The Gadd45 proteins play important roles in growth control, maintenance of genomic stability, DNA repair, and apoptosis. Recently, Gadd45 proteins have also been implicated in epigenetic gene regulation by promoting active DNA demethylation. Gadd45 proteins have sequence homology with the L7Ae/L30e/S12e RNA binding superfamily of ribosomal proteins, which raises the question if they may interact directly with nucleic acids.
Here we show that Gadd45a binds RNA but not single- or double stranded DNA or methylated DNA in vitro. Sucrose density gradient centrifugation experiments demonstrate that Gadd45a is present in high molecular weight particles, which are RNase sensitive. Gadd45a displays RNase-sensitive colocalization in nuclear speckles with the RNA helicase p68 and the RNA binding protein SC35. A K45A point mutation defective in RNA binding was still active in DNA demethylation. This suggests that RNA binding is not absolutely essential for demethylation of an artificial substrate. A point mutation at G39 impared RNA binding, nuclear speckle localization and DNA demethylation, emphasizing its relevance for Gadd45a function.
Citation: Sytnikova YA, Kubarenko AV, Schäfer A, Weber ANR, Niehrs C (2011) Gadd45a Is an RNA Binding Protein and Is Localized in Nuclear Speckles. PLoS ONE 6(1): e14500. https://doi.org/10.1371/journal.pone.0014500
Editor: Shuang-yong Xu, New England Biolabs, Inc, United States of America
Received: June 22, 2010; Accepted: December 9, 2010; Published: January 7, 2011
Copyright: © 2011 Sytnikova et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the Deutsche Forschungsgemeinschaft, Grant NI 286/12-1 to Christof Niehrs, and WE-4295/1-1 to Alexander Weber. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The Gadd45 genes are a family of stress response genes, which are involved in diverse processes, including cell growth, DNA repair, and apoptosis, and function as tumor- and autoimmune suppressors , . Expression of these genes is induced by DNA-damage and genotoxic stress, including hyperosmotic stress and UV irradiation. The three Gadd45 genes encode multifunctional, 18 kDa acidic proteins, which can homo- and heterodimerize and which are predominantly localized in the nucleus . Gadd45 proteins interact with many effectors, including Cdc2/CyclinB1 , , PCNA , , p21 , nuclear hormone receptors , histones  and MEKK4 , , to mediate cell cycle arrest, differentiation or apoptosis.
More recently Gadd45 proteins have been implicated in epigenetic gene regulation, promoting active DNA demethylation via a DNA repair mechanism. Gadd45a binds to the repair endonuclease XPG and initiates excision repair at methylated CpG motifs in Xenopus, Zebrafish, and mammalian cells –.
Gadd45 proteins exhibit sequence homology to the L7Ae/L30e/S12e superfamily . Members of this family are diverse proteins from archea, eubacteria and eucaryota, including ribosomal proteins (S12, L30e), proteins that bind guiding RNA (L7Ae, 15.5 kD, fibrillarin), as well as components of ribonuclease P. Many of these proteins bind functionally diverse RNAs, including ribosomal RNA, snoRNA, snRNA and mRNA. Rather than binding to a specific consensus sequence, these proteins recognize a common structural motif – the kink turn, formed by both canonical Watson-Crick base pairing as well as and non-canonical interactions .
The fact that Gadd45 proteins belong to the L7Ae/L30e/S12e superfamily raises the question whether they may also bind RNA. Importantly, RNAs have been repeatedly implicated in active DNA demethylation although their history in this process is confusing –. Most recently ROS3 has been described as an essential mediator of DNA demethylation in Arabidopsis. ROS3 resides in nuclear speckle-like structures and binds small RNAs. It was suggested that these RNAs may guide the DNA demethylase towards their substrate .
Gadd45a has been shown to associate with chromatin , , , however, it is unknown whether it directly interacts with nucleic acids. Here we provide evidence that Gadd45a has RNA binding properties and possesses characteristics of a ribonucleoprotein particle (RNP).
Expression constructs and antibodies
For Xenopus tropicalis (xt) Gadd45a overexpression in human cells and E.coli we used constructs containing xtGadd45a ORF in vectors pRKW2 and pET28a as well as N-EGFP tagged xtGadd45a in pCS2 . Point mutants of xtGadd45a, were obtained by circular PCR . The following antibodies were used: anti-hGadd45a (H165), anti-p68 (H144), anti-Brg1 (N-15) (Santa Cruz), anti-hnRNP A1 and anti-histone H3 (Abcam), anti-GFP (Dianova), anti-SC35 (Novus Biologicals).
Cell culture and transfections
HEK293T cells (ATCC CRL 11268) and RKO cells (ATCC CRL 2577) were grown at 37°C in 10% CO2 for 293T cells and 5% for RKO cells in Dulbecco's Modified Eagle's Medium (DMEM), 10% fetal calf serum, 2 mM L-Glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin. Transient DNA transfections were carried out using FuGENE6 (Roche), TurboFect™ (Fermentas) in case of HEK293T, and for RKO cells a combination of Lipofectamine and Plus reagents (Invitrogen) was used following the manufacturer's instructions.
Cell detergent extraction and RNase treatment was essentially as described . For detection of overexpressed Gadd45a (wild type and G39A mutant) RKO cells were transfected with pCS2-EGFP-Gadd45a. 24 h after transfection cells were extracted with 0.05% Triton X-100, washed with Hank's BSS 1x buffer (PAA laboratories GmbH); with or without subsequent treatement with RNase A for 7 min with (Roche Applied Science, 1 mg/ml). Cells were fixed with Dithiobis (succinimidyl propionate) (DSP) (Thermo Scientific) according to the manufacturer. Briefly, immediately before fixation DSP was added to a final concentration of 0.5 mM in 100 mM Hepes (pH 7.4) in Hank's buffer. Cells were fixed in the freshly prepared DSP solution for 90 minutes and then incubated for 30 minutes in quenching solution (50 mM monoethanolamin, 0.1% Triton X100 in Hank's buffer). Immunostaining was performed with anti-SC35 (Novus Biologicals) and anti-p68 (Santa Cruz Biotechnology) antibodies; immunofluorescent images were recorded on a Nikon confocal microscope. For statistical analysis, the nuclear pattern of EGFP-Gadd45a was assessed manually (n = 35–50).
For detection of endogenous Gadd45a HEK293T cells grown on coverslips and subjected to UV irradiation (40 mJ/cm2) were permeabilized with 0.2% Triton X-100 in 20 mM Tris-HCl (pH 7.4), 5 mM MgCl2, 0.5 mM EDTA, and 25% glycerol; with or without subsequent RNase treatment as described above. Cells were fixed with 2% formaldehyde and antigen retrieval was performed as described , except that microwave treatment was done at 450W. Immunofluorescence was performed using anti-Gadd45a H165 antibody (Santa Cruz). Statistical analysis of the endogenous Gadd45 nuclear patterns was performed manually (n = 50).
Nuclear extract preparation
HEK293T or RKO cells were harvested, washed twice with DPBS buffer, and homogenized in buffer A (0.3 M sucrose, 10 mM Tris-HCl pH 8.0, 3 mM CaCl2, 2.5 mM Magnesium acetate, 0.25% Triton X-100, 0.1 mM EDTA, 2 mM DTT, Complete™ proteinase inhibitors (Roche)) in a douncer homogeniser. Homogenates were mixed 1∶1 with buffer B (1.8 M sucrose, 10 mM Tris-HCl pH 8.0, 3 mM CaCl2, 2.5 mM magnesium acetate, 0.1% Triton X-100, 0.3 mM EDTA, 2 mM DTT, CompleteTM proteinase inhibitors (Roche)), and centrifuged through a buffer B cushion for 20 min at 13,000 g in a Beckman SW41 Ti rotor. The pellet was resuspended in nuclear extraction buffer (50 mM Hepes-NaOH (pH 7.8), 140 mM NaCl, 1 mM MgCl2, 1 mM DTT, 0.1 mM Vanadyl ribonucleosides (Sigma), proteinase inhibitors (Roche) and sonicated. Nuclear extract was centrifuged at 15000 g for 30 min, and the supernatant was used for further procedures.
Sucrose gradient sedimentation analysis
Sucrose gradient sedimentation analysis was performed using nuclear extract from 2×107 RKO cells. Samples were untreated or treated with 100 µg/ml of ribonuclease A (Roche) or 40 U/ml of DNase I (MBI) for 30 min at room temperature. Soluble nuclear proteins were applied to the top of a 8–40% sucrose gradient and centrifuged for 26 h at 50000 g at 4°C. Samples containing sedimentation markers thyroglobulin (19S), β-galactosidase (16.4 S), catalase (11S) or a cytoplasmic fraction containing ribosomal subunits were run separately. Proteins from gradient fractions were precipitated and analyzed on immunoblots.
Filter binding assay
Filter binding assays were performed essentially as described . Recombinant proteins used were bovine serum albumin (Fraction V, Sigma), His-Gadd45a and M-MLV-reverse transcriptase (Invitrogen). Binding reactions were performed in RNA binding buffer (10 mM Tris-HCl (pH 7.4), 10 mM KCl, 1 mM MgCl2). In reactions without competitor, 2.5 or 1.3 µM recombinant proteins were preincubated for 20 min with 10,000 cpm (approximately 2 ng) of 32P-labelled multiple cloning site (MCS) RNA of pCS2 and pXT1 plasmids. In competition assays, unlabeled competitor nucleic acids were preincubated for 10 min with recombinant proteins before addition of labeled RNA. Reactions were applied to nitrocellulose filters that were pre-blocked with 50 µg/ml BSA in RNA binding buffer, washed with RNA binding buffer and quantified by scintillation counting.
Structural data files
Apart from the crystal structures used as templates for xtGadd45a homology modelling, the crystal structures of human spliceosomal p15.5 kDa protein bound to a U4 snRNA fragment (PDB ID 1e7k, ), yeast L30e-mRNA complex (PDB ID 1t0k, ), Haloarcula marismortui ribosomal protein L7Ae-rRNA complex (PDB ID 1s72, ), yeast spliceosomal protein Snu13p dimer complex (PDB ID 1zwz, ) were used. For homology modeling the sequences of xtGadd45a (NCBI AN CAJ82672) and human SPB2 (NCBI AN NP_076982) were used.
Homology modeling and structure analysis
Two close homologues of xtGadd45a were used as templates for homology modeling: human (PDB ID 2wal, citation pending), and mouse (PDB ID 3cg6, ) Gadd45g. Initial sequence alignments were generated in ClustalX2  and manually refined from 3D alignments of available crystal structures for hsp15.5, scL30e, hmL7Ae, scSnu13p, hsGadd45g and generated models of hsSBP2_RBD and xtGadd45a. Modeling of xtGadd45a was carried out as described ,  using the MODELLER package . In the same way a model of the xtGadd45a mutant G39A was generated. For modeling of the human SPB2 RNA-binding domain the crystal structure of yeast spliceosomal protein Snu13p (PDB ID 1zwz; ) was used. Initial models were scored for energy content and sterical correctness and the best model further optimized using GROMACS molecular dynamics simulations  was used. All models were scored for energy and sterical correctness using the ANOLEA , VERIFY_3D and ERRAT (http://nihserver.mbi.ucla.edu) online servers. Structure analysis was carried out using SwissPBD Viewer  and PyMol (www.pymol.org). PDB2PQR , PropKa  and APBS  packages were used for charge surface calculations and the HotPatch web server  for hydrophobicity calculations. For protein-protein dockings the GRAMM package was used in hydrophobic mode .
Purification of recombinant his-xtGadd45a
pET28a vectors containing ORF of xtGadd45a and mutants were transformed into BL21DE3 E.coli. Bacteria were grown in 100 ml medium and induction was performed by IPTG for 4 h. Disruption was performed by French Press, and the lysate was cleared by centrifugation at 110.000 g for 30 min. Purification was performed by metal affinity chtomatography under native conditions on a Ni-NTA column (Qiagen) according to the manufacturer's instruction except that lysis buffer contained 1 mM of MgCl2 and 0.04% of NP40. Step elution with 60–100 mM imidazole was performed. Fractions were analyzed by SDS-PAGE with Coomassie staining (Thermo Scientific), and fractions eluting at 100 mM imidazole were dialyzed against lysis buffer and used for further experiments.
Luciferase reporter assay
Dual-Luciferase reporter assays (Promega) were performed as described in .
Southern blot methylation analysis
HEK293T cells were transiently transfected in 10 cm dishes with 5 µg HpaII in vitro methylated pOctTK-EGFP and 1.2 µg pBl-KS or xtGadd45a. Transfected plasmid DNA was recovered 72 h after transfection, digested with NotI and either HpaII or MspI and analyzed by Southern blot using a GFP probe. The expression of EGFP was additionally analyzed by SDS-PAGE and Western blot using anti-GFP antibody.
Gadd45a binds RNA in vitro and in vivo
To test whether Gadd45a binds RNA, we carried out filter binding assays using recombinant Gadd45a and radiolabeled synthetic vector RNA, which indicated significant RNA binding compared with M-MLV reverse transcriptase (Figure 1A). To further characterize nucleic acid binding, filter binding assays were performed by preloading Gadd45a with unlabeled nucleic acids followed by competition with labeled RNA from a plasmid multiple cloning site (Figure 1B, C). Since Gadd45a is implicated in DNA demethylation, we tested methylated as well as unmethylated DNAs. Neither unmethylated, nor methylated single- nor double stranded DNA efficiently competed for RNA binding. Poly-uridine was the best competitor among RNA homopolymers (Figure 1C). Other complex RNAs, including tRNA, vector derived RNA, and notably total cellular RNA, were also effective.
A, RNA filter binding assay using the indicated proteins and 32P-labeled RNA (multiple cloning site transcript, MCS). Co, no protein; BSA, bovine serum albumin; M-MLV RT - Moloney murine leukemia virus reverse transcriptase. B, C, RNA filter binding assays using 32P-labeled MCS RNA were performed with recombinant Gadd45a in the presence of the indicated unlabeled competitor nucleic acids. Data are shown as percentage of 32P bound in the absence of the competitor. Each sample was done in triplicate; average and standard deviation was generated; A representative experiment out of three is shown. U, unmethylated; M, methylated; U/U, unmethylated; U/M, hemimethylated; M/M, holomethylated; PolyA, polyC, polyG, polyU, homopolyribonucleotides; total RNA, RNA isolated from HEK293T cells; tRNA, yeast tRNA; MCS RNA, multiple cloning site RNA. Error bars, s.e.m. (n = 3). A representative experiment out of three is shown.
To examine if Gadd45a is bound to RNA in vivo, we analyzed its sedimentation behaviour in sucrose density gradients (Figure 2). As a source we used RKO cells, which express Gadd45a endogenously . Interestingly, the majority of Gadd45a sedimented in the ribosome-sized fractions, with S-values between 40 and 60 (Figure 2A). Significantly, RNase treatment shifted Gadd45a to lighter fractions, suggesting that Gadd45a may be present in an RNP-like particle (Figure 2C).
A–C Sedimentation analysis of RKO nuclear extracts in a linear 8–40% (top-bottom) sucrose gradient. Fractions were analyzed by Western blot for Gadd45a, hnRNP A1 and p68 (control RNA binding proteins) and Brg1 (negative control). Prior to sedimentation nuclear extracts were left untreated (A), DNAse treated (B), or RNAse treated (C). b, resuspended micro-pellet of tube. Representative experiment out of three is shown.
Two RNA binding proteins, ribonucleoprotein hnRNP A1, a component of ribonucleoprotein (RNP) particles  and RNA helicase p68 , were analyzed as positive controls. The ATPase Brg1 served as negative control protein. It is part of a nucleosome remodelling complex and not thought to bind RNA . The sedimentation profile of hnRNP A1 was broad, with a peak in the ribosomal fractions, like for Gadd45a (Figure 2A) and consistent with it being part of RNP particles. In contrast, p68 showed a bimodal distribution, fractionating as a very heavy and a light form, as reported previously . Brg1 was recovered only in heavy fractions. DNase pretreatment showed minor alterations in the sedimentation behavior of the proteins but these were within the margins of sample variability (Figure 2B). In contrast, RNAse pretreatment led to a reproducible shift in sedimentation of both RNA binding proteins hnRNPA1 and p68 to the light fraction, while it did not affect Brg1 (Figure 2C), indicating that the effect was indeed due to RNase and not contaminating protease.
The RNase sensitive sedimentaion profile of Gadd45a supports it being associated with RNA endogenously.
Gadd45a localizes in nuclear speckles in an RNase sensitive manner
To further examine whether Gadd45a is a RNA binding protein we analyzed the RNase sensitivity of its localization. Cells were transfected with Gadd45a and soluble proteins were detergent extracted with or without RNase treatment and analyzed by Western blot or immunofluorescence (IF) microscopy (Figure 3A–F). In Western blot analysis Gadd45a is removed from the detergent-resistant fraction upon RNase treatment (Figure 3B). IF analysis of detergent-extracted RKO cells showed that EGFP-Gadd45a is localized in nuclear speckles, as described previously . There it was colocalized with the nuclear speckle markers SC35 (Figure 3C) and p68 (Figure 3D). Nuclear speckles are the main repository for factors involved in transcription elongation, mRNA processing and export –. In some cells overexpressed Gadd45a strongly localized to the nuclear periphery (Figure S1). We tested if this localization is RNase sensitive. Indeed, RNase treatment reduced the number of cells where Gadd45a localized in nuclear speckles from 72% to 20% (Figure 3E, F). In contrast, RNase treatment did not affect localization of Gadd45a in the nuclear periphery (not shown) as well as SC35 staining.
A, Scheme of detergent extraction. B, RKO cells expressing EGFP-Gadd45a were subjected to detergent extraction with or without RNaseA treatment followed be Western blot analysis of the indicated proteins. A representative experiment out of three performed is shown. C, D Immunofluorescence confocal microscopy of detergent-extracted RKO cells. Cells were transfected with N-EGFP-Gadd45a and stained with antibodies against SC35 (C) and p68 (D). E, cells were treated as in B, but subjected to RNase treatment after extraction and before fixation. F, Statistical analysis for localization of EGFP-Gadd45a and SC-35 in nuclear speckles in cells with and without RNaseA treatment (n = 35 cells; n = 3 experiments; a representative experiment is shown). G, H Immunofluorescence confocal microscopy of endogenous Gadd45a in UV irradiated detergent-extracted HEK293T cells. Cells were stained with antibodies against Gadd45a and SC35. In (H), cells were subjected to RNase treatment after extraction and before fixation. I, Statistical analysis for localization of UV inducible Gadd45a in nuclear speckles in HEK293T cells with and without RNaseA treatment (n = 50 cells).
In HEK293T cells endogenous Gadd45a showed only a weak, homogeneous nuclear signal. However, following UV-irradiation, which induces Gadd45a expression, the protein was colocalized with SC35, but was also found in SC35 negative punctae and in the nuclear periphery (Figure 3G). Once again, RNaseA treatment removed Gadd45a from nuclear speckles (Figure 3H, I).
The nuclear speckle co-localization with RNP proteins SC35 and p68 and its RNase sensitivity support that Gadd45a is an RNA binding protein and may be part of an RNP.
Modeling of Gadd45a-RNA binding
To gain insight into the structural basis of Gadd45a-RNA interactions we inspected the three dimensional structures of Gadd45 as well as of structures of other L7Ae members, which were solved in complex with RNA.
First, we built a homology model for Xenopus tropicalis Gadd45a by employing the available crystal structures of Gadd45g . Next we compared the sequences (Figure 4A) and structures (Figure 4B–E) of the three L7Ae protein family members: human spliceosomal p15.5 kDa protein bound to a U4 snRNA fragment, yeast L30e-mRNA complex, and Haloarcula marismortui ribosomal protein L7Ae-rRNA complex. All three L7Ae family proteins are bound to the so-called kink-turn RNA motif. Analyzing general rules of recognition of this type of RNA, we identified two main patches on the RNA-binding surface of these proteins. In patch 1 (shades of blue in Figure 4) RNA-protein contacts are formed by positively charged amino acids from β-strand β1, helix α2 and several highly conserved amino acids. Patch 2 (green in Figure 4) represents a hydrophobic pocket able to accommodate a purine or pyrimidine base flipped from the kink-turn RNA. This pocket is formed by amino acids from different parts of the protein. Both of these patches are present in all L7Ae family member-RNA complexes. Patch 1 is relatively conserved in sequence and structure (Figure 4). Patch 2 varies in size and amino acid composition but is generally composed of hydrophobic core residues surrounded by polar amino acids. We propose that patch 1 plays a role in general protein-RNA interaction and patch 2 may be responsible for sensing the kink-turn RNA motif. Interestingly, the same conserved RNA-binding like patches are present on the surface of Xenopus Gadd45a (Figure 4E).
A, Sequence alignment of L7Ae family proteins: human hsp15.5 kDa protein, yeast ribosomal scL30e protein, Haloarcula marismortui ribosomal hmL7Ae protein, yeast spliceosomal protein scSnu13p protein, human hsSBP2_RBD (RNA-binding domain), human hsGadd45g and Xenopus tropicalis xtGadd45a, including secondary structure elements (above) and sequence conservation (below). Light and dark blue letters indicate backbone- and side chain RNA interacting residues from patch 1. Light and dark green letters indicate backbone and hydrophobic side RNA interacting residues from patch 2 (see text for details). Residues targeted by mutagenesis are marked. B–E, Comparison of the crystal structures of human hsp15.5 kDa protein (B), yeast ribosomal scL30e protein (C) and Haloarcula marismortui ribosomal hmL7Ae protein (D), and the homology model of Xenopus tropicalis xtGadd45a (E). Residue coloring as above. The red area denotes the ultra-conserved Gly residue (RNA guanine G-binding region) important for specific RNA binding and DNA demethylation.
Besides the two patches, the glycine residue homologous to G39 in Gadd45a (Figure 4A; red in Figure 4B–E) is highly conserved in all L7Ae family members and constitutes a third important structural determinant for kink-turn RNA binding. This residue is located at the beginning of helix α2, further referred to as guanine (G)-binding region, since in all L7Ae-RNA structures a guanine base is tightly bound in this region through an extensive hydrogen-bonding network (Figures S2 and S3). A glycine to alanine or lysine mutation of this residue completely abrogates RNA binding and protein function in human SBP2 and p15.5 kDa protein, respectively , . Indeed, modeling the G38A mutant of hsp15.5, we discovered that a G38A mutation should result in sterical clashes with bound RNA (Figure S3).
Taken together, our modeling and structural analysis suggests a rationale for the ability of Gadd45a to bind RNA despite its acidic pKa and the absence of a distinctly positively charged region.
Mutations affecting Gadd45a RNA binding and demethylation
To test the role of amino acid residues in patch 1 and of G39 in RNA binding and DNA demethylation, we generated four point mutations in Gadd45a, K45A, R34G, V49R and G39A (Figure 5A). RNA binding activity of purified recombinant proteins (Figure 5B) was tested by filter binding assay with radioactively labeled vector derived RNA. Nonspecific RNA binding ability was comparable for the three mutants K45A, R34G and G39A and wild type Gadd45a (Figure 5C). The V49R substitution showed three-fold higher RNA binding ability. This may reflect increased ionic interaction upon addition of an extra positive charge in this position.
A, general characteristics of Gadd45a point substitutions. B, SDS-PAGE analysis of His-tagged Gadd45a wild type and point mutant proteins produced and purified from E.coli. C, filter binding assay of Gadd45a wild type and point mutant proteins using multiple cloning site (MCS) 32P-RNA. D, RNA filter binding assays using 32P-labeled MCS RNA were performed with wild type or point mutant Gadd45a proteins in the presence of the indicated unlabeled competitor RNAs. Data are pooled from seven independent experiments. E, F DNA demethylation assays. E, Luciferase reporter assays of HEK293T cells transiently transfected with an M. SssI in vitro methylated SV40-luciferase reporter and the indicated constructs. Error bars, s.e.m. (n = 3). F, Methylation sensitive Southern blot of HpaII in vitro methylated pOctTK reporter recovered from HEK293T cells cotransfected with Xenopus Gadd45a wild type and mutants.
We next tested specific RNA binding by the RNA competition assay described in Figure 1C. For each mutant we compared the binding competition of labeld synthetic (vector) RNA with unlabeled cellular RNA. Since total RNA was a good competitor in the in vitro binding assays (Figure 1C), we reasoned that it contains relevant- but unknown RNAs, which physiologically bind to Gadd45a with high affinity. Wild type Gadd45a showed a three-fold difference in competition assays between vector- and total RNA (Figure 5D). This was similar for the R34G and V49R mutants, which harbor patch 1 substitutions naturally occurring in other L7Ae members. In contrast, substitution of either of the two ultra-conserved amino acids – K45A and G39A - caused a loss of discrimination between vector- and total RNA binding.
To test the activity of the mutants in DNA demethylation, we monitored Gadd45a-mediated re-activation of an in vitro methylated – and hence silenced - luciferase reporter plasmid. Gadd45a can demethylate and thus transcriptionally activate such reporters . Upon transfection in HEK293T cells wild type Gadd45a as well as R34G and V49R mutants equally activated the methylated reporter (Figure 5E). Notably the K45A mutant, which failed to discriminate between specific and non-specific RNA binding, was fully active in the demethylation assay, indicating that the two properties can be uncoupled. This already suggests that specific RNA binding is not absolutely essential for DNA demethylation, at least under these experimental conditions. In contrast, Gadd45a G39A was the only mutant inactive in the reporter assay as well as in specific RNA binding.
To test for gene specific demethylation, we transfected a methylated EGFP expression plasmid and monitored its methylation status by digestion with the methylation sensitive endonuclease HpaII. In parallel we measured its expression by detecting EGFP protein in the cell lysates. The analysis showed that cotransfection of wild type, K45A, R34G and V49R led to expression of EGFP protein and to the appearance of a HpaII cleavage product, indicative of demethylation (Figure S4 and Figure 5F). The G39A mutant, which failed to discriminate between specific and non-specific RNA binding, was also inactive in DNA demethylation as well as activation of EGFP expression.
Finally, we tested, if the G39A mutant protein still localizes to nuclear speckles. Nuclear localization of the G39A mutant was much reduced in general and less than 20% showed nuclear speckles (Figure 6A-C). Similarly, Western blot analysis of detergent treated cells showed that G39A was more sensitive to extraction than wild type Gadd45a (Figure 6D).
A,B IF microscopy comparison of nuclear pattern after detergent extraction of EGFP-Gadd45a wild type (A) and EGFP-G39A mutant (B). Experiments were done essentially as in Figure 3A. C, Statistical analysis of immunofluorescence patterns as in Figure 3F. D, Western blot analysis of RKO cells expressing EGFP-Gadd45a wild type (wt) or EGFP-G39A mutant harvested without or after detergent extraction. Scale bar, 4 µm.
From the mutant analysis we conclude that a) specific RNA binding is not absolutely essential for DNA demethylation and b) that G39 is a critical amino acid for the function and localization of Gadd45a.
The main findings of this study are that Gadd45a is an RNA binding protein and that it appears to be part of an RNP particle. This is in line with the function of other members of the L7Ae/L30e/S12e superfamily, which are either ribosomal components or associated with RNP particles. Gel filtration and cross linking analysis of recombinant Gadd45b,g indicates that the protein forms a dimer of 35 kDa , , while Gadd45a can oligomerize , suggesting that the cellular high molecular weight form of Gadd45a may contain multimers.
The conclusion that Gadd45a is in an RNP complex is supported by sucrose density gradient centrifugation and its localization in nuclear speckles. It is interesting that nuclear speckles are a site of active transcription, RNA splicing and processing. This raises the possibility that Gadd45a RNPs are associated with genes undergoing active DNA demethylation and transcriptional activation. Of note, p68/Ddx5, which colocalizes with Gadd45a in nuclear speckles, was previously described as a component of a DNA demethylase complex , .
RNP complexes play prominent roles in RNA processing, RNA transport and RNA translation (for review, see –). In light of our results it is interesting that overexpression of Gadd45 leads to a similar phenotype in Drosophila as mutation of squid, which encodes an hnRNP. In both cases the chorion of fly eggs is dorsalized due to defects in grk mRNA localization and translation –, supporting the idea that a Gadd45 RNP function is evolutionary conserved.
Our in silico modeling suggests a structural basis for the RNA binding of Gadd45a. Like other RNA binding proteins of the L7Ae family, Gadd45a contains two patches, which appear to be involved in RNA binding. The in vitro RNA binding assays and point mutagenesis data suggest that Gadd45a has moderate affinity for nonspecific RNAs and high affinity for specific RNAs. The V49R substitution increased non-specific RNA binding. Since arginine instead of valine 49 is a naturally occurring variant in some L7Ae superfamily proteins, we propose that such proteins have a higher general RNA binding propensity.
The specific RNA binding of Gadd45a is clearly not absolutely essential for its demethylating activity, as shown by the K45A mutant. However, our DNA demethylation assay chosen for convenience is rather artificial; it involves an abundant in vitro methylated reporter plasmid, which is demethylated by overexpressed Gadd45a. In contrast, locus-specific demethylation under physiological conditions may very well require its ability to bind specific RNAs as discussed below.
The G39A substitution in the G-binding region abolished both DNA demethylation as well as specific RNA binding ability. Since the K45A mutant still demethylates despite inactivated specific RNA binding, the RNA binding defect of G39A may not be the cause for the demethylation defect. This raises the possibility that G39A interferes with some other important property of Gadd45a. However, at least dimerisation of Gadd45a does not seem to be affected by G39A since by in silico docking of Gadd45a dimers , G39 was not found in the vicinity of the putative Gadd45a dimer interface (Figure S5).
Our study raises new questions concerning the biology and biochemistry of Gadd45 proteins. Which RNAs are physiologically bound to Gadd45? What other proteins are parts of the Gadd45 RNP particle? Is the role of Gadd45 bound RNAs purely structural or is RNA involved in e.g. specific targeting to demethylated DNA regions?
Perinuclear pattern of Gadd45a. Immunofluorescence confocal microscopy of detergent-extracted RKO cells. Cells were transfected with EGFP-xtGadd45a and developed with antibody against SC35; nuclei were stained with Hoechst. This pattern is observed in ∼10% of EGFP-Gadd45 positive cells. Scale bar, 5 µm.
(1.05 MB JPG)
Possible H-bonding networks in patch 2 and G-binding region in human hsp15.5 kDa protein (A), yeast ribosomal scL30e protein (B), Haloarcula marismortui ribosomal hmL7Ae protein (C) and the model of xtGadd45a (D). Residues colored in light and dark blue form patch 1 and those colored in light and dark green form patch 2, respectively (see also Figure 4 legend for details). The red area denotes the highly conserved Gly residue (RNA guanine G-binding region) found to be important for proper RNA binding. Left subpanels show how the flipped RNA base is sensed and accommodated in the patch 2 pocket by hydrophobic interactions of the purine or pyrimidine base with sidechains of hydrophobic residues (colored in green). The hydrogen bonding with backbone and/or sidechains of some charged amino acids surrounding the hydrophobic pocket is also shown. Right subpanels show the extensive hydrogen bonding network which could be formed by the RNA base (in all three discussed crystal structures it is always guanine). For xtGadd45a these small subpanels show modeled interactions of guanidine and uridine bases in the G-binding region and patch 2 hydrophobic pocket, respectively. RNA is shown in a semitransparent cartoon representation.
(1.92 MB JPG)
Modeling the Gly to Ala mutation on the basis of the hsp15.5-RNA complex. Panel A represents possible the hydrogen bonding network formed by the guanine base which is properly bound and oriented in the G-binding region. Panel B shows surface representations of the same structure illustrating that the guanine base perfectly fits into the G-binding region without any sterical clashes. Exchange of glycine residue for alanine leads would lead to considerable sterical clashes (C). To resolve these clashes we propose that the guanine base moves out from the G-binding site (D) leading to a loss of most hydrogen bonds (E).
(1.08 MB JPG)
Induction of EGFP expression from HpaII methylated promoter by xtGadd45a wild type and mutants. Western blot analysis of EGFP induction from HpaII methylated pOctTK-GFP reporter, as well as of xtGadd45a wild type and mutants expression. Loading was controlled using histone H3. A representative experiment out of three independent experiments is shown.
(0.49 MB JPG)
A, crystal structure of a yeast spliceosomal protein scSnu13p dimer complex with bound RNA molecules modeled on the hsp15.5-RNA complex crystal structure by fitting corresponding scSnu13p and hsp15.5 proteins in SwissPDB Viewer. B, model of xtGadd45a dimer complex obtained by GRAMM docking in hydrophobic mode with RNA molecules superimposed from hsp15.5-RNA complex crystal structure by fitting corresponding xtGadd45a and hsp15.5 proteins in SwissPDB Viewer. In both cases the red area on the protein surfaces represents the conserved glycine of the G-binding region. From these structures it is evident that dimerization and RNA-binding interfaces (patches 1 and 2) do not overlap.
(0.69 MB JPG)
We thank Gabi Döderlein for help with experiments and Ingrid Grummt for critical reading.
Conceived and designed the experiments: YAS CN. Performed the experiments: YAS AK AS. Analyzed the data: YAS AK AS AW CN. Contributed reagents/materials/analysis tools: AK CN. Wrote the paper: YAS AW CN.
- 1. Hollander MC, Fornace AJ Jr (2002) Genomic instability, centrosome amplification, cell cycle checkpoints and Gadd45a. Oncogene 21: 6228–6233.
- 2. Hoffman B, Liebermann DA (2009) Gadd45 modulation of intrinsic and extrinsic stress responses in myeloid cells. J Cell Physiol 218: 26–31.
- 3. Kovalsky O, Lung FD, Roller PP, Fornace AJ Jr (2001) Oligomerization of human Gadd45a protein. J Biol Chem 276: 39330–39339.
- 4. Zhan Q, Antinore MJ, Wang XW, Carrier F, Smith ML, et al. (1999) Association with Cdc2 and inhibition of Cdc2/Cyclin B1 kinase activity by the p53-regulated protein Gadd45. Oncogene 18: 2892–2900.
- 5. Vairapandi M, Balliet AG, Hoffman B, Liebermann DA (2002) GADD45b and GADD45g are Cdc2/CyclinB1 kinase inhibitors with a role in S and G2/M cell cycle checkpoints induced by genotoxic stress. J Cell Physiol 192: 327–338.
- 6. Azam N, Vairapandi M, Zhang W, Hoffman B, Liebermann DA (2001) Interaction of CR6 (GADD45gamma) with proliferating cell nuclear antigen impedes negative growth control. J Biol Chem 276: 2766–2774.
- 7. Vairapandi M, Azam N, Balliet AG, Hoffman B, Liebermann DA (2000) Characterization of MyD118, Gadd45, and proliferating cell nuclear antigen (PCNA) interacting domains. PCNA impedes MyD118 AND Gadd45-mediated negative growth control. J Biol Chem 275: 16810–16819.
- 8. Kearsey JM, Coates PJ, Prescott AR, Warbrick E, Hall PA (1995) Gadd45 is a nuclear cell cycle regulated protein which interacts with p21Cip1. Oncogene 11: 1675–1683.
- 9. Yi YW, Kim D, Jung N, Hong SS, Lee HS, et al. (2000) Gadd45 family proteins are coactivators of nuclear hormone receptors. Biochem Biophys Res Commun 272: 193–198.
- 10. Carrier F, Georgel PT, Pourquier P, Blake M, Kontny HU, et al. (1999) Gadd45, a p53-responsive stress protein, modifies DNA accessibility on damaged chromatin. Mol Cell Biol 19: 1673–1685.
- 11. Takekawa M, Saito H (1998) A family of stress-inducible GADD45-like proteins mediate activation of the stress-responsive MTK1/MEKK4 MAPKKK. Cell 95: 521–530.
- 12. Chi H, Lu B, Takekawa M, Davis RJ, Flavell RA (2004) GADD45beta/GADD45gamma and MEKK4 comprise a genetic pathway mediating STAT4-independent IFNgamma production in T cells. Embo J 23: 1576–1586.
- 13. Barreto G, Schafer A, Marhold J, Stach D, Swaminathan SK, et al. (2007) Gadd45a promotes epigenetic gene activation by repair-mediated DNA demethylation. Nature 445: 671–675.
- 14. Schmitz KM, Schmitt N, Hoffmann-Rohrer U, Schafer A, Grummt I, et al. (2009) TAF12 recruits Gadd45a and the nucleotide excision repair complex to the promoter of rRNA genes leading to active DNA demethylation. Mol Cell 33: 344–353.
- 15. Rai K, Huggins IJ, James SR, Karpf AR, Jones DA, et al. (2008) DNA demethylation in zebrafish involves the coupling of a deaminase, a glycosylase, and gadd45. Cell 135: 1201–1212.
- 16. Sen GL, Reuter JA, Webster DE, Zhu L, Khavari PA (2010) DNMT1 maintains progenitor function in self-renewing somatic tissue. Nature 463: 563–567.
- 17. Ma DK, Jang MH, Guo JU, Kitabatake Y, Chang ML, et al. (2009) Neuronal activity-induced Gadd45b promotes epigenetic DNA demethylation and adult neurogenesis. Science 323: 1074–1077.
- 18. Schäfer A, Schomacher L, Barreto G, Döderlein G, Niehrs C (2010) Gemcitabine Functions Epigenetically by Inhibiting Repair Mediated DNA Demethylation. PLoS ONE 5(11): e14060.
- 19. Koonin EV, Bork P, Sander C (1994) A novel RNA-binding motif in omnipotent suppressors of translation termination, ribosomal proteins and a ribosome modification enzyme? Nucleic Acids Res 22: 2166–2167.
- 20. Vidovic I, Nottrott S, Hartmuth K, Luhrmann R, Ficner R (2000) Crystal structure of the spliceosomal 15.5kD protein bound to a U4 snRNA fragment. Mol Cell 6: 1331–1342.
- 21. Weiss A, Keshet I, Razin A, Cedar H (1996) DNA demethylation in vitro: involvement of RNA. Cell 86: 709–718.
- 22. Fremont M, Siegmann M, Gaulis S, Matthies R, Hess D, et al. (1997) Demethylation of DNA by purified chick embryo 5-methylcytosine-DNA glycosylase requires both protein and RNA. Nucleic Acids Res 25: 2375–2380.
- 23. Jost JP, Fremont M, Siegmann M, Hofsteenge J (1997) The RNA moiety of chick embryo 5-methylcytosine- DNA glycosylase targets DNA demethylation. Nucleic Acids Res 25: 4545–4550.
- 24. Swisher JF, Rand E, Cedar H, Marie Pyle A (1998) Analysis of putative RNase sensitivity and protease insensitivity of demethylation activity in extracts from rat myoblasts. Nucleic Acids Res 26: 5573–5580.
- 25. Jost JP, Siegmann M, Thiry S, Jost YC, Benjamin D, et al. (1999) A re-investigation of the ribonuclease sensitivity of a DNA demethylation reaction in chicken embryo and G8 mouse myoblasts. FEBS Lett 449: 251–254.
- 26. Jost JP, Schwarz S, Hess D, Angliker H, Fuller-Pace FV, et al. (1999) A chicken embryo protein related to the mammalian DEAD box protein p68 is tightly associated with the highly purified protein-RNA complex of 5-MeC-DNA glycosylase. Nucleic Acids Res 27: 3245–3252.
- 27. Schwarz S, Bourgeois C, Soussaline F, Homsy C, Podesta A, et al. (2000) A CpG-rich RNA and an RNA helicase tightly associated with the DNA demethylation complex are present mainly in dividing chick embryo cells. Eur J Cell Biol 79: 488–494.
- 28. Vairapandi M, Liebermann DA, Hoffman B, Duker NJ (2000) Human DNA-demethylating activity: a glycosylase associated with RNA and PCNA. J Cell Biochem 79: 249–260.
- 29. Lin IG, Hsieh CL (2001) Chromosomal DNA demethylation specified by protein binding. EMBO Rep 2: 108–112.
- 30. Imamura T, Yamamoto S, Ohgane J, Hattori N, Tanaka S, et al. (2004) Non-coding RNA directed DNA demethylation of Sphk1 CpG island. Biochem Biophys Res Commun 322: 593–600.
- 31. Zheng X, Pontes O, Zhu J, Miki D, Zhang F, et al. (2008) ROS3 is an RNA-binding protein required for DNA demethylation in Arabidopsis. Nature 455: 1259–1262.
- 32. Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: a laboratory manual. Cold Spring Harbor: Cold Spring Harbor Laboratory Press.
- 33. Mayer C, Schmitz KM, Li J, Grummt I, Santoro R (2006) Intergenic transcripts regulate the epigenetic state of rRNA genes. Mol Cell 22: 351–361.
- 34. Bilic J, Huang YL, Davidson G, Zimmermann T, Cruciat CM, et al. (2007) Wnt induces LRP6 signalosomes and promotes dishevelled-dependent LRP6 phosphorylation. Science 316: 1619–22.
- 35. Stepanov AS, Voronina AS, Ovchinnikov LP, Spirin AS (1971) RNA-binding protein factor of animal cell extracts. FEBS Lett 18: 13–18.
- 36. Chao JA, Williamson JR (2004) Joint X-ray and NMR refinement of the yeast L30e-mRNA complex. Structure 12: 1165–1176.
- 37. Klein DJ, Moore PB, Steitz TA (2004) The roles of ribosomal proteins in the structure assembly, and evolution of the large ribosomal subunit. J Mol Biol 340: 141–177.
- 38. Oruganti S, Zhang Y, Li H (2005) Structural comparison of yeast snoRNP and spliceosomal protein Snu13p with its homologs. Biochem Biophys Res Commun 333: 550–554.
- 39. Schrag JD, Jiralerspong S, Banville M, Jaramillo ML, O'Connor-McCourt MD (2008) The crystal structure and dimerization interface of GADD45gamma. Proc Natl Acad Sci U S A 105: 6566–6571.
- 40. Larkin MA, Blackshields G, Brown NP, Chenna R, McGettigan PA, et al. (2007) Clustal W and Clustal X version 2.0. Bioinformatics 23: 2947–2948.
- 41. Kubarenko AV, Ranjan S, Colak E, George J, Frank M, et al. (2010) Comprehensive modeling and functional analysis of Toll-like receptor ligand-recognition domains. Protein Sci 19: 558–569.
- 42. Kubarenko A, Frank M, Weber AN (2007) Structure-function relationships of Toll-like receptor domains through homology modelling and molecular dynamics. Biochem Soc Trans 35: 1515–1518.
- 43. Sali A, Overington JP (1994) Derivation of rules for comparative protein modeling from a database of protein structure alignments. Protein Sci 3: 1582–1596.
- 44. Van Der Spoel D, Lindahl E, Hess B, Groenhof G, Mark AE, et al. (2005) GROMACS: fast, flexible, and free. J Comput Chem 26: 1701–1718.
- 45. Melo F, Devos D, Depiereux E, Feytmans E (1997) ANOLEA: a www server to assess protein structures. Proc Int Conf Intell Syst Mol Biol 5: 187–190.
- 46. Guex N, Peitsch MC (1997) SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis 18: 2714–2723.
- 47. Dolinsky TJ, Nielsen JE, McCammon JA, Baker NA (2004) PDB2PQR: an automated pipeline for the setup of Poisson-Boltzmann electrostatics calculations. Nucleic Acids Res 32: W665–667.
- 48. Li H, Robertson AD, Jensen JH (2005) Very fast empirical prediction and rationalization of protein pKa values. Proteins 61: 704–721.
- 49. Baker NA, Sept D, Joseph S, Holst MJ, McCammon JA (2001) Electrostatics of nanosystems: application to microtubules and the ribosome. Proc Natl Acad Sci U S A 98: 10037–10041.
- 50. Pettit FK, Bare E, Tsai A, Bowie JU (2007) HotPatch: a statistical approach to finding biologically relevant features on protein surfaces. J Mol Biol 369: 863–879.
- 51. Vakser IA, Aflalo C (1994) Hydrophobic docking: a proposed enhancement to molecular recognition techniques. Proteins 20: 320–329.
- 52. Aleman MJ, DeYoung MP, Tress M, Keating P, Perry GW, et al. (2005) Inhibition of Single Minded 2 gene expression mediates tumor-selective apoptosis and differentiation in human colon cancer cells. Proc Natl Acad Sci U S A 102: 12765–12770.
- 53. Dreyfuss G, Philipson L, Mattaj IW (1988) Ribonucleoprotein particles in cellular processes. J Cell Biol 106: 1419–1425.
- 54. Ogilvie VC, Wilson BJ, Nicol SM, Morrice NA, Saunders LR, et al. (2003) The highly related DEAD box RNA helicases p68 and p72 exist as heterodimers in cells. Nucleic Acids Res 31: 1470–1480.
- 55. Trotter KW, Archer TK (2008) The BRG1 transcriptional coregulator. Nucl Recept Signal 6: e004.
- 56. Fayolle C, Pourchet J, Cohen A, Pedeux R, Puisieux A, et al. (2006) UVB-induced G2 arrest of human melanocytes involves Cdc2 sequestration by Gadd45a in nuclear speckles. Cell Cycle 5: 1859–1864.
- 57. Shopland LS, Johnson CV, Byron M, McNeil J, Lawrence JB (2003) Clustering of multiple specific genes and gene-rich R-bands around SC-35 domains: evidence for local euchromatic neighborhoods. J Cell Biol 162: 981–990.
- 58. Brown JM, Green J, das Neves RP, Wallace HA, Smith AJ, et al. (2008) Association between active genes occurs at nuclear speckles and is modulated by chromatin environment. J Cell Biol 182: 1083–1097.
- 59. Lamond AI, Spector DL (2003) Nuclear speckles: a model for nuclear organelles. Nat Rev Mol Cell Biol 4: 605–612.
- 60. Nottrott S, Hartmuth K, Fabrizio P, Urlaub H, Vidovic I, et al. (1999) Functional interaction of a novel 15.5kD [U4/U6.U5] tri-snRNP protein with the 5′ stem-loop of U4 snRNA. Embo J 18: 6119–6133.
- 61. Allmang C, Carbon P, Krol A (2002) The SBP2 and 15.5 kD/Snu13p proteins share the same RNA binding domain: identification of SBP2 amino acids important to SECIS RNA binding. RNA 8: 1308–1318.
- 62. Tornatore L, Marasco D, Dathan N, Vitale RM, Benedetti E, et al. (2008) Gadd45 beta forms a homodimeric complex that binds tightly to MKK7. J Mol Biol 378: 97–111.
- 63. Kohler A, Hurt E (2007) Exporting RNA from the nucleus to the cytoplasm. Nat Rev Mol Cell Biol 8: 761–773.
- 64. Staley JP, Woolford JL Jr (2009) Assembly of ribosomes and spliceosomes: complex ribonucleoprotein machines. Curr Opin Cell Biol 21: 109–118.
- 65. Wahl MC, Will CL, Luhrmann R (2009) The spliceosome: design principles of a dynamic RNP machine. Cell 136: 701–718.
- 66. Kelley RL (1993) Initial organization of the Drosophila dorsoventral axis depends on an RNA-binding protein encoded by the squid gene. Genes Dev 7: 948–960.
- 67. Neuman-Silberberg FS, Schupbach T (1993) The Drosophila dorsoventral patterning gene gurken produces a dorsally localized RNA and encodes a TGF alpha-like protein. Cell 75: 165–174.
- 68. Peretz G, Bakhrat A, Abdu U (2007) Expression of the Drosophila melanogaster GADD45 homolog (CG11086) affects egg asymmetric development that is mediated by the c-Jun N-terminal kinase pathway. Genetics 177: 1691–1702.