreases the recruitment of splicing factors (such as transformer 2-beta homolog, Tra2b) and the rate of transcription elongation, thereby increasing intron retention. The two mechanisms involving BS69 and MeCP2 signify interdependency between intron retention and epigenetic control of gene expression. In contrast, Chtop has a unique mechanism whereby its binding to its own pre-mRNA causes the skipping of an internal intron during splicing [13].

Concluding remarks

It has been generally thought that intron retention occurs because of mis-splicing of a pre-mRNA that leads to failed intron excision, which often results in a premature stop codon and mRNA degradation via NMD [63,80,81]. This assumption is misleading, however, in that we now understand that intron retention is a common and distinct process that controls gene expression during many cellular events including differentiation of granulocytes, germ cells, megakaryocytes, and erythrocytes [63,82-85]. Indeed, intron retention controls the cellular level of Chtop, which is a versatile regulator of gene expression at multiple levels from transcription to mRNA export. Deregulation of the cellular level of Chtop is linked to human diseases including glioblastoma. Chtop controls intron retention of only its own mRNA. Utilizing this autoregulatory mechanism, we may be able to find reagents that allow specific control of intron retention of Chtop mRNA for the development of potential therapeutic agents for glioblastoma, anemia, or certain cancers.

Funding Statement

This work was supported by the Japan Science and Technology Agency (JP) (JPMJCR13M2) This work was funded by a grant for Core Research for Evolutionary Science and Technology (CREST) from Japan Science and Technology Agency (JPMJCR13M2), a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, & Technology of Japan (MEXT), and JSPS KAKENHI Grant Number 16K18489. Funding for open access charge: Core Research for Evolutionary Science and Technology (CREST), Japan Science and Technology Agency.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgment

We thank Drs. Toshiaki Isobe, Masato Taoka and Hiroshi Nakayama for discussions and support concerning this work.