The Stem Cell Pluripotency Factor NANOG Activates Transcription with Two Unusually Potent Subdomains at Its C Terminus*

Embryonic stem cells are pluripotent progenitors for virtually all cell types in our body and thus possess unlimited therapeutic potentials for regenerative medicine. NANOG, an NK-2 type homeodomain gene, has been proposed to play a key role in maintaining stem cell pluripotency presumably by regulating the expression of genes critical to stem cell renewal and differentiation. Here, we provide the evidence that NANOG behaves as a transcription activator with two unusually strong activation domains embedded in its C terminus. First, we identified these two transactivators by employing the Gal4-DNA binding domain fusion and reporter system and named them WR and CD2. Whereas CD2 contains no obvious structural motif, the WR or Trp repeat contains 10 pentapeptide repeats starting with a Trp in each unit. Substitution of Trp with Ala in each repeat completely abolished its activity, whereas mutations at the conserved Ser, Gln, and Asn had relatively minor or no effect on WR activity. We then validated the activities of WR and CD2 in NANOG by constructing a reporter plasmid bearing five NANOG binding sites. Deletion of both WR and CD2 from NANOG completely eliminated its transactivation function. Paradoxically, whereas the removal of CD2 reduced NANOG activity by ∼30–70%, the removal of WR not only did not diminish but actually enhanced its activity by ∼50–100% depending on the cell lines analyzed. These data suggest that either WR or CD2 is sufficient for NANOG to function as a transactivator.

Embryonic stem cells are pluripotent progenitors for virtually all cell types in our body and thus possess unlimited therapeutic potentials for regenerative medicine. NANOG, an NK-2 type homeodomain gene, has been proposed to play a key role in maintaining stem cell pluripotency presumably by regulating the expression of genes critical to stem cell renewal and differentiation. Here, we provide the evidence that NANOG behaves as a transcription activator with two unusually strong activation domains embedded in its C terminus. First, we identified these two transactivators by employing the Gal4-DNA binding domain fusion and reporter system and named them WR and CD2. Whereas CD2 contains no obvious structural motif, the WR or Trp repeat contains 10 pentapeptide repeats starting with a Trp in each unit. Substitution of Trp with Ala in each repeat completely abolished its activity, whereas mutations at the conserved Ser, Gln, and Asn had relatively minor or no effect on WR activity. We then validated the activities of WR and CD2 in NANOG by constructing a reporter plasmid bearing five NANOG binding sites. Deletion of both WR and CD2 from NANOG completely eliminated its transactivation function. Paradoxically, whereas the removal of CD2 reduced NANOG activity by ϳ30 -70%, the removal of WR not only did not diminish but actually enhanced its activity by ϳ50 -100% depending on the cell lines analyzed. These data suggest that either WR or CD2 is sufficient for NANOG to function as a transactivator.
Embryonic stem cells hold the key to regenerative medicine, which offers an alternative to classic treatments based primarily on operations or drugs (1)(2)(3)(4). The regenerative potential of stem cells may provide the "fountain of youth" for our ever aging population and has sparked tremendous interest both socially and scientifically in recent years. Yet conceptual as well as technical hurdles recognized in recent years may set back any realistic application of stem cell biology and regenerative medicine for decades if not centuries. One such hurdle is how to maintain stem cells in pluripotent states in vitro and trigger their differentiation toward a specific lineage suitable for transplantation (1)(2)(3)(4). Investigation into the pluripotency of stem cells should provide the necessary tools to both control and utilize the regenerative potential of embryonic stem cells for therapeutic purposes (1).
Our understanding of stem cell pluripotency has been focused on a couple of transcription factors that regulate both positively and negatively distinct sets of genes for pluripotency and differentiation of stem cells. Oct4 is the first such factor and has been studied extensively over the past decade (1,3,5). Interestingly, a novel homeodomain protein, NANOG, was also demonstrated to play a key role in maintaining pluripotency for mouse embryonic stem cells (6,7) consistent with the notion that the pluripotency of embryonic stem cells is regulated at the transcription level by transcription factors (1,3,5). In addition, factors such as STAT3 have also been implicated in stem cell pluripotency (5,8,9). However, neither STAT3 nor Oct4 is sufficient to maintain stem cell pluripotency (10,11), suggesting that additional factors such as NANOG may play a more important role for embryonic stem (ES) 1 cell self-renewal (6,7,11).
The mechanism through which NANOG regulates stem cell pluripotency remains entirely unknown. Based on the differences in gene expression between wild type and NANOG null cells, it has been proposed that NANOG regulates pluripotency mainly as a transcription repressor for downstream genes such as gata4 or gata6 (6,7). However, NANOG may function more broadly than originally proposed. First, NANOG appears to function in parallel with STAT3 and be sufficient for maintaining stem cell pluripotency without gp130/STAT3 activation (6,7). Secondly, NANOG not only inhibits the differentiation of stem cells into endoderm but also actively maintains pluripotency, in contrast to the role of Oct4 as a blocker of differentiation of inner cell mass and ES cells into trophectoderm (1, 5-7, 10, 12). Consequently, NANOG has been proposed as the missing determinant of pluripotency for inner cell mass and ES cells (7). Because differentiation and self-renewal are likely to be regulated through the expression of mutually exclusive genes, NANOG may assume a bifunctional role to repress those genes important for differentiation and activate the ones necessary for self-renewal (13).
NANOG is a multidomain protein with a well conserved Nk-2 homeodomain (6,7,13,14). By fusing various domains of NANOG with the DNA binding domain of the yeast transcription factor Gal4, we have previously identified two transactivators in mouse NANOG, the N-terminal transactivation domain and the C-terminal transactivation domain or CD (13). We assume that the signature 60-residue homeodomain should be able to bind DNA and interact with other proteins as dem-onstrated for Oct4 (1,13). The N-terminal domain contains 95 residues rich in Ser and Thr and acidic residues found in typical transactivators (6,7,13,15). The CD is 150 residues long with no apparent transactivation motifs (13). It remains unclear whether NANOG is a transcription activator by itself and which domain is required for its transactivation activity. Here we demonstrate for the first time that NANOG can transactivate a reporter plasmid bearing its cognate binding site. Furthermore, we have dissected two unusually strong transactivation domains in its C terminus that are required for its transactivation activity and thus may contribute to the expression of downstream genes critical for maintaining stem cell pluripotency.
Transfections, Western Blotting, and Reporter Assay-For Western blotting analysis, HEK293T cells cultured in 12-well tissue culture plates were transfected by the calcium phosphate co-precipitation method with expression plasmids (1 g each) as described (13). Cells were harvested 48 h after transfection. Western blotting analyses were performed as described (13, 16). For reporter assay, cells seeded in a 24-well plate were transiently transfected with reporters such as p5G-e1b-luciferase or p5N-tk-luciferase (0.1 g each) and effector plasmids (0.5 g) using Lipofectamine (Invitrogen) according to the manufacture's instruction. pCMV-Renilla (0.005 g per transfection, Promega) was co-transfected into each well as an internal reference, and the DNA concentrations for all transfections were normalized to equal amounts by adding pCR3.1 empty vector. 36 h later, cells were washed by phosphate-buffered saline and lysed in 70 l of 1ϫ passive lysis buffer (Promega). Luciferase activity was measured using the dual-luciferase reporter assay system (Promega) and TD2020 Luminometer (Turner Design) as described (17). Each transfection was carried out in duplicate and repeated at least twice. For ESCs transfection, mouse embryonic fibroblasts were removed as described above, and the ESCs were seeded on a gelatin-coated 24-well tissue culture plate and transfected with plasmids in the same manner as described above using Lipofectamine 2000 (Invitrogen).

The C Terminus of NANOG Encodes Two Unusually Potent
Transactivators-Although NANOG was originally proposed to function primarily to repress the expression of genes such as gata4 and gata6, we have generated preliminary evidence that NANOG contains transactivation domains by fusing both Nand C-terminal domains to the conserved homeodomain to the DNA binding domain of Gal4 and demonstrating that both fusions are able to transactivate a luciferase reporter construct bearing five copies of the Gal4 binding sites (13). Interestingly, the C-terminal domain or CD of NANOG possesses transactivation activity at least six times as active as that of its Nterminal domain (13). The most prominent feature of the CD in mouse NANOG is the presence of 10 pentapeptide repeats starting with a tryptophan or W residue, thus named W-repeat or WR (6, 7, 13) (Fig. 1, A and B). We reasoned that the WR may contribute to the unusually high activity for the CD and subsequently divided the entire CD into CD1  , and CD2 248 -305 as illustrated in Fig. 1A. We fused these three subdomains with the Gal4 DNA binding domain individually or in combinations as shown in Fig. 1B. Upon transfection into HEK293 cells, these fusion constructs expressed proteins of expected size (Fig. 1C, lanes 2-7). To evaluate their transactivation potentials, these constructs in three doses were co-transfected with the reporter plasmid, p5G-e1b-luciferase, and their activities were quantified as described (17) and presented in Fig. 1D. These results confirmed our initial reasoning that the WR is indeed a functional structure. First, the WR itself is a potent transactivator capable of activating the reporter in a dose-dependent fashion up to ϳ900-fold (Fig. 1D, lanes 8 -10). Surprisingly, CD2, which is downstream of WR, encodes an even stronger transactivator that activates the report up to ϳ3000-fold, apparently stronger than the viral VP16, arguably the best transactivator known so far (Fig. 1D, lanes 11-13  versus lanes 20 -22). The CD1, on the other hand, appears to be inactive, like the Gal4 DBD expressed alone (Fig. 1D, lanes  2-7). Interestingly, the C1W and WC2 combinations yielded activity lower than either WR or CD2 (Fig. 1D, lanes 14 -19). Nonetheless, these results revealed two potent transactivators embedded in the C terminus of NANOG.
Both WR and CD2 Are Active Both in Pluripotent and Nonpluripotent Cells-The cells used in Fig. 1D are HEK293 cells, which were derived from human embryonic kidney and are not known to be pluripotent. To assess the activities of both WR and CD2 in pluripotent cells, we co-transfected these constructs with the reporter plasmid into mouse embryonic stem cells, P19 germ tumor cells, and mouse NIH3T3 cells, and the results were obtained and shown in Fig. 2. Both WR and CD2 can mediate the trans-activation of the reporter gene in all three cell lines including pluripotent cells (ES cells and P19 cells) and non-pluripotent cells (NIH3T3 cells), confirming their strong and universal activities for these two domains as observed in different cells. Nevertheless, both constructs are much more active in mouse ES cells than P19 and NIH3T3 cells (ϳ300 versus 50 versus 20 for WR and ϳ1600 versus 300 versus 200 for CD2) (Fig. 2, A-C). As observed in HEK293 cells, the C1W and WC2 combinations are not active as WR or CD2 alone, suggesting that C1 may negatively regulate WR and WR in turn may negatively impact CD2 function as well.
The Tryptophan Residues Are Required for Transactivation Activity of the WR-The defining feature of the C-terminal domain of NANOG is the WR, which contains 10 pentapeptide repeats starting with the tryptophan residue. To test the role of the Trp residues in the function of WR, we mutated half or all of the 10 Trps into Alas as shown in Fig. 3A. The mutant proteins migrated slightly higher than the wild type proteins perhaps reflecting changes in the shapes of the mutant molecule (Fig. 3B). Both mutants lost almost all of their transactivation activity in all three cell lines tested (Fig. 3, C-E, lanes 4  and 5 versus 3). It is of interest to note that the mutant (Trp/Ala) 5 has some residual activity (Fig. 3, C-E, lanes 4  versus 2), suggesting that the remaining five Trps can help maintain part of the WR structure, thus, preserving part of its transcription activity. Nevertheless, these data demonstrate  that the Trps in the WR are required for its transactivation activity. In addition to Trp residues, there are several potentially important residues such as the Ser residues, which may be phosphorylated, and Asn and Gln repeated alternately in every two repeats (Fig. 3A). These residues may also play important roles in WR function. To test these possibilities, we mutated Ser, Asn, or Gln into Ala in this domain and also fused them to Gal4 DBD as illustrated in Fig. 3A. Each mutation construct produced the same protein size as wild type WR did as shown in Fig 3B. The data from the reporter gene assay in three different cell lines demonstrated that virtually all three mutants have the same transactivation activities as the wild type WR (Fig. 3, C-E, lanes 6 -8 versus lane 3). Taken together, these data suggest that the Trp residues play a vital role in maintaining the activity of the WR domain, whereas other conserved residues do not.
NANOG Transactivates a Reporter Bearing NANOG Binding Sites-So far, we have demonstrated the transactivation potential of NANOG through the DNA binding sites of Gal4. To prove that NANOG is a transactivator by itself, we designed a reporter construct as shown in Fig. 4B. First, we synthesized two NANOG binding sites, one wild type (1N) and one mutant (1Nmu), based on the consensus binding site obtained by the SELEX procedure as reported by Mitsui et al. (7). We then confirmed that the wild type binding site can bind to NANOG protein in a gel shift assay as shown in Fig. 4A (lane 3). The binding is specific because the mutant did not bind, and the unlabeled binding site can compete for binding from the P 32labeled binding site (Fig. 4A, lanes 4 and 5 versus 3), whereas the unlabeled mutant did not (data not shown). The binding sites were cloned into a reporter containing the minimal thymidine kinase promoter (17) and obtained multiple reporter constructs bearing two, four, and five copies for the wild type site (p2N, p4N, and p5N, respectively) and three copies for the mutant site (p3Nm) as shown in Fig. 4B. To see if NANOG can transactivate these reporters specifically, we co-transfected NANOG with these reporters into P19 cells and present the results in Fig. 4C. As expected, NANOG activated the reporters bearing wild type binding sites but failed to do so toward the reporter bearing three mutant NANOG sites (Fig. 4C,  lanes 6, 8, and 10 versus 4). Interestingly, the copy numbers did not appear to influence the activity of transactivation for NANOG significantly (Fig. 4C, lanes 6 versus 8 versus 10). These results demonstrated that NANOG is a transcription activator capable of activating reporters bearing its binding sites.
Either the WR or CD2 Is Required for NANOG to Function as a Transcription Activator-To test the role of WR or CD2 The cell lysates were probed with anti-FLAG antibody and were developed as described (13). C, transcription activities for W-repeat and derivatives as listed in A in HEK293 cells. The same constructs (0.5 g per construct) were co-transfected with p5xGal4-e1b-luc reporter (0.1 g per transfection) and an internal reference in duplicate into HEK293T cells and the activities were measured and analyzed as described (13). D and E, the same transfections as described in C were carried out in P19 (D) and NIH3T3 cells (E). in mediating transactivation in the context of native NANOG protein, we constructed deletion mutants lacking both subdomains (N1) or lacking either of them (N3 and N4) as shown in Fig. 5B. All of these constructs expressed proteins with the expected sizes when transfected into HEK293 cells as shown in Fig. 5C (lanes 2-4). To evaluate the transcription activity of these deletion mutants, we transected them with either pTK or p5N (Fig. 5A) into P19 cells. As shown in Fig. 5D, N1 is consistently negative in HEK293, NIH3T3, P19, and F9 cells (lanes 3 versus 2), suggesting that the WR and CD2 are required for NANOG to function as a transactivator. On the other hand, N3 can activate the reporter but less efficiently at ϳ30% in HEK293 cells, 60% in NIH3T3 cells, and ϳ50% in both P19 and F9 cells compared with full-length NANOG (Fig. 4D, lanes 4 versus 2). These data suggest that CD2 is required for NANOG to express its full activity and WR is sufficient to maintain ϳ30 -70% of its activity. Surprisingly, N4 appears to have better activity than not only N3 (Fig. 5D,  lanes 5 versus 4) but also the wild type NANOG (Fig. 5D,  lanes 5 versus 2), suggesting that CD2 alone is sufficient to maintain potent activity in the absence of WR, which in fact may interfere with the activity of CD2 in the native configuration. These findings are consistent with the results obtained with the WC2 combination in Fig. 1D. Nevertheless, these data also validate the finding in Fig. 1D that CD2 is a more potent activator than WR. Among the four different cell lines tested, P19 is the best (70-fold), followed by F9 (50-fold), then HEK293 (15-fold) and NIH3T3 (10-fold) in supporting NANOG function. Because both P19 and F9 are considered pluripotent whereas HEK293 and NIH3T3 are not, these differences in activity for NANOG may reflect a degree of cell type specificity toward pluripotent cells. Thus, this NANOG reporter should be a useful tool in future analysis of NANOG function. Taken together, both WR and CD2 should play important roles in NANOG-mediated transactivation. DISCUSSION We report here that NANOG is a transcription activator capable of transactivating a reporter plasmid bearing its cognate binding sites. Furthermore, we also demonstrated that NANOG employs dual activation domains, i.e. WR and CD2, at its C terminus to mediate transcription activation. Therefore, these findings provide a mechanistic understanding of NANOG as a stem cell pluripotency factor. The multiple and redundant nature of transactivation domains in NANOG (like the ones in Oct4) pose considerable challenge in our future investigations into to their precise roles in maintaining stem cell pluripotency (1,5). However, a better understanding of NANOG and other similar transcription factors may help us design effective tools to control stem cell pluripotency pharmacologically and achieve therapeutically favorable end points for many degenerative diseases in the near future.
NANOG as a Transcription Activator-OCT4 is the first homeodomain protein known to regulate stem cell pluripotency by both repressing and activating distinct sets of downstream genes (1,18). However, when NANOG was discovered, it was proposed that it should act as a transcription repressor for proteins such as gata4 and gata6 (6,7). However, it has also been noted that NANOG differs from OCT4 in that NANOG can positively maintain the status of pluripotency whereas OCT4 cannot (6,7,11), perhaps by activating genes critical for the maintenance of pluripotency. Therefore, our findings in binding site (p3Nm) were co-transfected with the control vector (CK) or 0.5 g of NANOGF (N) plasmids as indicated to P19 cells in a 24-well plate. Internal reference and luciferase assay were as described in Fig. 1.   FIG. 4. NANOG mediates transactivation through its cognate binding site. A, electrophoretic mobility shift assay of oligos containing NANOG consensus binding site (1N) or mutant binding site (1Nmu) with nuclear extracts of 293T cells transfected with NanF. Nuclear extracts of 293T cells transfected with NanF were incubated with double stranded 1N or 1Nmu labeled with 32 P as indicated and analyzed by 5% non-denature polyacrylamide gel followed by autoradiography. End-labeled 1N were incubated with no nuclear extracts or nuclear extracts of 293T cells transfected with control vector as control (lanes 1 and 2). Specific binding was analyzed by competition with an ϳ100-fold excess of unlabeled 1N as indicated (lane 5). B, construction of reporters containing NANOG binding sites. Annealed oligos containing the NANOG binding site or mutant binding site were inserted upstream of a luciferase gene under the control of a minimal thymidine kinase promoter. The copy numbers of the inserted binding site were determined by sequencing. C, NANOG can transactivate the reporter gene containing the wild type NANOG binding site but not its mutant. 0.2 g reporters containing two copies (p2N), four copies (p4N), and five copies (p5N) of the NANOG binding site or three copies of the mutant this study provide direct evidence for this possibility, i.e. NANOG being capable of activating genes bearing its cognate binding sites. We are currently searching for promoters bearing the cognate binding sites and characterizing the effect of NANOG toward these promoters.
WR as a Novel Transactivator-The WR was recognized as a structural feature in NANOG yet without any function prescribed (6,7). Here, we provide the first evidence proving that it functions as a potent transactivator. Remarkably, its activity is dependent on the Trp residues as shown in Fig. 3. Data base searches failed to identify similar motifs in other proteins so far. Therefore, it might have been evolved specifically in NANOG to activate genes in stem cell self-renewal and pluripotency. Although WR functions well in multiple cell types, WR exhibits higher activity in mouse ES cells than P19 and NIH3T3 cells (Fig. 2). Structurally, we demonstrated that the tryptophan residues are absolutely required, perhaps to maintain a unique structure for WR, most likely of a repetitive nature. One may argue that the unique structure of WR may represent a specific regulatory pathway for NANOG to interact with the transcription machinery, which regulates the expression of genes critical to stem cell pluripotency. However, deletion of WR from native NANOG appears to suggest that WR is not required for NANOG to exert robust activity (Fig. 5D). Instead, the N4 mutant without WR actually activates the reporter more strongly than the wild type molecule (Fig. 5D), consistent with the observation in Fig. 1 that the combination of WR and CD2 in the Gal4-based system also has lower activity than CD2 alone. It is possible that the close association between WR and CD2 may hinder their interactions with any potential co-activator or the general transcription machinery. Alternatively, simultaneous engagement of the transcription apparatus by WR and CD2 reduces the efficiency of transcription. Further studies are required to clarify these two scenarios.
The Redundancy of WR and CD2-Whereas WR appears to lower the activity of CD2 as discussed above, it is clear that both are strong activators by themselves and thus redundant FIG. 5. NANOG mediates transactivation through WR and CD2. A, schematic illustration of NANOG reporters used in this study, pTK and p5N as described in Fig. 4. B, schematic illustration of deletion mutants of NANOG. A FLAG was fused to the C terminus of each deletion for protein detection as described under "Materials and Methods." at the C terminus of NANOG. One may argue that WR and CD2 may be able to substitute each other functionally (at least partially) as demonstrated in Fig. 5. Given the structural difference, we propose that WR and CD2 regulate transcription activation through distinct pathways. At the present, we have little evidence to support this idea. We have initiated a yeast-based strategy to identify binding partners for WR and CD2. Should different binding partners be identified, we would be able to define distinct pathways that NANOG regulate gene activation. Alternatively, we should be able to define the downstream genes using N3 or N4 as activators in a DNA chip based survey of downstream genes in either ES or P19 cells. Nevertheless, further studies are required to sort out the mechanism through which different domains of NANOG regulate the expression of genes critical to stem cell pluripotency.