Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

© 2016. Published by The Company of Biologists Ltd.

Methyl-CpG-binding domain protein 3-like 2 (MBD3L2)

promotes Tet2 enzymatic activity for mediating 5mC oxidation

Lina Peng1,2,3¶, Yan Li1,2,3¶, Yanping Xi1,2, Wei Li1,2, Jin Li1,2,Ruitu Lv4, Lei

Zhang5, Qingping Zou1,2, Shihua Dong1,2, Huaibing Luo1,2, Feizhen Wu2*,

Wenqiang Yu1,2,3*

1 Laboratory of RNA Epigenetics, Institutes of Biomedical Sciences &

Department of Biochemistry and Molecular Biology, Shanghai Medical

College, Fudan University, 130 Dong-An Road, Shanghai, 200032, China.

2 Key Laboratory of Ministry of Education, Department of Molecular Biology,

Fudan University, 130 Dong-An Road, Shanghai 200032, China.

3 State Key Laboratory of Genetic Engineering, Collaborative Innovation

Center for Genetics and Development, School of Life Sciences, Fudan

University, Shanghai 200032, China

4 Epigenetics Laboratory, School of Basic Medicine and Institutes of

Biomedical Sciences, Shanghai Medical College ofin Fudan University,

Shanghai 200032, China.

5 Biomedical Core Facility, School of Basic Medicine and Institutes of

Biomedical Sciences, Shanghai Medical College of Fudan University,

Shanghai 200032, China.

* Corresponding author

Email: wenqiangyu@fudan.edu.cn

¶These authors contributed equally to this work.

Keywords: MBD3, MBD3L2, Tet2, Enzymatic activity, DNA demethylation,

Epigenetics

JCS Advance Online Article. Posted on 14 January 2016

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Abstract

Ten-eleven translocation (TET) proteins are key players involved in the

dynamic regulation of cytosine methylation and demethylation. Inactivating

mutations of TET2 are frequently found in human malignancies, highlighting

the essential role of TET2 in cellular transformation. However, the factors that

control TET enzymatic activity remain largely unknown. Here we found that

MBD3 and its analogue MBD3L2 can specifically modulate the enzymatic

activity of Tet2 protein, but not Tet1 and Tet3 proteins, in converting 5mC into

5hmC. Moreover, MBD3L2 is more effective than MDB3 in promoting Tet2

enzymatic activity via strengthening the binding affinity between Tet2 and the

methylated DNA target. Further analysis revealed pronounced decreases in

5mC levels at MBD3L2 and Tet2 co-occupied genomic regions, most of which

are promoter elements associated with either cancer-related genes or genes

involved in the regulation of cellular metabolic processes. Our data add new

insights into the regulation of Tet2 activity by MBD3 and MBD3L2 in

modulating its target gene activities in cancer development and have important

applications in understanding how dysregulation of TET2 may contribute to

human malignancy.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Introduction

The TET family including TET1, TET2, and TET3 are Fe (II) - and α-ketoglutarate

(α-KG) dependent dioxygenases which oxidize 5-methylcytosine (5mC) to

5-hydroxymethylcytosine (5hmC). 5hmC can be progressively oxidized into

5-formylcytosine (5fC) and 5-carboxycytosine (5caC), followed by efficient base

excision through TDG followed by base excision repair (BER), culminating in

cytosine replacement and DNA demethylation (He et al., 2011; Wu and Zhang, 2014).

As 5hmC prefers to enrich at promoters, enhancers and other active chromatin regions,

leading to gene activation (Fu and He, 2012; Perera et al., 2015) , which indicates that

5hmC catalyzed by TET play roles in gene expression regulation.

TET1 is highly expressed in embryonic stem (ES) cells, and TET3 is expressed

relatively high in germ cells/oocytes, whereas TET2 is most widely expressed in

somatic tissues, especially haematopoietic cells (Wu and Zhang, 2011). Importantly,

TET2 mutations are frequently found in myeloid malignancies, and TET2 inactivation

in hematopoietic progenitor cells blocks myeloid differentiation (Scopim-Ribeiro et

al., 2015). The reduction of TET2 activity is strongly correlated with the low 5hmC

level in leukemia (Solary et al., 2014). In addition, loss of TET2 function contributes

to melanoma progression (Lian et al., 2012). These results indicate that TET2 plays an

important role in tumorigenesis via DNA demethylation and activation of yet

unappreciated oncogenes. However, TET2 lacks the N-terminal CXXC domain, a

Zn-chelating domain that can bind with unmethylated CpG dinucleotides of the DNA

sequence (Ko et al., 2013). Previous studies have shown that VPRBP-mediated

monoubiquitylation can enhance the binding ability of TET2 and DNA (Nakagawa et

al., 2015), whereas WT1 can recruit TET2 to regulate its target gene expression and

suppress Leukemia cell proliferation (Wang et al., 2015). Nevertheless, the precise

mechanism by which TET2 regulates its target gene activation is still unclear.

MBD family proteins are highly conserved in all vertebrates, and capable of

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

selectively recognizing and binding to methylated CpG dinucleotides by the

methyl-CpG binding domain (Hendrich and Bird, 1998). However, an exception to

this general rule is the case of MBD3 due to lack of the conserved motif (Clouaire and

Stancheva, 2008). Originally, there are several conserved analogous genes of MBD3,

MBD3L1 and MBD3L2 that lack the MBD domain of MBD3 but share a 42%

similarity with MBD3 (Clouaire and Stancheva, 2008).The function of MBD family

proteins are mainly dependent on the MBD domain, which attribute to the genome

wide location of MBD proteins at methylated CpG-dense, and inactive regulatory

regions (Baubec et al., 2013). Yet, MeCP2 with the deletion of MBD domain, is

preferably located at open accessible chromatin regions (Baubec et al., 2013),

indicating that MBD family protein lacking MBD domain may converse their intrinsic

function from repression to activation. It has been reported that MBD3 family proteins

are associated with gene activation (Brown and Szyf, 2007). Recent study has shown

that MBD3 can bind 5mC oxidation derivatives (Iurlaro et al., 2013; Yildirim et al.,

2011), which are the intermediates in the demethylation process, inspiring us to

further explore whether MBD3 and its analogues function in the process of DNA

demethylation mediated by TET proteins.

In the present study, we found that both MBD3 and its analogue MBD3L2 enhance

the enzymatic activity of Tet2. In addition, MBD3L2 enhances Tet2-mediated DNA

demethylation most likely via reinforcing the accessibility of substrates to

Tet2/C-terminal cysteine-rich dioxygenase (CD) domain, leading to the activation of a

group of oncogene expression. Our findings provide a new model of TET regulation,

indicating a potential role of MBD3-TET2 axis in the epigenetic control of

tumorigenesis.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Results

MBD3 and MBD3L2 co-occupy with Tet2 on chromatin

In order to explore whether MBD3 family proteins function in the process of DNA

demethylation mediated by TET proteins, we performed immunofluorescence

experiment to show the localization of MBD3 and Tet2 in mammalian cells. We

transiently transfected HA-Tet2 and Myc-MBD3 to HEK293T cells, and the confocal

microscopic analysis of the IF staining revealed that both the exogenously expressed

Tet2 and MBD3 localized in the nucleus (Fig. 1A).

MBD3L2 was homologous to MBD3 with the absence of MBD domain, which is

specifically expressed in germ cells and related tumors (Iurlaro et al., 2013; Yildirim

et al., 2011). As MBD3L2 is less expressed in somatic cells (Jin et al., 2008),

including HEK293T cells, we constructed MBD3L2 plasmid with tag and transfected

it to HEK 293T cell to explore the localization of MBD3L2. Firstly, after 48 hours of

Myc-MBD3L2 and HA-Tet2 transient transfection, immunofluorescence staining of

the constructed tags and confocal microscopic image show that MBD3L2 localized in

the nucleus and more interestingly, it seems that MBD3L2 co-localized with Tet2 in

the nucleus (Fig. 1A).

To further define whether MBD3/MBD3L2 co-localize with Tet2, we investigated the

genome-wide distribution of MBD3 and Tet2 in mammalian cell by chromatin

immunoprecipitation-sequencing (ChIP-seq). Since Tet2 is poorly

chromatin-associated, it is challenging to successfully perform ChIP (Deplus et al.,

2013). Tet2 overexpressed cells were sequentially cross-linked with DSG and

formaldehyde and then we did Tet2 ChIP (Liu et al., 2013). ChIP-qPCR proved that it

is suitable for Tet2 ChIP (supplementary material Fig. S1A). Bioinformatics analysis

of ChIP-seq data revealed that there are 34821 Tet2 specific binding peaks genome

wide, which are enriched at promoter region (Fig. 1B). Analysis of Tet2 binding

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

peaks at each class of genomic elements (Wu et al., 2014) indicate that Tet2 is more

enriched at promoter (observed/random [obs/rand]=4.1) and exon (obs/rand=5.5)

relative to intron and intergenic regions (Fig. 1B). Recently, Shimbo et al have

successfully conducted MBD3 ChIP in human MCF7 cells using DSG and

formaldehyde for the crosslinking (supplementary material Fig. S1B) (Shimbo et al.,

2013). Although Tet2 and MBD3 ChIP-seq data are from different cell types, we

surprisingly found that 7504 MBD3 ChIP peaks (p-value< 10-16, hypergeometric test)

are overlapped with Tet2 binding peaks. KEGG (Kyoto Encyclopedia of Genes and

Genomes) pathway enrichment analysis showed that the overlapped genes are

enriched in cancer pathway (supplementary material Fig. S1C), which implies that the

regulation of Tet2 activity may be associated with cancer development.

Moreover, we completed MBD3L2 ChIP-seq in the HEK 293T cell after the

transfection of tagged MBD3L2, and found that MBD3L2 is more enriched at

promoter (observed/random [obs/rand] =5.8) and exon (obs/rand=7.3) relative to

intron and intergenic regions (Fig. 1C). Further analysis indicates that above half of

MBD3L2 binding peaks are overlapped with Tet2’s (Fig. 1D) and the overlapped

gene are also enriched in cancer pathway (supplementary material Fig. S1D).

Furthermore, MBD3L2 binding profile is similar to the genomic distribution observed

for Tet2 in HEK293T cell lines (Fig. 1E, F). We further conducted ChIP-qPCR

analysis and found that Tet2 and MBD3/MBD3L2 were enriched at Tet2 target sites

in HKE293T cells (supplementary material Figs. S1E, F, G), further validating the

co-localization of Tet2 and MBD3 family proteins. These results indicate MBD3L2

did co-localize with Tet2 in nucleus, which inspired us to further investigate the

biological effects of MBD3/MBD3L2 on Tet2 in mammalian cells.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

MBD3 and MBD3L2 enhance Tet2 enzymatic activity by mediating 5hmC

generation

MBD3L2 expression occurs in oocyte and germ cell tumors, while MBD3L1

specifically in haploid male germ cells (Jin et al., 2008). It has been reported that

MBD3L1 and MBD3L2 do not possess direct DNA demethylation function (Jin et al.,

2008), but we found that MBD3 and MBD3L2 co-occupy with Tet2 in mammalian

cells, it is still intriguing to explore whether they have regulatory effects on Tet2

enzymatic activity. Therefore, MBD3 family proteins, including MBD3, MBD3L1,

MBD3L2, and Tet2 with HA tags were transiently transfected to HEK 293T cells

independently. We used HA antibody to purify Tet2 and MDB3 family proteins and

eluted full functional Tet2 and MBD3 family proteins with HA peptide

(supplementary material Fig. S2A). We use high slat solution to wash the proteins

during the purification to remove the genomic DNA contamination as much as

possible. We used the parallel reactions without adding methylated oligo to do the

LC-MS/MS assay, and the results show that 5hmC was not detected and dC, 5mC and

dG were less detected. Moreover, the content of dG is only 1/100 of our methylated

oligo sample, which will not affect the conclusion (supplementary material Fig. S2B).

The 120bp DNA fragment containing completely methylated cytosine was artificially

made by PCR, owing to the substrate of cytosine replaced by 5mC during PCR

amplification. The purified Tet2 and MBD3 family proteins were incubated with

methylated DNA fragment at 37℃ for 1h. The DNA was extracted, digested and

analyzed by LC-MS/MS, and the enzymatic activity of Tet2 was measured by 5hmC

generation. 5hmC standard nucleoside was used to verify the effectiveness of

LC-MS/MS method. N-oxalylglycine (NOG, a 2-OG analog) was used to replace part

of the 2-OG for reaction and the results show that NOG can repress half of Tet2

enzymatic activity (Fig. 2A). Interestingly, MBD3 increases 5hmC level by about four

folds. However, MBD3L1 has no effects on modulating Tet2 activity (Fig. 2A).

Surprisingly, MBD3L2 together with Tet2 can increase the 5hmC production by

thirteen folds comparing to Tet2, suggesting that MBD3L2 has stronger effect on

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

regulating Tet2 enzymatic activity (Fig. 2A and supplementary material Fig. S2C).

Moreover, incubation of MBD3L2 with Tet1 or Tet3, we did not result in increased

Tet2 activity (supplementary material Fig. S3A), indicating that MBD3L2 functions

specifically with Tet2. In order to investigate whether recombinant MBD3L2 protein

enhances the activity of Tet2 protein, the recombinant proteins purified from E.coli

were used in the enzymatic analysis. Our data show that recombinant MBD3L2 can

also increase the enzymatic activity of recombinant TET2 in vitro (supplementary

material Fig. S3B). In addition, DNA fragment treated by CpG specific bacterial

methyltransferase M.SssI was used as the substrate of the enzymatic activity, to make

sure that the enhancement also play effects in the context of CpG dinucleotides

(supplementary material Fig. S3C).The full-length Tet2 protein purified is easy to

degrade in vitro, which has been reported. The Tet2 protein we used is purified from

HEK293T cell and the amount of Tet2 protein is limited. And the newly published

data (Guo et al., 2015) indicate that TET proteins are evolutionarily tuned to be less

reactive towards 5hmC, therefore the generation of 5fC and 5caC is far more less than

5hmC. When we increase the amount of Tet2 protein in enzymatic activity, the 5fC

and 5caC can be detected, moreover, the amount of 5fC and 5caC increased with the

addition of MBD3L2 in vitro.

Since MBD3L2 can enhance the enzymatic activity of Tet2 greatly, we attempted

to detect physical interaction between MBD3L2 and Tet2 in mammalian cells. We

transfected SBP-MBD3L2 and Flag-Tet2 into HEK 293T cell, and performed the

co-immunoprecipitation with antibodies specific to different tags. MBD3L2 can

specifically pull down the Tet2 protein and vice versa (Fig. 2B, C), indicating that

MBD3L2 can physically interact with Tet2 in mammalian cells. Empty vectors were

used in the co-IP experiment to prove the specificity of TET2 and MBD3L2 and

beta-actin was included as the negative control. It has been acknowledged that

C-terminal cysteine-rich dioxygenase (CD) domain is responsible for the enzymatic

activity of Tet2 (Hu et al., 2013). To confirm which part of Tet2 protein interacts with

MBD3L2, we transfected Tet2 CD domain or CD minus part of Tet2 together with

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

MBD3L2 into HEK 293T cells. MBD3L2 can clearly pull down the Tet2 CD domain

but failed to pull down CD minus part of Tet2 (Fig. 2D), which demonstrate that

MBD3L2 may enhance the enzymatic activity of Tet2 by directly interacting with the

CD domain.

To test this, we purified Tet2 CD proteins with HA tag from HEK 293T cells, then

incubated with MBD3L2 proteins and methylated DNA fragments. We measured

5hmC level by LC-MS/MS. MBD3L2 together with Tet2 CD domain increased 5hmC

level by about two folds (Fig. 2E). Moreover, when we measured the 5fC and 5caC

levels by LC-MS/MS, it is very clear that 5fC and 5caC levels increased accordingly

by MBD3L2 via Tet2 CD (Fig. 2E and supplementary material Fig. S2D). The above

results suggest that MBD3L2 promotes not only the first step of 5mC oxidation, but

also the subsequent oxidation mediated by Tet2 CD.

MBD3L2 enhances Tet2 binding affinity

As Tet2 lacks the DNA binding domain, we speculated that MBD3L2’s interaction

with Tet2 may either guide Tet2 to the targeted methylated DNA sequences or modify

the affinity of Tet2 protein to its substrate. We performed the EMSA (electrophoretic

mobility shift assay) experiments and find that Tet2/CD can bind with the substrate,

methylated or unmethylated DNA (Fig. 3). Moreover, the EMSA experiment of

Tet2/CD show that MBD3L2 increases Tet2 binding to the methylated DNA, and this

binding affinity increase is clearly in a dose-dependent manner (Fig. 3A). However,

MBD3L2 cannot bind with the substrate (Fig. 3B) (Jin et al., 2005). Moreover, the

binding affinity of Tet2 CD and unmethylated DNA cannot be enhanced by MBD3L2

(Fig. 3C), indicating that MBD3L2 plays specific enhancement on Tet2’s binding on

methylated DNA. Collectively, these results support that MBD3L2 mainly interacts

with Tet2/CD domain to increase its binding affinity to the methylated DNA for

regulating Tet2 oxidation.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

In order to investigate whether MBD3L2 increase Tet2’s binding affinity in vivo,

ChIP-qPCR analysis indicate that co-transfected MBD3L2 increased Tet2 enrichment

at target regions. In addition, we transfected MBD3L2 siRNA to knockdown the

overexpressed MBD3L2 and found that the enrichment of Tet2 binding at the target

regions decreased accordingly (supplementary material Fig. S1H), indicating that

MBD3L2 specifically increases Tet2’s binding to the target genes in HEK293T cells.

MBD3L2 promotes the demethylation process catalyzed by Tet2

In order to explore whether MBD3L2 enhances global DNA demethylation mediated

by Tet2 in mammalian cells, we transfected Flag-MBD3L2 plasmids with or without

HA-Tet2 into HEK 293T cell. The genomic DNA was extracted, digested and

detected with HPLC to measure 5mC and 5hmC levels for evaluating the enzymatic

activity of Tet2 protein. As expected, we discovered that MBD3L2 together with Tet2

significantly decreased global 5mC level comparing to that of Tet2 (Fig. 4A and

supplementary material Fig. S3D), and the expression level of Tet2 protein has no

difference between these two groups (Fig. 4A). Accordingly, 5hmC was significantly

increased in MBD3L2 plus Tet2 group (Fig. 4B and supplementary material Fig. S3E).

These preliminary results demonstrate that MBD3L2 can promote Tet2 enzymatic

activity by converting more 5mC to 5hmC in mammalian cells.

In order to verify the in vivo enhancement of MBD3L2 on genome-wide

demethylation via Tet2, we transfected MBD3L2 and Tet2 plasmids into HEK 293T

cells, and then performed Reduced Representational Bisulphite sequencing (RRBS) to

measure DNA methylation status among different groups (Gu et al., 2011). We found

that MBD3L2 together with Tet2 can decrease the general DNA methylation level

comparing with other groups (supplementary material Table S1). We compared the

CpG density of affected regions and the results show that the CpG density of common

RRBS regions tend to be higher than the CpGs in the human genome

(p-value<2.2e-16, Wilcoxon rank sum test), which is attributed to the enrichment of

CCGG site by the method of RRBS (Fig. 4C). Moreover, the demethylated regions

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

induced by Tet2 and MBD3L2 is preferably occurred in the relatively low CpG

density region (p-value<2.2e-16, Wilcoxon rank sum test) (Fig. 4C), which is

consistent with the report that 5hmCG is enriched in Low CpG Regions (Yu et al.,

2012). Moreover, in order to examine whether MBD3L2 together with Tet2 can

influence the DNA methylation, we analyzed the RRBS data for the methylation

status across the overlapped binding regions while compared with the non-overlapped

binding regions. As expected, the general DNA methylation level of the overlapped

binding regions by Tet2 and MBD3L2 decreased significantly (p-value=

8.210051e−51, Wilcoxon rank sum testing) than the non-overlapped counterpart (Fig.

4D), which further confirms that MBD3L2 can function as a modulator of Tet2 and

further affect the Tet2 target gene expression during tumor development.

As that we couldn’t distinguish 5mC and 5hmC with RRBS, we did comparative

hMeDIP-seq to detect the global 5hmC level for evaluating the efficiency of

demethylation under different conditions (supplementary material Fig. S4A). The four

samples were sonicated and ligated with adaptors containing a specific barcode

sequence each (Tan et al., 2013). hMeDIP and sequencing for the equal amount of

four samples mixing in one reaction system can reduce the experimental error and

variation among these four samples. The sequencing data can be separated by the

different barcodes sequence, and the 5hmC level difference of each sample leads to

the different amounts of sequencing reads, which makes it possible to compare the

5hmC level between different samples with sequencing data (Tan et al., 2013). After

normalized to the correspondent input, the percentage of sequencing data for each

sample can represent the percentage of 5hmC level for each sample in general (Tan et

al., 2013). The percentage of the sequencing reads in MBD3L2 together with Tet2

group is about 48.9%, almost two folds to Tet2 and three folds to the control (Fig. 4E).

Yet MBD3L2 alone has no effect on the 5hmC level (Fig. 4E). Further analysis

reflects that the increased 5hmC regions induced by MBD3L2 are enriched at TSS

and exon regions (supplementary material Fig. S4B). Moreover, we observed a

significantly increased 5-hmC level within the averaged gene bodies and TSS± 5kb

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

regions in MBD3L2 together with Tet2 group compared with other groups (Fig. 4F,G).

These results clearly demonstrate that MBD3L2 increases Tet2 demethylation activity

but cannot influence DNA methylation only by itself.

In order to investigate whether MBD3L2/MBD3 affect endogenous TET2, we

transfected MBD3L2 and MBD3 in human acute lymphoblastic leukemia Jurkat cells,

and we found that either MBD3L2 or MBD3 increases the global 5hmC level by

about two times in Jurkat cell (Fig. 4H), indicating that MBD3L2 and MBD3 enhance

the enzymatic activity of endogenous TET2.

MBD3L2 promotes Tet2 target genes expression by enhancing

demethylation

In order to investigate the biological function of MBD3L2 regulating Tet2

enzymatic activity in mammalian cells, we extracted RNA and perform microarray

profiling analysis, which show that MBD3L2 together with Tet2 can up-regulated

1212 genes and down-regulated 421 genes (GSE74915). GO analysis show that the

up-regulated genes are enriched at metabolic process (Fig. 5A), such as XRCC6 and

LDHB, which are also the target genes of Tet2 and MBD3L2. Moreover, the genes

up-regulated after MBD3L2 and Tet2 co-transfection not only take part in cellular

metabolic process but may also be related to cancer development, such as IRF4 and

HOXD1, which have been reported to be abnormally expressed in leukemia and

breast cancer (Faryna et al., 2012; So et al., 2014). The RT-qPCR experiment

validated that MBD3L2 and Tet2 further increased the cancer-related genes

expression comparing to Tet2 itself (Fig. 5B). We transfected HA-Tet2 and Tet2

siRNA sequentially in HEK293T cell, and the up-regulated gene expression were

reduced (supplementary material Fig. S4C). In addition, we found that catalytic

inactive mutant Tet2 didn’t affect Tet2 target genes expression (supplementary

material Fig. S4D). Furthermore, MBD3L2 co-transfected with mutant Tet2 didn’t

enhance these genes expression, indicating that MBD3L2 increased Tet2 target genes

expression were dependent on the catalytic activity of Tet2.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Since the 5hmC level of promoter and gene body regions were involved in

up-regulated gene expression (Song et al., 2011; Ficz et al., 2011), we considered

whether the up-regulated gene expression mediated by MBD3L2 and Tet2 was

dependent on the demethylation process. Accordingly, we analyzed the hMeDIP data

on the promoter regions of those selected target genes and found that the 5hmC level

of the target genes promoter increased significantly by MBD3L2 via Tet2 protein (Fig.

5C). To further quantify the 5hmC status at the specific locus within Tet2-target gene

promoters, we performed glucosylated hydroxymethyl-sensitive qPCR

(GluMS-qPCR), which enables a single-base-level quantification of the methylation

status of CCGG sites. Our data show that co-transfection of MBD3L2 and Tet2 in

HEK293T cells led to a significant increase of 5hmC at particular loci of the

promoters or genebody regions in Tet2-target genes (supplementary material Fig.

S4E). In sum, our results demonstrate that MBD3L2 reinforces the DNA

demethylation process of Tet2 target regions, and induces the up-regulation of target

gene expression that may be involved in tumorigenesis.

Discussion

Ten-eleven translocation (TET) dioxygenases are able to catalyze the oxidation of

5mC into 5hmC and further oxidize 5hmC to 5fC and 5caC, which has revealed new

pathways in the cytosine demethylation process. The TET2 inactivating mutations in

hematological malignancies and other solid tumors suggest that cellular

transformation is in part caused by the dysregulation of 5-mC conversion (Nakajima

and Kunimoto, 2014). The down-regulation of TET2 was suggested to be responsible

for the loss of 5-hmC in melanoma cells and overexpression of TET2 suppresses

melanoma growth, indicating that 5hmC regulated by TET2 affects melanoma

progression (Lian et al., 2012). TET2 lacks the CXXC domain, of which TET1/TET3

have been proven to bind methylated cytosine and non-CpG sequence (Wu and Zhang,

2014), facilitating the recruitment of TET proteins to their genomic targets. Thus, it is

likely that other factors would be involved in the demethylation process on TET2

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

target regions. Here we show that MBD3 and its analogue MBD3L2 co-localize with

Tet2 in the nucleus and exhibit the ability to increase the enzymatic activity of Tet2.

As MBD3 is preferentially associated with CpG rich promoters and enhancers marked

by H3K4me3 or H3K27ac (Shimbo et al., 2013), our results provide evidence that

MBD3 family may interact with Tet2 in fine tuning the regulatory elements for gene

expression. Jin’s report (Jin et al., 2008) indicates that MBD3L2 has no direct

demethylation effects and MBD3L2 knockout mouse is alive. In our experiments, we

also find that MBD3L2 and MBD3 did not have direct demethylation effects, however,

they can promote the demethylation effects mediated by Tet2. We propose that the

complementary effects of MBD3 can function when MBD3L2 is knocked out.

MBD3 is the only member of MBD proteins that lacks specificity towards

methyl-CpGs in vitro due to evolutionary mutations in the MBD domain (Baubec et

al., 2013). It has been reported that the 60 N-terminal amino acids of MBD3 are

important for Nanog-dependent MBD3’s ability to facilitate reprogramming (Dos et

al., 2014). Recent studies show that MBD3 can bind 5hmC and regulate gene

expression. Surprisingly, here we found that MBD3L2 was able to increase Tet2

enzymatic activity greatly, indicating that the C terminal of MBD3, as well, plays a

contributory role in the functions of MBD3. In turn, another analogue of MBD3,

MBD3L1, has no effects on Tet2 catalytic activity, revealing it is a particular

regulation between MBD3L2 and MBD3. Additionally, it has been reported that both

MBD2 and MBD3 possess the ability to induce DNA demethylation (Brown et al.,

2008; Wang et al., 2013), nevertheless, few of these studies provide solid evidence or

clarified the detailed mechanism. We found that MBD3L2 was a partner of Tet2 and

the interaction was mediated by the CD domain of Tet2. Our EMSA experiments

provided the evidence that MBD3L2 increases the binding affinity of Tet2/CD to their

substrates, methylated DNA. We propose that the binding affinity enhancement plays

roles on Tet2 activity, binding to methylated DNA is a prerequisite for the function of

Tet2 in catalyzing 5mC hydroxylation. We think that on the initiation of the enzyme

reaction, the increased binding affinity of Tet2 and methylated DNA may make more

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

5mC hydroxylated to generate 5hmC. However, at the end of the reaction, the

increased binding affinity of Tet2 and methylated DNA trapped the protein, and

limited the reaction of Tet2 and other substrate. Therefore, the increased binding

affinity may regulate Tet2 activity dynamically. The kinetic analysis indicates that

Tet2 enzymatic efficiency was defined as Kcat/Km. Although increased binding

affinity lead to Km increased, Tet2 catalyze more 5mC and generate more 5hmC per

second, which lead to Kcat increased. In our experiment, the content of 5mC was

excessive and the increased binding affinity makes the enzymatic reaction rate

increased, leading to the increase of Kcat/Km, which enhances the enzymatic

efficiency of Tet2. There are some evidence shown that TET2 activity can be

enhanced by other factors. For example, it has been reported that PRDM14 induces

DNA demethylation at germline-specific genes Piwil2 and Slc25a3, meanwhile,

PRDM14 can also enhance the binding of Tet2 at or near the TSS of these genes

(Okashita et al., 2014). Recent study has shown that histone H3 peptide induces the

activation of de novo DNA methyltransferase by removal of Dnmt3a autoinhibiton for

configuration change (Brown et al., 2008; Wang et al., 2013). Considering that

MBD3L2 enhances Tet2 binding affinity rather than binds with methylated DNA itself,

whether the configuration of Tet2 was altered by MBD3L2 needs further

investigation.

Traditionally, MBD3 is shown to be involved in transcriptional repression, however,

MBD3 is also required for embryonic development, pluripotent cell differentiation

and acts as a positive facilitator in the transcription-factor-mediated reprogramming of

NSCs (Dos et al., 2014; Kaji et al., 2007). Transcriptome analysis has shown that 61%

of differentially expressed genes are down-regulated after Mbd3 deletion in ESCs,

suggesting it to be a mediator of transcription-factor-induced gene activation. In

somatic cells, MBD3 is essential for rRNA promoter demethyaltion and its absence

results in the hypermethyaltion of rRNA gene promoter (Dos et al., 2014; Kaji et al.,

2007). Knowing this, it seems likely that MBD3 may be a mediator of

transcription-factor-induced gene activation. Despite knowing that MBD3L2 is one

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

analogue of MBD3, sharing 40% similarity with MBD3, yet the function of MBD3L2

remains unclear. Here we show that both MBD3L2 and Tet2 can up-regulate a group

of genes, which are associated with metabolic process, suggesting that MBD3L2 and

Tet2 may serve as general activators. Further experiments show that the up-regulation

of the Tet2-dependent gene was enhanced by MBD3L2, possibly due to effects on

demethylation process at Tet2 target regions. More importantly, most of the

up-regulated genes are cancer associated. Since MBD3L2 is specifically expressed in

germ cell tumors, it needs further research to explore whether MBD3L2 can play a

role on TET2 in regulating germ cell tumorgenesis.

Our study provides evidence that MBD3 and its analogue MBDL2 promote Tet2

enzymatic activity, and MBD3L2 enhances the demethylation process mediated by the

Tet2 CD domain, especially at overlapped binding regions of Tet2 and MBD3L2. We

found that the binding affinity of Tet2 and methylated DNA was increased by

MBD3L2, which contributes to Tet2 enzymatic activity enhancement. Moreover,

MBD3L2 increases the expression of Tet2 target genes, such as HOXD1 and IRF4,

which are closely linked with cancer development. Cumulatively, our results

demonstrate that MBD3L2 can enhance Tet2-dependent gene expression, providing

another regulatory pathway for Tet2 target genes and tumor genesis.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Materials and Methods

Antibodies

Antibodies against HA (Cell Signaling, C29F4), Flag (Abmart, M2008), Myc (Abcam,

ab9106), 5hmC (Active Motif, 39791), rabbit IgG (Santa Cruz, sc-2027), SBP (Santa

Cruz, SBC9-14) were purchased commercially. Endogenous TET2 antibody was

obtained from Dr. Dan Ye’s lab.

Plasmids

The cDNA encoding full-length human MBD3L2 and MBD3 were cloned into HA,

myc or GFP-tagged vectors (pCMV-HA; pCMV-Myc; pCDH-GFP; pcDNA3.1 with

SBP tag). DNA fragments of Tet2 catalytic domain (1099–1936) and catalytic minus

domain (1-1098) were sub-cloned into pCMV-HA vector.

Mouse Tet2-pCAG vector was obtained from Dr. Guoliang Xu’s lab. The catalytic

inactivie mutant (CM) vector of Tet2 was obtained from Dr. Dan Ye’s lab.

Cell culture and transfection

Human embryonic kidney 293T (HEK293T) cells were cultured in Dulbecco’s

modified Eagle’s medium (DMEM, HyClone) supplemented with 10% fetal calf

serum (HyClone) at 37°C in 5% CO2. Human acute lymphoblastic leukemia Jurkat

cells (Jurkat cells) were cultured in RPMI-1640 medium (HyClone) supplemented

with 10% fetal calf serum (HyClone) at 37°C in 5% CO2.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Plasmid transfection was carried out either by the calcium phosphate method

(Spandidos and Wilkie, 1984) or lipofectamine 2000 (Invitrogen), and the cells were

harvested 48 h after transfection.

Genomic DNA extraction

Genomic DNA was extracted with QIAamp DNA Mini Kit (Qiagen) according to the

manufacturer’s instructions. RNaseA/T1 was used in the process of DNA extraction

to remove the RNA contamination in genomic DNA. The concentration and quality of

genomic DNA were measured by NANODROP2000c (Thermo) and electrophoresis.

Co-Immunoprecipitation

Cells were lysed in ice-cold lysis buffer [20 mM HEPES, pH 7.9, 1.5 mM MgCl2,

0.15 M NaCl, 0.15% NP-40)] containing protease inhibitor cocktail (Roche).

Immunoprecipitation was carried either by incubating Flag beads (Sigma), HA beads

(Sigma), Dynabeads® M-280 Streptavidin (Invitrogen) and rabbit IgG beads (Sigma)

at 4°C with lysate for 6 hrs. Then the beads were washed for four times with ice-cold

lysis buffer.

Protein extraction for enzymatic analysis

Cells were lysed in ice-cold lysis buffer [20 mM HEPES, pH 7.9, 1.5 mM MgCl2, 0.5

M NaCl, 0.15% NP-40)] containing protease inhibitor cocktail (Roche).

Immunoprecipitation was carried either by incubating Flag beads (Sigma) or HA

beads (Sigma) at 4°C with lysate for 6 hrs. Then the beads were washed for four times

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

with ice-cold lysis buffer. The proteins were eluted with Flag peptide (Sigma) or HA

peptide (Sigma) at 4°C for 2hrs.

In vitro Tet2 enzyme assays

Tet2 proteins were extracted from HEK293T cells with ANTI-FLAG M2 Magnetic

Beads, and the proteins were eluted with anti-flag peptide. The 120 bp methylation

DNA fragment (500ng) was incubated with Tet2 (200ng) or Tet2 CD proteins (1ug) in

buffer containing 50 mM HEPES (pH 8.0), 100 mM NaCl, 2 mM ascorbate, 1mM

2-oxoglutarate, 100 mM Fe(NH4)2(SO4)2, 1mM DTT, and 1mM ATP at 37℃ for 1h

(He et al., 2011). The DNA was purified with PCR purification kit (Qiagen) and the

product DNA was denatured at 100℃ for 10 min and further digested to nucleosides

with 0.5 U nuclease P1 (Sigma Aldrich) at 37℃for 16h and 0.5 U CIP (NEB) at 37℃

for 1.5 h. The 120bp methylation DNA substrate contains 31 methyl-cytosine

nucleotides (one methyl-CpG dinucleotide). The recombinant TET2 and MBD3L2

proteins were obtained from Dr. Yanhui Xu’s lab at Fudan University.

M.SssI treatment

A 163bp PCR-amplicon was methylated in vitro with bacterial CpG methyltransferase

M.SssI (NEB, #M0226S). 1ug DNA was treated with 1ul (4U/ul) M.SssI in buffer

with SAM and incubated four hours at 37°C. Then the DNA was purified with the

PCR Purification Kit (Qiagen).

LC-MS/MS analysis

The digested samples were subjected to LC-MS/MS using a Shimadzu LC (LC-20AB

pump) system coupled with TSQ-vantage triple quadrupole mass spectrometer

(Thermo). A C18 column (250mm×2.1mm I.D., 5μm particle size, ULTIMATE) was

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

used. The mass spectrometer was optimized and set up in selected reaction monitoring

(SRM) scan mode for monitoring the [M+H+] of dC (m/z 228.2→112.2), 5mC (m/z

242.1→126.1), 5hmC (258.1→ 142.1), 5fC (256.1→140.1), 5caC (272.1→156.1) and

dG (268.1→152.3). Standard curves for 5hmC, 5fC, 5caC, and dG.were constructed

using a weighted (1/X) linear regression of the integrated areas of the analyte (Y)

against the corresponding 6 concentrations of the analyte (X, ng/ml). The Xcalibur 2.2

sp1 was used for integrated areas analysis. The concentration of samples was

calculated by interpolation from the standard curves. The standard nucleosides of dC,

5mC and dG were bought from Thermo, 5hmC was bought from Zymo Research. The

standard nucleosides of 5fC and 5caC were obtained from Dr. Yanhui Xu’s lab at

Fudan University.

HPLC analysis

The reaction products were then analyzed on a ShimazuLC (LC-20AD

pump) machine with an AQ-C18 column of 5-μm particle size, 25 cm x 4.6 mm. The

mobile phase was 10 mM KH2PO4, pH 3.7, running at 1 ml per min, and the detector

was set at 280 nm.

EMSA

The biotin-16bp DNA（upper strand: 5’-TA(Cm)CGTACAGTACACTA）(10pmol)

was incubated with Tet2 (100ng) or Tet2 CD (500ng) purified in a total 20 µl reaction

mixture. An increasing amount of MBD3L2 (100ng, 300ng, 600ng) were used in

different groups. The experiment was done according to the instructions of LightShift

Chemiluminescent EMSA Kit (Thermo). And the membrane was placed in a film

cassette and exposed to X-ray film.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

ChIP

Cells were first fixed with 2mM disuccinimidyl glutarate (DSG) (Thermo) for 45 min

at room temperature, washed twice with PBS and then double-fixed with 1%

formaldehyde for another 10 min at RT room temperature (Liu et al., 2013). The

reaction was stopped with 0.125M glycine solution for 5 min. Nuclei were extracted

with cell lysis buffer (10 mM HEPES, pH7.9, 0.5% NP40,1.5 mM MgCl2, 10 mM

KCl, and 1×protease inhibitor cocktail) and dissolved in nuclear lysis buffer (50mM

Tris, pH 8.1, 10mM EDTA, 0.3% SDS, and 1×protease inhibitor cocktail). After

sonication, the soluble chromatin was incubated with anti-HA antibody (Cell

Signaling; C29F4) for overnight followed by incubation with Protein A magnetic

beads (Millipore). The immunoprecipitated DNA was purified with the PCR

Purification Kit (Qiagen).

ChIP-seq analysis

Public MBD3 ChIP-seq data used for analyses was MCF-7_MBD3 (GSM1089817)

and MCF-7_MBD3_input (GSM1089816). Sequenced reads of Tet2 ChIP-seq,

MBD3L2 ChIP-seq and Input libraries were combined for replicate samples and

filtered based on a mean base quality score <20. Filtered reads were then aligned to

the human reference genome (UCSC assembly hg19, GRCh37) using the Bowtie

short-read alignment program (v0.12.7) to retain reads mapped to unique genomic

locations with at most 2 mismatches. Unique and monoclonal reads were used in

subsequent peak calling analyses. Since the average DNA fragment length used in

ChIP-seq was 400bp, each sequence was extended to 400 bp. Summary of ChIP-seq

data is shown in Supplementary Table S2. The peaks were identified using the MACS

software (version 1.4) with the P-value <1e-5 and FDR <0.01, IgG sample was used

as a control for ChIP samples. (Zhang et al., 2008).

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Heatmaps

Heatmaps were generated by plotting for all the annotated RefSeq genes enrichment

as a colour gradient at their relative location using the annotated gene TSS as the

reference. Genes were first ordered using the average tag density significance in the

1kb region around the TSS for Tet2 ChIP-seq in HEK293T cells.

The overlapped binding genes of MBD3L2 and Tet2

The overlapped binding regions of Tet2 and MBD3L2 are defined as the Tet2 peaks

whose overlapping ratios of MBD3L2 peaks are above 50%. The overlapped binding

genes are defined as the genes whose promoter (1000bp upstream and downstream the

TSS) located at the overlapped binding regions of Tet2 and MBD3L2. The

overlapping ratios were identified using bedtools IntersectBed (Quinlan and Hall,

2010).

CpG density

The CpG count is the number of CG dinucleotides in the genome. The CpG

percentage, i.e. CpG density, of a region is the ratio of CpG nucleotide bases (twice

the CpG count) to the length of the window centered on this region. The length of the

window is set as 200bp. Genomic regions are composed of all the CG dinucleotides in

the human reference genome (hg19). Demethylated regions include the CpG sites

whose methylation level decreased in RRBS_Tet2 and further decreased in

RRBS_Tet2+MBD3L2. Common RRBS regions consist of the common sites of

RRBS_Control, RRBS_Tet2 and RRBS_Tet2+MBD3L2.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

The analysis of methylation change

Firstly, we analyzed the ChIP-seq data of Tet2 and MBD3L2. The overlapped binding

regions of Tet2 and MBD3L2 are defined as the Tet2 peaks whose overlapping rates

of MBD3L2 peaks are above 50%. And the non-overlapped binding regions of Tet2

and MBD3L2 are the Tet2 peaks whose overlapping rates of MBD3L2 peaks are

below 10%. Secondly, for each region defined above, we calculated the average

methylation level of the two samples, Tet2 and (Tet2+MBD3L2), respectively. Then

we defined the methylation change between the two samples in a region as the ratio of

average methylation level of Tet2 to that of Tet2+MBD3L2. We further compared the

methylation change of the overlapped regions and non-overlapped regions by box plot

and Wilcoxon rank sum testing.

Gene ontology (GO) analysis and KEGG pathway annotation

GO analysis and KEGG pathway enrichment analysis for regulated genes were

performed by the database for annotation, visualization and integrated discovery

(DAVID) online tools (http://david.abcc.ncifcrf.gov/) (Quinlan and Hall, 2010).

Statistical Analysis

Statistical analysis were performed with a two-tailed unpaired Student’s t test. All data

shown represent the results obtained from triplicated independent experiments with

SEM (mean ± SD). The values of p < 0.05 were considered statistically significant.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Comparative hMeDIP-seq

Library construction (adding adaptor containing barcode sequence)

Genomic DNA was sonicated to <500 bp by Bioruptor sonicator (Diagenode) and

quantified using the NANODROP2000c (Thermo) according to the manufacturer’s

manual. The quantified DNA (1ug of each sample) were sonicated to less than 500bp

by Bioruptor sonicator (Diagenode). The sonicated DNA fragments were end-repaired

using the End-It DNA End Repair Kit (EPICENTRE Biotechnologies) according to

the manufacturer’s instructions, followed by treatment with Klenow fragment

3’-5’exo (NEB) and dATP to generate a protruding 3’A for ligation with the adaptor

containing a specific barcode sequence. The barcode sequence (four-base

index:TTAGGC, GATCAG, AGTTCC, GTCCGC) within the adaptor. Then the four

samples were mixed together.

hMeDIP

Equal amounts (1ug each sample) of barcode-tagged gDNA from four samples were

pooled together in one tube (Tan et al., 2013). The mixed DNA was denatured and

diluted by IP buffer (10% was taken off as input at this step). The denatured DNA was

incubated with 4ml anti-5hmC antibody (active motif) at 4℃ overnight.

Antibody–DNA complexes were captured by protein A/G beads, and the enriched

5hmC-containing DNA fragments (hMeDIP product) were purified by Qiagen Mini

Gel Purification Kit.

Library amplification

The hMeDIP product, as well as input DNA, was amplified for 16 cycles, and the

products range from 400 to 500bp were purified by Qiagen Mini Gel Purification Kit.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Next-generation sequencing

The amplified libraries were submitted for cluster growth and sequencing by the

Illumina Genome Analyzer II (GAII)

hMeDIP-seq data processing

We separated the raw sequence reads of hMeDIP and Input into different files

according to the specific barcode sequence of each sample. Sequence reads were

mapped onto the reference human genome using the Bowtie (v0.12.7) algorithm. The

peaks of hMeDIP-seq are identified using the MACS software (version 1.4) with the

default parameters and a P-value cut-off <10-5. hMeDIP and sequencing for the equal

amount of four samples mixing in one reaction system can reduce the experimental

error and variation among these four samples. The sequencing data can be separated

by the different barcodes sequence, and the 5hmC level difference of each sample

leads to the different amount of sequencing reads, which makes it possible to compare

the 5hmC level between different samples with sequencing data. After normalized to

the correspondent input, the percentage of sequencing data for each sample can

represent the percentage of 5hmC level for each sample in general.

Peak calling

The peaks of hMeDIP-seq are identified using the MACS software (version 1.4) with

the default parameters.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Reduced Representational Bisulphite sequencing (RRBS)

500 ng of genomic DNA was digested with the methylation insensitive restriction

digest enzyme Msp1 (NEB), which cleaves DNA at CCGG sites creating fragments

high in CpG content (Gu et al., 2011). The DNA was purified using the QIAquick

PCR Purification Kit (Qiagen). Samples were end-repaired and A-tailed using a

Klenow fragment (New England Biolabs, Inc). TruSeq™ adaptors (Illumina, Inc., San

Diego, CA) were ligated to the modified DNA ends using T4 DNA ligase (New

England Biolabs, Inc.). We excised the gel ranged from 150-400bp and purified the

gel with the QIAquick Gel Extraction Kit (Qiagen). The sample was bisulfite

modified and PCR amplified to enrich for fragments containing high CpG content.

Samples were submitted for sequencing on the Illumina HiSeq™ instrument.

RRBS data analysis

In brief, RRBS sequencing raw reads were trimmed with trim_galore (v0.4.0) with

default parameters to remove adaptor sequence and low-quality sequence, and then

the rest reads were aligned to human reference genome (hg19, downloaded from

http://genome.ucsc.edu) using BSMAP (v2.6) with default parameters (Xi and Li,

2009), finally a script, methratio.py was used for extracting methylation ratios from

BSMAP mapping results NE.Ref (Gu et al., 2011)

Microarray experiment

Total RNA was extracted and RNA quality was evaluated using NanoDrop ND-1000

and Denaturing Agarose Gel Electrophoresis. The Arraystar Human LncRNA

Microarray (v2.0) was used to detect RNA expressions, the Agilent Feature Extraction

software (v11.0.1.1) to analyze array images, and the GeneSpring GX software

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

package (v 11.5.1, Agilent Technologies) to analyze intensity and normalization.

Probes with 3 flags (Present or Marginal) in 3 samples were selected for downstream

analysis. Differential expression mRNAs were identified by cutoff of Fold-Change >=

2.

GluMS-qPCR Analysis

The 5hmC level in Tet2-binding regions were measured by a restriction-enzyme-based

assay (EpiMark kit, New England Biolabs). Genomic DNA was treated with or

without T4 Phage b-glucosyltransferase, and then digested by MspI or no enzyme.

The MspI-resistant fractions were quantified by qPCR and normalized to the mock

digestion. PCR primers and siRNA sequence are listed in Table S3.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Acknowledgements

We thank Dr. Guoliang Xu for providing the plasmids of Tet1, Tet2 and Tet3.

We thank Dr. Yanhui Xu for providing the standard nucleosides of 5fC, 5caC

and recombinant MBD3L2 and TET2 proteins. We thank Dr. Dan Ye for

providing endogenous TET2 antibody. We thank Qinhui Rao, Zhentian Wang

and Jianbo Diao for technical supports with in vitro enzymatic activity. We

thank Yue Yu from Carleton College, Liang Liu from Columbia University,

Taixing Cui from North Carolina University for critical reading of the manuscript.

We also thank Dan Ye and Fei Lan from Fudan University for valuable

discussion and suggestions.

Author contributions

LNP and YL performed the biochemical experiments and sequencing library

construction. YPX, LNP, JL, WL, RTL and FZW analyzed sequencing data. LZ

performed the LC-MS/MS/MS experiment. YL, SHD and HBL performed

sample preparation on sequencing. WQY and FZW supervised the

experiments. WQY, FZW and LNP wrote the manuscript.

Competing interests

The authors declare that they have no competing interests.

Funding

This work was supported by grants from National Natural Science Foundation

of China (31271355), National Basic Research Program of China 973 Program

(2012CB517606), National Natural Science Foundation of China (81272392),

The Major Project of Basic Research of Technology Committee in Shanghai of

China (12DJ1400200), National Basic Research Program of China 973

Program (2009CB825603).

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

References:

Baubec, T., Ivanek, R., Lienert, F. and Schubeler, D. (2013). Methylation-dependent and

-independent genomic targeting principles of the MBD protein family. Cell 153, 480-92.

Bock, C., Tomazou, E. M., Brinkman, A. B., Muller, F., Simmer, F., Gu, H., Jager, N.,

Gnirke, A., Stunnenberg, H. G. and Meissner, A. (2010). Quantitative comparison of genome-wide

DNA methylation mapping technologies. Nature Biotechnology 28, 1106-14.

Brown, S. E. and Szyf, M. (2007). Epigenetic programming of the rRNA promoter by MBD3.

Molecular and Cellular Biology 27, 4938-52.

Brown, S. E., Suderman, M. J., Hallett, M. and Szyf, M. (2008). DNA demethylation induced

by the methyl-CpG-binding domain protein MBD3. Gene 420, 99-106.

Clouaire, T. and Stancheva, I. (2008). Methyl-CpG binding proteins: specialized transcriptional

repressors or structural components of chromatin? Cellular and Molecular Life Sciences 65, 1509-22.

Deplus, R., Delatte, B., Schwinn, M. K., Defrance, M., Mendez, J., Murphy, N., Dawson, M.

A., Volkmar, M., Putmans, P. and Calonne, E. et al. (2013). TET2 and TET3 regulate

GlcNAcylation and H3K4 methylation through OGT and SET1/COMPASS. EMBO Journal 32,

645-55.

Dos, S. R., Tosti, L., Radzisheuskaya, A., Caballero, I. M., Kaji, K., Hendrich, B. and Silva,

J. C. (2014). MBD3/NuRD facilitates induction of pluripotency in a context-dependent manner. Cell

Stem Cell 15, 102-10.

Faryna, M., Konermann, C., Aulmann, S., Bermejo, J. L., Brugger, M., Diederichs, S., Rom,

J., Weichenhan, D., Claus, R. and Rehli, M. et al. (2012). Genome-wide methylation screen in

low-grade breast cancer identifies novel epigenetically altered genes as potential biomarkers for tumor

diagnosis. FASEB Journal 26, 4937-50.

Ficz, G., Branco, M. R., Seisenberger, S., Santos, F., Krueger, F., Hore, T. A., Marques, C. J.,

Andrews, S. and Reik, W. (2011). Dynamic regulation of 5-hydroxymethylcytosine in mouse ES cells

and during differentiation. Nature 473, 398-402.

Fu, Y. and He, C. (2012). Nucleic acid modifications with epigenetic significance. Current

Opinion in Chemical Biology 16, 516-24.

Gu, H., Smith, Z. D., Bock, C., Boyle, P., Gnirke, A. and Meissner, A. (2011). Preparation of

reduced representation bisulfite sequencing libraries for genome-scale DNA methylation profiling.

Nature Protocols 6, 468-81.

Guo, X., Wang, L., Li, J., Ding, Z., Xiao, J., Yin, X., He, S., Shi, P., Dong, L. and Li, G. et al.

(2015). Structural insight into autoinhibition and histone H3-induced activation of DNMT3A. Nature

517, 640-4.

He, Y. F., Li, B. Z., Li, Z., Liu, P., Wang, Y., Tang, Q., Ding, J., Jia, Y., Chen, Z. and Li, L.

et al. (2011). Tet-mediated formation of 5-carboxylcytosine and its excision by TDG in mammalian

DNA. Science 333, 1303-7.

Hendrich, B. and Bird, A. (1998). Identification and characterization of a family of mammalian

methyl-CpG binding proteins. Molecular and Cellular Biology 18, 6538-47.

Hu, L., Li, Z., Cheng, J., Rao, Q., Gong, W., Liu, M., Shi, Y. G., Zhu, J., Wang, P. and Xu, Y.

(2013). Crystal structure of TET2-DNA complex: insight into TET-mediated 5mC oxidation. Cell 155,

1545-55.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Huang, D. W. Sherman, B. T. and Lempicki, R. A. (2009). Systematic and integrative analysis

of large gene lists using DAVID bioinformatics resources. Nature Protocols 4, 44-57.

Iurlaro, M., Ficz, G., Oxley, D., Raiber, E. A., Bachman, M., Booth, M. J., Andrews, S.,

Balasubramanian, S. and Reik, W. (2013). A screen for hydroxymethylcytosine and formylcytosine

binding proteins suggests functions in transcription and chromatin regulation. Genome Biology 14,

R119.

Jiang, C. L., Jin, S. G., Lee, D. H., Lan, Z. J., Xu, X., O'Connor, T. R., Szabo, P. E., Mann, J.

R., Cooney, A. J. and Pfeifer, G. P. (2002). MBD3L1 and MBD3L2, two new proteins homologous to

the methyl-CpG-binding proteins MBD2 and MBD3: characterization of MBD3L1 as a testis-specific

transcriptional repressor. Genomics 80, 621-9.

Jin, S. G., Jiang, C. L., Rauch, T., Li, H. and Pfeifer, G. P. (2005). MBD3L2 interacts with

MBD3 and components of the NuRD complex and can oppose MBD2-MeCP1-mediated methylation

silencing. Journal of Biological Chemistry 280, 12700-9.

Jin, S. G., Tsark, W., Szabo, P. E. and Pfeifer, G. P. (2008). Haploid male germ cell- and

oocyte-specific Mbd3l1 and Mbd3l2 genes are dispensable for early development, fertility, and zygotic

DNA demethylation in the mouse. Dev Dyn 237, 3435-43.

Jin, S., Lee, Y. K., Lim, Y. C., Zheng, Z., Lin, X. M., Ng, D. P., Holbrook, J. D., Law, H. Y.,

Kwek, K. Y. and Yeo, G. S. et al. (2013). Global DNA hypermethylation in down syndrome placenta.

PLoS Genetics 9, e1003515.

Kaji, K. Nichols, J. and Hendrich, B. (2007). Mbd3, a component of the NuRD co-repressor

complex, is required for development of pluripotent cells. Development 134, 1123-32.

Ko, M., An, J., Bandukwala, H. S., Chavez, L., Aijo, T., Pastor, W. A., Segal, M. F., Li, H.,

Koh, K. P. and Lahdesmaki, H. et al. (2013). Modulation of TET2 expression and 5-methylcytosine

oxidation by the CXXC domain protein IDAX. Nature 497, 122-6.

Lian, C. G., Xu, Y., Ceol, C., Wu, F., Larson, A., Dresser, K., Xu, W., Tan, L., Hu, Y. and

Zhan, Q. et al. (2012). Loss of 5-hydroxymethylcytosine is an epigenetic hallmark of melanoma. Cell

150, 1135-46.

Liu, W., Ma, Q., Wong, K., Li, W., Ohgi, K., Zhang, J., Aggarwal, A. K. and Rosenfeld, M.

G. (2013). Brd4 and JMJD6-associated anti-pause enhancers in regulation of transcriptional pause

release. Cell 155, 1581-95.

Nakagawa, T., Lv, L., Nakagawa, M., Yu, Y., Yu, C., D'Alessio, A. C., Nakayama, K., Fan,

H. Y., Chen, X. and Xiong, Y. (2015). CRL4(VprBP) E3 ligase promotes monoubiquitylation and

chromatin binding of TET dioxygenases. Molecular Cell 57, 247-60.

Nakajima, H. and Kunimoto, H. (2014). TET2 as an epigenetic master regulator for normal and

malignant hematopoiesis. Cancer Science 105, 1093-9.

Okashita, N., Kumaki, Y., Ebi, K., Nishi, M., Okamoto, Y., Nakayama, M., Hashimoto, S.,

Nakamura, T., Sugasawa, K. and Kojima, N. et al. (2014). PRDM14 promotes active DNA

demethylation through the ten-eleven translocation (TET)-mediated base excision repair pathway in

embryonic stem cells. Development 141, 269-80.

Perera, A., Eisen, D., Wagner, M., Laube, S. K., Kunzel, A. F., Koch, S., Steinbacher, J.,

Schulze, E., Splith, V. and Mittermeier, N. et al. (2015). TET3 Is Recruited by REST for

Context-Specific Hydroxymethylation and Induction of Gene Expression. Cell Reports 11, 283-94.

Quinlan, A. R. and Hall, I. M. (2010). BEDTools: a flexible suite of utilities for comparing

genomic features. Bioinformatics 26, 841-2.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Scopim-Ribeiro, R., Machado-Neto, J. A., Campos, P. M., Silva, C. A., Favaro, P.,

Lorand-Metze, I., Costa, F. F., Saad, S. T. and Traina, F. (2015). Ten-Eleven-Translocation 2

(TET2) is downregulated in myelodysplastic syndromes. European Journal of Haematology 94, 413-8.

Shimbo, T., Du Y, Grimm, S. A., Dhasarathy, A., Mav, D., Shah, R. R., Shi, H. and Wade, P.

A. (2013). MBD3 localizes at promoters, gene bodies and enhancers of active genes. PLoS Genetics 9,

e1004028.

So, A. Y., Sookram, R., Chaudhuri, A. A., Minisandram, A., Cheng, D., Xie, C., Lim, E. L.,

Flores, Y. G., Jiang, S. and Kim, J. T. et al. (2014). Dual mechanisms by which miR-125b represses

IRF4 to induce myeloid and B-cell leukemias. Blood 124, 1502-12.

Solary, E., Bernard, O. A., Tefferi, A., Fuks, F. and Vainchenker, W. (2014). The Ten-Eleven

Translocation-2 (TET2) gene in hematopoiesis and hematopoietic diseases. Leukemia 28, 485-96.

Song, C. X., Szulwach, K. E., Fu, Y., Dai, Q., Yi, C., Li, X., Li, Y., Chen, C. H., Zhang, W.

and Jian, X. et al. (2011). Selective chemical labeling reveals the genome-wide distribution of

5-hydroxymethylcytosine. Nature Biotechnology 29, 68-72.

Spandidos, D. A. and Wilkie, N. M. (1984). Malignant transformation of early passage rodent

cells by a single mutated human oncogene. Nature 310, 469-75.

Tan, L., Xiong, L., Xu, W., Wu, F., Huang, N., Xu, Y., Kong, L., Zheng, L., Schwartz, L. and

Shi, Y. et al. (2013). Genome-wide comparison of DNA hydroxymethylation in mouse embryonic

stem cells and neural progenitor cells by a new comparative hMeDIP-seq method. Nucleic Acids

Research 41, e84.

Wang, L., Liu, Y., Han, R., Beier, U. H., Thomas, R. M., Wells, A. D. and Hancock, W. W.

(2013). Mbd2 promotes foxp3 demethylation and T-regulatory-cell function. Molecular and Cellular

Biology 33, 4106-15.

Wang, Y., Xiao, M., Chen, X., Chen, L., Xu, Y., Lv, L., Wang, P., Yang, H., Ma, S. and Lin,

H. et al. (2015). WT1 recruits TET2 to regulate its target gene expression and suppress leukemia cell

proliferation. Molecular Cell 57, 662-73.

Wu, H. and Zhang, Y. (2011). Mechanisms and functions of Tet protein-mediated

5-methylcytosine oxidation. Genes Dev 25, 2436-52.

Wu, H. and Zhang, Y. (2014). Reversing DNA methylation: mechanisms, genomics, and

biological functions. Cell 156, 45-68.

Wu, H., Wu, X., Shen, L. and Zhang, Y. (2014). Single-base resolution analysis of active DNA

demethylation using methylase-assisted bisulfite sequencing. Nature Biotechnology 32, 1231-40.

Xi, Y. and Li, W. (2009). BSMAP: whole genome bisulfite sequence MAPping program. BMC

Bioinformatics 10, 232.

Yildirim, O., Li, R., Hung, J. H., Chen, P. B., Dong, X., Ee, L. S., Weng, Z., Rando, O. J. and

Fazzio, T. G. (2011). Mbd3/NURD complex regulates expression of 5-hydroxymethylcytosine marked

genes in embryonic stem cells. Cell 147, 1498-510.

Yu, M., Hon, G. C., Szulwach, K. E., Song, C. X., Zhang, L., Kim, A., Li, X., Dai, Q., Shen, Y.

and Park, B. et al. (2012). Base-resolution analysis of 5-hydroxymethylcytosine in the mammalian

genome. Cell 149, 1368-80.

Zhang, Y., Liu, T., Meyer, C. A., Eeckhoute, J., Johnson, D. S., Bernstein, B. E., Nusbaum,

C., Myers, R. M., Brown, M. and Li, W. et al. (2008). Model-based analysis of ChIP-Seq (MACS).

Genome Biology 9, R137.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Figures

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Fig. 1. MBD3 and MBD3L2 co-localize with Tet2. (A) MBD3L2/MBD3

co-localize with Tet2 in the nucleus. The results of IF staining were observed

with the confocal microscopy. Scale bars shown are 10 um. (B) Genome-wide

distribution of Tet2 and MBD3L2 binding sites. Promoter is defined as −1kb to

+1kb relative to transcription start site (TSS). (C) The relative enrichment of the

binding peaks (blue) and corresponding randomly shuffled control regions

(gray), normalized to the total size of the element type (per mega base-pairs

[Mbp]). Random consists of 10 random sampling of specific genomic elements

in the human genome. The ratio between observed and random for each

genomic element is shown on the top (black). (D) Venn diagrams indicating

significant overlap of Tet2 and MBD3L2 bound regions (p-value< 10-16,

hypergeometric test). (E) Heatmap showing ChIP-seq in HEK293T. The tag

densities of Tet2 and MBD3L2 binding were profiled through -5kb to +5kb

relative to the TSS of each gene with bin of 100bp. Genes are ordered

according their Tet2 enrichment (average peak significance in 1kb region

surrounding the TSS). (F) Examples of MBD3L2 and Tet2 overlapped binding

genes on UCSC tracks. See also supplementary Fig S1E-G.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Fig. 2. MBD3 and MBD3L2 promote the enzymatic activity of Tet2/CD in

vitro. (A) MBD3 increases four times of 5hmC production catalyzed by Tet2

while the same amount of MBD3L2 increases about thirteen times in vitro. The

content of 5hmC level was measured by LC-MS/MS (*p < 0.01; ***p < 0.001).

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

The equal level of Tet2 protein in each group was analyzed by western blot.

The purified proteins were washed by high salt solution to get rid of the

genomic DNA contamination. The analysis of parallel reactions without adding

methylated oligo were used as control and 5hmC was not detected. See also

supplementary Fig S2. (B) MBD3L2 interacts with Tet2 as shown by co-IP

assay. HEK293T cells were transiently transfected with plasmids expressing

Flag-Tet2 or SBP-MBD3L2 as indicated. Flag and IgG beads were used for the

immunoprecipitation. Empty vector transfection and beta-actin were used as

negative controls. (C) HEK293T cells were transiently transfected with

plasmids expressing Flag-Tet2 or SBP-MBD3L2 as indicated. Dynabeads

Streptavidin was used for the immunoprecipitation. Empty vector transfection

and beta-actin were used as negative controls. (D) Tet2 can be divided into CD

minus and CD, which is the dioxygenase domain of Tet2. MBD3L2 interacts

with the CD, but not CD minus. Flag beads were used for the

immunoprecipitation. Beta-actin was included as a negative control. (E)

MBD3L2 enhances 5hmC, 5fC and 5caC production catalyzed by CD (*p <

0.05; **p < 0.01). See also supplementary Fig S2B.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Fig. 3. MBD3L2 increases the binding affinity of Tet2 and methylated DNA.

(A) EMSA assay shows that the binding affinity of Tet2/CD and methylated

DNA increased according to the increase of MBD3L2 proteins. HEK293T cells

were transfected with HA-tagged Tet2 or CD or MBD3L2 as indicated. Proteins

were purified and incubated with double stranded DNA oligonucleotides

containing a single methylated CpG. Protein-DNA binding was detected by

EMSA assay. Protein levels were analyzed by western blot. (B) MBD3L2

cannot associate with unmethylated or methylated DNA in vitro. HEK293T cells

were transfected with HA-tagged MBD3L2. Proteins were purified and

incubated with double stranded DNA oligonucleotides containing a single

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

methylated or unmethylated CpG. Protein-DNA binding was

detected by EMSA assay. (C) Increasing amount of MBD3L2 cannot enhance

the binding affinity of Tet2 CD and unmethylated DNA. HEK293T cells were

transfected with HA-tagged CD or MBD3L2 as indicated. Proteins were

purified and incubated with double stranded DNA oligonucleotides containing a

single unmethylated CpG. Protein-DNA binding was detected by EMSA assay.

Protein levels were analyzed by western blot.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Fig. 4. MBD3L2 enhances Tet2 mediated demethylation in HEK293T cell.

(A) The co-overexpression of MBD3L2 and Tet2 lead to the global 5mC level

decreases by HPLC analysis. Error bars indicate SD of three independent

experiments (*p < 0.05; ***p < 0.001). See also supplementary Fig S3D. (B)

The global 5hmC level increases with the co-transfection of MBD3L2 and Tet2

by LC-MS/MS/MS. Error bars indicate SD of three independent experiments

(*p < 0.05). See also supplementary Fig S3E. (C) The demethylated regions

induced by Tet2 and MBD3L2 is preferably occurred in the relatively low CpG

density region (p-value<2.2e-16, Wilcoxon rank sum test). (D) The level of

5mC decreases more significantly at Tet2 and MBD3L2 overlapped binding

regions by RRBS analysis (p-value= 8.210051e−51, Wilcoxon rank sum

testing). (E) Percentages of Normalized 5hmC reads in four samples show

5hmC level significant increase in (Tet2+MBD3L2) group. (F) Normalized

5-hmC tag density distribution across the gene body. Normalized Tag density is

plotted from 20% of upstream of TSSs to 20% downstream of TESs. (G)

Normalized 5-hmC tag density distribution across the promoter region. (H)

MBD3/MBD3L2 enhance global 5hmC level of Jurkat cell. Endogenous protein

level of TET2 in HEK293T and Jurkat cells were determined by western blot.

Beta-actin was used as a loading control.

Jour

nal o

f Cel

l Sci

ence

• A

dvan

ce a

rtic

le

Fig. 5. MBD3L2 increases Tet2 target genes expression by demethylation

process. (A) Bar plot of the GO analysis of genes up-regulated by Tet2 and

MBD3L2. The y axis shows the GO terms, and the x axis shows the

enrichment significance p values. The top ten most enriched GO terms are

shown (p < 0.05). (B) MBD3L2 increases the expression of Tet2 target genes.

The expression level of target genes were analyzed by RT-qPCR. Error bars

indicate SD of three independent experiments (**p < 0.01; ***p < 0.001). (C)

The 5hmC level of Tet2 targeted genes are enhanced by MBD3L2 (UCSC

tracks), as shown by the analysis of comparative hMeDIP-seq data. See also

supplementary Fig S4G.