The defining activity of the homeodomain protein Nanog is the ability to confer cytokine-independent self-renewal upon ES (embryonic stem) cells in which it is overexpressed. However, the biochemical basis by which Nanog achieves this function remains unknown. In the present study, we show that Nanog dimerizes through a functionally critical domain. Co-immunoprecipitation of Nanog molecules tagged with distinct epitopes demonstrates that Nanog self-associates through a region in which every fifth residue is tryptophan. In vitro binding experiments establish that this region participates directly in self-association. Moreover, analytical ultracentrifugation indicates that, in solution, Nanog is in equilibrium between monomeric and dimeric forms with a Kd of 3 μM. The functional importance of Nanog dimerization is established by ES cell colony-forming assays in which deletion of the tryptophan-repeat region eliminates the capacity of Nanog to direct LIF (leukaemia inhibitory factor)-independent self-renewal.
ES (embryonic stem) cells present an interesting paradox because they can both self-renew and differentiate into derivative cells of all three primary germ layers . Indeed, the simultaneous possession of these properties defines ES cells and promises to make them useful for regenerative therapies. It is therefore important that the mechanisms directing optimal ES cell self-renewal are elucidated fully to allow rapid expansion of pluripotent cells.
Nanog was identified on the basis of its functional capacity, when overexpressed, to enhance ES cell self-renewal efficiency to the point of cytokine-independence [2,3]. Conversely, deletion of Nanog from ES cells co-cultured with fibroblasts produced Nanog-null colonies that expanded into primitive endodermal cell lines . Subsequent studies also found an increase in the propensity of ES cells to differentiate upon knockdown of Nanog [4–6]. However, our recent studies show that loss of Nanog expression is not inextricably linked to commitment to differentiation and that cells lacking detectable Nanog expression can revert their Nanog expression status . Moreover, Nanog−/− ES cells can be established and retain pluripotency . Nanog−/− ES cells do have an increased propensity to differentiate, and the stepwise reduction in the number of wholly undifferentiated colonies formed by Nanog+/+, Nanog+/− and Nanog−/− ES cells underscores the quantitative effect of Nanog upon self-renewal efficiency . Altogether, these data suggest that, rather than being essential for self-renewal, Nanog regulates ES cell self-renewal efficiency by acting as a rheostat to confer a variable resistance to differentiation. However, the biochemical basis of these observations remains unclear.
The 305-residue Nanog protein contains a typically folded HD (homeodomain) (residues 96–155) . Outwith the HD, the most notable sequence encompasses residues 198–243 in which every fifth residue is tryptophan [the WR (tryptophan repeat) region] [9,10]. In the present paper, we report a recombinant expression system that has allowed us to begin biophysical characterization of Nanog. Together with ES cell studies, this approach has identified the WR region as a functionally critical dimerization domain.
Full-length Nanog and Nanog1−160 were expressed in pET15b. To produce MBP (maltose-binding protein), the ORF (open reading frame) in pMALC2E (NEB) was truncated by mutagenesis immediately after the enterokinase cleavage site. To produce MBP–WR, codons for Ser192–Gly252 of Nanog were amplified and cloned into pMALC2E. Trimerized HA (haemagglutinin) or FLAG tags followed by a glycine spacer were added at the 5′-end of ORFs immediately upstream of the initiator methionine codon and cloned into the episomal vector pPyCAGIP . Truncation constructs were prepared by introducing stop codons following Trp159 (NanogΔC) or Leu256 (NanogΔC49). In Nanog-C, codons 1–155 were deleted from (HA)3Nanog. In NanogΔWR, residues Leu188–Leu248 were replaced with a linker encoding Ala–Ser. All constructs were checked by sequencing, and details are available from I.C. on request.
Expression and purification
Full-length Nanog and Nanog1−160 were expressed in Escherichia coli BL21(DE3)pLysS and MBP constructs in E. coli BL21. Bacteria were grown to a D600 of ∼0.6 before induction with 1 mM IPTG (isopropyl β-D-thiogalactoside) at 30 °C for 3 h. Nanog1−160 was purified from soluble lysate using a 1 ml His-trap column (GE Healthcare). Full-length Nanog was solubilized from inclusion bodies using 25 mM Hepes (pH 7.4), 300 mM NaCl, 8 M urea and 10 mM 2-mercaptoethanol, and loaded on to a nickel–Sepharose column equilibrated in the same buffer. Protein was refolded by removal of urea over 40 column vol. and eluted with 250 mM imidazole. For further manipulation, NP-40 (Nonidet P40) was added (to 0.025%, v/v). Detergent-free rNanog (recombinant Nanog) was prepared by removing NP-40 with Extractigel (Pierce). MBP proteins were purified from soluble lysates on amylose resin (MBP) or by a combination of nickel and amylose resin (MBP–WR).
AUC (analytical ultracentrifugation)
All studies were performed at 20±0.5 °C in a Beckman XL-A analytical ultracentrifuge (detailed protocols are available in the Supplementary Material at http://www.BiochemJ.org/bj/411/bj4110227add.htm).
For functional assessment of mutant Nanog molecules, ES cells were transfected and processed as described in . On day 12, cells were stained using a leucocyte alkaline phosphatase kit (Sigma). Transfection and processing for co-immunoprecipitation is described in the Supplementary Material.
Binding of rNanog to ES cell lysates
ES cells were transfected and lysates prepared at 24 h post-transfection as described above. Then, 5 μg of rNanog was added to each lysate and incubated at 4 °C overnight. Finally, 5 μg of anti-FLAG M2 antibody (Sigma) or anti-Nanog  was added and incubated for 3 h at 4 °C. Immune complexes were collected and analysed as described in the Supplementary Material.
Binding of rNanog to MBP-fusion proteins
MBP and MBP–WR were immobilized on a 0.2 ml amylose column pre-equilibrated in 25 mM Hepes (pH 7.4), 300 mM NaCl and 0.025% NP-40. Then, 5 μg of rNanog was loaded on to the column (in the same buffer) and incubated at 4 °C for 1 h before washing with 50 column vol. of the same buffer. Protein fractions eluted with 10 mM maltose were separated by SDS/PAGE (10% gels) and immunoblotted using an anti-Nanog antibody .
Expression of full-length Nanog
Attempts to express full-length Nanog in E. coli resulted in the accumulation of large amounts of insoluble material, with little soluble Nanog, under a wide range of conditions (variations in cell type, temperature, IPTG concentration, time of induction, etc.). An on-column refolding protocol was therefore developed (see the Experimental section). Soluble protein eluting from this column (rNanog) (yield of 100–300 μg/l of culture) was sufficiently pure for further studies (Figure 1B). Under approximately physiological conditions (pH 7.4, 200–300 mM NaCl), rNanog was difficult to handle with losses accompanying further manipulation, e.g. dialysis or buffer exchange. Similar observations were made with the small amounts of soluble Nanog obtained both from E. coli and from expression in insect cells, indicating that this behaviour is a feature of the protein rather than the production protocol. A low concentration of detergent (0.025% NP-40) stabilized rNanog.
Hydrodynamic properties of Nanog
rNanog has biological activity
To investigate whether rNanog adopts a native conformation, binding to known Nanog targets was assessed. A target site identified using SELEX (systematic evolution of ligands by exponential enrichment)  was used to examine DNA binding of rNanog. A mobility shift was observed that could be specifically competed with unlabelled oligonucleotide (Supplementary Figure 1A at http://www.BiochemJ.org/bj/411/bj4110227add.htm). We next investigated binding of rNanog to known protein partners, Sall4 and SMAD1 [10,13]. Addition of rNanog to lysate from ES cells expressing (FLAG)3SMAD1-MH2 allowed co-immunoprecipitation of SMAD1-MH2 and Nanog (Supplementary Figure 1B). Similarly, when rNanog was incubated with lysate from ES cells expressing (FLAG)3Sall4, Nanog and Sall4 could be co-immunoprecipitated (Supplementary Figure 1C). Together, these data indicate that rNanog has a native structure in terms of its ability to interact with Sall4, SMAD1 and DNA.
Dimeric nature of Nanog
AUC was used to further characterize rNanog. Sedimentation velocity measurements indicated that rNanog is predominantly a single oligomeric species (>90%) with a sedimentation coefficient of 4.5 S (2.25 S in the presence of NP-40). This eliminates any concern that rNanog may contain a significant proportion of non-specifically aggregated material. A value of 4.5 S corresponds to a molecular mass of approx. 74 kDa, consistent with a dimeric form for Nanog.
By sedimentation equilibrium, excellent fits were obtained (Figures 1C–1E), giving estimates of Kd for Nanog dimerization of 12 μM (with NP-40 in ‘ordinary’ buffer), 60 μM (with NP-40 in H218O buffer) and 3.2 μM (without NP-40). These results suggest that, under all conditions studied, Nanog will be in equilibrium between dimer and monomer, with the dimer predominating.
Hydrodynamic behaviour of Nanog
To further characterize rNanog, SEC (size-exclusion chromatography) was performed. Rather than migrating in the expected position for a dimer, rNanog is present in early eluting fractions (Figure 1F). Interestingly, and consistent with previous observations , we found that Nanog is present in ES cell lysate fractions corresponding to apparent molecular masses ranging from approx. 2 MDa to below that expected for monomeric Nanog (35–40 kDa) (Figure 1F). This behaviour had previously been considered to indicate that Nanog interacted with a range of complexes of differing composition in ES cells. However, the SEC profile of purified rNanog (Figure 1F) implies that the SEC profile of Nanog in ES cell lysates is not due solely to interaction of Nanog with other proteins; rather, anomalous migration of Nanog in SEC is likely to be a consequence of a combination of Nanog dimerization and the possession of a flexible non-globular conformation.
To examine further the hydrodynamic properties of Nanog, we expressed Nanog1−160 in E. coli in a soluble form. Nanog1−160 extends from the N-terminus through the HD and binds DNA specifically according to EMSA (electrophoretic mobility-shift assay) results (not shown). SEC indicates that Nanog1−160 is monomeric with an apparent molecular mass close to its calculated mass (20.5 kDa) (Figure 1F). Thus the anomalous migration of Nanog on SEC is due to sequences C-terminal to residue 160.
Mapping of the Nanog dimerization domain
To determine the location of the Nanog dimerization domain, co-transfections were performed using two forms of Nanog tagged with different epitopes (Figure 1A) in COS cells (Figure 2) and ES cells (Supplementary Figure 2 at http://www.BiochemJ.org/bj/411/bj4110227add.htm). Co-immunoprecipitation of (HA)3Nanog with (FLAG)3Nanog indicated that this approach could detect Nanog dimerization and that deletion of the N-terminal region or the HD, either individually (Figure 2A and Supplementary Figure 2A) or together (Figure 2B) did not prevent co-immunoprecipitation. Thus residues 160–305 are sufficient to mediate Nanog dimerization. Deletions from the C-terminus resulted in a failure to co-immunoprecipitate (FLAG)3Nanog upon deletion of the entire 146-residue C-terminal domain, but not upon deletion of the C-terminal 20 or 49 residues (Figure 2A and Supplementary Figure 2A). This suggests that the dimerization domain is located between residues 159 and 256. To pinpoint further the dimerization domain, residues 188–248, encompassing the WR region, were deleted. The failure of (HA)3NanogΔWR to co-immunoprecipitate full-length (FLAG)3Nanog (Figure 2C and Supplementary Figure 2B) establishes that the WR region is required for Nanog dimerization. Moreover, the similarity in the results obtained in ES cells and COS cells indicates that ES-cell-specific proteins do not provide additional bridging interactions between Nanog monomers independently of the WR region.
The WR region is involved in Nanog–Nanog interactions
Analysis by SEC of ES cell lysates expressing the ΔWR mutant show that (HA)3NanogΔWR runs at a position substantially different from that of (HA)3Nanog and more consistent with its calculated monomeric molecular mass (Figure 1F).
Dimerization of Nanog is mediated directly through the WR region
To determine whether the WR region is sufficient to mediate direct dimerization, rNanog was passed over an amylose column to which the fusion protein MBP–WR or MBP alone had previously been immobilized. Bound material eluted with maltose and analysed by immunoblotting showed that rNanog bound specifically to immobilized MBP–WR (Figure 2D). This indicates that the WR region mediates direct Nanog dimerization.
The WR region is required to confer cytokine-independent self-renewal
Finally, transfection of E14/T ES cells with episomal DNAs  was used to examine the effect of deletion of the dimerization domain on Nanog function. Deletion of sequences C-terminal to the HD abolished cytokine-independent colony formation, whereas deletion of sequences C-terminal to the WR region did not (Figure 2E). Deletion of the WR region resulted in a failure to confer cytokine-independent colony formation (Figure 2E). Similar results were obtained from transfection of LIF (leukaemia inhibitory factor) receptor-null LRK1 ES cells, eliminating any concern that the functionality of (HA)3NanogΔC49 could be due to an indirect effect upon LIF signalling in the E14/T cells (Supplementary Figure 3 at http://www.BiochemJ.org/bj/411/bj4110227add.htm). Comparative immunoblotting and immunofluorescence indicated that functional impairment of (HA)3NanogΔWR was not due to reduced expression or mislocalization (Supplementary Figure 3). These results indicate that the dimerization domain is required for the biological activity of Nanog.
ES cell self-renewal efficiency is proportional to the expression level of Nanog [2,7]. The discovery that the Nanog WR region is a dimerization domain necessary for cytokine-independent ES cell self-renewal is an important step towards understanding how Nanog works. Previous biochemical studies of Nanog function suggested that the sequence C-terminal to the WR region [CD (C-terminal domain) 2] and, to a lesser extent, the WR region itself could both act as transactivation domains . Furthermore, recent analysis of ES cells carrying integrated Nanog transgenes has suggested a role for CD2 in ES cell self-renewal . This contrasts with our episomal expression analysis, performed in two ES cell lines, in which deletion of the C-terminal 49 residues encompassing CD2 did not eliminate the ability of Nanog to confer cytokine-independent ES cell self-renewal. This apparent discrepancy is likely to be related to the higher expression level achieved by episomal compared with integrative transgenic expression. Nevertheless, our results clearly indicate that it is the WR region dimerization domain rather than CD2 that has the major role in conferring cytokine-independent self-renewal, the defining biological activity of Nanog, on transfected ES cells (Figure 2).
Most non-plant HD proteins that oligomerize form homodimers (or heterodimers with other HD proteins) through interactions involving their HDs . Thus their cognate DNA sequences normally consist of two regions, each recognized by the HD of a different subunit, separated by just a short linker. In contrast, Nanog dimerization is mediated not by HD, but by the WR region, which is separated from the HD by 42 residues. Thus the cognate DNA sequence of Nanog may also consist of a pair of DNA-recognition sites, but these may be positioned further apart, especially if the WR–WR interaction were head-to-tail. Among the few non-plant HD proteins that dimerize using non-HD interactions are HNF1α (hepatocyte nuclear factor 1α)  and PRH (proline-rich homeodomain)/Hex . However, the influence of dimerization on target site selection in these cases is unknown.
Dimerization is likely to affect the interaction of Nanog with identified Nanog protein-binding partners [10,13,14,19], because binding surfaces available on monomers will be different from those on dimers. Therefore Nanog may interact with some partner proteins only when dimeric. In this context, it is notable that the interaction of HNF1α with the transcriptional coactivator DCoH (dimerization cofactor of HNF1α) requires contributions from both HNf1α subunits . It is also possible that interactions between Nanog and other binding partners are mediated through monomeric Nanog and that deleting the WR region destroys the ability of Nanog to form homodimers or heterodimers. In either case, the observations that NanogΔWR does not form a dimer and that it cannot confer cytokine-independent self-renewal suggests that dimerization is a key regulator of Nanog activity. In this regard, the AUC-based evidence that Nanog exists in equilibrium between monomeric and dimeric forms in vitro suggests an intuitively attractive mechanism by which Nanog function could be modulated. Covalent modification of Nanog might alter the steady state of the monomer–dimer equilibrium and so promote or suppress self-renewal activity. Further studies will be required to determine whether Nanog phosphorylation , or some additional covalent modification, could play such a regulatory role.
We thank Dr R. Poot (Rotterdam) for comments on the manuscript. This research was supported by the Biotechnology and Biological Sciences Research Council and the Medical Research Council of the U.K., and by the Wellcome Trust and Juvenile Diabetes Research Foundation.
- ES cell
embryonic stem cell
hepatocyte nuclear factor 1α
leukaemia inhibitory factor
open reading frame