i
The identification and
characterization of novel JMJD5 &
Hdac1/Hdac2 interacting proteins
2012
The Netherlands Cancer Institute – Antonie van Leeuwenhoek (NKI - AvL) Department of Gene Regulation
ii
The identification and characterization of novel JMJD5 &
Hdac1/Hdac2 interacting proteins
Final Report
August 2012, Amsterdam
I.A.M. Lodewijk Counselling professor M.C.M. Verschuren
Avans University of Applied Sciences, Breda School of Health and Environment Technology
Supervisor M.R. Heideman JH. Dannenberg
The Netherlands Cancer Institute – Antonie van Leeuwenhoek (NKI - AvL) Department of Gene Regulation
1
Abstract
Cancer is a genetic disease initiated by alterations in, for example, genes that regulate cell proliferation. These alterations may arise from genetic aberrations, which alter the function of gene encoding proteins, and aberrant epigenetic modifications such as acetylation, methylation and phosphorylation, which alter gene-expression patterns. The role of epigenetic modifications in tumorigenesis is poorly understood. Understanding of the epigenetic mechanisms, controlled by cancer-relevant histone modifying enzymes which counterbalance epigenetic modifications, will provide insight into the development of human diseases, including cancer. In order to obtain insight into the role of histone modifying enzymes in cancer, this study is focused on histone deacetylases HDAC1 and HDAC2, and histone demethylase JMJD5.
In this study, we have identified and partly characterized novel Hdac1 interacting proteins Wdr5, Npm1 and Dlat1 using a combination of GFP-tagged Hdac1/2 interactor overexpression, GFP-Trap® pull-down and western blot analysis. We showed that Dlat is a substrate for HDACs, which provides insight into a role for Hdac1 in regulation of the pyruvate decarboxylation step that links glycolysis to the citric acid cycle.
Furthermore, we have identified novel JMJD5 interacting proteins using a mass spectrometry-based proteomic approach. Many of the identified interactors have a role in metabolism, such as serine biosynthesis. In addition, we identified the JMJD5 interacting protein HCLS1-associated protein X-1, which is involved in the regulation of carcinoma cell migration and invasion. Future studies should provide insight into a possible role for JMJD5 in metabolism and/or carcinoma cell migration and invasion.
2
Index
Abstract ... 1
Introduction ... 4
Theoretical Background ... 5
Histones & Post-translational modifications ... 5
Histone acetylation... 6 Lysine acetyltransferase ... 6 Histone deacetylases ... 7 HDAC1&HDAC2 ... 7 Histone methylation ... 9 Lysine methyltransferases ... 9 Lysine demethylases ... 10 JMJD5 ... 12
Material and Methods ... 13
Cloning using Gateway technology ... 13
Generation of pDONR223-JMJD5 and pDONR223-interactor ... 13
Generation of pDEST47-JMJD5 and pDEST47-interactor ... 13
Generation of retroviral plasmids pQCXIB-JMJD5-GFP and pQCXIB-interactor-GFP ... 14
Generation of Hdac1-/-;Hdac2-/- double knock-out cells ... 14
Cell culture and retroviral infection ... 14
Nuclear extraction ... 15
Purification of Green Fluorescent Protein (GFP-) tagged proteins ... 15
Western Blot analysis ... 15
Coomassie staining of the SDS gel ... 16
Mass spectrometry ... 16
Results Validation and characterization of novel Hdac1 and Hdac2 interactors ... 17
Identification of novel Hdac1 and Hdac2 interactors... 17
Generation of Hdac1/Hdac2 interactor-GFP fusion proteins ... 17
Expression analysis and cellular localization of interactor-GFP fusion proteins ... 18
Expression of interactor-GFP fusion proteins in MEFs ... 19
Validation of the interaction between Hdac1 and GFP fusion proteins... 19
Validation of GFP fusion proteins as substrates of Hdac1 and Hdac2 ... 20
3
Generation and expression of the JMJD5-GFP fusion protein ... 21
Identification of novel JMJD5 interactors ... 21
Discussion and conclusions ... 24
Histone deacetylases HDAC1 and HDAC2 ... 24
Expression and cellular localization of interactor-GFP fusion proteins ... 24
Validation of the interaction between Hdac1 and GFP fusion proteins ... 24
Validation of GFP fusion proteins as substrates of Hdac1 and Hdac2 ... 24
Histone demethylase JMJD5 ... 25
Expression of JMJD5-GFP fusion protein in primary and tumor cells ... 25
Identification of novel JMJD5 interactors ... 25
4
Introduction
Cancer is a genetic disease initiated by alterations in, for instance, genes that regulate cell proliferation, survival, and other homeostatic functions. These alterations may arise from genetic aberrations, which alter the function of gene encoding proteins. In addition, tumors often display aberrant epigenetic modifications of DNA and histones, resulting in altered gene-expression patterns. DNA and histone proteins, collectively named chromatin, are subject to a variety of epigenetic modifications such as acetylation, methylation and phosphorylation. These modifications are counterbalanced by numerous histone modifying enzymes, like demethylases and deacetylases. Over time, accumulation of genetic and epigenetic alterations will result in uncontrolled proliferation of cells and ultimately dissemination of cancerous cells to other organs.
The role of epigenetic modifications in tumorigenesis is poorly understood. Since these modifications regulate gene transcription, it is thought that aberrant epigenetic modifications will result in altered expression patterns of tumor suppressor genes and oncogenes. For example, DNA methylation induces ‘epigenetic silencing’ of tumor suppressor genes, causing normal cells to be transformed into cancer cells. In tumors, histone modifying enzymes are known to be recruited by oncogenic fusion proteins, like PML-RAR-α, AML1-ETO and MLL-AF9, inducing epigenetic alterations at the wrong place at the wrong time (Harris et al. 2012; O’Connell et al. 2011; Petrie & Zelent 2007). Moreover, through the efforts of full-genome sequencing of various tumors, genetic mutations in histone modifying enzymes, like MLL2, UTX, EZH2 and CBP/p300, have been identified in hematological cancers (Grasso et al. 2012; Hyndman et al. 2012; Zhang et al. 2012). Despite numerous studies, little is known about the mechanisms controlled by histone modifying enzymes during normal development and tumorigenesis. Understanding of the epigenetic mechanisms, controlled by cancer-relevant histone modifying enzymes, will provide insight into the development of human diseases, including cancer (Egger et al., 2004; Novak 2004).
In order to obtain insight into the role of histone modifying enzymes in cancer, this study is focused on histone deacetylases and histone demethylases, two classes of enzymes which are linked to tumorigenesis. HDAC1 and HDAC2 are well-known histone deacetylases and proved to control a variety of cellular processes by deacetylation of lysine residues in histone and non-histone substrates (Brunmeir et al. 2009; Hagelkruys et al. 2011). Identifying interaction partners of HDAC1/2 was shown to provide some insight into the function of these proteins. Although many HDAC1/2 interactors have already been identified, Heideman et al. revealed many novel Hdac1/2 protein interactors using a SILAC-based proteomic approach. The purpose of this study is to characterize these novel Hdac1/2 interacting proteins.
Histone demethylase JMJD5 is another histone modifying enzyme that has been implicated in cancer. To discover more about the function and mechanisms controlled by JMJD5 during normal development and tumorigenesis, this research is focused on the identification of JMJD5 interacting proteins using a mass spectrometry-based proteomic approach.
5
Theoretical Background
Histones & Post-translational modifications
The genomic DNA of eukaryotes is wrapped around histones, which are basic proteins in the nucleus of the cell. The resulting, highly structured, complex is called chromatin. The basic units of this chromatin are nucleosomes, which consist of a histone octamer, wrapped by 146 base pairs of DNA (Fig. 1). The histone octamer
contains four core histones and is constructed of two H3-H4 histone dimers, resulting in a stable tetramer, flanked by two separate heterodimers of histones H2A and H2B. Another histone, linker histone H1, is bound to the entry-exit point of the nucleosome to protect internucleosomal linker DNA. Protruding from the nucleosome is the amino-terminal (N-amino-terminal), which appears to be unstructured (Campos & Reinberg, 2009; Cohen et al., 2011). Although these tails are unnecessary for nucleosome formation, they seem to be required for nucleosome-nucleosome interaction (Luger et al. 1997). Angelov et al. (2001) suggested
that they participate in the formation of higher-order chromatin structure by binding to linker DNA rather than intranucleosome core DNA (Cohen et al., 2011; Zhang & Reinberg, 2001). Although linker histone H1 is far less investigated, it is known that also this histone is a target for multiple post-translational modifications (PTMs). PTMs are relatively small chemically and structurally additions, like acetyl, methyl and phosphate groups, to a subset of amino acid residues especially in the histone tails, including lysine (K), arginine (R), serine (S), threonine (T), tyrosine (Y), histidine (H) and glutamic acid (E) (Taverna et al., 2007). PTMs are crucial in the determination of the chromatin structure. Globally, two types of chromatin can be identified: i) euchromatin, which is transcriptionally active chromatin correlated with an open structure and ii) heterochromatin, which is transcriptionally repressive chromatin associated with a compacted structure (Grunstein et al. 1995). Chromatin structure can be changed by adding or removing PTMs by so-called ‘histone writers’ and ‘histone erasers’ (Fig. 1). A well-established example involves histone acetylation. In general, acetylation of the histone tail correlates with an open chromatin structure and active transcription (Roth et al., 2001). Removal of acetyl groups induces a more compact chromatin structure and therefore suppresses transcription.
Figure 1. DNA is wrapped around histone proteins to form a nucleosome and nucleosomes are further compacted to form chromatin. Protruding from the nucleosome is the N-terminal, which can be modified by post-translational modifications (PTMs) like acetylation and methylation. Proteins that attach acetyl or methyl groups are termed ‘writers’ (like histone acetylases and histone methylases). Proteins that recognize and bind to histone modifications are termed ‘readers’ (like proteins that include bromodomains) and the proteins that remove the histone marks are called ‘erasers’ (like histone deacetylases and histone demethylases) (Arrowsmith et al., 2012).
6 The distinct effects of PTMs on chromatin can be classified in intrinsic, extrinsic and effector-mediated effects. Intrinsic effects alter histone-DNA interaction, while extrinsic effects modulate nucleosome interactions. Effector-mediated effects change chromatin structure by modification-dependent recruitment of proteins, such as transcription factors or histone modifying enzymes. The histone code, proposed by Strahl & Allis (2000), suggest that a combination of histone modifications on one or more histone tails are read by other proteins to specify unique downstream functions. Nearly 80 different sites on histone tails and the core domains seem to be a target for more than 10 different types of modifications, which indicates the complexity of PTMs in regulating chromatin-dependent processes. The regulation of chromatin-chromatin-dependent processes becomes even more complex by cross-talk between histone marks, in which PTMs regulate each other (Lee et al., 2010). Additionally, many conserved protein domains posses the ability to bind specific histone PTMs, a process dependent on both modification state and position within the histone sequence. The proteins, containing such a domain, are termed ‘histone readers’ (Fig. 1). Some well-known histone reader domains are bromodomain, chromodomain and PHD motif. The bromodomain is a well-known reader of acetylated lysine marks, while chromodomains promote protein binding to methylated lysines in the tail of histone H3 (H3K9me2/3 and H3K27me2/3). The PHD motif recognizes the unmodified histone H3 tail (H3K4), methylated lysine residues in the histone H3 tail (H3K4me1/2/3, H3K9me3), and acetylated lysine marks in the histone H3 and histone H4 tail (H3K14ac, H3K9ac, H4K5ac, H4K8ac, H4K12ac, H4K16ac) (Musselman & Kutateladze, 2011; Taverna et al., 2007; Zeng & Zhou, 2002). Interestingly, the presence of different reader domains, within a histone modifying enzyme or transcription factor, aids to the binding specificity of such proteins to chromatin containing a specific combination of PTMs.
Histone acetylation
Histone acetylation of lysine residues is regulated by the counteracting activities of lysine acetyltransferases (KATs) and histone deacetylases (HDACs).
Lysine acetyltransferase
KATs catalyze the transfer of acetyl groups from acetyl-CoA to the ε-group of lysine residues, which results in acetylated lysine (AcK) residues (Fig. 2). Such modified residues were first identified in histone proteins and about a decade later, AcK residues were also found in non-histone proteins (Aka et al., 2011).
KATs can be classified into different families, of which the best-characterized are the GNAT (Gen5-related N-acetyltransferase) superfamily, the MYST (MOZ, Ybf2/Sas3, Sas2 and Tip60) superfamily, the p300 (E1A-associated protein of 300 kDa)/CBP (CREB-binding protein) family and
the Rtt109 (regulator of ty 1 transposition 109) family. Since p300, CBP and Rtt109 do not share any similarity in sequence with the GNAT and MYST superfamilies, these KATs are classified into one superfamily. The largest KAT-family is the GNAT superfamily, in which the members share many sequence motifs including the one essential for acetyl-CoA binding, which is also conserved in the MYST family (Aka et al., 2011). KATs are found to be mutated in tumors. For example, mutations in
Figure 2. The transfer of an acetyl group from acetyl-CoA to the ε-group of lysine residues, catalyzed by lysine acetyltransferases (KATs) (Aka et al., 2011).
7 the p300 gene have been identified in several tumors, including cancers of the colon and rectum, stomach, breast and pancreas. Chromosomal rearrangements involving chromosome 22, on which the p300 gene is localized, have also been associated with certain types of cancer, like acute myeloid leukemia (AML). Rubinstein-Taybi syndrome patients harbor p300 mutations, resulting in the loss of one copy of the gene in each cell. The amount of p300 protein is in these patients reduced by half, leading to disruption of normal development and a higher risk of developing tumors like lymphomas and leukemia.
Histone deacetylases
Removal of acetyl moieties from lysines is performed by histone deacetylases (HDACs) (Fig. 3). Although acetylation of lysine residues was initially discovered in histones, hence their name, HDACs also target acetylated lysines in non-histone proteins (Hasan & Hottiger, 2002). Deacetylation changes the charge of lysine residues from neutral to positive. As DNA is negatively charged, this will result in a more tight interaction between histones and DNA, and therefore in a more compact chromatin structure.
Based on domain homology and sequence similarity to yeast orthologs, phylogenetic analysis showed that eukaryotic HDACs can be divided into four classes (class I – IV) (Aka et al., 2011; Gregoretti et al., 2004). Class I consists of HDAC1, HDAC2, HDAC3 and HDAC8, which are relatively small proteins consisting of 350 – 500 amino acids. Class II HDACs are composed of 660 – 1250 amino acids and can be subdivided in two classes: Class IIa includes HDAC4, HDAC5, HDAC7 and HDAC9, and class IIb contains HDAC6 and HDAC10. Class III
consists of SIRT1-7, which require NAD+ as a cofactor in their catalytic mechanism, while HDAC1 – 11 are zinc dependent enzymes. Class IV includes HDAC11, which is with 347 amino acids the smallest known HDAC and related to both class I and class II (Aka et al., 2011) (Annex 1).
Many of the histone and non-histone substrates of HDAC are involved in crucial cellular processes in normal development and cancer, such as cell cycle regulation, haematopoiesis, proliferation, development and differentiation (Brunmeir et al., 2009; Hagelkruys et al., 2011; Wilting et al., 2010). In cancer, HDACs are often overexpressed or aberrantly recruited, thereby blocking differentiation and apoptosis or inducing proliferation, metastasis and angiogenesis.
HDAC1&HDAC2
HDAC1 and HDAC2 are well-characterized histone deacetylases. They contain 82% sequence similarity and are predominantly nuclear proteins.
Overlapping functions for HDAC1 and HDAC2 have been identified in cell cycle regulation and haematopoiesis (Wilting et al., 2010). In primary cells, Hdac1 and Hdac2 regulate collectively the cell cycle progression. Simultaneous removal or inactivation of Hdac1 and Hdac2 resulted in a senescence-like arrest at the G1-phase of the cell cycle, indicating that maintaining either Hdac1 or
Hdac2 is crucial for the proliferative potential of primary cells. Furthermore, they observed that bone marrow specific deletion of Hdac1 and Hdac2 results in anemia and thrombocytopenia indicative for
Figure 3. Acetylation and deacetylation of histone lysine residues, catalyzed by HATs and HDACs. HATs catalyze the transfer of an acetyl group to lysine residues in histones using acetyl-CoA. HDACs remove the acetyl group, thereby changing the charge of the lysine residue from neutral to positive. (Hasan & Hottiger, 2002).
8 overlapping functions of Hdac1 and Hdac2 in
hematopoietic stem cell survival. They showed that removal of Hdac1 or Hdac2 resulted in up-regulation of respectively Hdac2 or Hdac1. The levels of either Hdac1 or Hdac2 are sufficient for haematopoiesis, suggesting that the total levels of Hdac1 and Hdac2 are critical in haematopoiesis. A decrease in these levels led to differentiation defects and T cell lymphomagenesis, indicative for a correlation between HDAC activity and tumorigenesis. This suggestion was confirmed when mice harboring a conditional knock-out allele for Hdac1 or both Hdac1 and Hdac2 developed T-cell lymphomas of CD4 and CD8, both
lacking Hdac1. These results indicated for the first time a tumor suppressor role for Hdac1 and Hdac2, and suggested a dosage effect in suppressing lymphomagenesis (Fig. 4) (Heideman et al., unpublished data).
HDACs participate in large regulatory complexes that both suppress and enhance transcription. HDAC1 and HDAC2 can function as catalytic subunits of protein complexes that interact with transcription factors, resulting in transcription repression (Gregoretti et al., 2004). Alterations in HDAC1/HDAC2 expression or aberrant recruitment of HDAC1/HDAC2 causes different transcription patterns, which may result in tumorigenesis as we described previously. To date, HDAC1 and HDAC2 are identified in three repressor complexes, called the Sin3, NuRD and CoREST complexes (Fig. 5). To repress gene transcription, these protein complexes are recruited to gene promoters by DNA binding complex members. A ‘core complex’ is conserved between the Sin3 and NuRD complex, containing the histone binding proteins HDAC1, HDAC2 and RbAp46/48. Next to the core proteins, each complex contains specific co-repressor proteins, along with their own function and activity (Grozinger & Schreiber, 2002). Besides binding to numerous complex members, HDACs target a multitude of substrates, including histones. Still, many HDAC substrates remain unknown. Therefore a (SILAC) based proteomic approach was developed to identify non-histone substrates of HDAC1 and HDAC2 (Heideman et al, unpublished data). This approach revealed besides all known HDAC1 and HDAC2 protein complex members also novel interactors, including potential substrates.
The aim of this study was to characterize novel HDAC1/HDAC2 interactors and verify whether these proteins are HDAC1/HDAC2 substrates.
Figure 5. Protein repressor complexes containing HDAC1 and HDAC2. To date, HDAC1 and 2 are found in three main complexes, called the Sin3, NuRD and CoREST complexes. (Grozinger & Schreiber, 2002).
Figure 4. Hdac1 and Hdac2 dosage dependent lymphomagenesis suppression (Heideman et al., unpublished data).
9
Histone methylation
Methylation is a PTM that occurs on arginine and lysine residues. Methylation of histones can result in either transcriptional repression or activation, dependent on the site and degree of methylation. Methylation is a unique modification among the histone PTMs, since up to three methyl groups (me1, me2 and me3 respectively) can be added to a single lysine residue. Arginine residues however can only be monomethylated and dimethylated, of which dimethylation can occur in a symmetric or asymmetric way (Fig. 6A) (Latham & Dent, 2007; Cohen et al., 2011).
Lysine methyltransferases
Lysine methylation is regulated by histone methyltransfersases (HMTs) and histone demethylases (HDMs). Methylation of arginine residues is catalyzed by type I protein arginine methyltransferases (PRMTs), which catalyze the monomethylation and asymmetric dimethylation of arginine residues, and type II PRMTs, which carry out monomethylation and symmetric dimethylation (Fig. 6A).
Methylation of lysine residues is performed by histone lysine methyltransferases (HKMTs) (Fig. 6B) (Margueron et al., 2005). A general characteristic of this class of HMTs is the presence of a conserved SET domain, which consists of 130 amino acids and is crucial for the catalytic activity of the methyltransferase. Moreover, since the SET-domain requires adjacent cysteine-rich domains, it has been suggested that a combination of the SET-domain and those adjacent domains modulate the substrate specificity of HMTs. (Lachner & Jenuwein, 2002; Dillon et al., 2005; Zhang & Reinberg, 2001). SET-domain proteins belong to a family arising from AdoMet-dependent methyltransferases (MTases). Five MTase classes are known (Class I – V), each containing a unique structure. AdoMet-dependent enzymes use S-adenosyl-L-methionine (AdoMet) as a substrate for methyltransfer,
thereby producing S-adenosyl-L
-homocysteine (AdoHcy) (Fig. 6B). SET-domain proteins belong to class V MTases and differ from class I – IV, especially by the location of their binding sites for histone substrates and AdoMet (Schubert et al., 2003). Since methylation is considered to be an essential step in many processes, like maintaining genome integrity, regulating embryonic development and cellular differentiation, it is no surprise that methylation is
Figure 6. A) Schematic of arginine mono- and di-methylation. Two types of protein arginine methyltransferases, type I and II, catalyze arginine methylation. Monomethylation is catalyzed by PRMT type I and both, type I and type II, catalyze dimethylation, respectively asymmetric and symmetric dimethylation. B) Schematic of lysine mono-, di-, trimethylation. HKMTs are AdoMet-dependent enzymes and catalyze the methylation of lysine residues by using S-adenosyl-L-methionine (AdoMet) as a
substrate for methyltransfer, leaving the product S-adenosyl-L-homocysteine (AdoHcy)
10 associated with cancer and other diseases, such as the Hutchinson-Gilford Progeria Syndrome (HGPS) and Immunodeficiency, centromeric instability and facial anomalies (ICF) Syndrome (Gopalakrishnan et al. 2008; Shumaker et al. 2006).
Furthermore, a correlation between histone methylation and cancer has been suggested, since some histone methyltransferase genes have been shown to be overexpressed in human cancer. For example, enhancer of zeste homolog 2 (EZH2) is found to be overexpresssed in many types of aggressive cancers such as the breast, lung, liver as well as in many other types of carcinomas (Cohen et al., 2011). Moreover the histone methyltransferase DOT1L is recruited by leukemic mixed lineage leukemia (MLL) fusion proteins, thereby mistargeting histone methylation and altering gene expression (Okada et al., 2005).
Lysine demethylases
Histone lysine methylation has been considered as a permanent chromatin mark until 2004 when the first histone lysine demethylase was discovered (Shi et al., 2004). Since then, many demethylases have been identified.
Two classes of histone demethylases have now been discovered, Lysine specific demethylase 1 (LSD1) and Jumonji C-terminal domain (JmjC)-containing demethylase enzymes. LSD1 is proposed to mediate demethylation of mono- and dimethylated lysine residues by an oxidative demethylation reaction that uses flavin (FAD) as a cofactor. JmjC-domain-containing histone demethylases (JHDMs) can demethylate mono-, di- and trimethylated lysine residues by an oxidative mechanism that requires iron Fe(II) and α-ketoglutarate (αKG) as cofactors (Klose et al., 2006; Upadhyay et al., 2011). Direct hydroxylation of the methylgroup results in an unstable hydroxymethyl product (1, Fig. 7), which is spontaneously released as formaldehyde (2, Fig. 7) and therefore provides a demethylated lysine residue as a final product. Coordinated binding of the Fe(II) and αKG cofactors is performed by the 3-dimensional folding of the enzymatic active pocket, involving the JmjC domain. Three amino-acids bind to the Fe(II) cofactor and two additional amino-amino-acids bind to αKG, both located within the JmjC domain. Activation of the Jumonji histone demethylases (JHDMs) seems to occur by conserved residues within the predicted cofactor-binding sites. The cofactor-bound JmjC domain produces highly reactive oxoferryl species, which provides hydroxylation of the methylated lysine residues. Variation in residues within the predicted cofactor-binding sites can inhibit enzymatic activity.
Both, the JmjC domain and additional domains, such as PHD fingers and Tudor domains, found within each enzyme seem to determine substrate specificity for JHDMs. JmjC-domain-containing proteins are classified into seven groups based on JmjC-domain homology and full-length protein domain structure (Fig. 8). Six groups contain besides the JmjC domain at least one additional protein domain of which many, like PHD fingers and Tudor domains, bind specifically to methylated lysine residues. The six groups with an additional protein domain are termed JHDM1, PHF2/PHF8, JARID, JHDM3/JMJD2, UTX/UTY and JHDM2. The seventh group, named JmjC domain only, contains proteins
Figure 7. Schematic of the demethylation reaction of lysine residues by JmjC-domain-containing histone demethylases. Direct hydroxylation of the
methylgroup results in an unstable
hydroxymethyl product (1), which is
spontaneously released as formaldehyde (2) and therefore provides a demethylated lysine residue as final product (Klose et al., 2006).
11 with only the JmjC domain and no additional
recognizable protein domains (Klose et al., 2006; Cohen et al., 2011). This group is divided in the sub-groups FIH, MINA53/NO66, PLA2G4B, HSPBAP1, PTDSR, JMJD5, JMJD4 and L0C339123, based on homology within the JmjC domain.
Histone demethylases seem to be important enzymes in developmental processes and have been associated with tumorigenesis and aggressiveness of tumors (Terndrup Pedersen & Helin, 2010; Yoshimi & Kurokawa, 2011). In addition, several demethylases, like UTX and JMJD3, have been linked to human diseases such as cancer. UTX (Ubiquitously transcribed tetratricopeptide repeat, X chromosome), a member of the Jumonji C family of proteins, is a di- and trimethyl H3K27 demethylase and has been associated with mixed-lineage leukemia (MLL) 2/3 complexes (Lee et al., 2007). UTX inactivating mutations were moreover found in multiple cancer types. The highest prevalence was observed in multiple myeloma. Other cancer types with inactivating mutations in UTX included myeloid leukemia, breast and colorectal cancers and glioblastoma (Van Haaften et al., 2009). JMJD3 (also known as KDM6B) is a trimethyl H3K27 demethylase and contributes to the activation of the CDKN2A locus, expressing p16Ink4a and p14Arf/p19Arf, upon oncogenic BRAF and RAS expression. Human ARF is implicated as a tumor suppressor gene, mainly in association with the simultaneous deletion of INK4A, indicating that inactivation of H3K27me3 demethylases may inactivate a tumor protective fail-safe mechanism (Agger et al., 2009). Although reduced levels of JMJD3 were indeed found in various cancers, including lung and liver carcinomas, as well as numerous hematological malignancies, like a subset of lymphomas and leukemias (Agger et al., 2009), increased levels of
Figure 8. Schematic of the seven groups of JmjC-domain-containing proteins. Six groups, termed JHDM1, PHF2/PHF8, JARID, JHDM3/JMJD2, UTX/UTY and JHDM2, have at least one additional protein domain besides the JmjC domain. The seventh group, named JmjC domain only, contains proteins with no additional recognizable protein domains besides the JmjC domain. This group is divided in the sub-groups FIH, MINA53/NO66, PLA2G4B, HSPBAP1, PTDSR, JMJD5, JMJD4 and L0C339123, based on homology within the JmjC domain (Klose et al., 2006).
12 JMJD3 were observerd in Hodgkin’s lymphoma and prostate cancer progression (Anderton et al., 2011; Xiang et al., 2007).
JMJD5
The lysine demethylase JMJD5 (also known as KDM8) is classified in the JmjC-domain only group. JMJD5 is unique as the first αKG-binding site contains a threonine instead of the common serine residue. Although the molecular composition and charge from the complex remain similar by this substitution, there seems to be a preference for inclusion of threonine instead of a serine residue. This may be related to the formation of an enzymatic competent JmjC-domain active site (Klose et al., 2006).
Although it has been shown that JMJD5 is a H3K36me2 demethylase, our knowledge of the role of JMJD5 in specific biological processes is limited. JMJD5 has been implicated in both the plant and human circadian systems (Jones et al., 2010). JMJD5 and TIMING of CAB1 EXPRESSION 1 (TOC1) show a co-operative genetic interaction, which in turn promotes expression of the morning-phased clock genes CIRCADIAN CLOCK ASSOCIATED 1 (CCA1) and LATE ELONGATED HYPOCOTYL (LHY). CCA1 and LHY in turn repress the expression of TOC1 by binding to a conserved motif within its promoter, known as the evening element, forming a negative feedback loop.
JMJD5 is also reported to attenuate osteoclastogenesis as a post-translational co-repressor for Nuclear factor of activated T-cells calcineurin-dependent 1 (NFATc1), a transcriptional inducer of several osteoclast-specific genes, which are required for maturation and bone resorption activity of osteoclasts (Youn et al., 2012). JMJD5 co-represses transcriptional activity by destabilizing NFATc1 protein, using its hydroxylase activity, and induces the association of hydroxylated NFATc1 with E3 ubiquitin ligase VHL, thereby facilitating proteasomal degradation of NFATc1 via ubiquitination. Furthermore, Jmjd5 was identified as a potential tumor suppressor gene (TSG) in murine B-cell lymphomas (Suzuki et al., 2006). Loss-of-function of Jmjd5 was shown to impair DNA mismatch repair allowing the accumulation of mutations, which ultimately can drive tumorigenesis. In contrast, others identified a putative oncogenic function for Jmjd5 (Hsia et al., 2010). JMJD5 was found to be overexpressed in tumors arising from breast tissue, bladder, liver, thyroid, uterine and adrenal glands. Ablation of JMJD5 in breast cancer cells resulted in cessation of proliferation and indicated that JMJD5 is required for progression through the G2/M phase of the cell cycle. The mechanism of cell cycle regulation by JMJD5 seems to rely on the ability of JMJD5 to regulate cyclin A1 transcription. JMJD5 is recruited to H3K36me2, located on the cyclin A1 coding region, and demethylates this mark. Demethylated H3K36 prevents the recruitment of histone deacetylases (HDACs), which subsequently leads in hyper-acetylated histone tails and up-regulated cyclin A1 transcription. An oncogenic function for Jmjd5 was supported by genetic studies, showing that Jmjd5 is essential for embryonic development as Jmjd5 knock-out embryos die at ~11.0 days of embryonic development, probably due to upregulation of the cell cycle inhibitor Cdkn1a (p21) (Suzuki et al., 2012)
Despite these studies, little is known about signal transduction pathways that activate or inhibit JMJD5 function or contribute to specificity in gene regulation. To learn more about the function and mechanisms controlled by JMJD5 during normal development and tumorigenesis, this study is focused on the identification of JMJD5 interacting proteins.
13
Material and Methods
Cloning using Gateway technology
Generation of pDONR223-JMJD5 and pDONR223-interactor
JMJD5 sequence was amplified by polymerase chain reaction (PCR) from plasmid pLKO.1 using Phusion® DNA polymerase (Finnzymes) and two primers (JMJD5-202 & JMJD5-001) (Invitrogen) (annex 2). Sequences of Hdac interactors of our interest were PCR-amplified from cDNA using Phusion DNA polymerase and two specific generated primers for each interactor, containing attB recombination sequences (Table 1) (annex 2). PCR products were purified using MinElute® PCR Purification Kit (Qiagen), and recombined with attP sites of the Gateway pDONR223 vector (Invitrogen) (annex 3) using BP clonase. Using heat shock transformation, the resulting constructs were introduced in Escherichia coli (E. coli) strain DH5α, after which the bacteria were grown on LB (1% tryptone, 0.5% yeast extract, 1% NaCl), 1.5% agar plates containing 50 µg/ml spectinomycin for one night at 37°C. Subsequently, single colonies were picked and incubated in liquid LB medium containing the same antibiotic for one night at 37°C while shaking. Plasmids were then purified from these liquid cultures, using PureLinkTM Quick Plasmid Miniprep Kit (Invitrogen). DNA concentrations were measured using Nanodrop 1000 Spectrophotometer (Thermo Fisher Scientific) and the resulting constructs were verified by restriction enzyme reactions using HpaI and XbaI, and sequencing, for which a PCR was performed using M13F and M13R primers and BigDye Terminator sequence reaction mixture (Applied Biosystems). Products were analyzed by the NKI Sequence facility.
Table 1. Specific generated primers for several Hdac interactors. Full sequences of the primers are shown in annex 2.
Generation of pDEST47-JMJD5 and pDEST47-interactor
JMJD5 and Hdac interactor sequences, flanked by attL sites in pDONR223, were recombined with attR sites into Gateway destination vector pDEST47 vector (Invitrogen) (annex 3) using LR clonase. The resulting construct was transformed in competent E. coli DH5α cells by heat shock transformation and the bacteria were grown on selective LB culture plates containing 50 µg/ml ampicillin for one night at 37°C. Plasmids were then purified from liquid cultures and analyzed by restriction enzyme reactions using specific enzymes for each construct (Table 2) and sequencing using T7 promotor and GFP RV primers. Bacteria from the remainder culture of correct analyzed plasmids were grown again for one night at 37°C while shaking in 250 ml LB medium with 50 µg/ml ampicillin. Plasmids were finally purified from these liquid cultures using NucleoBond® Xtra Midi/Maxi NucleoSpin® Plasmid kit (Macherey-Nagel GmbH & Co. KG).
Hdac interactor FW primer FW primer pDEST47-X-GFP RV primer
Npm1 B1.1F_Npm1 XIB F Npm1 B2.1R_Npm1
Wdr5 B1.1F_Wdr5 XIB F Wdr5 B2.1R_Wdr5
Phf23 B1.1F_Phf23 XIB F Phf23 B2.1R_Phf23
Bbx B1.1F_Bbx XIB F Bbx B2.1R_Bbx
Zbtb20 B1.1F_Zbtb20 XIB F Zbtb20 B2.1R_Zbtb20
14
Plasmid Restriction enzymes
pDEST47-JMJD5 ClaI XhoI
pDEST47-Wdr5 ScaI XhoI
pDEST47-Zbtb20 HindIII XhoI
pDEST47-Npm1 BglII XhoI
pDEST47-Dlat BamHI XhoI
pDEST47-Bbx BglII XhoI
pDEST47-Phf23 NheI XhoI
Table 2. The restriction enzymes used in control digests to analyze constructed
plasmids pDEST47-JMJD5 and pDEST47-interactor.
Generation of retroviral plasmids pQCXIB-JMJD5-GFP and pQCXIB-interactor-GFP
JMJD5-GFP sequence was amplified by PCR from the resulting construct pDEST47-JMJD5 using Phusion DNA polymerase and two primers (JMJD5-202 & RV11 GFP) from which the forward primer contained a NotI restriction site and the reverse primer an EcoRI restriction site (annex 2). The interactor-GFP sequence was amplified by PCR from the resulting construct pDEST47-interactor using Phusion DNA polymerase and gene specific forward primers (Table 1). As a reverse primer, a GFP primer (RV11 GFP) was used for all reactions (annex 2). Forward primers contained a NotI restriction site and the RV GFP primer contained an EcoRI restriction site.
PCR products were purified using MinElute® PCR Purification kit, digested with NotI and EcoRI
restriction enzymes and subsequently T4 DNA ligase (5 U/µl) (Roche) ligated into pQCXIB vector (annex 3). Ligation mixtures were introduced in chemically competent E. coli DH5α cells by heat shock. Bacteria were grown on selective LB culture plates containing 50 µg/ml ampicillin for one night at 37°C, after which plasmids were purified from liquid cultures. Bacteria from the remainder culture of correct plasmids were grown again for one night at 37°C while shaking in 250 ml LB medium containing 50 µg/ml ampicillin. Plasmids were finally purified from these liquid cultures using NucleoBond® Xtra Midi/Maxi NucleoSpin® Plasmid kit. Products were analyzed by sequencing using T7 promotor and GFP RV primers, and restriction enzyme digests using NotI and EcoRI restriction enzymes.
Generation of Hdac1
-/-;Hdac2
-/-double knock-out cells
In order to generate Hdac1-/-;Hdac2-/- double knock-out (DKO) Mouse Embryonic Fybroblasts (MEFs), Hdac1L/L;Hdac2-/- MEFs harboring the tamoxifen inducible Rosa26CreERT2 allele were treated with 200 nM tamoxifen (4-OHT) in order to activate Cre-recombinase, resulting in deletion of the Hdac1 conditional allele.
Cell culture and retroviral infection
Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen) containing 10% fetal bovine serum (FBS) (Invitrogen), 2 mM L-glutamine (Invitrogen), 50 U Penicillin (Invitrogen) and 50 µg Streptomycin (Invitrogen) at 37°C and 5% CO2. 0.05% Trypsin-EDTA (1x) (Gibco) was used to generate
single cell suspensions.
For retrovirus production, 12.5 µg pCL helper plasmid (Naviaux et al., 1996) and 12.5 µg retroviral plasmid were added to 1 ml 0.25 M CaCl2, after which 1 ml 2x HEPES-buffered Saline (HBS) (pH
6.95-7.05) (40 mM HEPES, 270 mM NaCl, 10 mM KCl, 10 mM D-Glucose and 1.5 mM Na2HPO4∙2H2O) was
added drop-wise, while vortexing. The solution was added to the medium of a subconfluent 100x20 mm cell culture dish with Human Embryonic Kidney (HEK) 293T cells and incubated overnight. Cell
15 culture medium was refreshed the next day. For retroviral infection, viral supernatant was filtered (0.22 µm) 48 hrs after transfection and added along with polybrene (1.6 mg/ml) to the target cells (Table 3). Fresh medium was added to the transfected 293T cells and fresh viral supernatant was collected the next day to be used for a second viral infection of the target cells. After a 6 hrs incubation at 37°C and 5% CO2, infected cells were selected using blasticidin (8 µg/ml) for at least 2
days.
Nuclear extraction
Cells were detached and single cell suspension were generated by treating cell cultures with 1x 0.05% Trypsin-EDTA. Cells were pelleted by centrifugation (5 min, 1500 rpm). Cell pellets were washed in 10 volumes 1x ice-cold phosphate buffered saline (PBS) and gently resuspended in 2 volumes of hypo-buffer 1 (10 mM Tris-Cl (pH 7.8), 5 mM MgCl2, 10 mM KCl, 0.1 mM EDTA, 300 mM sucrose, 5 mM beta-glycerol phosphate, 0.5 mM dithiothreitol and protease inhibitors (Complet@, Roche)). Cell suspensions were kept on ice for 10 min. followed by the addition of 10% NP-40 and centrifugation for 3 min. at 2600 rpm at 4°C. The resulting supernatant is the cytoplasmic fraction. The pellet was washed in 1x ice-cold PBS, resuspended in 2 volumes of buffer C (420 mM NaCl, 20 mM Hepes KOH (pH 7.9), 20% v/v glycerol, 2 mM MgCl2, 0.2 mM EDTA, 0.15% NP-40, 0.5 mM DTT and protease inhibitor) and subsequently incubated at 4°C for 1 hr, while rotating. The nuclear fraction was obtained by collecting the supernatant upon centrifugation for 30 min. at 14000 rpm at 4°C. Protein concentration of both the cytoplasmic and nuclear fractions were measured using the Bio-Rad Protein Assay (Bio-Rad).
Purification of Green Fluorescent Protein (GFP-) tagged proteins
In order to purify GFP-tagged proteins, GFP-Trap® coupled agarose beads (Chromotek) were washed
with 1 ml buffer C (420 mM NaCl, 20 mM Hepes KOH (pH 7.9), 20% v/v glycerol, 2 mM MgCl2, 0.2 mM EDTA, 0.15% NP-40, protease inhibitor 1x and 0.5 mM DTT) for 3 times, thereby centrifuged at 3000 rpm. 25 µl beads were subsequently equally aliquted for each sample and another washing step was performed. A maximum of 1000 ug protein and 2 µl of 10 mg/ml Ethidium Bromide (EtBr) (Invitrogen) were then added onto the beads and the samples were incubated for 90 minutes on a rotating wheel at 4°C and centrifuged for 1 min. at 3000 rpm. The pellet, consisting of agarose beads, was washed 2 times with 1 ml buffer C and once with 1 ml 1x PBS containing 0.5% NP-40, followed by the addition of 50 µl 0.1 M glycine (pH 2). The samples were incubated for 5 min. in a shaker at 1200 rpm at room temperature, after which the beads were spinned down for 30 sec. at 14000 rpm. The supernatant was then transferred to a new tube and 10 µl 1M Tris (pH 8.5) was added. Protein concentrations were determined using the Bio-Rad Protein Assay.
Western Blot analysis
To generate protein lysates, cells were washed in 1x PBS and lysed in RIPA lysis buffer (150 mM NaCl, 1.0% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris (pH 8.0) and a protease inhibitor cocktail tablet (Complet@, Roche) for 30 minutes on ice. Cell lysate was subsequently sonicated for 5 min. with 25 sec. intervals at 200W using a Bioruptor (Diagenode), after which the sample was
Cell line Organism
Human Embryonic Kidney 293T HEK 293T Human Michigan Cancer Foundation – 7 MCF-7 Human Mouse Embryonic Fibroblast MEF Mouse
16 centrifuged for 30 min. at 14000 rpm and 4°C. Protein concentration was measured using the Bio-Rad Protein Assay.
To prepare protein samples for SDS-Poly Acrylamide Gel Electrophoresis (PAGE), a 4x loading buffer (Invitrogen) containing 2.5% v/v β-mercaptoethanol was added to the cell lysates. Protein lysates were size separated on a NuPAGE 4-12% gradient Bis-Tris polyacrylamide gel (Invitrogen) in 1x MOPS or MES electrophoresis buffer (Invitrogen). As a marker for protein size we used SeeBlue®Plus2 Prestaind Standard (1x) (Invitrogen). For Western blot analysis, size separated proteins were transferred to Whatman® Protan® Nitrocellulose Transfer Membrane (0.45 µm) (Whatman) in transfer buffer (50 mM Tris, 380 mM glycine, 20% v/v methanol). Membranes were subsequently blocked for 1 h at room temperature in blocking buffer consisting of 5% milk powder in 1x TBST (10 mM Tris-HCl, 100 mM NaCl and 0.1% Tween 20), after which primary antibody was diluted in blocking buffer and incubated with the membrane overnight at 4°C (Table 4). Primary antibodies were then removed and the membrane was washed 3 times for 10 minutes in 1x TBST. The membrane was subsequently incubated with blocking buffer diluted Horse Radish Peroxidase (HRP-) conjugated goat anti-rabbit or goat anti-mouse secondary antibodies (Dako) (Table 3), dependent on the primary antibody, for 1 h at room temperature. Upon washing the membranes 3 times in 1x TBST the antibodies were visualized using SuperSignal® West Pico Chemiluminescent Substrate (Thermo Fisher Scientific) and the ChemiDocTM XRS+ System (Bio-Rad Laboratories), and analyzed using Image LabTM software.
Table 4. The primary and secondary antibodies used in this research.
Coomassie staining of the SDS gel
To visualize proteins separated by SDS-PAGE, 4-12% gradient Bis-Tris polyacrylamide gels were stained after electrophoresis for 30 min. using Coomassie Brilliant Blue R-250 (Invitrogen). The gels were destained using a 40% methanol/ 10% acetic acid solution.
Mass spectrometry
Single proteins, directly or indirectly bound to JMJD5, were identified by Alex Fish (Department of Biochemistry, NKI) using mass spectrometry. Therefore, trypsin digestion and sample extraction from the 4-12% gradient Bis-Tris polyacrylamide gels were performed. Peptides were subsequently separated and data were collected on a LTQ-Orbitrap Discovery mass spectrometer (Thermo Scientific) coupled to a nano-LC system (Proxeon). The data were ultimately analyzed using Sequest and Mascot algorithms software.
Primary Antibody Company
α-GFP Santa Cruz
α-HDAC1 Santa Cruz
α-HDAC2 Santa Cruz
α-pan-acetyl-lysine Abcam
Secondary Antibody
Goat α-Rabbit HRP Dako Denmark A/S Goat α-Mouse HRP Dako Denmark A/S
17
Results
Validation and characterization of novel Hdac1 and Hdac2 interactors
Identification of novel Hdac1 and Hdac2 interactorsTo identify novel non-histone substrates of Hdac1 and Hdac2, a SILAC (Stable Isotope Labeling by Amino acids in Cell culture) based proteomic approach was developed (Heideman et al, unpublished data). In their study, Heideman et al. expressed GFP-tagged versions of Hdac1 or Hdac2, in Hdac1-/- or Hdac2-/- knock-out mouse embryonic fibroblasts (MEFs) by retroviral infection. Cells were differentially labeled by growing them in light medium with normal arginine or medium with heavy arginine. Metabolic incorporation of the amino acids into the proteins results in a mass shift of the corresponding peptides. A GFP pull-down was then performed on nuclear extracts for the purification of Hdac1/Hdac2 along with their interaction partners, followed by the identification of these interactors using mass spectrometry (MS). Mass shifts of the proteins were detected by a mass spectrometer and when both samples, incorporated with normal or heavy arginine, are combined, the ratio of peak intensities in the mass spectrum reflects the relative protein abundance. This approach revealed besides known Hdac1 and Hdac2 interactors also novel interactors (table 5). Using a database harboring proteins with known post-translational modifications (www.phosida.de), several interactors were identified as acetylated proteins suggesting that these interactors may be potential Hdac1 and or Hdac2 substrates (Table 5 & Annex 4).
Table 5. Novel Hdac1/2 interactors,
identified by Heideman et al.
(unpublished data). Proteins indicated in bold are identified as acetylated proteins.
Generation of Hdac1/Hdac2 interactor-GFP fusion proteins
In order to validate the interaction between identified novel Hdac1/2 interacting proteins and Hdac1/2, and characterize the identified novel Hdac1/2 interactors as potential substrates for Hdac1/2, we generated GFP-tagged versions of Hdac1/2 interactors in retroviral vectors. Therefore, the interactors: WD repeat domain 5 (Wdr5), Bobby box homolog (Bbx), PHD finger protein 23 (Phf23), Dihydrolipamide S-acetyltransferase (Dlat), Nucleophosmin 1 (Npm1) and Zinc finger and BTB domain containing 20 (Zbtb20), were amplified from mouse tissue cDNA. PCR fragments with sizes corresponding to length of the known coding sequence were purified and recombined into pDONR223. The presence of the interactor sequence in pDON223 was confirmed by restriction enzyme digests and sequencing. The resulting plasmids pDON223-interactor were subsequently recombined with the destination vector pDEST47 to generate C-terminal interactor-GFP fusion sequences.
Zmym3 Dnttip1 Znf827 C16orf87 Znf516
Sf3b1 Snd1 Foxk1 Sall3 Zfp521
Bcl11b Zmynd8 Foxk2 Bahd1 Zfp532
Npm1 Pwwp2b Zbtb20 Zmym2 Znf592
Trerf1 Bahcc1 Phf23 Zmym4 Znf608
Bbx Sfmbt1 Rreb1 Ruvbl2 Znf609
Wdr5 Cdc5l Bnc2 Znf217 Ilf3
Son Znf521/Evi3 Trps1 Znf219 Tnrc18
Dlat Znf326 Znf687 Alyref Zfp496
18
Expression analysis and cellular localization of interactor-GFP fusion proteins
In order to validate the generation of interactor-GFP fusion proteins, the expression and cellular localization of interactor-GFP fusion proteins were checked by transfection of HEK 293T cells with the resulting pDEST47-interactor plasmids (Fig. 9). Since Hdac1 is a nuclear protein, we expected expression of the Hdac1/2 interacting proteins also in the nucleus. Transfection of pDEST47-Wdr5-GFP, pDEST47-Bbx-GFP and pDEST47-Phf23-GFP constructs revealed exclusively green fluorescent nuclei, indicative for nuclear expression of Wdr5-GFP, Bbx-GFP and Phf23-GFP fusion proteins. pDEST47-Dlat-GFP transfection into HEK 293T cells resulted in green fluorescent mitochondria and nuclei, suggesting the presence of Dlat-GFP in mitochondria and nucleus. Cells transfected with pDEST47-Npm1-GFP showed green fluorescent nucleoli, indicating expression of the Npm1-GFP fusion protein. Surprisingly, pDEST47-Zbtb20-GFP transfection into HEK 293T cells resulted in green fluorescent dots in the cytoplasm, adjacent to the nucleus.
Figure 9. Validation of the expression and cellular localization of GFP-tagged HDAC interactors Wdr5, Bbx, Phf23, Dlat, Npm1 and Zbtb20 by transfection of HEK 293T cells.
In addition, to validate the expression of full length interactor-GFP fusion proteins, we performed western blot analysis on lysates of pDEST47-interactor-GFP transfected cells using an antibody against GFP. This analysis confirmed the presence and the expected size of Zbtb20-GFP (110,9 kDa), Npm1-GFP (62,4 kDa), Wdr5-GFP (66,5 kDa), Bbx-GFP (130,6 kDa) and Phf23-GFP (73,4 kDa) fusion proteins. Surprisingly, the Dlat-GFP fusion protein was expressed with a molecular weight of approximately 67 kDa, which is smaller than the expected size of 97,8 kDa. (Fig. 10). In conclusion, we successfully constructed Wdr5-GFP, Bbx-GFP, Phf23-GFP and Zbtb20-GFP fusion proteins which were, except for Zbtb20, expressed in the nucleus.
Wdr5
Dlat
Bbx Phf23
Zbtb20 Npm1
19
Figure 10. Expression analysis of interactor-GFP fusion proteins by western blot using an antibody against GFP.
Expression of interactor-GFP fusion proteins in MEFs
In order to validate the interaction between novel identified Hdac1/2 interactors and Hdac1, and characterize the novel identified Hdac1/2 interacting proteins as potential substrates for Hdac1/2, the interactor-GFP fusion proteins are expressed in wild-type (wt) mouse embryonic fibroblasts (MEFs) and tamoxifen (4-OHT) treated RCM2;Hdac1L/L;Hdac2-/- MEFs (referred to as DKO MEFs). DKO cells lack Hdac1 and Hdac2, which induces hyperacetylation of Hdac1/2 substrates.
In order to express Hdac1/2 interacting proteins, we generated retroviral vectors which allow expression of a Green Fluorescent Protein (GFP)-tagged Hdac1/2 interactor. A PCR was therefore performed on plasmid pDEST47-interactor-GFP, using primers harboring a NotI or a EcoRI restriction site. The DNA fragments interactor-GFP, digested with NotI and EcoRI, were subsequently cloned into the retroviral vector pQCXIB. Resulting plasmids pQCXIB-interactor-GFP were checked by sequencing and restriction enzyme reactions (data not shown).
In order to express the fusion proteins Wdr5-GFP, Zbtb20-GFP, Npm1-GFP and Dlat-GFP in both wt MEFs and DKO MEFs, we generated retroviral supernatants harboring Wdr5-GFP, pQCXIB-Zbtb20-GFP, pQCXIB-Npm1-GFP and pQCXIB-Dlat-GFP viruses. Transfection of HEK 293T cells using pQCXIB-interactor-GFP constructs resulted in successful expression of the fusion proteins (Fig. 11). Retroviruses were subsequently harvested and used to infect MEFs. Retroviral infection with empty vector pQCXIB was included as a negative control. As a positive control, wt MEFs were treated with HDAC inhibitors (HDACi), which induces hyperacetylation of HDAC substrates.
Dlat Npm1
Zbtb20 Wdr5
Figure 11. Validation of the expression of fusion proteins Wdr5-GFP, Zbtb20-GFP, Npm1-GFP and Dlat-GFP from retroviral vectors pQCXIB – interactor – GFP in HEK 293T cells.
20
Validation of the interaction between Hdac1 and GFP fusion proteins
In order to validate the Hdac1/Hdac2 interactors, we performed nuclear extractions and enriched interactor-GFP fusion proteins by GFP-Trap® pull-down. Western blot analysis on GFP lysates using an antibody against GFP confirmed the presence of fusion proteins Wdr5-GFP, Zbtb20-GFP, Npm1-GFP and Dlat-GFP (Fig. 12). As expected, no GFP was detected in the negative control.
To validate the interaction between Hdac1 and interactor-GFP fusion proteins, western blot analysis on GFP pull-down lysates using an antibody against Hdac1 was performed. This analysis revealed the presence of Hdac1 along with Wdr5 and Dlat in wt MEFs, thereby validating the interaction between these proteins (Fig. 12). No Hdac1 was detected in GFP-pull down lysates from wt MEFs expressing Zbtb20-GFP and Npm1-GFP. Notably, when cells were treated with 10 µM of the HDAC inhibitor SAHA, Hdac1 interacted with Npm1. The absence of Hdac1 in GFP-pull down lysates from DKO MEFs indicates the specificity of the interaction with Hdac1 in Hdac1-proficient MEFs.
Validation of GFP fusion proteins as substrates of Hdac1 and Hdac2
To validate whether the identified Hdac1 and Hdac2 interacting proteins are genuine substrates of these deacetylases, we performed western blot analysis on GFP pull-down lysates of wt and DKO MEFs expressing GFP-fusion proteins using an antibody against acetylated lysines (acK). In addition, as a positive control, we performed a similar analysis on HDACi treated wild-type MEFs expressing GFP fusion proteins (Fig. 12). This analysis revealed acetylation of Dlat only in the presence of HDACi, indicating that this protein is a substrate for HDACs. Since we did not observe acetylated Dlat in the absence of Hdac1 and Hdac2, we could not confirm that this protein is a genuine substrate of Hdac1 and Hdac2.
In conclusion, we have identified and partly characterized novel Hdac1 interacting proteins Wdr5, Npm1 and Dlat1.
Figure 12. Western blot analysis on GFP pull-down lysates from wt MEFs, tamoxifen (4-OHT) treated
RCM2;Hdac1L/L;Hdac2-/- MEFs (referred to as DKO MEFs) and HDAC inhibitor (10 µM SAHA) treated wt MEFs expressing empty vector, Wdr5-GFP, Zbtb20-GFP, Npm1-GFP and Dlat-GFP.
21
Identification of novel JMJD5 interactors
Generation and expression of the JMJD5-GFP fusion protein
Several studies suggest the involvement of histone demethylase JMJD5 in cancer cell proliferation (Suzuki et al., 2006; Hsia et al., 2010; Suzuki et al., 2012). However, despite these proposed models, it is poorly understood how JMJD5 is regulated and how it is connected to known signal transduction pathways. To learn more about the function and mechanisms controlled by JMJD5 during normal development and tumorigenesis, this study is focused on the identification of JMJD5 interacting proteins.
We generated a retroviral vector which allows expression of a Green Fluorescent Protein(GFP)-tagged human JMJD5. To clone human JMJD5, we performed a PCR to amplify the JMJD5 sequence from plasmid pLKO.1. PCR fragments with size corresponding to length of the known coding sequence were purified and a recombination reaction was performed to clone JMJD5 into the pDONR223 donor vector. The presence of the JMJD5 sequence in pDONR223 was confirmed by a restriction enzyme reaction and sequencing (data not shown). The resulting plasmid pDONR223-JMJD5 was subsequently recombined into the destination vector pDEST47 to generate a C-terminal JMJD5-GFP fusion sequence. Next to verification of this construct by sequencing, we tested the expression of the JMJD5-GFP fusion protein by transfection of this construct into 293T human embryonal kidney (HEK). pDEST47-JMJD5-GFP transfection into 293T HEK cells resulted in green fluorescent nuclei indicating the expression of nuclear JMJD5-GFP fusion protein (Fig. 13A). Furthermore, western blot analysis on lysates of pDEST47-JMJD5-GFP transfected cells using an antibody against GFP confirmed the presence of the fusion protein JMJD5-GFP (Fig. 13B).
Identification of novel JMJD5 interactors
JMJD5 was found to be overexpressed in tumors arising from breast tissue, bladder, liver, thyroid, uterine and adrenal glands (Hsia et al., 2010). In this study, we expressed JMJD5-GFP in both primary (HEK 293T) and tumor cells (MCF-7) to compare interacting proteins.
In order to express JMJD5-GFP in HEK 293T and MCF-7 cells, we cloned the JMJD5-GFP sequence into a retroviral vector. A PCR was therefore performed on pDEST47-JMJD5-GFP plasmid template, using primers harboring a NotI or a EcoRI restriction site. A JMJD5-GFP DNA fragment, digested with NotI and EcoRI, was subsequently cloned into the retroviral vector pQCXIB. pQCXIB-JMJD5-GFP plasmids were verified by sequencing and restriction enzyme digests (Fig. 14). In order to identify JMJD5 interaction partners, we overexpressed JMJD5-GFP fusion protein in HEK 293T and MCF-7 cells by retroviral infection using pQCXIB-JMJD5-GFP plasmids (Fig. 15). As a negative control, we used HEK293T and MCF-7 cells infected with an empty retrovirus. Fluorescence microscopic analysis of the JMJD5 overexpression cells indicated expression of JMJD5-GFP in the nucleus. Since JMJD5 is
Figure 13. Validation of the presence of JMJD5-GFP in pDEST47 by A) transfection of 293T cells and B) Western Blot analysis.
22 described as a nuclear protein, this analysis confirms a successful expression of the JMJD5-GFP fusion protein in both HEK 293T cells and MCF-7 cells.
In order to purify JMJD5 and interacting proteins from JMJD5-GFP expressing cells, we generated nuclear extracts of the JMJD5-GFP expressing cells and analyzed both cytoplasmic and nuclear fractions for JMJD5-GFP expression. Surprisingly, JMJD5-GFP was detected in the cytoplasmic fraction derived from both 293T and MCF-7 cells (Fig. 16). To exclude a technical error during the extraction, we determined whether HDAC2, a protein that is exclusively located in the nucleus, was present in the nuclear fractions. As expected, HDAC2 was detected in de nuclear fraction, which excludes a technical error in the nuclear extraction and confirms the localization of JMJD5-GFP in the cytoplasm. Since western blot analysis showed that JMJD5-GFP is present in the cytoplasmic extract, this fraction was used for the GFP-Trap® pull-down to enrich interactor-GFP fusion proteins along with their interactors. Using GFP-Trap® coupled to agarose beads, JMJD5-GFP fusion protein was highly enriched from the cytoplasmic fraction (input), while no JMJD5-GFP was detected in the non-bound flow through fraction (Fig. 17).
Figure 14. Agarose gel electrophoretic analysis of a control digestion on the designed plasmid pQCXIB-JMJD5-GFP.
bp 2 000 - 10 000 - 1 000 - 7 000 bp; 2 000 bp; JMJD5-GFP backbone vector
Figure 16. Detection of JMJD5-GFP after a nuclear extraction by Western blot analysis. The nuclear and cytoplasmic fraction of both 293T cells and MCF-7 cells were collected and loaded. HDAC2 served as a technical control.
Figure 17. Western blot analysis after a GFP pull-down on JMJD5-GFP and the negative control. Input, non-bound and resulting eluate were collected and loaded, derived from both 293T cells and MCF-7 cells.
Figure 15. Green fluorescence in MCF-7 cells after retroviral infection and selection.
23 In order to analyze GFP pull-down lysates from JMJD5-GFP overexpressing cells by mass spectrometry, GFP pull-down lysates were size separated on a NuPAGE 4-12% gradient Bis-Tris polyacrylamide gel. As a negative control, GFP pull down lysates from cells expressing empty vector were loaded. Analysis using Coomassie Brilliant Blue R-250 revealed the presence of JMJD5-GFP in the GFP pull-down lysates from JMJD5-GFP overexpressing cells (Fig. 18).
In order to identify the proteins in the GFP pull-down lysates from JMJD5-GFP expressing cells and empty vector expressing cells, each lane was chopped and proteins were identified by mass spectrometry. Proteins specifically interacting with JMJD5 in HEK 293T cells are shown in table 6 (Annex 5 for a more detailed overview of the interacting proteins). The most frequent protein identified specifically in JMJD5-GFP overexpressing cells was JMJD5 (KDM8) itself, indicating that the approach worked. GFP pull-down lysates from MCF-7 cells are not yet identified by mass spectrometry.
Specific interactors of JMJD5
GFP pull-down lysate from 293T cells Number of identified peptides
Lysine-specific demethylase 8 36
HCLS1-associated protein X-1 2
D-3-phosphoglycerate dehydrogenase 3
ATP synthase subunit beta, mitochondrial 2
CD226 antigen 1
B-cell receptor-associated protein 31 1
ADP/ATP translocase 3 1
RCC1 domain-containing protein 1 1
Squalene monooxygenase 1
Phosphatidylinositol 4-kinase type 2-alpha 1
ATP synthase subunit alpha, mitochondrial 1
CAD protein 3
DNA-dependent protein kinase catalytic subunit 1
Table 6. Proteins specifically interacting with JMJD5-GFP, identified in GFP pull-down lysates from HEK 293T cells expressing JMJD5-GFP.
Figure 18. Protein analysis on a 4-12% gradient Bis-Tris polyacrylamide gel using Coomassie Brilliant Blue R-250.
24
Discussion and conclusions
Histone modifying enzymes are crucial in the regulation of many biological processes and involved in many diseases including cancer. Moreover, over the past decade targeting histone modifying enzymes using small molecule inhibitors has gained considerable attention in the treatment of hematological malignancies and neurological disorders. To harness the therapeutic potential of these inhibitors, knowledge of the function of histone modifying enzymes is critical. Despite numerous studies, the mechanisms controlled by histone modifying enzymes, like histone deacetylases and histone demethylases, during normal development and tumorigenesis are poorly understood.
Histone deacetylases HDAC1 and HDAC2
Little is known about non-histone substrates of HDAC1 and HDAC2. Here we have used a proteomic approach to identify interaction partners and substrates of histone deacetylases Hdac1/2 to provide insight into the function of these proteins. Although many Hdac1/2 interactors have already been identified, a SILAC-based proteomic approach revealed many novel Hdac1/2 protein interactors. In this study, we validated and characterized these novel Hdac1/2 protein interacting proteins.
Expression and cellular localization of interactor-GFP fusion proteins
Transfection of “interactor”-GFP fusion proteins in 293T cells resulted in a functional GFP fusion expression. Western blot analysis on lysates of pDEST47-interactor-GFP transfected cells confirmed the presence and expected sizes of Wdr5-GFP, Bbx-GFP, Phf23-GFP and Zbtb20-GFP.
As expected, based on the fact that Hdac1 is a nuclear protein, fluorescence microscopic analysis confirmed the expression of Wdr5-GFP, Bbx-GFP, Phf23-GFP, Dlat-GFP and Npm1-GFP fusion proteins in the nucleus. However, Zbtb20-GFP fusion protein was localized in the cytoplasm, adjacent to the nucleus. Since Zbtb20 is a nuclear protein and unknown to be localized in the cytoplasm, the observed presence of Zbtb20-GFP in the cytoplasm is probably caused by the GFP-tag. In order to check this hypothesis, Zbtb20 could be fused with another peptide tag, like the Human influenza hemagglutinin (HA)-tag. Western blot analysis on cytoplasmic and nuclear extracts of transfected cells using an antibody against HA, would reveal the localization of the fusion protein. A better approach would be the expression of non-tagged Zbtb20. The localization of this protein can be determined by western blot analysis on cytoplasmic and nuclear extracts of transfected cells using an antibody against Zbtb20.
Validation of the interaction between Hdac1 and GFP fusion proteins
To further validate the fusion proteins Wdr5-GFP, Zbtb20-GFP, Npm1-GFP and Dlat-GFP, we expressed these proteins successfully in MEFs by retroviral infection and checked for Hdac1 interaction using GFP down. Western blot analysis showed the presence of Hdac1 in GFP pull-down lysates from wild-type but not DKO (Hdac1 and Hdac2 deficient) MEFs expressing Wdr5-GFP and Dlat-GFP, validating the interaction between these proteins. Npm1-GFP interacted with Hdac1 only in the presence of HDAC inhibitor. Since HDAC inhibitors occupy the active site within the deacetylase domain of Hdac1, Hdac1 probably interacts with Npm1 through a different domain.
Validation of GFP fusion proteins as substrates of Hdac1 and Hdac2
In the case of Dihydrolipamide S-acetyltransferase(Dlat), a subunit of the pyruvate dehydrogenase enzymatic complex, we observed an interaction with Hdac1 which was lost in the presence of HDAC
