Identification of new human mastermind proteins defines a family that consists of positive regulators for notch signaling.

Mastermind (Mam) is one of the evolutionarily conserved elements of Notch signaling. Genetic analyses in Drosophila implicated it as an important positive regulator of the pathway. We show here identification of two new members of human Mam family (human Mastermind-2 (hMam-2) and human Mastermind-3 (hMam-3)), which retain characteristics similar to human Mastermind-1 (hMam-1) and Drosophila Mastermind. Both hMam-2 and hMam-3 stabilize and participate in the DNA-binding complex RBP-J/CBF-1 protein and the Notch intracellular domains that serve as intermediates of the signaling. Both hMam-2 and hMam-3 enhanced the activation of transcription from a target promoter by Notch signaling. However, we also show evidence that the activation of the target promoter by Notch3 and Notch4 is more efficiently potentiated by hMam-2 than by hMam-1 or -3. The multiplicity of Mam proteins in the mammalian system may help provide divergence to the strength of the Notch signals in different cell types.

Notch signaling is an evolutionarily conserved mechanism that mediates cell-cell communications required for cell fate decisions in metazoans (1,2). In vertebrate systems, Notch signaling has been shown to mediate cell type specification in multiple tissues (3)(4)(5)(6)(7)(8). Furthermore, abnormalities in Notch signaling were linked to human diseases such as neoplasia, stroke, syndromic intrahepatic cholestasis, an axial skeletal defect, and a congenital heart disease (9 -13). In the vertebrates, four Notch genes have been identified (Notch1, Notch2, Notch3, and Notch4) (14). Each of the four Notch genes is expressed in distinct spatiotemporal patterns and has been shown to play independent and overlapping roles in the mammalian systems (15)(16)(17)(18)(19).
The products of Notch genes are receptor type molecules that are first synthesized as large type I transmembrane proteins.
During its maturation, Notch receptor is cleaved once in the extracellular region (site 1) and noncovalently reattached (2). Upon binding to ligands that are expressed on neighboring cells, Notch is cleaved sequentially at an extracellular juxtamembrane region (site 2) and in its transmembrane domain (site 3) (2). The intracellular (IC) 1 domain of the receptor is then released from the membrane and transported to the nucleus (2,20,21), where it participates in transcriptional activation through association with promoter elements via CBF-1, Su(H), Lag-1 (CSL) DNA-binding proteins (22)(23)(24)(25). This signaling process can be experimentally mimicked by expression IC domains of Notch (2,20,21). Enhancer of split in Drosophila or Hairy Enhancer of split (HES)-1 and HES-5 in mammals are among the primary target genes of this signaling. It has been shown that mice homozygous for the Notch1 allele deficient for the processing of the site 3 exhibit embryonic lethality similar to those homozygous for its null allele (26), exemplifying the relevance of the nuclear Notch model at least in the mammalian system. Mastermind (Mam) was first identified in Drosophila and implicated as an important positive regulator of Notch signaling pathway by genetic analyses (27)(28)(29)(30). Recently, we and others have identified mammalian Mam and have shown that both the mammalian and Drosophila Mam proteins stabilize and participate in the DNA-binding complex of the intracellular domains of Notch (NotchIC) and CSL proteins during the activation of the target promoters (25,31,32).
We show here identification of two new members of the human Mam protein family. These proteins have properties similar to the previously characterized human Mam. They exhibited stabilization and participation to the NotchIC/CSL DNA-binding complex and enhancement of the activation of transcription from the target promoter. However, we also show evidence that some members of the Notch receptors have a preference for the three human Mam proteins as a partner to activate the target. Thus, the multiplicity of Mam proteins may help provide divergence to the strength of the Notch signals in various cell types.
Northern Blotting-Multiple tissue Northern blot and multiple tissue Northern blot II (Clontech) membranes that had been immobilized with polyadenylated RNAs from various human tissues were hybridized with 32 P-labeled fragments from hMam-1 (2.4 kb; XhoI-XhoI) (25), hMam-2 (1.2 kb; KpnI-KpnI), or hMam-3 (2.2 kb; SalI-BamHI) cDNA. Prehybridization, hybridization, and washing were done in stringent conditions under standard procedures. The washed membranes were analyzed with BAS2000 image analyzer (Fuji, Japan) without modification. After the analysis, the membranes were stripped out of the radioactivity, hybridized again with human ␤-actin probe (Clontech), and analyzed as above. All of the experiments for the three probes were performed concomitantly.
Antibody Production-pGEX-hMam-1 (amino acids 1-344) was constructed by introducing the coding sequence of hMam-1 into pGEX-4T (Pharmacia Corp.). Glutathione S-transferase fusion protein was expressed in the BL21 strain of Escherichia coli. The rabbits were immunized by the purified protein with a standard procedure to raise antisera against hMam-1 protein. Immunoglobulin was purified with protein G column. Two of the antibody preparations (2-4 and 3-1) cross-reacted with hMam-2 protein.
Electrophoretic Mobility Shift Assay (EMSA)-EMSA using transfected 293T cells were detailed in Ref. 25. Briefly, cells (2 ϫ 10 6 cells/ 10-cm dish) were transfected with 5 g each of the expression vectors for RBP-J, NotchIC, and Mam constructs or their empty counterparts. Total amount of the plasmid DNA was kept constant (15 g). DNAprotein binding reactions were done by incubation of the whole cell extracts (20 g, equivalent to the protein amount) in a solution (15 l) containing 13 mM Hepes-NaOH (pH 7.9) buffer supplemented with 8% glycerol, 50 mM NaCl, 0.4 mM MgCl 2 , 0.5 mM dithiothreitol, 66.6 g/ml poly(dI-dC):poly(dI-dC), and 33.3 g/ml salmon sperm DNA for 15 min on ice, followed by an additional 30-min incubation with 32 P end-labeled synthetic double-stranded oligonucleotide probe (0.1-0.2 ng in 1 l, 5-20 nCi) at room temperature. The final concentration of the probe is 0.25-0.5 fmol/l. Half of the mixture was loaded on polyacrylamide gels (5%) in 0.5ϫ Tris-borate-EDTA buffer to separate the DNA-protein complexes. The complexes were detected by exposing the dried gels to x-ray films. For competition analysis to define sequence specificity of DNA-binding complexes, the molar excess of unlabeled double-stranded oligonucleotides was included in the binding reaction. The sequences of the oligonucleotides for the labeled probe were from the murine HES-1 gene that include the RBP-J-binding site (Ϫ91 to Ϫ56; HES-1 probe) (25). The competitors were the unlabeled HES-1 probe itself or m67 probe that binds specifically to activated STAT1 and STAT3 (36,37).
Transcriptional Activation Assays-U2OS cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. For transfection, the cells were seeded in 12-well dishes (0.3 ϫ 10 5 cells/well). For assay with hNotch1IC, the cells were cotransfected with 0.3 g of pHES-5luc reporter plasmid (23), 0.4 g of pcDNA1/Amp with or without hNotch1IC cDNA, 0.3 g of pEF-BOS with or without Mam cDNAs, and 10 ng of Renilla luciferase internal control plasmid (pRL-CMV; Promega). For assay with the four murine NotchICs, the cells were cotransfected with 0.3 g of pHES-5luc reporter plasmid, 0.3 g of pEF-BOSneo with or without NotchIC cDNAs, 1 ng of pRL-CMV, and 0.4 g of mixtures of pEF-BOS and the expression vectors for the three Mams in various ratio. The transient transfection was done using LipofectAMINE (Invitrogen). Two days after transfection, firefly and Renilla luciferase activities were determined using a dual luciferase assay kit (Promega) and a Turner Designs TD20/20 dual luminometer. Firefly luciferase activities were normalized with the Renilla luciferase control activities.
Analysis of Subcellular Localization of GFP-tagged Proteins-293T cells were seeded on the Lab-Tek chamber slides (Nalge NUNC, Naperville, IL). The cells were transfected with plasmids by calcium phosphate method. Two days after the transfection, the cells were analyzed with a confocal laser scanning microscope (LSM 410, Carl Zeiss).

RESULTS
During a project to clone long human cDNAs, we have come across two cDNA sequences (KIAA1816 and KIAA1819) (38) that potentially encode proteins with similarity to hMam-1 (Fig. 1A). The overall identity between hMam-1 protein and each of the proteins encoded by these cDNAs is 30 and 20%, respectively ( Table I). The arrangements of basic and acidic amino acid clusters in the Mam proteins including Drosophila Mam are well conserved, implying that their higher order structures are related (Fig. 1B). Based on the results reported below, we have tentatively named the protein encoded by KIAA1816 hMam-2 and that encoded by KIAA1819 hMam-3.
The expression of hMam-2 and hMam-3 mRNA in various human tissues was examined by Northern blotting analysis. For reference, hMam-1 mRNA was also analyzed in the same conditions. As shown in Fig. 2, hMam-2 mRNA exists as a single 7.9-kb transcript, and hMam-3 mRNA exists as a single 7.5-kb species. For hMam-3 mRNA, an additional 5.6-kb species was found in peripheral blood leukocytes and in placenta. Fig. 2 also shows that all three Mam mRNAs are expressed in many tissues with distinct patterns. Furthermore, it also indicates that multiple Mam mRNA species are expressed in a number of tissues.
We cloned each of the open reading frames of hMam-2 and hMam-3 cDNAs into a mammalian expression vector. When we blotted extracts of 293T cells transfected with these vectors, we found one of antisera raised against the N-terminal portion of hMam-1 protein cross-reacts with hMam-2. As shown in Fig.  3A, in addition to hMam-1 protein, which has an apparent molecular mass of 140 kDa, a single protein with an apparent molecular mass of 169 kDa was detected in the extract from the cells transfected with hMam-2. In contrast to hMam-2 protein, none of the anti-hMam-1 antisera reacted to hMam-3 protein (Fig. 3A). Thus, we attached a Myc tag to the C terminus of its coding region to identify the product of hMam-3 cDNA. Using an anti-Myc antibody, a single protein with an apparent molecular mass of 186 kDa was detected by Western blotting analysis of 293T cells transfected with this vector (Fig. 3B).
To test a hypothesis that these two proteins have functions similar to hMam-1, we transfected the expression vectors to 293T cells in combinations with vectors for the intracellular domain of human Notch1 (hNotch1IC) and a major form of mammalian CSL proteins, RBP-J/CBF-1 (39). Using EMSA, we analyzed binding activities in the extracts of the cells to a sequence from the HES-1 promoter, which has been shown to be essential for activation by the Notch1IC domain (22,23,25). As shown in Fig. 3C (lane 2), an extract from the cells transfected with RBP-J exhibited two specific bands that had been shown to contain RBP-J (25). Cotransfection of hNotch1IC with RBP-J induced another RBP-J/hNotch1IC complex that mi- grates more slowly than the bands mentioned above (Fig. 3C, lane 3) (25). Coexpression of hMam-1 with these two proteins reduced the RBP-J-specific bands and induced RBP-J/ hNotch1IC/hMam-1 bands that migrate more slowly (Fig. 3C, lane 4) (25). Expression of hMam-2 with RBP-J and hNotch1IC results in activities similar to that of RBP-J/hNotch1IC/ hMam-1 transfection (Fig. 3C, lane 5). Furthermore, as was the case for hMam-1 (25), expression of hMam-2 without hNotch1IC does not significantly alter the DNA binding activity of RBP-J or endogenous RBP-J-like factor (Fig. 3C, lanes 1,  2, 6, and 7). Fig. 3D shows the analysis with the hMam-3 vector. Expression of hMam-3 with RBP-J and hNotch1IC again results in similar binding activities to the RBP-J/hNotch1IC/hMam-1 and RBP-J/hNotch1IC/hMam-2 transfection (Fig. 3D, lanes  4 -6). Expression of hMam-3 without hNotch1IC also does not significantly alter the DNA binding activity of RBP-J or endogenous RBP-J-like factor (Fig. 3D, lanes 1, 2, 7, and 10).
To verify the DNA binding specificities of these complexes, we examined the effects of unlabeled competitors on the DNA binding reaction. As shown in Fig. 3E, the fast migrating complex in the vector-transfected extract and the two specific complexes in the RBP-J-transfected extracts were competed by molar excess of the unlabeled HES-1 probe but not by the same molar excess of a double-stranded oligonucleotide with unrelated sequence. Furthermore, all of the slowly migrating complexes in the RBP-J/hNotch1IC, RBP-J/hNotch1IC/hMam-1, RBP-J/hNotch1IC/hMam-2, and RBP-J/hNotch1IC/hMam-3- 293T cells were transfected with expression vectors for hMam-1, hMam-2, or hMam-3. As a control, the empty vector was transfected. Whole cell extracts prepared from these cells were analyzed by immunoblotting using an anti-hMam-1 antibody that cross-reacts with hMam-2. B, identification of the hMam-3 protein. 293T cells were transfected with an expression vector for hMam-3-Myc or the empty control vector. Whole cell extracts prepared from these cells were analyzed by immunoblotting using an anti-Myc antibody. C, hMam-2, RBP-J, and hNotch1IC forms RBP-J/hNotch1IC/hMam-1-like DNA-binding complex. 293T cells were transfected with the expression vectors for the indicated proteins (ϩ or numbers) or empty vector controls (Ϫ). Extracts of the transfected cells were examined for binding activities to the RBP-J element of HES-1 promoter by EMSA. D, hMam-3, RBP-J, and hNotch1IC forms RBP-J/hNotch1IC/hMam-1-like DNA-binding complex. The analysis was performed as described for C. E, sequence-specific binding of the complexes involving RBP-J. Unlabeled HES-1 or unrelated (m67) oligonucleotides were included in the binding reaction in molar excess as indicated in numbers. NS, nonspecific complex. F, sequence-specific binding of the RBP-J/hNotch1IC and RBP-J/hNotch1IC/hMam-1 complexes. Analysis was performed as described for E. G, sequence-specific binding of the RBP-J/hNotch1IC/hMam-2 and RBP-J/hNotch1IC/hMam-3 complexes. Analysis was performed as described for E. H, physical association of human Mam proteins with RBP-J and hNotch1IC. Extracts of 293T cells transfected with indicated vectors or empty control vectors were immunoprecipitated by an anti-Notch1 antibody or an anti-Posh antibody. The precipitates were separated on SDS gels, blotted onto membranes, and then stained sequentially with the anti-Notch1, the anti-RBP-J, the anti-hMam-1 that cross-reacts hMam-2, the anti-c-Myc, and   (38) transfected extracts were competed out by molar excess of the unlabeled HES-1 probe but not by the unrelated oligonucleotide (Fig. 3, F and G). These results indicate that RBP-J complex and all of the apparent multimeric complexes bind to the RBP-J site in the HES-1 promoter in a sequence-specific manner. We next analyzed the complex formation by coimmunoprecipitation. This assay revealed that both hMam-2 and hMam-3 are, like hMam-1, specifically coprecipitated with hNotch1IC and RBP-J by an anti-Notch1 antibody but not by an anti-Posh antibody, which has been raised against the C-terminal portion of Posh (a downstream signaling molecule for Rac GTPase) protein (40) by the same manufacturer (Santa Cruz) using the same species (goat) as the anti-Notch1 antibody (Fig. 3H, lanes  3-5 and 11-13) (25). These results indicate that both hMam-2 and hMam-3 associate with hNotch1IC and RBP-J in the absence of the binding site of DNA as shown for hMam-1 (25). This assay also disclosed that both hMam-2 and hMam-3 associate with hNotch1IC only in the presence of RBP-J (Fig. 2E,  lanes 3-8). In EMSA, expression of neither hMam-1, hMam-2, nor hMam-3 without hNotch1IC significantly altered the DNA binding activity of RBP-J (Fig. 3, C, lanes 2 and 7; and D, lanes  2 and 7) (25), indicating that all three human Mam proteins associate with RBP-J only in the presence of hNotch1IC. These results suggest that both hMam-2 and hMam-3, just like hMam-1, associate with the complex of the two proteins but not FIG. 4. The basic domains of the human Mam proteins are important for the complex formation. A, hMam-2 and its truncations containing the basic domain can associate with RBP-J and hNotch1IC within ternary complexes. The hMam-2 truncations, characterized in B, were tested for ability to form the ternary complex that is observed with full-length hMam-2. 293T cells were transfected with the indicated combinations of the expression vectors, and their extracts were analyzed for binding to the RBP-J element. The binding complexes migrate at a rate proportional to the length of the hMam-2 truncation. B, expression of the truncated forms of hMam-2 protein. Extracts of 293T cells transfected with the vectors for indicated proteins were blotted with the anti-hMam-1 antibody that cross-reacts hMam-2. To detect hMam-2 Met156stop protein, an x-ray film was exposed for longer period. C, sequence-specific binding of the complexes involving the truncated forms of hMam-2 protein. Analysis was performed as described for Fig. 3E. D, C-terminal truncations of hMam-3 can associate with RBP-J and hNotch1IC within ternary complexes. The hMam-3 truncations, characterized below, were tested for ability to form the ternary complex as described for A. The binding complexes migrate at a rate proportional to the length of the hMam-3 truncation. E, expression of the short forms of hMam-3 protein.
Extracts of 293T cells transfected with the vectors for indicated proteins were blotted with the anti-c-Myc antibody. F, sequence-specific binding of the complexes involving the truncated forms of hMam-3 protein. Analysis was performed as described for Fig. 3E.

FIG. 3-continued
New Human Mastermind Proteins Augment Notch Signaling 50617 with the single protein species. The coimmunoprecipitation assays further reveal that expressions of all three human Mam proteins enhance the physical association of hNotch1IC and RBP-J (Fig. 2E, lanes 2-5) (25). These results indicate that hMam-2 and hMam-3 have characteristics very similar to hMam-1 in terms of complex formation with hNotch1IC and RBP-J. We next examined the effect of hMam-2 and hMam-3 expression in comparison with that of hMam-1 in a transcriptional activation assay using the HES-5 promoter (25). As shown in Fig. 3I, the HES-5 promoter was activated by the expression of hNotch1IC alone but not by expression of hMam-1 alone. Coexpression of hNotch1IC and hMam-1 augmented the activation of the HES-5 promoter (25). Both hMam-2 and hMam-3 enhanced the hNotch1IC-induced activation of the HES-5 promoter to a similar extent to hMam-1, and these two new Mams did not activate the promoter in the absence of hNotch1IC (Fig.  3I). Because multiple Mam mRNAs are expressed in a number of tissues, we further examined whether the three Mam proteins exhibit synergism when expressed in combinations on the transactivation assay system. Fig. 3I shows, however, that no   (42) combinations of the proteins elicited synergistic or combinatorial effects on the transactivation, if total amount of the transfected vectors of the three was kept constant. In hMam-1 protein, the N-terminal basic domain has been shown to be essential for the complex formation with RBP-J and Notch1IC (25,32). We examined the effect of C-terminal truncations of hMam-2 on the DNA-binding complexes involving RBP-J and hNotch1IC. As shown in Fig. 4A (lanes 3-5), all of the truncations up to amino acid 156 reduced the complexes involving RBP-J only and induced more slowly migrating complexes whose mobility correlates with the molecular mass of each truncation (Fig. 4B). Fig. 4C shows that the complexes involving the hMam-2 truncations could be competed out by the HES-1 oligonucleotide but not by the unrelated oligonucleotide, verifying their binding specificity. These results support the idea that the slowly migrating complexes contain hMam-2 protein. Furthermore, ϳ155 amino acids from the N terminus are sufficient to alter the mobility of the DNA-binding complexes, and a deletion construct lacking amino acids 67-95 exhibits virtually no activity on the complexes (Figs. 4A, lane 6, and 1B). Thus, the N-terminal region of hMam-2 is necessary and sufficient to mediate the physical association. Similarly, C-terminal truncations of hMam-3 up to amino acid 202 formed the slowly migrating complexes whose mobility correlates with the molecular mass of each truncation (Figs. 4, D and E, 3B, and 1B). Again, all the slowly migrating complexes could be competed out by the addition of the HES-1 oligonucleotide but not by the unrelated oligonucleotide (Fig. 4F). These results indicate that the N-terminal basic domain, a well conserved region in the three proteins (Fig. 1A), is important for the complex formation in all three.
It has been indicated that all four mammalian Notch proteins undergo proteolytic processing during signaling and produce Notch1IC-like molecular species as signaling intermediate (20,21). We examined the complex formation between the three Mam proteins with RBP-J and four murine NotchICs that extend from juxtamembrane RAM domains to the C termini (20,34). As shown in Fig. 5 (A-D), any combinations of Mam and the NotchIC can be parts of complexes that resemble those involving hNotch1IC (Fig. 3, C and D). We next investigated whether expression of the three Mam proteins could synergistically activate the transcription of the HES-5 promoter with the four NotchICs. As shown in Fig. 5E, Notch1ICinduced activation was, in agreement with the experiment shown in Fig. 3I, enhanced by coexpression of any of the three Mams to similar degrees. Activation by Notch2IC was also strongly activated by any of the three Mams to a comparable extent (Fig. 5E). However, Notch3IC-induced activation was not efficiently up-regulated by either hMam-1 or hMam-3. In contrast, hMam-2 showed significant augmentation of the Notch3IC-induced activation (Fig. 5E). Notch4IC-induced activation was relatively resistant to up-regulation by Mams (Fig.  5E). Only a high dose of hMam-2 showed significant augmentation (Fig. 5E). These results indicate that hMam-2 has a unique role for the Notch3-and Notch4-mediated signaling.
Finally, to obtain an insight into subcellular localization of hMam-2 and hMam-3, we attached GFP tags to the C termini of their coding regions cloned in the mammalian expression vector. When these vectors were transfected into 293T cells, each produced a single protein with an apparent molecular mass of 200 kDa as detected by Western blotting analysis with an anti-GFP antibody (Fig. 6A). Because the tag attached to hMam-2 is larger in molecular mass, it is consistent that the fusion proteins were presented as the similarly sized bands. When the transfected cells were analyzed with a confocal laser scanning microscopy, both of the proteins were found as the nuclear dots (Fig. 6B). These results suggest that as is the case for hMam-1, both hMam-2 and hMam-3 are nuclear protein and may localize in nuclear bodies (25,32). DISCUSSION We have described the identification of two new human Mam proteins that possess characteristics similar to hMam-1 and Drosophila Mam (25). Both hMam-2 and hMam-3 contribute to the formation of a ternary complex along with a CSL protein and IC domain of the Notch receptors, and this complex associates with an HES promoter sequence. The expression of hMam-2 and hMam-3 augments Notch pathway-mediated activation of the HES target depending on its context. The reliance on the N-terminal basic domain for the complex formation is also a conserved feature of the proteins.
It may be worth noting that Drosophila Mam, along with hNotch1IC and RBP-J, forms a heterologous ternary complex that cannot bind to the HES promoter. 2 Furthermore, Drosophila Mam cannot augment the transcriptional transactivation induced by hNotch1IC. 2 Thus, the complexes that result from promiscuous interaction between the mammalian versions of Mam proteins and NotchICs seem to be evolved under functional restraint and may be significant.
As paralogous proteins in a mammalian species, the three Mams, especially hMam-3, seem to be significantly diverged from each other in their primary structures compared with the other groups of paralogues (Table I). For instance, one class of the association partners of Mam, Notch proteins show 30 -51% identities in pairwise comparisons among the mammalian sequences (Table II). One group of the mammalian ligands for Notch, Dll1, Dll3, and Dll4 proteins exhibit 31-51% identity (41). Another group of the Notch ligands, Jagged1 and 2 exhibit 54% identity in a mammalian species (42). Three mammalian Fringe proteins that are glycosyltransferases that modify the ligand interaction of Notch (43, 44) exhibit 47-64% identity (45). Thus, one may feel it unexpected that the three mammalian Mam proteins manifest such similar characteristics in their function.
It is still to be elucidated why the mammalian system requires multiple Mam species. In this sense, the preferences that are shown by some NotchICs to Mam proteins as interacting partners for the transactivation (Fig. 5E) may be significant. Because the multiple Notch species have unique roles in the mammalian systems (15)(16)(17)(18), the plurality of Mam proteins may have a role in the system providing divergence to the strength of the Notch signals in various cell types in the mammalian systems.