Direct binding of DNA by tumor suppressor menin.

Menin is a tumor suppressor that is mutated in patients with multiple endocrine neoplasia type I (MEN1), an inherited tumor-prone syndrome. Because there is no obvious conserved structural domain in menin that suggests a biochemical function, little is known as to how menin suppresses tumorigenesis. Although menin interacts with a variety of nuclear proteins including transcription factors, it is unknown whether menin itself can directly bind DNA. Here we show that menin directly binds to double-stranded DNA. It also binds a variety of DNA structures, including Y-structures, branched structures, and 4-way junction structures. The COOH terminus of menin mediates binding to DNA, but MEN1 disease-derived mutations in the COOH terminus abolish the ability of menin to bind DNA. Importantly, these MEN1 disease-related menin mutants also fail to repress cell proliferation as well as cell cycle progression at the G2/M phase. Furthermore, detailed mutagenesis studies indicate that positively charged residues in two nuclear localization signals mediate direct DNA binding as well as repression of cell proliferation. Collectively, these results demonstrate, for the first time, a novel biochemical activity of menin, binding to DNA, and link its DNA binding to the regulation of cell proliferation.

Menin, a tumor suppressor consisting of 610 amino acid residues, is primarily a nuclear protein (1). Men1, the gene encoding menin, is frequently mutated in patients with multiple endocrine neoplasia type I (MEN1), 1 an inherited tumor syndrome characterized by multiple endocrine tumors (2,3). There are no obvious motifs in menin that suggest a tumorsuppressing function. Consequently, how menin suppresses tumorigenesis is poorly understood (2,3). Identification of menininteracting proteins has provided new insights into the function of menin (4). For example, menin interacts with a diverse group of transcription factors and co-regulators, including JunD, NF-B, Pem, Smad3, and mSin3A-histone deacetylase (5)(6)(7)(8)(9). These reports suggest a role for menin in the regulation of gene transcription.
Menin has also been shown to participate in the regulation of cell proliferation and genome stability. For instance, ectopic expression of menin in Ras-transformed NIH3T3 cells represses cell proliferation as well as tumorigenesis (10). In addition, overexpression of menin inhibits the proliferation of a rat insulinoma cell line (11). Numerous reports also indicate a potential role for menin in the regulation of DNA repair and genome stability (12)(13)(14)(15). Consistent with these reports, menin has recently been shown to interact with FancD2, a protein that participates in DNA repair (16,17), as well as with RPA2, a critical component of the single-stranded DNA-binding protein RPA (18,19). Many tumor suppressors are involved in the regulation of multiple cellular processes. For example, Brca1, a tumor suppressor mutated in patients with inherited breast cancer, is also involved in the regulation of gene transcription, cell proliferation, and genome stability (20,21). Recently, Brca1 has been shown to directly bind DNA in a sequenceindependent manner (22).
Given the fact that menin associates with various DNAbinding proteins as well as chromatin (4,18), we sought to test whether menin directly binds DNA and, if it does, whether binding to DNA is crucial for the function of menin in regulating cell proliferation. In the current studies, we demonstrate that menin directly binds DNA. Menin also binds a variety of other DNA structures and the COOH terminus of menin is responsible for direct binding to DNA. Importantly, MEN1 patient-derived mutations in the COOH terminus abolish the ability of menin to bind DNA and also fail to repress cell proliferation. These results uncover a mechanistic link between DNA binding by menin and its role in repressing cell proliferation.
Protein Expression and Purification-Exponentially growing DH5␣ E. coli cells harboring various GST-menin constructs were induced with 150 M isopropyl 1-thio-␤-D-galactopyranoside at room temperature. The bacteria were collected and lysed at 4°C in 1ϫ phosphate-buffered saline with protease inhibitors and Triton-X-100 (1%), followed by sonication. The supernatant from the lysates was incubated with glutathione-Sepharose TM 4B beads (Amersham Biosciences) and the beads were washed extensively with 1ϫ phosphate-buffered saline with protease inhibitors prior to elution with the elution buffer containing 10 mM glutathione. To express the full-length menin, with or without small deletions, the corresponding constructs were transformed in E. coli strain BL21(DE3) and the proteins were expressed as COOH-terminal His 6 -tagged proteins. Proteins were purified to homogeneity over a nickel-nitrilotriacetic acid-agarose resin (Qiagen), followed by anion exchange chromatography (Source-15Q) and gel filtration (Superdex-200, Amersham Biosciences).
Analysis of Apoptosis and Cell Cycle Using Fluorescence Activating Cell Scanning (FACS)-For propidium iodide/5Ј-bromo-2Ј-deoxyuridine 5Ј-triphosphate (BrdUrd) staining, on day 0, cells were seeded at a density of 2 ϫ 10 5 cells/100-mm dish. On day 2, cells were pulsed for 45 min with 10 M BrdUrd (Sigma). Following harvesting, 10 6 cells per sample were fixed in 70% ethanol while vortexing. Cell pellets were washed in washing buffer (phosphate-buffered saline plus 0.5% bovine serum albumin), pelleted, and denatured in denaturing solution (3 N HCl and 0.5% Tween 20) for 10 min. After pelleting, cells were then incubated with anti-BrdUrd antibody (BD Pharmingen) for 20 min, followed by washing and pelleting. Next, fluorescein isothiocyanateconjugated goat anti-mouse IgG (BD Pharmingen) antibody was added for 15 min. After another washing and pelleting, cells were resuspended in 0.5 ml of propidium iodide (Sigma, 10 g/ml in phosphate-buffered saline) and incubated for 30 min. Samples were analyzed on a FACS-Calibur (BD Pharmingen).
For annexin V staining, cells were seeded at a density of 2 ϫ 10 5 cells/100-mm dish on day 0. On day 2, cells and supernatant were collected and stained with annexin-V-fluorescein isothiocyanate, using the annexin V-Fluorescein Isothiocyanate Apoptosis Detection Kit (MBL, Inc., Woburn, MA). Duplicate samples were analyzed on a FAC-SCalibur (BD Biosciences).
Western Blotting and Immunofluorescent Staining-Proteins from cell lysates or various cellular fractions were separated on SDS-PAGE gels, transferred to Hybond C ϩ membranes (Amersham Biosciences), and blotted with various primary antibodies, followed by incubation with appropriate horseradish peroxidase-conjugated secondary antibodies. The blotted proteins were visualized using the enhanced chemiluminescence detection system (Amersham Biosciences). For immunofluorescent staining, cells were set up on coverslips and processed for incubation with the primary antibody (M 2 ) and the fluorescein isothiocyanate-conjugated secondary antibody (Jackson ImmunoResearch Laboratories, Inc.) as previously described (18) and the stained cells were photographed under a fluorescence microscope (Nikon).
[ 3 H]Thymidine Incorporation Assay-On day 0, cells in 2 ml of normal medium were seeded in each well of a 6-well plate. On day 4, [ 3 H]thymidine (ICN Biomedicals Inc.) was added to the medium. [ 3 H]Thymidine was first diluted in Opti-MEM (Invitrogen) and then was added to each well. Triplicate samples were incubated at 37°C, 5% CO 2 for 4 h before being trypsinized and harvested. Incorporation of [ 3 H]thymidine into cells was determined using a model 1450 Microbeta Trillux Scintillation Counter (Wallac), as previously described (26).

RESULTS
Menin Binds Double-stranded DNA-Menin is primarily a nuclear protein that interacts with a variety of other DNAbinding proteins as well as chromatin (1,18). The tumor suppressor gene Brca1, which also represses cell proliferation (27), binds DNA in a sequence-independent manner (22). However, it is unclear whether menin itself binds DNA, either in a sequence-specific or nonspecific manner. To examine whether menin binds DNA directly, we incubated full-length menin with either a radiolabeled single-stranded DNA (ssDNA) probe or a dsDNA probe, each of which is 65 base pairs (bp) in length. These probes were used to characterize the DNA binding by tumor suppressor Brca1 (22). To detect the potential menin-DNA interaction, the reaction was performed in the absence of nonspecific competitive DNA, and separated on agarose gels. Fig. 1A shows that neither of the single-stranded DNA probes binds to menin (lanes 2 and 4); however, the dsDNA probe is shifted by menin (lane 6). These results indicate that menin binds dsDNA. Consistent with this, an affinity-purified antimenin antibody supershifts the menin-bound DNA probe (Fig.  1B, lane 3), although the control IgG fails to supershift meninbound DNA (lane 6). Collectively, these experiments demonstrate that menin binds dsDNA. As a note, menin-DNA complexes were unable to migrate out from the wells of polyacrylamide gels (data not shown), 2 which have small pore sizes, whereas they were able to do so in agarose gels, which have larger pores. This is similar to what was described for the DNA binding by another tumor suppressor gene, Brca1 (22). Collectively, these results suggest that multiple molecules of menin may bind to a DNA probe.
Menin Binds DNA in a Sequence-independent Manner-There are various DNA-binding proteins with functions ranging from the regulation of gene transcription to the regulation of DNA replication. For example, transcription factors usually bind to sequence-specific DNA (24). However, proteins involved in DNA replication or repair, including human replication origin binding proteins, Brca1 and Rad50, bind DNA in a sequence-nonspecific manner (22,28,29). To determine whether menin binds DNA in a DNA sequence-specific or nonspecific fashion, we synthesized a pool of 65-mer oligonucleotides that harbor random nucleotide sequences in the middle region. The oligonucleotides were converted to dsDNA, and then used as a DNA probe for gel shift assays. Fig. 2 shows that menin binds to the DNA with random sequences (lanes 5-7). Notably, the ability of menin to bind to the random DNA is competed by poly(dI/dC), a homopolymer of dsDNA (lanes 9 -14). These results indicate that menin not only binds to DNA in a nonsequence-specific fashion, but that the homopolymer of DNA competes with this non-sequence-specific binding, further suggesting that the DNA binding by menin is not sequence-specific in this condition. To further demonstrate that menin is indeed able to bind DNA in sequence-independent manner, we incubated full-length menin with linearized and radiolabeled Puc19, a plasmid of 2.6 kb (kilobases), and examined whether menin is also able to bind this long stretch of DNA. The results show that menin indeed shifts the plasmid DNA probe (data not shown). 2 The COOH Terminus of Menin Mediates Binding to DNA-Because there are no motifs or domains in menin that suggest any obvious DNA binding activity, we sought to determine whether such a domain exists in menin. To this end, we generated constructs to express various GST-menin fusion proteins that span three various regions: the NH 2 terminus, the middle region, and the carboxyl terminus of menin (Fig. 3A). The full-length menin and its various fragments were expressed and purified from E. coli, and their sizes and purity were determined on SDS-PAGE by Coomassie Blue staining (Fig. 3B). These various menin fusion proteins were incubated with a plasmid DNA probe for gel shift assays (Fig. 3C). As expected, full-length menin binds the probe DNA (Fig. 3C,  lanes 2 and 3). The NH 2 terminus does not appear to dramat-ically shift the DNA probe (lanes 4 and 5). The middle region of menin, spanning from amino acid residues 219 to 395, does not shift the DNA probe (Fig. 3C, lanes 6 and 7). Notably, even at the lowest concentration of the protein used the carboxyl terminus of menin shifts the DNA probe completely (Fig. 3C, lanes  8 and 9). These results demonstrate that the carboxyl terminus is the major domain in menin that mediates binding to DNA in a sequence-independent fashion.  1-7). The binding to DNA by menin was competed with increasing concentrations of sequence-nonspecific poly(dI/dC) (lanes 9 -14) in the gel shift assay. Menin Binds to Various DNA Structures-DNA structures encountered in DNA metabolism include regular dsDNA, Ystructures, branched structures, and 4-way junction structures, a 4-strand cross-structure analogous to Holiday structure (30,31). BRAF35, a protein associating with the tumor suppressor Brca2, preferentially binds to 4-way structures (32) and plays a critical role in regulating cell proliferation (32). Thus, we examined whether menin binds to these various DNA structures. Fig. 4A shows that the carboxyl terminus of menin containing amino acid residues 396 -610 binds linearized dsDNA (lanes 3 and 4), Y-structures (lanes 6 and 7), branched structures (lanes 9 and 10), and 4-way structures (lanes 12 and  13). Similarly, full-length menin also binds these four DNA structures (data not shown). To further support the binding of menin to these various structures, a GST-menin (F3) fusion protein was incubated with the Y structure DNA probe and its binding to the probe was competed by each of the following 4 various DNA structures: linearized, Y-shaped, branched, and 4-way junction structures. Fig. 4B shows that each of the four structures competed with the DNA probe for binding to menin. However, at the lowest concentration of the competitor DNA used (36 nM), the 4-way structure DNA competed away ϳ60% of the bound probe DNA, whereas the other three structures only competed away 20 -25% of the bound DNA probe (Fig. 4B,  lanes 4, 8, 12, and 16). Similar results were obtained when the full-length menin was used in the gel shift assays with various competitive DNA structures (data not shown). The stronger competition for binding to the probe by the 4-way structure may reflect the higher mass of the 4-way structure at an equal molar concentration, because the molecular weight for the 4-way structure is more than twice that of the Y structure probe (78,839 versus 30,540). Together, these results indicate that menin can bind to various DNA structures.

MEN1-related Mutations in the COOH Terminus Compromise the Ability of Menin to Bind DNA as Well as Inhibit Cell
Proliferation-Over 63% of mutations in the Men1 gene eventually lead to loss of the carboxyl terminus (33). Thus, we sought to determine whether a common disease-related mutation at the carboxyl terminus, R527X, and a second mutation, M558X, interfere with the ability of menin to bind to DNA. Both mutants retain the first nuclear localization signal (NLS1, 479 -497), but lose NLS2-(588 -608). The wild type COOH terminus of menin (GST-F3) and its two truncation mutants, 527X and 558X, were expressed and purified from E. coli (Fig. 5B). The purified menin and its mutants were incubated with a short dsDNA probe. Fig. 5C shows that, as expected, the wild type COOH terminus binds the probe (lanes 2 and 3). However, the two mutants fail to bind to the DNA (lanes  4 -7). These results indicate that two common mutations from MEN1 patients lose the ability to bind DNA, suggesting that disruption of menin-DNA binding may be at least in part related to the development of MEN1.
To further examine the effects of these MEN1-related menin mutants on the regulation of cell proliferation, we introduced wild type menin or the two mutants into a menin-null cell line (18) via retroviral infection. Expression of the wild type and mutant menin was confirmed by Western blotting analysis (Fig. 6A, top panel). Equal loading of the cell lysates is confirmed by detection of the levels of actin by Western blot analysis (Fig. 6A, bottom panel). These various cell lines were seeded for culture and the number of cells for each of the cell lines was counted on day 4 of culture, to determine the rate of cell proliferation (Fig. 6B). The results indicate that complementation of the menin-null cells with wild type menin represses cell proliferation. However, the complementation of the menin-null cells with the MEN1-related menin mutants, 527X and 558X, fails to repress cell proliferation (Fig. 6B). These results suggest that DNA binding by menin may play a crucial role in repressing cell proliferation.
We previously observed that menin potentiates tumor necrosis factor-␣-induced apoptosis (34). Thus we sought to determine whether menin and its MEN1-related mutants alter the basal level of apoptosis and steady state of cell cycle distribution in our culture conditions. To this end, control cells or cells expressing wild type menin or mutants were examined for apoptotic cells, using annexin V staining followed by flow cytometry analysis. Fig. 7A shows that control cells (vector) contain 8% apoptotic cells, menin-expressing cells contain 9% apoptotic cells, whereas both mutant 527X and 558X-expressing cells contain 7% apoptotic cells. These results suggest that menin does not obviously up-regulate the basal level of apoptosis. However, we cannot completely rule out the possibility that a slight increase of apoptosis in menin-expressing cells could in part contribute to the menin-mediated inhibition of accumulation of cell numbers, because a low level of rapidly degrading apoptotic cells may escape detection.
We further determined the steady-state distribution of various cell cycle phases in the above cell lines. Control cells or cells expressing wild type or mutant menin were pulsed with BrdUrd and processed for analysis with flow cytometry. Fig. 7B shows that control cells in G 1 , S, and G 2 /M phases account for 40.5, 35.4, and 24.1%, respectively. Distribution of menin-expressing cells is 30.5, 35.5, and 34%, respectively. These results show that menin decreases the number of cells in G 1 but increases the number of cells in G 2 /M, suggesting a potential block in G 2 /M phase. Consistent with this, both menin 527X and 558X mutants have a cell cycle distribution similar to that in menin-null cells: 36, 39 -40, and 25% for G 1 , S, and G 2 /M, respectively (Fig. 7B). We also sought to determine the cell cycle progression in these cell lines by synchronizing the cells using serum starvation or various cell cycle blocking drugs. were infected with control retroviruses or retroviruses expressing each of the following forms of menin: menin (wild type), 527X, and 558X. The expression of the wild type and mutant menin, as indicated, was detected by Western blot analysis using an anti-menin antibody (number 77). As a control, the amount of actin in lysates was detected using an anti-actin antibody (bottom panel). B, the MEFs expressing the wild type or mutant menin were seeded at a density of 2 ϫ 10 5 cells per 100-mm dish in duplicate on day 0, and counted on day 4. However, serum starvation failed to synchronize the cells and various drugs including mimosine and hydroxyurea were highly toxic to the cells. 3 Collectively, these results suggest that menin affects cell proliferation by affecting cell cycle progression, and its ability to bind DNA is crucial for the role of menin in suppressing cell proliferation and MEN1.
Positively Charged Residues in the Two NLSs in the COOH Terminus of Menin Are Essential for Direct DNA Binding as Well as Repression of Cell Proliferation-Because the two NLS are located in the COOH terminus (1), which binds DNA, we examined whether the two NLSs play a role in binding to DNA.
Thus, menin proteins containing either a truncation of the COOH terminus, or internal deletions of small regions of the NLSs were expressed and purified from E. coli, and quantity and purity of the proteins were verified by Coomassie Blue staining (Fig. 8, A and B). These various menin proteins were incubated with a DNA probe for gel shift assays. As expected, the full-length menin binds to the DNA probe (Fig. 8C, lanes 2  and 3). In contrast, truncation of the COOH terminus (lanes 4 and 5), deletion of NLS1 (lanes 6 and 7), and deletion of NLS2 (lanes 8 and 9), abolish the ability of menin to bind DNA.
Because the deletion of several amino acid residues in the NLS sequences could affect the conformation of the COOH terminus that alters DNA binding, we sought to introduce 3 R. Schnepp, unpublished data.

FIG. 7. Analysis of apoptosis and cell cycle profiles in cells expressing wild type or mutant menin. A, annexin V staining of menin-null
MEFs expressing vector, menin, 527X, or 557X. Cells were seeded at a density of 2 ϫ 10 5 cells/100-mm dish and allowed to grow for 48 h. Cells and supernatant were then harvested and stained with annexin V as described under "Experimental Procedures." Data are presented as an average of two independent experiments. B, cell cycle profiles of menin-null MEFs expressing vector, menin, 527X, or 557X. Cells were seeded at a density of 2 ϫ 10 5 cells/100-mm dish and allowed to grow for 48 h. Cells were then processed and stained with propidium iodide and anti-BrdUrd antibody as described under "Experimental Procedures." This is representative of two independent experiments. point mutations, replacing the positively charged residues arginines or lysines with alanines, into the NLSs as shown in Fig. 9A. The point mutations should generate more subtle changes to the protein. The resulting GST fusion proteins with various point mutations were expressed and purified for gel shift assays (Fig. 9B). Fig. 9C shows that the wild type COOH terminus of menin binds the DNA probe, as expected (lanes 3 and 4), but introduction of point mutations into NLS1-(479 -497) (lanes 5 and 6) or NLS2-(588 -608) (lanes 7 and 8) dramatically decreases its ability to bind DNA. Combined point mutations in both NLS1 and NLS2 abolish the ability of menin to bind DNA. These results demonstrate that the NLS1 and NLS2 are essential for the role of menin in direct DNA binding, and these short sequences may directly contact DNA structure.
We further determined the effect of the NLS alterations on the nuclear localization of menin and its role in regulation of cell proliferation. Menin-null cells were infected with retroviruses expressing wild type menin or its two mutants (Fig. 10A). The expression of the various proteins was confirmed by Western blotting analysis (Fig. 10B). Additionally, nuclear localization of the wild type and menin mutants was examined by immunofluoresence staining (Fig. 10C). Mutations in either NLS1 or NLS2 do not affect the nuclear localization of menin (Fig. 10C). We examined whether mutation of these NLSs, which compromise DNA binding by menin, affects the role of menin in repression of cell proliferation. To this end, [ 3 H]thymidine incorporation was used to examine the rate of cell proliferation in menin-null MEFs complemented with either wild type menin or the various mutants described above. Fig.  10D indicates that complementation of the menin-null cells with wild type menin suppresses cell proliferation. Conversely, mutation in either NLS1 or NLS2, although still localizing to the nucleus (Fig. 10C), fails to suppress cell proliferation. To further strengthen this conclusion, the same numbers of cells for each of these cells were seeded, and the number of the cells from each of the 4 cell lines was counted on day 4 of culture (Fig. 10E). These results indicate that loss of either the NLS1 or NLS2 sequence results in failure of menin to repress cell proliferation. Therefore, although mutation in one of the two positively charged NLSs does not disrupt targeting to the nucleus, it does deprive menin of the ability of binding DNA as well as suppressing cell proliferation. Collectively, these results indicate that association of menin with DNA, at least in part, plays a crucial role in suppressing cell proliferation.

DISCUSSION
The current studies have uncovered the ability of menin to directly bind DNA, and further identified the COOH terminus of menin and the positively charged residues in the NLSs as the domain responsible for direct DNA binding. Importantly, two MEN1-related mutations in this domain fail to bind DNA as well as to repress cell proliferation (Figs. 5 and 6). These results link the role of menin in DNA binding to the suppression of cell proliferation. They suggest a model in which the interaction between menin and DNA allows menin to coordinate DNA metabolism and the regulation of cell proliferation.
Menin Directly Binds DNA-DNA-binding proteins can generally be categorized as sequence-dependent or sequence-independent proteins. For example, transcription factors are usually sequence-dependent DNA-binding proteins (24). On the other hand, the interaction of other DNA-binding proteins with DNA is dependent on overall DNA structure, but not on the precise sequence. In other words, they bind DNA in a sequencenonspecific fashion. These include proteins that regulate DNA repair and gene transcription, such as Brca1 (20,27). Rad 50, which is involved in DNA repair, and human DNA replication origin binding proteins also bind DNA in a manner that is independent of the precise DNA sequence (28,29). There are yet other DNA-binding proteins, such as p53, which possesses domains that bind DNA in both a sequence-dependent and -independent manner (35,36). Menin has been shown to bind a variety of proteins including transcription factors, such as JunD, NF-B, Smad3, Pem (4), as well as proteins involved in DNA repair, such as RPA and FancD2 (18,19). However, it has not previously been shown to directly bind DNA.
We show that menin binds directly to dsDNA (Fig. 1), but this binding does not favor a specific DNA sequence (Figs. 2  and 3). Further experiments indicate that menin binds to a variety of DNA structures including regular dsDNA, Y-structures, branched structures, and 4-way junctions. Thus, menin may associate with various DNA structures to regulate gene transcription, cell proliferation, and DNA repair.
Although it is shown here that menin binds DNA in a sequence-independent manner, these findings do not rule out the possibility that menin also binds some DNA sequence specifically, using either the COOH terminus or other regions. As described above, the tumor suppressor p53 (also a transcription factor) was originally described as a protein that bound to DNA non-specifically (35,37,38). In addition, p53 was later found to possess domains for sequence-specific and sequence-nonspe- FIG. 8. Mutations in each of two NLS of the C terminus of menin abolish binding of menin to DNA. A, a diagram of the mutations in the two NLSs of menin. B, full-length menin and various mutants (5 g each) as shown in A were stained with Coomassie Blue. C, menin (0, 0.5, and 1.0 g) and its various mutants (0.5, and 1.0 g) were incubated with a Y-structure probe for gel shift assays.
cific DNA binding (36). Therefore, it cannot be ruled out that under certain circumstances menin may be able to bind to specific DNA sequences to regulate gene transcription and/or cell proliferation.
MEN1-related Mutations in Menin Abolish the Ability of Menin to Bind DNA as Well as Suppress Cell Proliferation- The COOH terminus of menin mediates binding to DNA (Fig.  3). In support of the importance of menin in DNA binding, mutations in the COOH terminus of menin, R527X and M558X, abolish the ability of menin to bind to DNA (Fig. 5). These mutants, which have lost positively charged residues in the second NLS, still retain the first NLS, and thus should be able to target to the nucleus, because NLS1 alone is sufficient for targeting menin to the nucleus (Fig. 10). Given the fact that the majority of mutations in the Men1 gene from MEN1 patients lead to the loss of the carboxyl terminus, this correlation between the loss of the COOH terminus and the inability for the mutants to bind DNA reinforces the importance of menin-DNA binding in the function of menin. Importantly, these mutants lose their ability to suppress cell proliferation (Fig. 6).
Several reports show that ectopic expression of menin leads to reduced accumulation of cell numbers, but it is not clear in these experimental conditions whether menin affects either cell cycle progression or apoptosis, or both (7,10,11,39). In our system, menin represses cell proliferation and increases the number of G 2 /M cells, but does not markedly increase basal level apoptosis (Figs. 6 and 7). Nevertheless, it is noteworthy that MEN1-related menin mutants not only fail to repress the accumulation of cell numbers but also fail to increase cell numbers in G 2 /M (Fig. 7B). Together, these results suggest that interaction of menin with DNA is crucial for the role of menin in suppressing cell proliferation through a potential G 2 /M block. However, these results do not exclude the possibility that menin may also regulate cell proliferation by other mechanisms, such as blocking G 1 /S phase. A recent interesting report shows that knock-down of menin by an antisense approach in the rat intestinal epithelial cell line, IEC-17, leads to increased expression of cyclin D1 and CDK4 and increased number of S-phase cells (40). This is consistent with a model that menin down-regulates the CDK4/6 and pRB pathway and inhibits cell cycle transition for G 1 to S phase. However, it is not clear whether ectopic expression of menin in the menin knock-down cells can truly inhibit cell cycle progression from G 1 to S phase in the cells. In our MEF system, we observed that expression of menin increases the number of cells in G 2 /M, but slightly decreases the number of cells in G 1 (Fig. 7B). This difference in the effect of menin on cell cycle is likely due to the different cell lines used, because our MEFs cells were immortalized by papillomaviral E6 and E7, which antagonizes the pRB pathway (18,41). The E7-immortalized MEFs are supposed to be resistant to many signals that induce G 1 arrest. Furthermore, the COOHterminal part of menin also functionally interacts with the activator of S-phase kinase (ASK) (42), an essential component of Cdc7/ASK that regulates DNA replication (43). Because menin may regulate cell proliferation by several mechanisms, it is possible that menin-DNA interaction plays a crucial role in regulating cell proliferation not only by G 2 /M blocking, but also G 1 /S phase. However, it remains to be tested how the menin-DNA interaction affects cell proliferation in other systems, and whether the point mutations in the COOH terminus of menin, which fail to bind DNA, also fail to functionally interact with activator of S-phase kinase, which induces cell proliferation (42).
It is noteworthy that primary MEFs with the Men1 loci targeted by a neoTK cassette (Men1 T/T ), which fail to transcribe menin mRNA and synthesize the menin protein, grow as fast as their wild type counterpart (44). However, Men1-null MEFs that were immortalized in our studies grow faster than the wild type control cells. This discrepancy may be attributed to the difference in response to menin between immortalized cells and primary cells, which usually grow more slowly. Alternatively, because our MEFs were immortalized by introduction of human papillomaviral E6 and E7 (18), it is also possible that the role of menin in repressing cell proliferation becomes more prominent in repressing oncogene-enhanced cell proliferation. Consistent with this notion, proliferation of oncogenic Rastransformed MEFs is inhibited by ectopic expression of menin (10).
The Arginines and Lysines in the Two NLSs in the COOH Terminus of Menin Mediate Direct DNA Binding and Repression of Cell Proliferation-To further study the ability of menin to bind DNA, we made even smaller alterations in the COOH terminus sequence, namely internal deletion of NLS1 or NLS2.

FIG. 10. Mutations of two NLS in the COOH terminus of menin abolish menin-induced inhibition of cell proliferation.
A, a diagram of menin and its various NLS mutants. NLS1⌬, deletion of amino acids 479 -497. NSL⌬, deletion of amino acids 588 -608. B, Western blot analysis of wild type and various menin mutants from the meninnull MEFs that were infected with retroviruses expressing these proteins. Anti-FLAG M 2 antibody was used. C, nuclear localization of menin and its NLS mutants. Menin-null cells complemented with menin or its various NLS mutants were seeded and processed for immunofluorescence staining with an anti-FLAG antibody. D, the NLS mutants fail to repress cell proliferation. The menin-null MEFs complemented with wild type menin or its various NLS mutants. These cells were seeded in 6-well plates and cell proliferation was assayed by [ 3 H]thymidine incorporation at day 4 of culture. E, menin-null cells infected were with various retroviruses as indicated, seeded on day 0, and counted on day 4.