SEARCH

SEARCH BY CITATION

Keywords:

  • NANOG;
  • Human embryonic stem cells;
  • Transcription

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

NANOG is a key transcriptional regulator of pluripotent stem cell (PSC) self-renewal. NANOG occupies promoters that are active and others that are repressed during self-renewal; however, the mechanisms by which NANOG regulates transcriptional repression and activation are unknown. We hypothesized that individual protein domains of NANOG control its interactions with both the promoters and its coregulators. We performed a detailed characterization of the functional domains in the human (h) NANOG protein, using a panel of deletion-mutant and point-mutant constructs. We determined that six amino acids in the homeodomain (136YKQVKT141) are sufficient for the nuclear localization of hNANOG. We also determined that the tryptophan-rich region (W) of hNANOG contains a CRM1-independent signal for nuclear export, suggesting a possible cellular shuttling behavior that has not been reported for hNANOG. We also show that at least four tryptophans are required for nuclear export. We also determined that similar to murine (m) NANOG, the W region of hNANOG contains a homodimerization domain. Finally, in vitro transactivation analyses identified distinct regions that enhance or diminish activity at gene promoters that are active during self-renewal. Specifically, the N-terminal region interferes with transcription and removal of this region that produced a “super-active” hNANOG with enhanced transcriptional activity. We also confirmed that the transcriptional activator in hNANOG is contained in the C-terminal region, similar to murine NANOG. In summary, this study has characterized the structure and function of hNANOG protein leading to an increased understanding of the mechanism by which hNANOG regulates both transcriptional activation and repression during PSC self-renewal. STEM CELLS 2009;27:812–821


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Pluripotent stem cells (PSC) have the capacity to self-renew and differentiate into cells from all three embryonic germ layers. They can be derived from the inner cell mass of blastocysts (embryonic stem cells [ESCs]) or generated through reprogramming somatic cells, for example to produce induced pluripotent stem cells (iPSCs) [1–3]. Regardless of the derivation method, PSC can serve as tools to study early human embryonic development and as a basis for regenerative medicine applications.

Many transcription factors are considered to regulate human PSC self-renewal, including OCT4, SOX2, and NANOG. Although reduced expression of any of these factors in PSC leads to differentiation [4–9], overexpression of NANOG can maintain feeder-independent self-renewal [10–12]. Although NANOG is not required for somatic cell reprogramming into PSC, selection of clones with high levels of NANOG expression increases the efficiency and pluripotentiality of the resulting iPSCs [13–17]. In addition, although murine NANOGnull ESCs can propagate, they proliferate poorly and are prone to differentiation [18]. Collectively, these studies demonstrate that NANOG plays a central role in regulating PSC self-renewal.

At a molecular level, NANOG has been shown to occupy promoters of genes that are expressed and others that are repressed during human- and murine-PSC self-renewal [19–24]. The mechanism regulating expression or repression at these loci is not completely understood, although it is likely related to the coregulators, the number of factors bound and the specific chromatin structure at each locus [25–27]. Orkin and coworkers have developed a protein interaction network for murine ESCs in their recent studies [24, 25]. These studies have demonstrated that NANOG associates, directly or indirectly, with transcriptional activators (such as OCT4 and SALL4), corepressors (such as HDAC2), and chromatin remodeling complex complexes (such as polycomb, SWI/SNF, and NuRD complexes). However, the specific protein domains in NANOG that interact with either the promoters or transcriptional activators are not well-characterized.

The general structure of murine (m) NANOG has been described with a serine-rich N terminal (N), NK-2 type homeobox (H), and C-terminal (C) domains (Fig. 1) [10, 11]. Heterologous GAL4 transactivation assays have identified transcription activities in both the N-terminal (N) and C-terminal regions [28, 29]. There is also a tryptophan-rich region in C which is required for murine NANOG homodimerization [30, 31].

thumbnail image

Figure 1. Intracellular localization of overexpressed hNANOG fragments. (A): Schematic representation of hNANOG deletion mutants generated via PCR cloning into a cytomegalovirus-driven expression plasmid containing a C-terminal V5-epitope tag. (B): Immunofluorescent staining of CV1 fibroblasts transiently transfected with V5-epitope tagged wild-type and mutant hNANOG (shown in green). The nuclei are stained with DAPI (shown in blue). Putative sequences encoding nuclear localization signal and nuclear export signal were identified to lie within the homeodomain (H, panel 9) and tryptophan-rich (W, panel 11) regions, respectively. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; hNANOG, human NANOG.

Download figure to PowerPoint

Human (h) NANOG has 58% amino acid homology with murine (m) NANOG and shares similar structural domains as the murine ortholog (Fig. 1). The H domain in hNANOG has been shown to localize to the nucleus and likely contains a nuclear localization signal (NLS) [32]. The transcriptional activity of hNANOG is in C-terminal and there is no function described for the N-terminal domain [33]. However, the detailed protein structure and protein interactions have not been described for human NANOG. Given the differences in the pathways maintaining human and murine PSC self-renewal, it is critical to understand the mechanism by which hNANOG regulates human PSC self-renewal. In these studies, we characterized the specific functional domains and sequences in the hNANOG protein to provide insight into the molecular mechanism by which human NANOG contributes to regulate PSC self-renewal.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Cells and Reagents

Monkey kidney CV1 fibroblasts and human embryonic kidney 293A cells were purchased from ATCC (Manassas, VA, http://www.atcc.org) and cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum, 2 nM L-glutamine, and penicillin–streptomycin antibiotics (50 μg/ml). Anti-V5 mouse monoclonal antibody was from Invitrogen (Carlsbad, CA, http://www.invitrogen.com). Anti-FLAG M2 mouse monoclonal antibody was purchased from Sigma-Aldrich (St.Louis, MO, http://www.sigmaaldrich.com). The fluorescent Alexa Fluor 488-conjugated secondary antibodies were purchased from Invitrogen. The horseradish peroxidase-conjugated secondary antibodies and isotype specific control IgG were purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, http://www.scbt.com). SuperSignal West Pico Chemiluminescent Substrate for Western blotting was purchased from Pierce (Rockford, IL, http://www.piercenet.com). Leptomycin B (LMB) was purchased as a ready-made solution (70% methanol) from Sigma.

Construction of Plasmids

Full-length hNANOG was amplified by RT-PCR from total RNA lysate of H1 human ESCs and cloned into cytomegalovirus (CMV)-driven, eukaryotic expression vectors pCDNA3.1D/V5-TOPO (Invitrogen). Subsequently, this construct was used as a template to generate deletion-mutation and point-mutation constructs via PCR. Gene reporter mOct4pr-Luc was generated by cloning the 3-kb gene fragment upstream of mouse Pou5f1 (Oct4) into pGL3-Basic vector (Promega, Madison, WI, http://www.promega.com). A 406-bp DNA fragment 5′ upstream of mouse Nanog gene was cloned into pGL3-Basic to generate mNanogpr-Luc construct. Similarly, mRex1pr-Luc luciferase reporter was made by cloning a 500-bp DNA fragment 5′ upstream of mouse Rex1 gene into pGL3-Basic vector. Enhanced green fluorescent protein (EGFP)-linked constructs were generated by PCR-based subcloning strategy, using pEGFP-1 vector (Clontech, Mountain View, CA, http://www.clontech.com) as a template. N-terminal three adjacent FLAG epitope-tagged p3xFLAG-hNANOG expression plasmid was generated by transferring full-length hNANOG fragment from pCDNA-hNANOG-V5 via restriction digestion followed by ligation into p3xFLAG-CMV-10 vector (Sigma). Reverse transcription was conducted using SuperScript III First-Strand Synthesis SuperMix kit from Invitrogen. PCR was performed using Phusion high-fidelity DNA polymerase from New England Biolabs (Ipswich, MA, http://www.neb.com). All constructs were verified by DNA sequencing.

Immunofluorescent Staining

CV1 monkey kidney fibroblasts (ATCC, American Type Culture Collection, Manassas, VA, http://www.atcc.org) grown overnight on four-well chamber slides (R&D Systems Inc., Minneapolis, MN, http://www.rndsystems.com) were transfected with V5-tagged expression plasmids using FuGENE HD (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com) reagent according to manufacturer's protocol. After 48 hours, cell were fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton-X and 0.05% Tween-20, then blocked with 10% normal donkey serum and 1% bovine serum albumin in PBS for 45 minutes. Primary antibody incubation was done using monoclonal anti-V5 antibody (1:500; Invitrogen) for 2 hours, followed by 1-hour incubation with Alexa Fluor 488-conjugated donkey anti-mouse IgG (1:1,000), and preserved with VECTASHIELD Mounting Medium with 4′,6-diamidino-2-phenylindole nucleus counterstain (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Cells overexpressing EGFP-fused constructs were fixed but not stained with antibodies. Images were taken using either Leica DM RXA2 epifluorescence microscope or Zeiss LSM 510 confocal microscope.

Luciferase Reporter Assays

CV1 cells were grown overnight in Optilux 96-well plates (BD Biosciences, San Jose, CA, http://www.bdbioscience.com). Transfection of plasmids into CV1 cells were performed using FuGENE HD (Roche Diagnostics) reagent according to manufacturer's recommendation. A series of luciferase assays were performed by transient transfection in combination of CMV-driven expression constructs (10 ng/well), luciferase reporter (1 ng/well), and empty expression vector pCDNA3.1-V5 (10 ng/well) to a balanced total of 21 ng of plasmids per well of the 96-well plate. Cotransfection experiments were performed in triplicate and repeated at least three times. Luciferase assays were performed following the protocol of Bright-Glo Luciferase Assay System (Promega).

Coimmunoprecipitation and Western Blot Analysis

Human embryonic kidney 293A cells seeded overnight were transiently cotransfected with FLAG-tagged and V5-tagged expression constructs using FuGENE HD (Roche Diagnostics) reagent, following manufacturer's recommendation. After 48 hours of incubation, nuclear extracts were harvested using the NE-Per kit (Pierce, Rockford, IL, http://www.piercenet.com), according to manufacturer's protocols. To perform coimmunoprecipitation (co-IP), nuclear lysates were precleared with ImmunoPure Protein A/G agarose resin (100 μl/ml, Pierce) for 1 hour, with rotation in the cold room. Approximately 200 μg of nuclear extract, diluted to 1 ml in IP binding buffer (25 mM Tris, 0.15 M NaCl, pH 7.2, Pierce), was incubated with primary antibodies against epitope-tagged proteins (anti-FLAG, 5 μg/ml; anti-V5, 10 μg/ml) or isotype-specific control IgGs (10 μg/ml), with rotation in the cold room overnight. Protein A/G agarose resin was then added (50 μl/ml) for secondary incubation, rotating in the cold room for 1 hour. Beads were precipitated by centrifuge at 2,500 rpm for 5 minutes at 4°C and washed with IP binding buffer extensively. Coimmunoprecipitated proteins were resolved on 4-20% SDS-polyacrylamide gel, followed by Western immunoblot analysis (anti-FLAG, 1:10,000 and anti-V5, 1:5,000).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Generation of hNANOG Deletion-Mutant Expression Constructs

As shown in Figure 1A (panel 1), human NANOG encodes a protein of 305 amino acid residues which can be generally divided into three regions: N-terminal (N, aa 1-94), DNA-binding homeodomain (H, aa 95-154), and C-terminal (C, aa 155-305). The C-terminal region can further be subdivided into a tryptophan-rich domain (W, aa 196-240) and the regions upstream (C1, aa 155-195) or downstream (C2, aa 241-305) of W. Using an expression plasmid with the full-length hNANOG cDNA as a template, we generated a series of 17 deletion-mutant expression constructs with V5-tags. The first set of deletion mutants had the full-length hNANOG with each specific protein domain removed individually (ΔN, ΔH, ΔC1, ΔW, ΔC2, and ΔC; Fig. 1A, panels 2-7). The second set was for each specific domain alone (N, H, C1, W, C2, and C; Fig. 1A, panels 8-12). The third series spanned regions of hNANOG with several adjacent domains (NHC1, WC2, HC1, HC1W, and C1W; Fig. 1A, panels 13-18). This repertoire of expression constructs allowed us to systematically examine the function of each individual protein domain within the context of the protein.

In the first experiments, V5-tagged expression vectors containing either the full-length hNANOG or deletion mutants were transfected into CV1 fibroblasts. In this overexpression system, full-length hNANOG, homeodomain (H) alone, and any mutant with homeodomain linked to it (ΔN, ΔC1, ΔW, ΔC2, ΔC, NHC1, HC1, and HC1W) were found localized exclusively to the nucleus (Fig. 1B, panels 1, 2, 4, 5, 6, 7, 9, 13, 16, and 17), suggesting the presence of a NLS in H. Two mutants, W and C1W, were specifically localized to the cytoplasmic compartment and excluded from the nucleus (Fig. 1B, panels 11 and 18), suggesting that there was a nuclear export signal (NES) within the W region. Other mutants displayed a pan-cellular distribution pattern (ΔH, C1, C2, C, and WC2; Fig. 1B, panels 3, 10, 12, 14, and 15, respectively), indicating the absence of a NLS or NES in these mutants.

Identification of the NLS Sequence in the hNANOG Homeodomain

To function as a gene transactivator, hNANOG must be transported to the nucleus where it binds to its target elements, as shown by the nuclear localization of the wild-type NANOG (Fig. 1B, panel 1). However, the mechanism by which hNANOG is transported to nuclei was undefined. As the construct that contained only the homeobox domain (H) of hNANOG was found to localize solely to the nucleus (Fig. 1B, panel 9), and removal of this DNA-binding segments (ΔH) resulted in a loss of its exclusive nuclear localization (Fig. 1B, panel 3), we hypothesized that a putative NLS was located in H.

The best characterized nuclear targeting sequence is the classic NLS for nuclear protein import, which consists of a stretch of basic amino acid residues with a loose consensus sequence of K(K/R)X(K/R) (reviewed in [34]). Using comparative genomics to scan for conserved stretches of lysine rich residues within the homeobox domain of NANOG from six-mammalian species, three-NLS candidates were identified (K9597: 93VKKQKT98; K13740: 136YKQVKT141; and K1495155: 148MKSKRWQKN156; Figure 2A). These candidate sequences were subcloned as in-frame C-terminal fusions to the EGFP in an expression vector. In transfected CV1 cells, overexpressed EGFP-K9597 or EGFP-K1495155 fusion proteins were localized to both the nuclear and cytoplasmic compartments (Fig. 2B, panels 1 and 2), similar to the subcellular distribution pattern of EGFP alone (Fig. 2B, panel 5). In contrast, EGFP fused to the six amino acid residues, YKQVKT (K13740), localized exclusively in the nucleus (Fig. 2B, panel 3), comparable with nuclear localization seen in EGFP fused with full-length hNANOG (Fig. 2B, panel 4). This demonstrates that this six amino acid peptide alone is sufficient to direct nuclear transport of EGFP.

thumbnail image

Figure 2. Identification of an nuclear localization signal (NLS) in human NANOG. (A): Alignment of the hNANOG homeodomain with murine, canine, bovine, and porcine orthologs. Locations of three highly conserved, lysine-rich NLS candidate sequences are underlined and highlighted: K9597 (VKKQKT), aa 93-98; K13740 (YKQVKT), aa 136-141; and K1495155 (MKSKRWQKN), aa 148-156. The alignment was generated with Clustal W: (*), identical; (:), strongly similar; and (.), weakly similar. (B): CV1 cells were transiently transfected with EGFP expression constructs fused to sequences encoding VKKQKT (K9597, panel 1), YKQVKT (K13740, panel 2), MKSKRWQKN (K1495155, panel 3), full-length hNANOG (panel 4), or EGFP alone (panel 5). Nuclei were stained with DAPI (shown in blue). (C): CV1 fibroblasts overexpressing V5-tagged mutant hNANOG constructs were immunostained with antibodies against V5 (shown in green). The mutants carried double point-mutations (K9597A and K13740A, panels 1 and 3, respectively), triple point-mutations (K1495155A, panel 2), or single point-mutations (Y136A, K137A, Q138A, V139A, K140A, and T141A, panels 4-9, respectively). Lysine residues in each of the constructs were specifically altered to become alanine residues. Nuclei were stained with DAPI (shown in blue). Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; EGFP, enhanced green fluorescent protein; hNANOG, human NANOG.

Download figure to PowerPoint

To confirm that this putative NLS is necessary to target hNANOG to the nucleus, we generated site directed point mutations by replacing each basic lysine residue with a neutral alanine. The hNANOG double-point mutant K9597A (Fig. 2C, panel 1), in which 95K and 97K were substituted with alanine residues, had a nuclear localization comparable with that of the wild-type hNANOG (Fig. 1B, panel 1). Similarly, the K1495155 triple-point mutant (Fig. 2C, panel 2), where 149K, 151K, and 155K were all changed to alanines, also showed an exclusive nuclear distribution. This demonstrated that lysine residues 95K, 97K, 149K, 151K, and 155K were not critical for nuclear localization. However, when lysine residues 137K and 140K were both mutated, the hNANOG double-point mutant K13740A did not retain its nuclear exclusive distribution and had a pan-cellular localization pattern (Fig. 2C, panel 3). Moreover, single alanine substitution of either lysine residue at either position 137 (K137A) or position 140 (K140A) also resulted in pan-cellular distribution (Fig. 2C, panels 5 and 8, respectively). In contrast, we did not observe any alteration in the subcellular localization pattern of hNANOG when tyrosine at position 136 (Y136A), glutamine at position 138 (Q138A), valine at position 139 (V139A), or threonine at position 141 (T141A) were replaced with alanine (Fig. 2C, panels 4, 6, and 7, respectively), indicating that these residues were not critical for nuclear localization. In summary, a stretch of six amino acid residues (YKQVKT, K13740) was identified that is sufficient to transport NANOG to the nucleus and, specifically, the two lysine residues (K137 and K140) are necessary and critical to maintain its nuclear distribution.

Identification of a NES in the Tryptophan-Rich Region

The studies presented in Figure 1 demonstrated that the hNANOG deletion mutant containing the W region alone localized exclusively to the cytoplasm (Fig. 1B, panel 11). The small size of this mutant (∼6 kDa) predicts that it should freely transport through the nuclear pore complex and have a pan-cellular distribution. Thus, we hypothesized that the W region contained a NES. Examination of the sequence in the W region did not reveal any similarity with known NES such as the CRM-1-dependent leucine-rich motif (Fig. 3A; [35]). To determine whether the W region used the CRM-1 nuclear transport pathway despite the lack of the leucine-rich motif, we expressed an EGFP-W fusion protein in CV1 cells and evaluated whether LMB, a specific inhibitor of the CRM-1 transport pathway [36], could eliminate the nuclear export (Fig. 3B, panels 1 and 2). EGFP-PKINES fusion protein containing the CRM-1-dependent NES of the protein kinase inhibitor α (LALKLAGLDI; [37]) was used as a control (Fig. 3B, panels 1 and 2). EGFP-W was found to localize exclusive to the cytoplasm despite the addition of LMB (10 ng/ml for 1 hour), confirming that the nuclear export was not CRM-1 mediated. The controls confirmed that LMB abolished CRM-1 nuclear export, as control EGFP-PKINES was altered to a pan-cellular distribution on treatment with LMB. This experiment indicated that the NES in the W region did not utilize the CRM-1 mediated mechanism for nuclear export, suggesting that the tryptophan-rich region has a novel uncharacterized NES.

thumbnail image

Figure 3. Identification of a nuclear export signal (NES) in human NANOG. (A): Alignment of the human NANOG tryptophan-rich region with murine, canine, bovine, and porcine orthologs. Location of the eight highly conserved, pentapeptide tryptophan repeats are highlighted green and numbered 1-8. The alignment was generated with Clustal W: (*), identical; (:), strongly similar; (.), weakly similar. (B): CV-1 cells were transiently transfected with EGFP fusion plasmids containing either the W-domain of hNANOG (EGPF-W, panels 1 and 2) or the NES of protein kinase inhibitor (EGFP-PKINES, panels 3 and 4). After 48 hours, cells were treated for 1-hour with (panels 2 and 4) or without (panels 1 and 3) the CRM-1 exportin-specific inhibitor, LMB (10 ng/ml). Nuclei were stained with DAPI (shown in blue). (C): Immunofluorescent staining of CV-1 fibroblasts overexpressing V5-tagged (shown in green) wild-type W-region of hNANOG (W, panel 1), mutant harboring eight W[RIGHTWARDS ARROW]A substitution (W1-8pm, panel 2), or mutant with alanine-substitutions either in the first four tryptophan residues (W1-4pm, panel 3) or the last four tryptophan residues (W5-8pm, panel 4). Nuclei were stained with DAPI (shown in blue). Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; EGFP, enhanced green fluorescent protein; LMB, leptomycin B.

Download figure to PowerPoint

The conservation of the unique pentapeptide tryptophan-repeat motif (WXXXX) in the W region across species suggested that this sequence was important for NANOG structure, function, or possibly localization (Fig. 3A). Thus, we generated W mutants with either alanine substitution of all eight tryptophans (W1-8pm, Fig. 3C, panel 2), or the four N-terminal (W1-4pm, panel three) or the four C-terminal (W5-8pm, panel three) tryptophans altered. When we evaluated the cellular localization of these mutants on expression in CV1 cells, all three mutant proteins had a pan-cellular distribution, indicating that more than half of the tryptophan residues are required for nuclear export of the W domain.

hNANOG Self-Associates Through the Tryptophan-Rich Domain

Several homeodomain proteins have been shown to function as homodimers, with dimerization critical for their transcriptional regulation [38–41]. Recent studies have demonstrated that murine NANOG forms dimeric complexes [31, 42]. Furthermore, homodimerization is required for biological function of mNANOG [31]. To examine whether hNANOG proteins physically and specifically interact with each other, we cotransfected FLAG epitope-tagged full-length hNANOG expression plasmid with V5 epitope-tagged hNANOG expression constructs into 293A cells. Nuclear extracts were harvested and subjected to co-IP with an anti-FLAG antibody (Fig. 4). A reciprocal co-IP setup using anti-V5 antibody was conducted in parallel to confirm the results (supporting information Fig. 1). This study determined that hNANOG coassociates with itself (Fig. 4, panels 1). hNANOG mutants with either a N-terminal truncation (ΔN, panel 2) or homeodomain removal (ΔH, panel 3) continued to associate with the full length hNANOG, suggesting that these regions were not involved in dimerization. In contrast, hNANOG deletion mutants lacking the C-terminal domain (ΔC, panel 4) and, specifically, the tryptophan-rich region (ΔW, panel 6) failed to interact with the full-length hNANOG, suggesting that the tryptophan rich region is involved in homodimerization. More importantly, we determined that the tryptophan-repeat subdomain alone was sufficient to coassociate with full-length hNANOG (panel 9). This study confirms that, similar to mNANOG, hNANOG dimerizes through its W region.

thumbnail image

Figure 4. Tryptophan-rich region is critical in NANOG homodimerization. Human 293A cells were transiently cotransfected with p3xFLAG-hNANOG and V5-tagged expression constructs as indicated. Nuclear lysates were subjected to coimmunoprecipitation assays using either anti-FLAG antibody (IP) or an isotype matched IgG negative control (IgG, −). Immune complexes or 10% of input nuclear extracts as a control (NE, +) were separated by 4-20% SDS-PAGE gels and immunoblotted (WB) with antibodies against V5 or FLAG epitope tag. Abbreviations: IP, immunoprecipitation; NE, nuclear extracts; WB, Western blot.

Download figure to PowerPoint

Human NANOG is a Bivalent Transcription Regulator

To analyze the transcriptional activity of hNANOG, we performed transient cotransfection experiments with mutant hNANOG expression constructs and a murine Oct4 gene promoter-luciferase reporter construct (mOct4pr-Luc, Fig. 5A) in CV1 cells. The promoter/enhancer region in this reporter construct is well-characterized and contains three conserved NANOG binding sites [21]. Cotransfection of the full-length hNANOG and the mOct4pr-Luc reporter resulted in approximately fivefold increase in activation of the mOct4 promoter above the basal level (Fig. 5B, lanes 2 and 1, respectively). None of the mutants lacking the DNA-binding homeodomain (ΔH, C, WC2, C1W; lanes 4, 10, 11, and 14, respectively) activated transcription of the reporter gene. This is not surprising because, to function properly, a nuclear transcription factor, such as NANOG, requires binding to its cognate DNA elements. The C-terminal region of hNANOG was also critical for the transactivation of mOct4pr. As shown in Figure 5B, truncation of the individual C-terminal subdomains alone (ΔC1, ΔW, and ΔC2) or in combination (NHC1 and ΔC) resulted in diminished reporter activity (Fig. 5B, lanes 5, 6, 7, 9, and 8, respectively). This demonstrates that each of the C-terminal subdomains (C1, W, and C2) contributes to the transcriptional activity of hNANOG.

thumbnail image

Figure 5. N-terminal region of hNANOG (aa 1-95) possesses gene interference function, whereas the transactivation domain is located within C-terminus (aa 156-305). (A): Schematic illustration of mOct4pr-Luc reporter construct. A 3-kb DNA fragment 5′ upstream of mouse Oct4 gene, containing three putative NANOG binding sites, was cloned into pGL3-Basic luciferase reporter. (B): CV1 fibroblasts were transiently cotransfected with mOct4pr-Luc reporter and cytomegalovirus (CMV)-driven NANOG mutant expression constructs as indicated (lanes 2-14). Luciferase activities were normalized to baseline reporter gene activity in presence of empty backbone pcDNA control vector (lane one) as fold activation, with error bars representing SEM. (C): CV1 fibroblasts were transiently cotransfected with the mOct4pr-Luc reporter, the full-length hNANOG and a CMV-driven NANOG deletion-mutant expression constructs as indicated (lanes 1-14). Luciferase activities were normalized to baseline reporter gene activity in presence of empty back-bone pcDNA control vector (lane 15) as fold-activation on, with error bars representing SEM. Abbreviations: hNANAOG, human NANOG.

Download figure to PowerPoint

In contrast, elimination of the N-terminal region (ΔN) of hNANOG greatly enhanced the transactivation of the mOct4pr by 18-fold (Fig. 5B, lane 3), suggesting that the N-terminal domain may interfere with gene activation. Furthermore deletion of either the C2 or WC2 regions from the superactive hNANOG mutant ΔN, resulted in an incremental decrease in reporter gene expression (HC1W and HC1, Fig. 5B, lanes 13 and 12, respectively). This data demonstrate that all of hNANOG's C-terminal subdomains contribute to transactivation of the mOct4pr, and that even though deletion of the N region results in an overall higher level of transactivation, the C-terminus is still necessary for hNANOG's transactivation potential. Similar transactivation trends were observed when we tested with other hNANOG target gene reporter constructs such as mNanogpr-Luc and mRex1pr-Luc (supporting information Fig. 2).

We next evaluated whether the nine hNANOG deletion mutants identified earlier with reduced reporter gene activity were due to null mutation or interference with gene activities. In these experiments, we cotransfected each of the deletion mutants with the full-length hNANOG cDNA and mOct4pr-Luc reporter constructs. Addition of twice the amount of full-length hNANOG expression plasmid produced approximately ninefold induction of the reporter activity, two times more active than a single unit of hNANOG at approximately fourfold above the basal level (Fig. 5C, lanes 2 and 1, respectively). Expression of six deletion-mutant hNANOG genes did not affect expression from the cotransfected full length hNANOG, indicating that they were null mutants (ΔH, ΔC1, ΔW, C, WC2, and C1W; Fig. 5C, lanes 4, 5, 6, 10, 11, and 14, compared with lane 1). In contrast, three deletion mutants (ΔC2, ΔC, and NHC1) resulted in a suppression of mOct4-Luc activity, compared with expression of the full-length hNANOG alone, demonstrating an interference with normal NANOG gene transactivation (Fig. 5C, lanes 7, 8, and 9, compared with lane 1). These three interference mutants (ΔC2, ΔC, and NHC1) carried the DNA-binding homeodomain suggesting that they may have competed with the wild-type hNANOG for binding to DNA targets. Moreover, the presence of N-terminal region in each of these deletion-mutant genes further supports the interference role of the N-region of hNANOG. Taken together, this data demonstrate that hNANOG is a bivalent protein with the N-terminal region containing a transcriptional interference domain and the C-terminal region with a transcriptional activator.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

NANOG is a key core regulator of the transcriptome in PSC; however, the molecular mechanism by which NANOG regulates transcription is unknown [43]. Furthermore, the few studies have focused on murine NANOG with few reports on human NANOG structure and function. Given the differences between the pathways maintaining human and murine PSC, study of human NANOG is critical for understanding the molecular basis of NANOG function in human PSC. In these studies, we performed a detailed structure–function analysis of the hNANOG protein to gain insight into the molecular mechanism by which NANOG regulates self-renewal [44]. Our studies have identified several previously unrecognized protein domains in murine or human NANOG, as well as confirmed or further elucidated domains, motifs, and functions previously identified or predicted for NANOG.

Transcription factors must localize to the nucleus to actively regulate transcription. It has previously been determined that hNANOG has a NLS in the homeodomain that is required for nuclear localization [32]. We confirmed these studies and further identified the specific lysine residues that are necessary and sufficient for the nuclear localization of human NANOG. Surprisingly, we also identified that hNANOG contains a NES motif in the tryptophan rich (W) region in the C-terminus of the protein. Approximately half of the proteins that are exported from the nucleus use the CRM-1 transport pathway and the others utilize some other undefined transport pathway (reviewed in [45]). The NES identified in the tryptophan-rich region is CRM-1 independent and thus utilizes some other shuttling mechanism for nuclear export. Although the mechanism of nuclear export is unclear, we demonstrated that mutagenesis of half of the tryptophans in this region abolished nuclear export. Thus, it is possible that the specific protein structure of this domain and/or its interactions with unknown shuttling proteins are required for nuclear export. Future studies are required to determine the shuttling mechanism of this region and whether there biological consequences of disrupting this shuttling behavior.

Recent studies have demonstrated that murine NANOG homodimerizes through the tryptophan rich region. Homodimerization is required for its role in regulating self-renewal of murine PSC [30, 31]. Our studies demonstrated that human NANOG also homodimerizes through the tryptophan rich region. The amino acid conservation between the murine and human proteins in this W region suggests that homodimerization will also be required for human NANOG function although future studies will be required to conclude this.

As a transcription factor, NANOG binds to promoters and regulates transcription. ChIP-on-chip studies have demonstrated that NANOG binds to promoters that are active and others that are inactive, during self-renewal, suggesting a bivalent activity [19–24]. Using a heterologous GAL4 binding assay, several studies have demonstrated that the C-terminal region contains the majority of the transcriptional activation for NANOG [28, 29]. It was also shown that there was a low level of transactivation in the N-terminal region of murine NANOG but not human [28, 33]. However, these studies evaluated the transactivation of the NANOG domains using a heterologous GAL4 DNA binding domain, rather than within the context of the NANOG protein and the native homeodomain. We chose to evaluate the transcriptional activity of hNANOG protein domains on a genomic regulatory region which contains NANOG binding sites-the Oct4 promoter/enhancer. In these studies, we evaluated the in vitro transcriptional activity of a panel of hNANOG mutants with each protein domain alone and deletion of each protein domain singly or in combination from the full-length protein. This analysis confirmed that the transcriptional activation of hNANOG is in the C-terminal region. Our studies also identified a novel transcriptional interference domain in the N-terminus. This interference domain is trans-dominant as it inhibited the transcriptional activity of the full-length hNANOG protein. However, our studies evaluated the transcriptional activity of human NANOG using an in vitro system and it is unclear what the biological effects of expressing the N-terminal deletion mutant in human PSC. Our data suggest that the C-terminus recruits transcriptional coactivators and the N-terminus recruits transcriptional corepressors which ultimately determines whether a promoter is active or not in PSC.

Recent studies demonstrated that one of the earliest steps in the cell fate decision to differentiate may be the upregulation of caspases which in turn cleave NANOG and other factors [46]. Interestingly, the caspase cleavage site in NANOG is at the N-terminal domain and generates a partial N-terminal deletion of NANOG. The predicted caspase-3 cleavage site of hNANOG generates a truncated protein similar to our N-terminal deletion mutant (ΔN) that increased the transactivation of mOct4 promoter. These results, however, are not mutually exclusive. Our study determined the transcriptional activity of NANOG in a simplified system on one promoter to identify the specific protein domains in fibroblasts without caspase activation. In contrast, the study by Fujita et al. evaluated the biological response of differentiating murine PSC to caspase activation, which cleaves NANOG and a variety of other cellular proteins and transcription factors [46, 47]. Future studies will be required to determine whether the N-terminal deletion NANOG has biological function in PSC.

The studies reported here provide detailed structure-function information on the human NANOG protein. Figure 6A shows a summary of the protein domains, sequences, and functions in the human NANOG protein identified in our study, which together provide insight into the molecular mechanism of hNANOG regulated transcription in human PSC. We propose a model which incorporates both the known information on the promoters and proteins with which NANOG interacts with our data of the specific protein domains in the NANOG protein, as outlined in Figure 6B. In brief, NANOG occupies a series of promoters which are either transcriptionally active or repressed during PSC self-renewal [19–24]. NANOG has also been shown to interact with transcriptional corepressors such as the polycomb group of proteins and Sp1 or other coactivators [24, 25,48–50]. Based on the published data on NANOG with our characterization of NANOG protein structure, we propose that NANOG differentially recruits transcriptional coactivators or corepressors to each of these loci through interactions with its C-terminus and N-terminus, respectively. In turn, recruitment of these transcriptional coregulators ultimately determines whether each promoter is transcriptionally active or repressed and thereby regulates the global transcriptome in self-renewing PSC.

thumbnail image

Figure 6. Summary of human NANOG protein domains and model of its molecular function. (A): Summary of functional and structural domains in human NANOG protein identified in this study. (B): Model of the interactions of NANOG protein domains at active and repressed gene promoters with transcriptional coactivators and corepressors. Abbreviations: NES, nuclear export signal; NLS, nuclear localization signal.

Download figure to PowerPoint

In summary, our detailed analyses provide valuable insight into the structure and function of protein motifs in human NANOG. This information may lead to further understanding of the mechanism by which NANOG regulates transcription in PSC, by recruiting corepressors and coactivators to promoters through interactions with the N- and C-terminal domains, respectively.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

The authors wish to thank Drs. Donald B. Kohn, Gay M. Crooks, Paula Cannon, and Elizabeth Lawlor for thoughtful discussions on these studies. This study was supported by funds from the Saban Research Institute and Division of Research Immunology and Bone Marrow Transplantation at Childrens Hospital Los Angeles. S.C.T. was supported by a fellowship from the Saban Research Institute at Children Hospital Los Angeles. D.F.C. was supported on a training grant to CHLA from the California Institute of Regenerative Medicine (T2-00005).

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Additional supporting information available online.

FilenameFormatSizeDescription
STEM_20080657_sm_suppinfofigure1.tif2069KSupporting Information Figure 1. Tryptophan-rich region is critical in NANOG homodimerization. Human 293A cells were transiently co-transfected with p3xFLAG-hNANOG and V5-tagged expression constructs as indicated. Nuclear lysates were subjected to co-immunoprecipitation assays using either anti-V5 antibody (IP) or an isotype matched IgG negative control (IgG, -). Immune complexes or 10% of input nuclear extracts as a control (NE, +) were separated by 4-20% SDS-PAGE gels and immunoblotted (WB) with antibodies against V5 or FLAG epitope tag.
STEM_20080657_sm_suppinfofigure2.tif721KSupporting Information Figure 2. Human NANOG displayed similar target gene transactivation in mouse Nanog and Rex1 gene reporter assays. A.CV1 fibroblasts were transiently co-transfected with the mNanogpr-Luc reporter and CMV-driven NANOG expression constructs (full-length and deletion mutants) as indicated (lane 1 to 5). Luciferase activities were normalized to baseline reporter gene activity in presence of empty back-bone pcDNA control vector (lane 15) as fold-activation on, with error bars representing SEM. Right panel is a schematic illustration of mNanogpr-Luc reporter construct. A 406-bp DNA fragment 5′ upstream of mouse Nanog gene, containing 1 putative NANOG binding sites, was cloned into pGL3-Basic luciferase reporter. B. CV1 fibroblasts were transiently co-transfected with the mRex1pr-Luc reporter and CMV-driven NANOG expression constructs (full-length and deletion mutants) as indicated (lane 1 to 5). Luciferase activities were normalized to baseline reporter gene activity in presence of empty back-bone pcDNA control vector (lane 15) as fold-activation, with error bars representing SEM. A 500-bp DNA fragment 5′ upstream of mouse Rex1 gene, containing 1 putative NANOG binding sites, was cloned into pGL3-Basic luciferase reporter.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.