JHDM3A Module as an Effector Molecule in Guide-directed Modification of Target Chromatin*

With the objective of returning cells to their undifferentiated state through alteration of epigenetic states, small molecules have been used that specifically inhibit proteins involved in sustaining the epigenetic system. However, this chemical-based approach can cause chaotic epigenomic states due to random actions of the inhibitors. We investigated whether JHDM3A/JMJD2A, a trimethylated histone H3-lysine 9 (H3K9me3)-specific demethylase, could function as an effector molecule to selectively demethylate target chromatin, with the aid of a guide protein to serve as a delivery vehicle. JHDM3A, which normally locates in euchromatin, spread out to heterochromatin when it was fused to heterochromatin protein-1α (HP1α) or HP1β; in these cells, demethylation efficiency was also markedly increased. Two truncated modules, JHDM3AGFP406 and JHDM3AGFP701, had contrasting modes and efficiencies of H3K9me3 demethylation; JHDM3AGFP406 showed a very uniform rate (∼80%) of demethylation, whereas JHDM3AGFP701 had a broad methylation range of 4–80%. The methylation values were highly dependent on the presence of the guide proteins OCT4, CTCF, and HP1. Chromatin immunoprecipitation detected reduced H3K9me3 levels at OCT4 regulatory loci in the cells expressing OCT4-tagged JHDM3AGFP701. Derepression of the Sox2 gene was observed in JHDM3AGFP701OCT4-expressing cells, but not in cells that expressed the JHDM3AGFP701 module alone. JHDM3AGFP701-assisted OCT4 more efficiently turned on stem cell-related microRNAs than GFP-OCT4 itself. These results suggest that JHDM3AGFP701 is a suitable catalytic module that can be targeted, under the control of a guide protein, to specific loci where the chromatin H3K9me3 status and the milieu of gene expression are to be modified.

With the objective of returning cells to their undifferentiated state through alteration of epigenetic states, small molecules have been used that specifically inhibit proteins involved in sustaining the epigenetic system. However, this chemicalbased approach can cause chaotic epigenomic states due to random actions of the inhibitors. We investigated whether JHDM3A/JMJD2A, a trimethylated histone H3-lysine 9 (H3K9me3)-specific demethylase, could function as an effector molecule to selectively demethylate target chromatin, with the aid of a guide protein to serve as a delivery vehicle. JHDM3A, which normally locates in euchromatin, spread out to heterochromatin when it was fused to heterochromatin protein-1␣ (HP1␣) or HP1␤; in these cells, demethylation efficiency was also markedly increased. Two truncated modules, JHDM3A GFP 406 and JHDM3A GFP 701 , had contrasting modes and efficiencies of H3K9me3 demethylation; JHDM3A GFP 406 showed a very uniform rate (ϳ80%) of demethylation, whereas JHDM3A GFP 701 had a broad methylation range of 4 -80%. The methylation values were highly dependent on the presence of the guide proteins OCT4, CTCF, and HP1. Chromatin immunoprecipitation detected reduced H3K9me3 levels at OCT4 regulatory loci in the cells expressing OCT4-tagged JHDM3A GFP 701 . Derepression of the Sox2 gene was observed in JHDM3A GFP 701 OCT4-expressing cells, but not in cells that expressed the JHDM3A GFP 701 module alone. JHDM3A GFP 701 -assisted OCT4 more efficiently turned on stem cell-related microRNAs than GFP-OCT4 itself. These results suggest that JHDM3A GFP 701 is a suitable catalytic module that can be targeted, under the control of a guide protein, to specific loci where the chromatin H3K9me3 status and the milieu of gene expression are to be modified.
A widely accepted hypothesis about the origins of cancer proposes that it is caused, at least in part, by the silencing of genes implicated in cell death and differentiation (1). Similarly, induced pluripotent stem cells (iPSCs) 2 (2) and somatic nuclear transfer clones are preceded by derepression of groups of genes responsible for stem cell properties and dedifferentiation, which are otherwise tightly suppressed in somatic cells (3). Fundamental changes in the nature of cells generally stem from perturbations in the transcriptional regulatory machinery; for example, changes in DNA methylation states at the regulatory DNA regions, mutations in chromatin remodeling complexes such as SWI/SNF, and/or changes in post-translational modifications of histone tails (4). Attempts to modulate DNA methylation by DNA methyltransferase inhibitors (5,6), histone deacetylation by HDAC inhibitors (7)(8)(9), and histone methylation by histone demethylase inhibitors (10,11) have received a great deal of attention. However, treating cells with small molecules such as these can lead to total inactivation of the target proteins, which could cause a chaotic alteration of epigenetic states by influencing the chromatin in its entirety and driving it to a structurally and functionally disorganized state. To avoid this, the epigenetic state of chromatin must be controlled at a local level. One means of achieving this is by targeting proteins that affect chromatin structure to specific genomic loci.
Histone lysine demethylases (12) have roles in transcriptional regulation, cancer cell proliferation (13), and reprogramming of differentiated cells (14); this suggests that they could be used to alter the histone methylation states of specific chromatin regions. In an attempt to selectively modulate the epigenetic states of target loci, we tested the ability of histone H3-lysine 9 (H3K9)-specific demethylase JHDM3A/ JMJD2A (15,16) to function as an effector molecule, together with various guide proteins that served as delivery vehicles to targeted chromatin regions.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-NIH3T3 and HEK293T cells were cultured in DMEM (Invitrogen) with 10% heatinactivated fetal bovine serum (Invitrogen) and 50 g/ml of streptomycin/penicillin (Invitrogen). Culture was kept in a humidified atmosphere with 5% CO 2 at 37°C. Plasmids were transfected into NIH3T3 cells using Lipofectamine TM (Invitrogen) according to the manufacturer's instructions. 24 or 40 h after transfection, the cells were fixed with 4% (w/v) formaldehyde solution (Sigma Aldrich).
Mouse Embryo Culture and Microinjection-For embryo collection, female BDF1 mice at 5 weeks of age were injected with 5 international units of pregnant mare serum gonadotrophin (Sigma Aldrich) followed by 5 international units of human chorionic gonadotropins (Sigma Aldrich) 48 h (h) apart and mated with male mice. Successful mating was determined at the following morning by detection of a vaginal plug. 18 to 20 h after human chorionic gonadotropin injection, mouse zygotes were collected from mouse oviduct into M2 medium (Sigma Aldrich) and then were treated with 0.1% (w/v) hyaluronidase to remove cumulus cells. Embryos were cultured in M16 medium (Sigma Aldrich) at 37°C and 5% CO 2 in air. For microinjection, mouse zygotes showing pronuclei were microinjected with JHDM3A GFP 21 h after human chorionic gonadotropin injection. Microinjected eggs were allowed to develop in vitro in M16 media. Two-to eight-cell stage embryos were fixed and immunostained with anti-H3K9me3 antibody (Upstate).
Vector Constructions-For construction of the JHDM3A Fg expression vector, cDNA obtained from human embryonic stem cells (17,18) were amplified by PCR. Primers we used are listed in supplemental Table 1. The PCR products were digested with NotI (Roche Applied Science) and BamHI (Roche Applied Science) and cloned into pCMV-Tag1 vector (Stratagene). For JHDM3A GFP expression vector, PCR was performed using JHDM3A Fg plasmid as template. The PCR products were digested with HindIII (Roche Applied Science) and BamHI (Roche Applied Science) and cloned into pEGFP-C2 vector (Clontech). For constructions of JHDM3A Fg HP1␣, JHDM3A Fg HP1␤, and JHDM3A Fg HP1␥ expression vectors, mouse cDNA was amplified by PCR. The PCR products were digested BglII (NEB) and HindIII (New England Biolabs) and cloned into pCMV-Tag1 vector. For JHDM3A GFP HP1␣ vector, the PCR products were digested BamHI (Roche Applied Science) and XbaI (Roche Applied Science) and cloned into JHDM3A GFP . For truncated JHDM3A constructs, PCR amplification was done using JHDM3A GFP plasmid as template. The PCR products were digested HindIII (NEB) and SalI (Roche Applied Science) and cloned into pEGFP-C2 vector. For amplifications of cDNA-encoding tagging proteins such as CCCTC-binding factor (CTCF), HP1␣, and OCT4, cDNAs were amplified by PCR. The PCR products were digested SalI (Roche Applied Science) and BamHI (Roche Applied Science) and cloned into JHDM3A GFP 406 and JHDM3A GFP 701 vectors. To obtain JHDM3A H188A GFP vector, we performed two separate PCR using two sets of primers. The resulting PCR products were mixed and used as template for secondary PCR. The PCR products were digested with HindIII (NEB) and BamHI (Roche Applied Science) and cloned into pEGFP-C2 vector. All vectors constructed were confirmed by enzyme digestion and sequencing analysis.
Immunocytochemistry-NIH3T3 cells or mouse embryos were fixed in 4% (w/v) formaldehyde for 15 min at room temperature and washed in PBS with 0.05% Tween 20 (PBST, Fisher Scientific). After permeabilization for 30 min in PBS containing 1% Triton X-100 (MP Biomedicals), samples were blocked in 2% BSA-PBS (Sigma Aldrich) for 1 h at room temperature and incubated for 1.5 h at 37°C with primary antibodies. The secondary antibodies that we used were chick anti-rabbit conjugated Alexa Flour 488 or 594 or chick anti-mouse conjugated Alexa Flour 488 or 594 (1:300, Molecular Probes). Stained samples were mounted on slide glass with mounting media containing DAPI (Vectashield). Antibodies recognizing H3K9me1, H3K9me2, and H3K9me3 were purchased from Upstate Biotechnology, and FLAG was purchased from Sigma Aldrich. Samples were observed with Carl Zeiss Axiovert 200 M fluorescence microscope equipped with an Apotome apparatus. Images were captured digitally and merged using Axiovision (version 4.7) or Adobe Photoshop software (version 7.0). Fluorescence intensity profile was measured using the profiling tool in Axiovision software.
RT-PCR-Total RNAs were isolated from FACS-sorted NIH3T3 cells using an RNeasy mini kit (Qiagen). Complementary DNAs (cDNA) were synthesized with SuperScript TM II first-strand synthesis system (Invitrogen) using oligo(dT) primers according to the manufacturer's instructions. For amplifications of human OCT4, mouse Sox2, and Gapdh cDNAs, PCR was performed with sets of primers listed in supplemental Table 1.
Chromatin Immunoprecipitation Assay-2 ϫ 10 8 HEK293T cells were cross-linked with 1% (w/v) formaldehyde (Sigma Aldrich) for 15 min at room temperature, and then formaldehyde was inactivated by the addition of 125 mM glycine (Sigma Aldrich). Chromatin extracts containing DNA fragments with an average size of 500 bp were immunoprecipitated using anti-H3K9me3 (Upstate) and anti-H3K4me3 (Upstate) antibodies. Precipitated DNA was purified with PCR purification kit (Geneall) and analyzed by quantitative PCR with primers (supplemental Table 1).
MicroRNA Microarray and Real-time PCR-Plasmids based on eGFP expression vector were transfected into HEK293T cells using Lipofectamine TM according to the manufacturer's instructions. Forty-eight hours after transfection, the transfected cells were treated with 0.25% trypsin-EDTA (invitrogen) and washed with PBS. 5 ϫ 10 5 GFP-positive cells were collected by FACSAria cell sorter (BD Biosciences) and sent for requesting microRNA expression analysis in a microarray planting 124 probes of stem cell related microRNAs (PANArray TM miRNA expression profiling kit). To obtain small RNA for real-time PCR analysis, total RNA was sieved through RNeasy mini column (Qiagen). Flow-through including small RNA was precipitated by 10 M ammonium acetate and 100% ethanol and eluted by RNase-free water. Collected small RNA was converted into cDNA using miScript reverse transcription kit and miScript SYBR Green PCR kit (Qiagen) according to the manufacturer's instructions. Real-time quantitative PCR was performed using standard protocols on an Applied Biosystems 7900HT with 2ϫ SYBR Green PCR Master Mix (Applied Biosystems). The reactions were 40 cycles of 94°C for 15 s, 60ϳ62°C for 30 s, and 72°C for 30 s with 10ϫ miScript Universal Primer (Qiagen) and microRNA forward primer. The microRNA-specific primers used in this work were 5Ј-GTAAACATCCTACACTCTC-AGC-3Ј for miR-30c, 5Ј-CAGCATTGTACAGGGCTATCA-3Ј for miR-107, 5Ј-TTAAGGCACGCGGTGAATGCCA-3Ј for miR-124a, 5Ј-CCTGAGACCCTTTAACCTGTG-3Ј for miR-125a, 5Ј-TTGGTCCCCTTCAACCAGCTA-3Ј for miR-133b, and 5Ј-CATTCATTGTTGTCGGTGGGTT-3Ј for miR-181d. Reactions are typically run in duplicate. To ensure the specificity and integrity of the PCR products, melting curve analyses were performed on all amplified products. The MicroRNA copy numbers were normalized using RNU6b small nuclear RNA. Fold changes were calculated according to the 2 Ϫ⌬⌬CT method (19).

RESULTS AND DISCUSSION
We constructed FLAG-and GFP-tagged JHDM3A expression vectors (JHDM3A Fg and JHDM3A GFP , respectively). Transient expression of JHDM3A Fg in NIH3T3 cells revealed that FLAG signals coincided with the signals obtained using the anti-JHDM3A antibody (Fig. 1A, a-c). We confirmed that JHDM3A Fg demethylated H3K9me3 but not H3K9me2 or H3K9me1 (Fig. 1, A, d, g, and j, respectively) as reported pre- ), which lacks the ability to demethylate H3K9me3. B, Jhdm3a expression during early mouse development. Shown is the RT-PCR result. Total RNAs obtained from mouse oocytes and embryos at given stages (20 per stage) were used to estimate expression levels of Jhdm3a. H2A.z, loading control. MII, mature oocytes; 1-c to 8-c, one-to eight-cell stages; Mo, morula; Bl, blastocyst; (Ϫ), no template; NIH, NIH3T3 cells. C, preimplantation stage mouse embryos expressing GFP-tagged JHDM3A (JHDM3A GFP ). After microinjection of JHDM3A GFP expression plasmids into the pronucleus of mouse zygotes, resulting transgenic embryos at given cleavage stages were collected and stained for H3K9me3. Arrowheads indicate JHDM3A GFP -expressing nuclei. Con, transgenic embryo expressing GFP protein. Note the compartmentalized two-cell nuclei for H3K9me3, as reported previously (29,30). Nuclei were counterstained with DAPI (blue). Microscopic observations were made using Axiovert 200 M equipped with ApoTome (Carl Zeiss). Image intensity profile was obtained using Axiovision software (version 4.5). FEBRUARY 11, 2011 • VOLUME 286 • NUMBER 6
We examined Jhdm3a expression during early mouse development. RT-PCR results showed that the Jhdm3a transcript was detected at a low level in mature oocytes and was almost undetectable in cleavage stage embryos, including blastocysts (Fig. 1B). Immunostaining did not reveal clear signals positive for endogenous JHDM3a proteins in mouse embryos (data not shown). We expressed GFP-tagged JHDM3A (JHDM3A GFP ) in early mouse embryos by microinjecting the expression construct into the pronucleus of mouse zygotes. When we delved into the resulting transgenic embryos (n ϭ 134) at the four-and eight-cell stages, we were unable to detect any clue for the occurrence of global demethylation in JHDM3A GFP -expressing nuclei (Fig. 1C). This result contrasted with the one for the cultured cells above, suggesting that JHDM3A alone is insufficient for global demethylation of H3K9me3 and that some unknown factor(s) is needed to assist the enzymatic action of JHDM3A in early mouse embryos.
H3K9me3 signals remained detectable in some JHDM3A Fgexpressing cells, suggesting that they somehow avoided JHDM3A-induced demethylation (Fig. 1, A, d, arrow). This led us to assess the degree and frequency of demethylation in JHDM3A-overexpressing cells. We first categorized the cells expressing JHDM3A GFP into three subgroups depending on the extent of H3K9me3 demethylation (Fig. 3A): group 1-type cells, no visible pericentric H3K9me3 speckles due to severe demethylation (Fig. 3, A, a); group 2-type cells, diminished H3K9me3 speckles (Fig. 3, A, b); and group 3-type cells, an almost normal level of H3K9me3 speckles (Fig. 3, A, c). As shown in Fig. 3B, group 1-type cells comprised only 32.6% of JHDM3A GFP -expressing cells 24 h after transfection (hptx; see also Table 1). This value increased to 80.1% when cells expressing JHDM3A GFP HP1␣ were observed, suggesting that the inefficient chromatin accessibility of JHDM3A alone limits full-scale demethylation. An HP1-mediated increase in the demethylation rate was also observed in JHDM3A Fg HP1␣and JHDM3A Fg HP1␤-expressing cells (supplemental Fig. S2). These results suggest that JHDM3A demethylates H3K9me3 more efficiently and rapidly when it is tagged with HP1, most likely because of enhanced access to otherwise inaccessible chromatin regions. Meanwhile, the extent of demethylation was not linked with the expression levels of exogenous JHDM3A. When the fluorescence signal intensity for GFP, H3K9me3, and DAPI was measured for comparison in individual cells, the levels of H3K9me3 relative to DAPI signals did not differ between the cells with lower GFP levels and those with higher GFP levels (p Ͻ 0.001; Fig. 3C).
We attempted to make the best use of JHDM3A as an effector molecule or a "demethylating warhead," by decreasing its size and moderating its rather indiscriminate activity toward chromatin. We made shorter JHDM3A proteins, JHDM3A GFP 406 and JHDM3A GFP 701 , by removing the C-terminal regulatory domain (Fig. 4A); JHDM3A GFP 701 contained the region encompassing the coiled-coil domain, which was absent from JHDM3A GFP 406 . When we assessed the demethylation frequency in cells transfected with either construct, the proportion of group 1-type cells was significantly higher at 24 hptx among the JHDM3A GFP 406 transfectants than among the JHDM3A GFP 701 transfectants (p Ͻ 0.005; Fig. 4C); the demethylation frequency in JHDM3A GFP 701 -transfected cells was lower than that in cells expressing JHDM3A GFP whole protein (Fig. 3B). Meanwhile, the overall level of H3K9me2 was not diminished in JHDM3A GFP 701 -expressing cells (Fig. 4, B, m-p). We next addressed whether the low demethylation frequency that resulted when the JHDM3A GFP 701 module was used could be modulated by a guide protein. We found that when  Table 1 for detailed statistics. C, measurement of fluorescence intensity for H3K9me3 versus GFP signals. JHDM3A GFP expressing cells (n ϭ 58 in total) were divided, according to GFP signal intensity, into three subgroups: 1) cells with weak GFP signals (Ͻ100); 2) those with moderate GFP signals (100 -300); and 3) those with high GFP levels (Ͼ300). Relative levels of H3K9me3/DAPI signals are not different between the three groups of cells with different levels of GFP signals. n, the number of cells counted. Fluorescence signal was quantified using the Profiling function in Carl Zeiss Axiovision software (version 4.5).

TABLE 1 Assessment of H3K9me3 demethylation efficiency in cells transiently transfected with various JHDM3A expression vectors
The degree of demethylation (mean Ϯ S.D.) was assessed 24 hptx. Group 1-, 2-, and 3-type cells denote severely, moderately, and weakly (or no) demethylated GFP-positive cells, respectively (see Fig. 3A). Significant differences existed between no. 1 versus 2 (p Ͻ 0.0001), 1 versus 3 (p Ͻ 0.001), 1 versus 5 (p Ͻ 0.0001), 2 versus 6 (p Ͻ 0.05), and 3 versus 4 (p Ͻ 0.001), but not between 1 vs. 4 and 7 versus 8. n, total number of cells counted. cells were transfected with JHDM3A GFP 701 OCT4 and the demethylation frequency was 24 hptx, group 1-type cells made up 4.3% of JHDM3A GFP 701 OCT4-positive cells (Fig. 5B). The value was lower than the percentage of group 1-type cells in cell cultures transfected with the OCT4-untagged version (JHDM3A GFP 701 , 20.8%; Fig. 4C and Table 1). Expression of both fusion proteins, JHDM3A GFP 406 OCT4 and JHDM3A GFP 701 OCT4, was exclusively nuclear (Fig. 5, A, a and e). As shown in Fig. 4,  B (e and i), when the modules alone were expressed, the GFP signals were not confined to the nucleus, probably because of the lack of the rear nuclear localization signal inside the plant homeodomain (Fig. 4A). This suggests that the OCT4 tag causes the JHDM3A GFP 701 and JHDM3A GFP 406 modules to be confined to the nucleus by providing an effective nuclear localization signal of its own.

No. Vectors
We also tested whether CTCF, a widely expressed 11-zinc finger nuclear protein that serves as a chromatin insulator protein with an unusual ability to recognize multiple target sites (21), could be used as a guide protein. CTCF tagging greatly increased the demethylation frequency in cells trans-  Table 1 for detailed statistics.  Table  1 for detailed statistics.
In this study, we tested whether JHDM3A modules could be utilized as demethylating warheads when they were fused with guide proteins with delivery function to specific chromatin targets. Used alone, JHDM3A can indiscriminately demethylate chromatin as a whole. We observed this in JHDM3Aexpressing cells, in which there was universal demethylation. Such a drastic change in the histone methylation level may threaten the integrity of the chromatin structure and lead to disastrous consequences for the cell. Here, we investigated candidate guide proteins to curb the random access of JHDM3A to chromatin and direct the module to predefined target chromatin loci. HP1␣ and HP1␤ directed the otherwise euchromatin-posited JHDM3A to also localize to pericentric heterochromatin ( Fig. 2A). Moreover, OCT4-tagged JHDM3A GFP 701 induced demethylation of loci regulated by OCT4 (Fig. 6A) and derepressed the SOX2 gene (Fig. 6B). Furthermore, JHDM3A 701 -guided OCT4 more efficiently turned on stem cell-related microRNAs than did OCT4 itself (Fig. 7).
When we structurally dissected the JHDM3A GFP protein by generating the JHDM3A GFP 406 and JHDM3A GFP 701 constructs, we found that their demethylation activities toward H3K9me3 were markedly different (Fig. 4). Compared with JHDM3A GFP , the JHDM3A GFP 406 module had stronger demethylation activity; conversely, JHDM3A GFP 701 had weaker demethylation activity than the full-length JHDM3A. It is possible that the coiled-coil domain or cryptic domain(s) in the 407-701-amino acid stretch present in JHDM3A GFP 701 could interfere with its access to chromatin. This suggested that the JHDM3A GFP 701 module was suitable for further modification and had potential as a tool for target-specific demethylation. We found that the efficiency of JHDM3A GFP 701 -catalyzed demethylation varied depending on the guide proteins attached. As shown in Fig. 5B, cells expressing JHDM3A GFP 701 OCT4 exhibited 4% demethylation frequency, whereas the demethylation frequency of those expressing JHDM3A GFP 701 CTCF reached 80%. The low demethylation frequency observed in JHDM3A GFP 701 OCT4 cells was likely because of stringent guidance by OCT4, which directs the catalytic module to only a handful of target loci. Unlike OCT4, CTCF is well known for its unusual ability to bind to multiple target sites (21); this could account for the high demethylation frequency of JHDM3A GFP 701 CTCF even though it had the same catalytic module as JHDM3A GFP 701 OCT4. According to these results, JHDM3A GFP 701 shows promise as an effector module that can be delivered to target loci by guide proteins.
Immunocytochemical observation showed that the overall levels of H3K9me3 varied among cells overexpressing the JHDM3A protein and that in a significant proportion of JHDM3A-positive cells, the H3K9me3 levels appeared not to be affected by demethylation. In JHDM3A GFP 701 OCT4transfected cells in particular, the percentage of group 3-type cells showing a normal level of H3K9me3 reached almost 80%. H3K9me3 levels of these group 3-type cells appeared unchanged on the surface, but we propose that demethylation indeed occurred in these cells, although the changes were too small to be visualized by immunostaining. This is supported by the ChIP results (Fig. 6A), which showed changes in H3K9me3 levels at OCT4 regulatory loci. If only a very limited portion of the cells had been demethylated, the ChIP results would not have been as robust.
Our guide-driven JHDM3A protein system would be particularly useful in the research fields of somatic cell nuclear transfer (24) and induced pluripotent stem cells (2). Despite their unquestionably promising future in the biomedical industry and regenerative medicine, the somatic nuclear transfer and iPSC fields have long been adversely affected by extremely low efficiency. There is universal agreement that the primary cause of this lies in inefficient dedifferentiation or reprogramming, by which differentiated cells can ultimately be brought back to undifferentiated states (25)(26)(27). Therefore, the greatest priority for these research areas is to facilitate the de-differentiation process in their donor cells or patient cells. In this regard, use of the guide JHDM3A fusion system, which aims to draw out efficient reprogramming in differentiated cells by inducing the demethylation of specific chromatin regions, would be a better choice of strategy for dedifferentiation. For example, expression of the JHDM3A module tagged by a stemness-related protein such as OCT4 and SOX2 in donor cells before or during the somatic nuclear transfer procedure would lead to an initial modification of target chromatin, which could make the donor cells primed and ready for further modification in the oocyte cytoplasm, consequently transfiguring the chromatin to be far more receptive to incoming reprogramming events. Such a pretreatment could be important, given that JHDM3A activity was hardly detectable in early stage embryos (Fig. 1B) and also that no demethylation was induced in embryonic nuclei in which JHDM3A was ectopically expressed (Fig. 1C). In a similar vein, we could expect that linking iPSC-inducing transcription factors or Yamanaka factors (2) as guide proteins to the JHDM3A module would increase the derivation efficiency of iPSCs. The outcome would be the derepression of a group of target genes under the control of the guide protein and then the activation of their downstream-level genes. Such a target-based approach could demethylate chromatin regions in a highly controlled and selective fashion. By doing this, we can avoid a potentially chaotic epigenomic situation of a state of whole genome undermethylation that could inevitably emerge from employing chemical inhibitors to, for example, Suv39h1, which is implicated in the maintenance of global H3K9 methylation patterns (28).
In conclusion, we found that JHDM3A showed a degree of flexibility with regard to localization and demethylation frequency depending on the guide proteins to which it was fused. The two truncated JHDM3A modules, JHDM3A GFP 406 and JHDM3A GFP 701 , exhibited different demethylation efficiencies; the former was associated with relatively high and constant demethylation frequencies (ϳ80%) regardless of the guide protein, whereas the latter exhibited a broad range of demethylation rates (4 -80%). ChIP analysis demonstrated a reduction in H3K9me3 levels at OCT4 regulatory gene loci in cells expressing JHDM3A GFP 701 OCT4. Expression of the OCT4 target gene SOX2 was detected in JHDM3A GFP 701 OCT4-expressing cells. JHDM3A 701 -assisted OCT4 turned out to be more efficient in turning on the stem cell-related microRNAs than OCT4 alone. Our results suggest that the JHDM3A GFP 701 module is a promising effector molecule; when it is fused to a guide protein, it can effectively and specifically alter the histone methylation state of target loci.