Introduction

Kleefstra syndrome (KS) (Mendelian Inheritance in Man (MIM) # 610253), also known as 9q34 subtelomeric deletion syndrome, is a rare autosomal dominant disorder characterized by the phenotype with manifestations such as the developmental delay, intellectual disability, hypotonia at childhood, and distinct facial features. A majority of the patients with KS had a microdeletion at chromosomal region 9q34.3. However, certain population of patients had a mutation only in the euchromatin histone methyltransferase-1 (EHMT1, GLP, or KMT1D) gene mapped at 9q34.3. Thus, the absence or inactivation of EHMT1 at a single allele is speculated to cause the major symptoms of the syndrome [1,2,3,4].

Histone H3 lysine 9 (H3K9) methylation, especially H3K9 di- and tri-methylation (H3K9me2 and me3), are well-conserved transcriptionally repressed epigenetic marks [5, 6]. G9a (also known as EHMT2 or KMT1C) and G9a-like protein (GLP) are SET-domain-type lysine methyltransferases (KMTs) in mammals and regulate H3K9 methylation of euchromatin [7]. G9a and GLP can form a homo or heterodimer; however, they mostly exist as a G9a/GLP heterodimer in the cells [8]. As (i) the constitutive levels of H3K9me1 and me2 are severely diminished in Ehmt1 or Ehmt2 single knockout (KO) cells [8, 9], (ii) the further reduction of H3K9 methylation is not recognized in Ehmt1 and Ehmt2 double KO (Ehmt1 and 2 DKO) cells [8], and (iii) the dimer formation is essential for SET-domain-mediated intrinsic histone methyltransferase (HMT) activities of G9a and GLP [8, 10], it is suggested that the G9a/GLP heterodimer complex is the major functional form for constitutional H3K9 methylation regulation in vivo, even though recombinant G9a or GLP can independently methylate H3K9 in vitro. Furthermore, dimer formation-competent but enzymatically inactive mutant of GLP but not G9a can rescue the specific KO cells from defective H3K9 methylation, suggesting that the enzymatic activity of G9a is more important for the in vivo HMT function than that of GLP [10].

Regarding the previously reported EHMT1 mutations in individuals with KS, most of them are frameshift or nonsense mutations. However, two individuals with KS were reported to possess two distinct missense EHMT1 mutations [2, 4]. Although these two missense mutations were predicted to show defective enzymatic function, precise biochemical validation was not conducted. Therefore, we performed validation of these two missense mutations.

Materials and methods

Plasmids

Mouse GLP (mGLP) full-length complementary DNA (cDNA) was subcloned into pCDNA3 with FLAG tag sequence. Three mutations (C1201A, R1197W, and C1071Y) were generated by PCR (Table 1). FLAG-tagged mouse GLP wild-type (WT) and mutant cDNAs were subcloned into pCAGGS vector for generation of mGLP that stably express in embryonic stem (ES) cell line. C-terminal truncated mGLP (1009–1296 aa) WT and mutant cDNAs were subcloned into pF3A WG (BYDV) Flexi vector (Promega) for in vitro translation. Human GLP (hGLP) catalytic domain (982–1266 aa) expression vector (pET28a) was a kind gift from Dr. Masoud Vedadi.

Table 1 Primers for generation of KS mutants
Full size table

Cell lines

Ehmt1 KO mouse embryonic stem cells (mESCs) (CD12) or Ehmt1 and 2 DKO mESCs (248-6) that stably express WT or mutant mGLP were established by transfection of pCAGGS/FLAG-mGLP vectors. At 2 days after transfection, the cells were treated with puromycin (1 µg/ml) and the cells that stably express mGLP were selected and cloned. All mESCs were cultured in Dulbecco’s modified Eagle's medium (Merck) containing 10% KnockOut serum replacement (Gibco), 2% fetal calf serum (Biowest), and non-essential amino acids (Gibco).

Antibodies

Antibodies used in this study were: anti-GLP (clone#0422) [8], anti-G9a (clone#8620) [8], anti-Histone H3 (CT, pan) (#07–690, Millipore), anti-Histone H3K9me1 (clone#2F7a) [9], anti-Histone H3K9me2 (clone#6D11) [9], anti-Histone H3K9me3 (clone#2F3) [9], anti-FLAG M2 (Merck), anti-FLAG (polyclonal) (#F7425, Merck), anti-α-Tubulin (#B5–1–2, Merck), and anti-His (#291–3, MBL).

Preparation of human and mouse GLP proteins for HMT assay

To produce human His-tagged hGLP (982–1266 aa) proteins, Escherichia coli BL21 (pLysS) strain transformed with pET28a plasmid was cultured in LB medium with kanamycin (50 µg/ml) and 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) for 4 h at 37 °C. The cells were pelleted by centrifugation and suspended in the lysis buffer (20 mM sodium phosphate buffer (pH 6.8), 0.25 M NaCl, and protease inhibitor cocktail (EDTA free) (Nacalai)). The suspension was incubated on ice for 15 min with lysozyme (1 mg/ml). Then, cells were lysed by freezing in liquid N2 and thawing thrice on ice. The samples were centrifuged at 40,000×g for 1 h. The supernatants were loaded onto 1 ml Ni-NTA agarose beads packed in Econo-Column (Bio-Rad) and were equilibrated with the lysis buffer. The column was washed with the lysis buffer and 20 mM imidazole. The bound proteins were eluted using 50 mM, 100 mM, and 250 mM imidazole in the lysis buffer. WT and mutant His-tagged hGLP were eluted with 100 mM imidazole. The fractions containing GLP proteins were dialyzed against the solution containing 20 mM sodium phosphate buffer (pH 6.8), 0.15 M NaCl, 10% glycerol, and 1 mM dithiothreitol, and stored at −80 °C.

C-terminal truncated FLAG-tagged mGLP (1009–1296) proteins were produced using TnT SP6 high-yield wheat germ protein expression system (Promega).

In order to prepare the full-length FLAG-mGLP, Ehmt1 and 2 DKO ES cells that stably express WT and mutant mGLP were lysed in RIPA buffer (20 mM Tris-HCl (pH 7.5), 420 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate, 1 mM EDTA, and protease inhibitor cocktail). FLAG-tagged mGLP proteins in the lysate were immunoprecipitated by anti-FLAG M2 affinity gel (Merck). After incubation of the lysates with the FLAG beads at 4 °C, beads were washed twice with RIPA buffer and were immediately used for HMT assay.

HMT assay

In vitro histone methylation assay was performed as previously described with minor modification [11]. Human or mouse WT and mutant GLPs were incubated with recombinant Histone H3 (NEB) and 14C-SAM for 30 min at 30 °C. The reaction products were resolved on sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and 14C-methylated histone H3 was detected by autoradiography.

Binding assay of G9a and GLP

WT (TT2) and Ehmt1 KO (CD12) mESCs that stably express WT and mutant mGLPs were lysed in RIPA buffer as described in the above section. The lysates were incubated with anti-GLP in RIPA buffer for 1 h at 4 °C. Later, Dynabeads protein G (Invitrogen) were added to the reaction and incubated for 30 min. The beads were washed twice with RIPA buffer and the bound proteins were resolved on SDS-PAGE which was followed by western blot analysis.

Results and discussion

Location of the two missense-derived amino-acid changes in GLP of patients with KS

Two missense mutations of EHMT1 were previously reported in patients with KS, (i) c.3125G>A (p.C1073Y) [2] and (ii) c.3589C>T (p.R1197W) [4]. Those two patients were both female and had the core KS phenotypes, including developmental delay, severe mental retardation, and distinct facial features plus heart defect. C1073 is located in a cysteine cluster that surrounds three zinc ions in the pre-SET domain. It was predicted that C1073Y mutation will abolish the cysteine and zinc interaction, and induce conformational change of the loops which interact with the SET domain of the other monomer [2] (Fig. 1a). In contrast, R1197 is located in the catalytic pocket of the I-SET domain and interacts with histone H3 N-terminal peptide [12]. It was predicted that R1197W mutation affects the H3 N-terminal tail binding and the catalysis of lysine methylation reaction.

Fig. 1
figure 1

Location of KS missense mutations. a Crystal structure of human GLP (PDB id#3hna) monomer (left, 982–1266 aa) and homo-dimer (right). Amino acids C1073, R1197, and NHHC (1199–1203) are shown in blue, red, and pink, respectively. GLP-binding factors are also shown in green (histone H3 peptide), yellow (SAH), and orange (Zn ion), respectively. The structures are designed using UCSF Chimera (http://www.rbvi.ucsf.edu/chimera). b Alignment of human and mouse GLP proteins. Identical residues are boxed in black. Ankyrin repeats and SET domain are boxed in green and blue, respectively. Amino acids C1073, R1197, and C1203 in hGLP are shown in blue, red, and pink, respectively. NHHC, GLP enzymatic active motif is underlined in pink. (Color figure online)

Full size image

In vitro HMT activity of the two missense mutations (C1073Y and R1197W) in hGLP

We introduced these missense mutations in the C-terminal part that comprises pre-SET and SET-domain (hGLP982–1266) of EHMT1 cDNA (Fig. 2a) and performed in vitro validation of the HMT activity of WT and two missense mutants of hGLP982–1266 that were produced in E. coli. As seen in Fig. 2b, H3 is efficiently methylated by WT hGLP; however, it is poorly and not methylated by C1073Y and R1197W mutants, respectively. Therefore, these two mutants show defective HMT function in vitro.

Fig. 2
figure 2

Validation of enzymatic activities of KS mutants in vitro. a Schematic representation of His-tagged hGLP truncated protein (982–1266) and FLAG-tagged mGLP full-length and truncated (1009–1296) proteins. Cysteine-rich domain, ankyrin repeats, and SET domain are boxed in yellow, green, and blue, respectively. b–d Histone methyltransferase activity of recombinant hGLP (982–1266) (b), in vitro translated mGLP (1009–1296) (c), and full-length mGLP expressed in Ehmt1 and 2 DKO mESC lines (d). Methylated histone H3 was detected by autoradiography. Histone H3 was detected by Coomassie brilliant blue staining. Human and mouse GLP were detected by western blotting using anti-His and anti-FLAG, restrictively. (Color figure online)

Full size image

In vivo HMT activity of the two missense mutants of mGLP

To further validate these two missense mutations in vivo, we switched to mouse system as (i) mouse Ehmt1 KO ESCs are available, (ii) human and mouse GLPs are highly homologous (88.5 and 96.2% identical for full-length and the C-terminal catalytic regions, respectively), and (iii) C1073 and R1197 locations are also conserved in mGLP (Fig. 1b). After these substitution mutations were introduced in mGLP (Fig. 2a), we examined the in vitro HMT activity of these mutants. As seen in Fig. 2c, d, the FLAG-tagged minimal C-terminal catalytic region (1009–1296) as well as full-length mGLP have methylated H3; however, C1071Y and R1195W mutants of C-terminal region as well as full-length have unmethylated H3. Additionally, in our study, we included a known catalytically defective but dimerization-competent mGLP mutant (C1201A) [10] and confirmed that C1201A mutation in full-length as well as C-terminal region showed no HMT activity.

Heterodimer formation activity of the two mutants of mGLP with mG9a

Finally, we validated these mutants for their heterodimer formation capacity with G9a and the subsequent HMT function in vivo. To this purpose, the WT, C1201A, and two mGLP mutants were introduced into the Ehmt1 KO ES cells. After the stable transfectants with similar expression of each construct were established, we assessed the cellular quantity of mouse G9a (mG9a) in each cell line as mG9a is destabilized in the Ehmt1 KO cells due to its inability to form heterodimer with mGLP [8] (Fig. 3a, left panel, first and second lanes from left). As seen in Fig. 3a, b, cellular quantity of mG9a is clearly recovered by the expression of WT and C1201A mutant, but not by C1071Y and R1195W mutants of mGLP. Furthermore, endogenous mG9a was efficiently co-immunoprecipitated with FLAG-tagged WT and C1201A mutant, but not with these two substitution mutants of mGLP (Fig. 3b). Then, we examined the status of H3K9me1, -2, and -3 in the Ehmt1 KO ES cells that stably express WT or mutant mGLP. As seen in Fig. 4, all the forms of H3K9 methylation that are diminished in Ehmt1 KO ES cells are recovered by WT mGLP expression. Moreover, the catalytically inactive but dimerization-competent C1201A mutant had rescued H3K9me1 and me2 methylation as described in a previous study [10], but the level of H3K9me3 was not recovered which was unexpected. Conversely, both the substitution mutants (C1071Y and R1195W) could not rescue any form of constitutive H3K9 methylation in Ehmt1 KO cells. These data strongly suggest that the two missense mutants show defective dimerization potential as well as HMT function in vivo. Based on above results, Fig. 5 shows a prediction of G9a/GLP complex status and levels of H3K9 methylation in two KS patients with missense mutation of EHTM1 (C1073Y or R1197W).

Fig. 3
figure 3

Validation of KS mutants for the G9a/GLP heterocomplex formation in vivo. a mG9a protein level in Ehmt1 KO mESCs (CD12) expressing FLAG-tagged mGLP WT and mutant (C1201A, R1195W, and C1071Y). Relative amount of mGLP and mG9a was quantified by LI-COR system and normalized by GLP/G9a protein level in WT mESCs (TT2) as 1 (n = 6 independent experiments, means ± SD, *p < 0.05, **p < 0.01 Student’s t-test). b Binding ability of mGLP WT or mutants to endogenous mG9a in CD12 cell. Mouse GLP WT and mutants expressed in CD12 or TT2 were immunoprecipitated by anti-GLP antibody. Blots were quantified by LI-COR system and normalized by the amount of mG9a bound to FLAG-mGLP WT as 1

Full size image
Fig. 4
figure 4

Impact of KS mutations on H3K9 methylation in vivo. Methylated histone H3K9 level in TT2 and CD12 expressing mGLP WT and mutants. H3K9me1, me2, and me3 were detected by western blotting with specific antibodies against each modification and pan-H3 antibody. Amount of methylated histone H3 was quantified by LI-COR system and normalized by the intensity of methylated H3K9 in TT2 as 1 (n = 4 – 5 independent experiments, means ± SD *p < 0.05, **p < 0.01, ***p < 0.001 Student’s t-test)

Full size image
Fig. 5
figure 5

Illustrative prediction of G9a/GLP complex status and levels of H3K9 methylation in two KS patients with missense mutation of EHTM1 (C1073Y or R1197W). Cellular amounts of G9a, GLP, and G9a/GLP complex and levels of H3K9me1,2 in WT (upper panel) and Ehtm1 KO (middle panel) mESCs are shown. Bottom panel depicts the G9a/GLP complex status and cellular contents, and levels of H3K9 methylation, which are thought to occur in two KS patients with EHMT1 missense mutation. Note that G9a/GLP heterocomplex is functional HMT in vivo. (Color figure online) 

Full size image

In conclusion, our biochemical characterization of the previously reported missense mutations in patients with KS further supports the concept that impairment of the HMT function of GLP is crucial for causing major symptoms of KS. However, at present we cannot determine whether the reduction of the G9a/GLP (KMT) complex formation or the intrinsic HMT activity of GLP is crucial for the development of KS. Various other important questions remain to be elucidated. Among them, it is important to clarify the spatio-temporal pattern of epigenetic dysregulation that is induced by EHMT1 haploinsufficiency, and to examine the degree of epigenetic changes that are crucial for the manifestation of symptoms in patients with KS. After addressing these issues, we could advance on to a new stage for developing therapeutic measures, which may harness “the reversibility of epigenetic modifications” to alleviate symptoms of patients with KS.