The migration of cells is a phenotypic response to changes in a number of cellular processes which leads to tumor invasion and metastasis.1 Tumor invasion and metastasis typically lead to the spread of cancer cells and can eventually result in destruction of the host.2 Thus, the ability of mammalian cells to regulate the processes which lead to cell migration is crucial for survival of the host. One of the cellular proteins identified as a master regulator of tumor migration and metastasis is Nm23-H1.3 Nm23-H1 is a nucleoside diphosphate kinase (NDPK, NDP kinase) involved in the biogenesis of NTPs from NDPs,4 through transfer of a phosphate group from ATP to NDP.5 NDP kinases are ubiquitous enzymes that are transiently phosphorylated on a conserved histidine residue during the catalytic reaction.6, 7 Nm23-H1 is encoded by the nm23-H1 gene located at position q21.3 on human chromosome 17.3, 8, 9 Additional functions of 2 known isoforms, namely Nm23-H1 and Nm23-H2, are independent of their NDP kinase activity.9, 10 They also form homo- and hetero-hexamers by randomly associating with each other.11
Nm23-H1 expression has been shown to be localized primarily to the cytoplasm, whereas Nm23-H2 was shown to be localized in the nucleus, and have DNA binding capabilities.12 Importantly, Nm23-H1 expression has been shown to indirectly correlate with the degree of tumor invasiveness and metastasis.13 Highly metastatic and invasive tumors consistently have low levels of nm23-H1 transcripts suggesting that Nm23-H1 is a negative regulator of cell migration.14 Indeed, exogenous introduction of nm23-H1 into breast carcinoma, melanoma and other invasive tumor cells resulted in reduced in vitro cell migration to chemoattractants as well as in vivo metastasis.14–17 It should also be noted that a positive correlation with metastasis has been observed in some human tumors such as neuroblastoma, osteosarcoma and pancreatic carcinoma suggesting alternate roles in different types of cancers.18–21 A more recent report has also shown that Nm23-H1 may also have a DNAse activity in response to cellular stress that leads to apoptosis.22, 23
Cytoskeletal changes leading to cell migration and invasion are regulated by a number of cellular molecules including the Rho family of GTPases.24 These GTPases are involved in a variety of biological processes including membrane trafficking, translation regulation, transcription activation, DNA synthesis and cytoskeleton reorganization.24, 25 The GTPases have been known to function as molecular switches, cycling between an inactive form when in a GDP bound state, and an active state which is typically bound to GTP.26 The balance of GDP-GTP binding is regulated by both guanine exchange factors (GEF) that enhance the exchange of GDP for GTP, and GTPase activating proteins (GAPs) that increase the intrinsic rate of hydrolysis of bound GTP.27, 28 A third group of molecules, the guanine nucleotide dissociation inhibitors (GDIs), inhibit both the exchange of GTP and the hydrolysis of bound GTP.25
Dbl-1 was identified from a diffuse B-cell lymphoma DNA library, and was the first member of the Dbl-1 family of mammalian GEFs.29–31 The original study identifying Dbl-1 (onco-Dbl) from the lymphoma library resulted in the cloning of the proto-oncogene pDbl.29 The Dbl-1 oncoprotein lacks the N-terminal 497 amino acid residues of pDbl, which is replaced by a short unrelated polypeptide sequence.29 These missing 497 residues may exert a downregulatory effect on the oncogenic potential of Dbl-1 as the oncoprotein has been shown to have a higher transforming potential as compared to pDbl.29, 32 The criteria for the Dbl family requires the presence of 2 domains in tandem, the Dbl homology domain (DH) which shares significant similarity with yeast cell division cycle protein Cdc24 and DH domain is the catalytic domain that stimulates GDP/GTP exchange. The other domain is the pleckstrin homology domain (PH)33, 34 whose function may be regulated by interaction with phosphoinositides, can regulate the DH domain catalytic activity and can also serve to promote Dbl-1 protein association with the plasma membrane. Analysis and screening for similar molecules using the Dbl-1 amino acid sequence has led to the identification of 45 members of the Dbl-1 GEF family31 and many DH domain containing proteins have been isolated bringing the number of GEFs to about 60 in human.35, 36 In this report we show that the suppressor of tumor metastasis protein Nm23-H1 is capable of interacting with, and regulating the activity of the GEF, Dbl-1. Thus regulation of the GEF activity by Nm23-H1 interferes with Dbl-1/Cdc42 signaling that ultimately alters cell migration.
Material and methods
Plasmids, antibodies and reagents
WT-nm23-H1, H118F and P96S are clones in pCMV and obtained from Dr. Patricia Steeg (NIH, Bethesda, MD). Using PCR techniques nm23-H1 was cloned into pA3M at the BamHI-NotI sites,37 H118F and P96S into pCDNA 3X HA- at the BamHI-EcoRI sites. The pCDNA Dbl-1 and pDbl constructs were obtained from Dr. Margaret Chou (University of Pennsylvania, Philadelphia, PA). Dbl-1 and pDbl were cloned into pCDNA 3X HA- at the EcoRI-XhoI sites. GST fusion constructs were made by PCR cloning of cDNAs into the BamHI-XhoI of pGEX2T. Cdc42 cDNA was obtained from Dr. Chou and cloned into pCDNA 3X HA- at BamHI-XhoI sites. COS-7 and 293T cells were obtained from the ATCC (Manassas, MD). TLC PEI-Cellulose plates (cat. No. Z122882) were purchased from Sigma (St. Louis, MO). Phosphorus-32 nucleotide (NEX 053001MC) was purchased from Perkin Elmer (Wellesley, MA). Rabbit anti-Dbl antibody was purchased from Santa Cruz (Santa Cruz, CA), anti-Nm23 antibody was obtained from SEIKAGAKU Corp. (Tokyo, Japan), anti-Myc antibody 9E10 and anti-HA antibody 12CA5, were culture products of hybridoma cell lines. The nuclear stain DAPI was purchased from Promega (Madison, WI).
Typically, cells were grown on 12 mm cover slips. Cover slips for 293T cells were coated with poly-L-lysine (1 μg/μl) prior to plating of cells. Cells were fixed in 3% paraformaldehyde (PFA) for 10 min at 4°C, 48 hr posttransfection using Fugene 6 from Roche (Indianapolis, IN) using manufacturer's suggestions. After fixation cells were extensively washed in PBS and incubated in blocking buffer (0.1% triton-X 100, 0.2% fish skin gelatin (Sigma, Cat no. G7765), in PBS). After a minimum of 5 min blocking, cover slips were incubated for 1 hr with primary antibodies diluted at 1:100 in blocking buffer. Cover slips were then washed thrice in PBS and incubated in corresponding secondary antibody for 30 min in blocking buffer. Secondary antibodies Alexa Fluor 488 and 594 were purchased from Molecular Probes (Carlsbad, CA). Cover slips were incubated for 10 min with DAPI nuclear stain diluted at 1:10,000 in PBS. Cover slips were then washed in PBS and mounted using Prolong anti-fade (Molecular Probes, Carlsbad, CA). Fluorescence was viewed by confocal microscopy and analyzed with Fluoview 300 software from Olympus (Melville, NY).
Yeast 2-hybrid and GST pulldown assay
The yeast 2-hybrid assay was performed as previously described.37 GST fusion constructs were transformed into BL21pLysS cells purchased from Invitrogen (Carlsbad, CA). Cells were grown overnight transferred to 250 ml of LB broth including 100 μg/ml of ampicillin and grown to an OD at 600 nm of 0.7. The culture was induced using 0.5 μM IPTG for 4 hr at 30°C. Cells were pelleted, resuspended in E. coli lysis buffer [50 mM Tris pH = 7.6, 0.1 mM EDTA, 2 mM DTT, protease inhibitors (BD Bioscience, San Jose, CA), lysozyme (0.4 mg/ml), in water], and sonicated 3 times for 15 sec pulses at 40% output (Fisher, Model 500). Lysates were clarified at 11,960g an in SS-34 rotor (Sorvall, Asheville, NC). Proteins were bound to Glutathione beads as per manufacturer's instructions in 500 μl total volume with the addition of binding buffer (1× PBS, 0.1% NP40, 10% glycerol, 0.5 mM DTT, 1 mM PMSF, 1 g/ml Aprotinin and 1 g/ml Pepstatin).
Proteins were radiolabeled with 35S methionine using Promega's in vitro TNT kit (Promega, Madison, WI). All samples were first precleared with Glutathione beads, split into 2 and incubated with either GST-fusion proteins or GST control bound to Glutathione beads. After overnight incubation with rotation, beads were washed 5 times in binding buffer (spun at 350g for 1 min in microcentrifuge at 4°C), and resuspended in Laemmlli's buffer prior to running on gels. Gels were dried and exposed to a PhosphorImager cassette (Amershan Biosciences, Piscataway, NJ) and analyzed with a Storm Phosphorimager (Molecular Dynamics, Piscataway, NJ).
293T cells (1 × 107 cells) were transfected with the corresponding plasmids containing myc-nm23-H1, HA-Dbl-1, HA-pDbl and pA3M by electrophoration using a Bio-Rad electroporator at 210V and 975 μF.38 After 48 hr, the cells were collected and washed with phosphate buffered saline before they were lysed in radio-immunoprecipitation assay buffer (RIPA buffer; 50 mM Tris pH 7.6, 150 mM sodium chloride, 15 mM MgCl2, 2 mM EDTA, 1% NP-40, 1 mM PMSF, 1 μg/ml aprotinin and 1 μg/ml pepstatin). Samples were vortexed every 15 min while incubated for 1 hr on ice. The cell pellet was removed by centrifugation and the supernatant was incubated with fusion protein bound Glutathione beads. After overnight incubation with rotation at 4°C, beads were washed 5 times in binding buffer, spun at 4°C, 350g for 2 min in a microcentrifuge. Proteins bound to the GST-fusion protein Glutathione bead complex were solubilized with Laemmli buffer transferred to nitrocellulose and western blot assay performed for detection of specific proteins.
293T cells (1 × 107) were transfected with the corresponding plasmids containing Myc tagged nm23-H1 and HA tagged Dbl-1 by electroporation using a Bio-Rad electroporator at 210 V and 975 μF.38 After transfection cells were cultured for 48 hr, collected and washed with phosphate buffered saline before they were lysed in radio-immunoprecipitation assay buffer as described earlier with the exception that 15 mM MgCl2 was used in RIPA to purify Dbl-1, pDbl and Nm23-H1. The cell pellet was removed by centrifugation and the supernatant was incubated with normal mouse serum or normal rabbit serum and a mixture of Protein A/G-Sepharose beads to preclear. The beads were collected, washed 5 times with RIPA buffer and the lysates were incubated with anti-myc ascites or rabbit anti-Dbl (Santa Cruz, CA) overnight with rotation at 4°C. Lysates were then incubated with Protein A/G-Sepharose beads at 4°C for 3 hr. The immunoprecipitates were washed 5 times with RIPA buffer solubilized with Laemmlli's buffer, transferred to nitrocellulose and western blotted to detect the specific antigens.
Cell motility assays
Ten gram of pA3M vector, pA3M nm23-H1, pCDNA Dbl and pCDNA-pDbl were transfected into MDA-MB-435 cell line by electroporation using a Bio-Rad electroporator at 210 V and 975 μF.39 Cells were selected with G418 at 1 mg/ml and replenished with fresh medium including G418 every 3 days. Individual colonies were cloned, checked for protein expression and used in motility assays. Cell motility assays were done using a 96-well Boyden ChemoTx® System (Neuro Probe Bethesda, MD). The chemoattractant fetal bovine serum was used in DMEM medium containing 0.1% bovine serum albumin, 10 mM HEPES and 100 U/ml penicillin and streptomycin. Serial dilutions of chemoattractants were placed in the lower part of the chamber and an 8-μm pore size polycarbonate polyvinylpyrrolidine-free membrane was sandwiched between the upper and the lower chamber. Selected cell clones were washed in PBS, resuspended in DMEM medium containing 0.1% BSA, 10 mM HEPES and 100 U/ml of penicillin/streptomycin. 7.5 × 104 selected cells were added to the upper wells of the chamber. The chamber was incubated for 2 hr at 37°C in a humidifying CO2 incubator. After the removal of cells from the upper side of the membrane, cells were stained using Gill's hematoxylin. The migrated cells were counted at sites in each wells at 40× magnification with an Olympus BX40 light microscope.37 The data presented at each concentration is the mean of 3 separate experiments.
GTP/GDP loading-exchange assays
COS-7 cells (50,000/well in a 6-well plate) were transfected with the corresponding plasmids containing HA-Cdc42, HA-Dbl, HA-pDbl, myc-nm23-H1 and balanced with control vector pA3M. After 48 hr the cells were washed 3 times in phosphate-free media (Invitrogen, Carlsbad, CA), and incubated in phosphate-free media for 2 hr at 37°C with the addition of 50 μCi orthophosphate 32P. At 2 hr cells were washed and lysed in 32P lysis buffer (20 mM Tris 7.5, 150 mM NaCl, 20 mM MgCl2, 1 mM Na3VO4, 0.5% Triton X-100, 1X protease inhibitor cocktail, 1 mM PMSF), and incubated on ice for 20 min. The lysates were clarified at 13,400g for 10 min. Samples were incubated in Protein A/G beads to preclear, followed by incubation with HA-antibody conjugated beads. Incubations were performed with rotation head-over-head, at 4°C. Samples were then washed in PBS 5 times, spun at 350g for 1 min per wash. The final sample of beads was resuspended in 32P elution buffer (20 mM Tris 7.5, 20 mM EDTA, 2% SDS).
Samples were spotted on polyethyleneimine (PEI)-cellulose thin-layer chromatography (TLC) plate were purchased from Sigma (St. Louis, MO) and allowed to dry. The plate was then incubated in a TLC chamber for 5 min in 0.1 M LiCl, 15 min in 1.0 M LiCl and finally in 1.6 M LiCl until the front reaches 1 inch from the top of the plate per the manufacturer's instruction (Sigma Aldrich St. Louis, MO). The plate was allowed to dry flat, and exposed to PhosphorImager cassette followed by analysis on a Storm PhosphoImager scanner. For Dbl phosphorylation analysis, 25% of the beads were resuspended in 2× Laemmlli's buffer and fractionated on a 10% SDS-PAGE gel. The gel was dried and exposed to PhosphorImager, and analyzed using Storm PhosphoImager scanner (Molecular Dynamics, Sunnyvale, CA).
Cell lysates were electrophoresed on SDS-PAGE gels and transferred to a nitrocellulose membranes. Blots were then probed using specific primary antibodies diluted 1:1,000, followed by fluorescently labeled secondary antibodies, Alexa Fluor 680 and Alexa Fluor 800 (Molecular Probes, Carlsbad, CA, and Rockland, Gilbertsville, PA, respectively) diluted at 1:20,000. Blots were visualized and analyzed using an Odyssey imaging system and Odyssey software (Li-Cor Lincoln, NE).
Nm23-H1 associates with the guanine exchange factor Dbl-1 in vitro
A yeast 2-hybrid screen was performed to identify the cellular binding partners of the suppressor of metastasis protein Nm23-H1. Briefly, a yeast library that contains a LEU-selectable marker and cDNAs from EBV infected LCL fused to the GAL4 acidic transactivation domain was transformed into the Y190 expressing the GAL4-DNA binding domain-Nm23-H1 fusion. The positive clones were selected and the cDNA inserts was sequenced and homology searches done through the GenBank and EMBL sequence libraries. Sequence analysis of our positive interacting clones identified the Dbl-1 oncoprotein as a strong binding partner, as determined by beta-galactosidase positive results within 30 min.
To confirm the yeast 2-hybrid interaction we did in vitro binding assays. A GST fusion protein of wild-type Nm23-H1 (GST-Nm23-H1) was generated and incubated with in vitro- transcribed/translated 35S-labeled Dbl-1 or pDbl (Fig. 1a). The original study identifying the oncoprotein Dbl-1, from the lymphoma library resulted in the cloning of the proto-oncogene pDbl.29, 32 Therefore, we used both expression constructs for comparison in this manuscript. Although a GST control protein was unable to bind either Dbl-1 or pDbl, GST-Nm23-H1 was able to bind to both proteins in an in vitro reaction. However, the Dbl-1 binding showed a relatively higher affinity compared to pDbl binding (Fig. 1a). Importantly, bacterially-generated GST-Nm23-H1 also bound Dbl-1 expressed from a heterologous promoter in the context of 293T-transfected cell extracts (Fig. 1b, left panel). Again, the binding to pDbl was relatively lower under similar conditions in vitro (Fig. 1b, right panel). To confirm this result, we performed immunoprecipitation assays using lysates from 293T cells co-transfected with expression constructs for myc-tagged Nm23-H1 (pA3M-nm23-H1) and either HA-tagged Dbl-1 (pcDNA3HA-Dbl-1) or pDbl (pcDNA3HA-pDbl). Indeed, the Dbl-1 protein was detected in the anti-myc immunoprecipitated sample (Fig. 1c left panel), and the reverse experiment where Nm23-H1-myc was detected in the anti-Dbl-1 immunoprecipitated complex (Fig. 1d left panel). Interestingly, pDbl was strongly detected in the anti-myc immunoprecipitation assay (Fig. 1c right panel), however a much reduced level of Nm23-H1-myc was detected by immunoprecipitation using an anti-Dbl antibody (Fig. 1d right panel).
Nm23-H1 associates with endogenous Dbl-1 in Burkitt's lymphoma cells
We showed interaction of Nm23-H1 with Dbl-1/pDbl using heterologous expression systems. To further support these results, we tested the endogenous interaction of Nm23-H1 with Dbl-1 and pDbl. Dbl-1 was identified from a diffuse B-cell lymphoma DNA library; therefore, we used cell lysates collected from the Burkitt's lymphoma cell line DG75. Endogenous expression of pDbl-1 in B lymphoma cell lines was low or undetectable; however, Dbl-1 expression was at relatively higher levels resulting in a strong signal as determined by Western blot analysis (data not shown). Lysates of DG75 cells were incubated with GST beads or GST-Nm23-H1 fusion protein bound to beads. The results showed that Dbl-1 was pulled down with a GST-Nm23-H1 fusion protein but not with the GST alone control beads (Fig. 2a). However, pDbl was not seen by pull down assays as determined by Western blot analysis (data not shown). To further support the pull down results, we performed immunoprecipitation assays in both directions using endogenously expressed proteins. The results showed that endogenous Dbl-1 and Nm23-H1 formed a complex with each other in the B cell line DG75 (Figs. 2b and 2c). The levels of endogenous Nm23-H1 were relatively low in B lymphoma cell lines compared to Dbl-1, and so the resulting Nm23-H1 signals were also lower than for Dbl-1 (Figs. 2b and 2c).
Nm23-H1 colocalizes to similar compartments with Dbl-1 and pDbl in cells
To determine whether Nm23-H1 and Dbl-1 or pDbl can localize to similar cellular compartments in vivo, we performed confocal immunofluorescence analysis with Nm23-H1 and Dbl-1 expressed from heterologous promoters in a human cell line HEK-293T cells. The results showed that Nm23-H1 staining was primarily cytoplasmic with some punctuate signals also noted at the periphery of the cells suggesting cell membrane localization (Fig. 3a left panels). Notably, the Dbl-1 staining pattern was also a distinct punctate pattern and was primarily at the periphery of the cells with enhanced cell membrane ruffling (Fig. 3a middle panel arrow). In contrast, pDbl staining was present diffusely throughout the cytoplasm and lacked a focused pattern at the periphery or plasma membrane of the cells (Fig. 3a right panels arrow). In cotransfected cells, there was clear overlap of Dbl-1 and Nm23-H1 signals at the edge of the cell which is consistent with a potential colocalization and interaction within membrane-associated compartments (Fig. 3b). Interestingly, in the presence of Nm23-H1, the Dbl-1 signal was predominantly at the cell membrane indicating a potential role in focusing Dbl-1 to sites at the cell membrane (Fig. 3b). Importantly, this overlap at the membrane was not observed when Nm23-H1 was cotransfected with pDbl but showed a diffused distribution with some overlap in the cytoplasm (Fig. 3c arrow). The presence of Dbl-1 clearly induced cell ruffling, which was reduced in the presence of Nm23-H1 (Figs. 3a compare with 3b).
Nm23-H1 kinase deficient mutants colocalizes with Dbl and pDbl in vivo
Experiments performed in the mammalian cell line COS-7 cells for Nm23-H1, Dbl-1 and pDbl staining showed similar results to that seen above in HEK 293T cells. The enhanced cell membrane ruffling was also clearly observed in HEK-293T cells as well as in COS-7 cells (Fig. 4). Previously, a number of Nm23-H1 point mutants have been generated with specific kinase phenotypes.40 The H118F mutant of Nm23-H1 lacks the ability to transfer a phosphate to NDP to generate NTP in addition to its inability to autophosphorylate at histidine-118 (NDP kinase deficient). The P96S mutant exhibits normal autophosphorylation and lower histidine kinase activity, but is deficient in its phosphotransfer activity, and has lost the ability to suppress motility.41, 42 We hypothesized that if Nm23-H1 kinase activity is important for Dbl-1 GEF activity, then these Nm23-H1 mutants may not colocalize with Dbl-1 in the same way as wild-type Nm23-H1. To this end, immunofluorescence assays were performed with the Nm23-H1 mutant H118F. The results showed that H118F stained similar to wild-type Nm23-H1 in its membrane localization (Fig. 4b, left panels), while the Nm23-H1 mutant P96S localized strongly to the nucleus as well as the periphery of the cells (Fig. 4b, right panels). As seen with the wild-type Nm23-H1, Dbl-1 signal overlapped with both the H118F and P96S mutants at the cell membrane although some nuclear signal was clearly seen for the P96S mutant (Fig. 4b). The staining of pDbl was diffusely cytoplasm, lacking any specific signal at the cell membrane or in the nucleus; however, the signals for P96S mutant with pDbl was typically more cytoplasmic with punctate signals (Fig. 4c, left and right panels, respectively).
Nm23-H1 interferes with the ability of Dbl-1 to load GTP onto CDC42
The interaction of Nm23-H1 with Dbl-1 and pDbl in vivo and the ability of Nm23-H1 to function as a metastatic suppressor, led us to explore the possibility that an effector molecule downstream of Dbl-1 and pDbl involved in cytoskeletal rearrangement may be affected. Dbl-1/pDbl have been shown to function as a guanine exchange factor for human Cdc42,43 a Ras family member involved in a number of biological processes including cytoskeleton rearrangement.24 To determine if Nm23-H1 can regulate the activity of the GEF Dbl-1, we utilized a TLC-based assay designed to test the level of radiolabeled GTP loaded onto a terminal molecule, in this case Cdc42. Indeed, both Dbl-1 and pDbl are able to load GTP onto Cdc42 in multiple assays (Fig. 5a lanes 2 and 3). Interestingly, Nm23-H1 interferes with the loading of GTP onto Cdc42 by Dbl-1 with an almost 8-fold reduction in activity (Fig. 5a, lane 4). To a lesser extent the loading of GTP onto Cdc42 by pDbl was only mildly affected (Fig. 5a, lane 5). The Nm23-H1 mutants also prevented the loading of GTP onto Cdc42 (data not shown). This is consistent with the ability of the mutants to colocalize with Dbl-1 similar to that seen with wild-type Nm23-H1. This further suggests that the loss of NDP and histidine kinase activity is not critical for the inhibitory function of Nm23-H1 on Dbl-1 GEF activity.
Nm23-H1 abrogates the phosphorylation of Dbl-1
Similar to other oncoproteins, Dbl-1 is actively regulated by phosphorylation.46 We therefore analyzed the level of phosphorylation of Dbl-1 and pDbl in the context of our TLC assays (Fig. 5b). Indeed, Dbl-1 was phosphorylated when transfected into COS-7 cells along with Cdc42 (Fig. 5b, lane 2). This phosphorylation was not observed if Nm23-H1 is cotransfected with Dbl-1 and Cdc42 (Fig. 5b, lane 4). It should also be noted that, consistent with previous data,46 phosphorylation of pDbl was almost undetectable when compared to that of Dbl-1 (Fig. 5b, compare lanes 2 and 3).
Expression of Dbl-1 and pDbl can reverse the ability ofNm23-H1 to suppress cell migration
A well-described in vitro phenotype of the suppressor of metastasis Nm23-H1 is its ability to inhibit cell motility when stably expressed in the breast cancer cell line MDA-MB435.47 Given the interaction between Nm23-H1 and Dbl-1, and the well-described role of Dbl-1 as an upstream regulator of cell migration, we asked whether Dbl-1 might rescue cell motility in Nm23-H1-suppressed cells. To this end, we established Nm23-H1, Dbl-1 and pDbl-expressing MDA-MB435 clones, either individually or in combination. Specific protein expression for the individual molecules was confirmed by Western blotting (data not shown). The results of cell motility assays are shown in Figure 6. As was demonstrated previously, the Nm23-H1 expressing clone showed significantly reduced cell motility when compared with wild-type (Fig. 6, lane 2). The Dbl-1 expressing clone had enhanced cell motility as compared to vector control, and importantly Dbl-1 did in fact rescue the ability of Nm23-H1 to suppress motility in the dual-expressing cell line (Fig. 6, lane 3 and lane 4). The pDbl expressing clone behaved similarly to Dbl-1 in that it rescued Nm23-H1 suppression of motility, although it had little effect on MDA-MB435 cell motility when expressed alone (Fig. 6, right panel).
Using a yeast-2-hybrid screen, we initially identified Dbl-1 as an Nm23-H1 binding partner. We confirmed the interaction of these proteins both in vitro and in vivo. Overall, our data suggest that the affinity of Nm23-H1 for Dbl-1 was greater than the interaction seen with pDbl. To confirm the interaction between endogenous Dbl-1 and Nm23-H1, lysates from Burkitt's lymphoma cell line DG75 was tested which showed that these 2 molecules can form a complex in cells. Dbl-1 was originally identified from a diffused B cell lymphoma (DBCL) in humans. We could demonstrate the endogenous association between Dbl-1 and Nm23-H1 in the B lymphoma cell line DG75. The in vitro binding data showed that Dbl-1 had a stronger binding to Nm23-H1 than pDbl. This suggested that its interaction with pDbl was affected by the amino terminal domain of pDbl. Oncogeninc Dbl-1 has a deletion of 497 amino acids in amino terminus of pDbl sequence.48 Therefore, less of this amino terminal domain allowed for greater access to the carboxy terminal domain which is likely the binding domain. The expression of amino terminus of pDbl-1 also led to attenuation of GEF activity, suggesting an auto-inhibitory regulatory mechanism.49 This auto-inhibitory regulation of pDbl is negatively regulated by the heat shock cognate protein 70 (Hsc70), which interacts with the amino terminus of pDbl (spectrin domain) as well as the PH domain.50 Also, Hsp90 interacts with the spectrin and PH domains of pDbl, leading to protection of pDbl from ubiquitination and degradation.51 Interaction of these proteins at the PH domain may be responsible for the reduced binding activity between Nm23-H1 and pDbl. Alternatively, Nm23-H1 may bind to oligomers of Dbl-1. The amino terminus deleted Dbl-1 can form homo-oligomers as well as hetero-oligomers with a close relative, but not with other GEFs.48 Mutants that can no longer oligomerize still possess GEF activity in vitro, but are less potent at activating Cdc42 and Rho in vivo.36 It is possible that the formation of multiple complexes may be required for generation and regulation of multiple signaling pathways involved in the biochemical activities of Nm23-H1.
This report suggests that the PH domain can play an important role in activation and stabilization of pDbl. Importantly, Nm23-H1 interacts with Dbl-1 and these oncogenic Dbl family members are constitutively activated GEF.36, 52 However, these amino terminus mutants could not be reconstituted in vitro.53 The Dbl family members have DH and PH domain likely to be important for GEF activities and the homologous oncogenic mutations are located outside the catalytic domain which is in the amino terminus of the proto-oncoprotein.50–55
Nm23-H1 was expressed primarily in the cytoplasm with some intense signal at the cell membrane with some nuclear staining. However, Dbl-1 was expressed with a defined signal at the edge of the cells suggesting a predominantly plasma membrane localization. We also observed a punctate staining pattern of Dbl-1 at membrane foci in HEK-293T cells. This membrane localized foci were lost when Nm23-H1 was co-expressed with Dbl-1, suggesting that Nm23-H1 can interact with Dbl-1 at these focal adhesion junctions along the cell membrane to regulate its function. Moreover, in the coexpressed HEK-293T cell Nm23-H1 was also observed in the periphery of the nuclear membrane. Although Nm23-H1 localization also overlaps with pDbl in the cytoplasm, there were little or no specific colocalization signals at the cell membrane, suggesting that the PH domain of pDbl may be blocked or was no longer available to associate with the cell membrane when Nm23-H1 was expressed.
Dbl-1 activates Cdc42 by exchanging the bound GDP for GTP, while the activation of the RhoA and JNK/Cyclin D1 is mediated by the PH domain of Dbl-1.56 Our data demonstrated that Nm23-H1 can interfere with the ability of the Dbl-1 oncoprotein to load GTP onto Cdc42. This effect was more dramatic with Dbl-1 as compared to pDbl indicating a potential regulatory domain in pDbl which is likely located within the amino terminal 500 amino acids. This result suggests that Nm23-H1 inhibits Dbl-1/Cdc42 signaling in B cell lymphoma cells. Further analysis of the Nm23-H1 mutant activities is ongoing which should provide new information as to relevant functions in the context of cell migration.
Cell migration is an important feature in normal development and is associated with numerous pathological processes. We wanted to determine whether or not the ability of Nm23-H1 to function as a metastasis suppressor is influenced by its interaction with the Dbl-1 oncoprotein. Our motility assay results showed that in the presence of Nm23-H1, Dbl-1 rescued the cell migration activity. Dbl-1 activates the Rho GTPase family member, Cdc42 which regulates cytoskeletal reorganization and cell motility and cell adhesion36 as well as the production of reactive oxygen species.57, 58 Activated Rac1 and Cdc42 interact with IQGAP1, an effector of Rac159, 60 and these interaction can lead to strong adhesive activity.61 The overexpression of IQGAP1 reduces E-cadherin mediated cell–cell adhesion by interacting with β-catenin, causing α-catenin to dissociate from the cadherin/catenin complex.62 Dbl family member Tiam1 activates the Rho GTPase Rac, and this interaction results in suppression of cell invasion by increasing E-cadherin mediated cell-to-cell adhesion in epithelial cell.63 However, both Tiam1 and Rac proteins can induce invasion of T lymphoma cells.64, 65 Our results shows that Nm23-H1 can regulate the signaling events which was strongly inhibited by Dbl-1 GEF activity on the Cdc42, but the suppression of cell motility activity mediated by Nm23-H1 was offset by this interaction. Therefore, this interaction can lead to enhancement of cell migration. Dbl-1 has transforming activity,66 and can regulate cell growth through downstream pathways which include Cdc42-PAK, and the Rho-ROK pathways.25 Recent studies have implicated focal adhesion kinase (FAK) as a positive regulator of cell migration as a nonreceptor protein tyrosine kinase. Autophosphorylation and activation of FAK are critical for enhanced adhesion induced FAK activation and cell migration responses.67 Integrin mediated cell adhesion can increase tyrosine phosphorylation of FAK by stimulating cells with a variety of soluble growth factors, neuropeptides and bioactive lipids.68 The small GTPase Arf6 has been shown to regulate endocytic trafficking and inhibition of Arf6 function impairs both cell adhesion and motility.69 Previously, we reported that expression of integrin is modulated by Nm23-H1 and the Epstein-Barr virus nuclear antigen 3C through interaction with the GATA-1 and Sp1 transcription factors.70 Additionally, Nm23-H1 has nucleoside diphosphate kinase (NDPK) activity, DNA transactivation activity, DNA nuclease activity and serine or histidine protein kinase activity.7 Nm23-H1 phosphorylated on S122 interacts with h-prune which is associated with an increase in metastatic activity in a number of tumors. Interestingly, phosphorylation or mutation of this residue affects the ability of Nm23-H1 to suppress metastasis.71 However, these Nm23-H1 associated functions are still somewhat unclear. The regulation of cell migration activity of Nm23-H1 and its interaction with Dbl-1 is potentially important for regulation of the ability of Dbl-1 to affect cell growth and transformation.
It was previously demonstrated that phosphorylation is necessary for the guanine exchange activity of another Dbl family member Ect2.72 Interestingly, in our studies Nm23-H1 expression resulted in loss of phosphorylation of Dbl-1 and this lack of phosphorylation correlated with the lack of GTP exchange to Cdc42. Taken together, these data suggest that as with Ect2,72 Dbl-1 phosphorylation is necessary for its nucleotide exchange activity. Surprisingly, we did not detect any significant level of phosphorylation of pDbl. However, a previous study had shown that pDbl was phosphorylated at a substantially lower level than that seen for Dbl-1.44 This may suggest that accessibility of the phosphorylation sites of pDbl is likely to be protected by folding and interaction of the amino terminal auto-inhibitory or spectrin domain with the PH domain and so blocking access to the phosphorylation site, possibly in complex with HSC70 as an inactive complex.48 It is most likely that the phosphorylation sites lie within the interacting PH domain. Future work will be involved in identification of the specific kinase as well as mutagenesis of the specific residues for loss of function. These studies are currently ongoing in the laboratory.
A recent study suggested a possible interaction of Nm23-H1 with Cdc42.73 In this study, the authors showed a reduction of Cdc42 activities in the presence of N-myc, and suggested that this reduction is likely due to the expression and possibly interaction of Cdc42 with Nm23-H1.73 This study showed an association from a GST pull-down assay in cells. Nm23-H1 may also function as a GTPase activating protein (GAP), maintaining Cdc42 in a GDP bound state.73 Moreover, our data suggested that the presence of Nm23-H1 can lead to an increase in Cdc42-GDP bound compared to the GTP bound molecules, but only in the presence of the Dbl-1 GEF.
This study now shows that Nm23-H1 forms a stable complex with the GEFs Dbl-1, and that this interaction regulates the activation of Cdc42 by blocking the GDP/GTP exchange activity of Dbl-1. These activities induced by Nm23-H1 expression resulted in modulation of membrane ruffling and focal adhesion activities of these cells. Thus, these data strongly suggest that Nm23-H1 is capable of regulating tumor metastasis and invasiveness by regulating the activities of cellular molecules involved in the control of cytoskeletal rearrangements. In the plasma membrane, Dbl-1 can exchange GDP with GTP on Cdc42. Cdc42 would then be activated leading to stimulation of cellular pathways leading to membrane trafficking, DNA synthesis, transcription activation, translation regulation and cytoskeletal reorganization. Nm23-H1 may disrupt the activation of Cdc42 by modulating the GDP/GTP exchange by Dbl-1. Our model shows that Nm23-H1 is bound to Dbl-1 on the cell membrane (Fig. 7). This block in ability to exchange GDP/GTP is due to the physical interaction of Nm23-H1 with Dbl-1, as a mutant Nm23-H1 lacking NDP kinase, autophosphorylation or histidine kinase activities. This interaction might account for both the loss of GEF function, and the loss of Dbl-1 phosphorylation. Most importantly, reduction of cell motility activity by Nm23-H1 is rescued by expression of Dbl-1. Further studies have been pursued to show the downstream activities affected due to the modulation of the functions of these signaling molecules.
We are thankful to Dr. Patricia S. Steeg (NIH) for providing nm23-H1, H118F and P96S, Dr. Margaret Chou (University of Pennsylvania, PA) for providing Dbl, pDbl, Cdc42 cDNA constructs and for technical suggestions. We also thank members of Robertson laboratory for discussion and suggestions with these studies. S.C.V. is supported by the Lady Tata Memorial Trust. K.L. is a special fellow and E.S.R. is a Scholar of the Leukemia and Lymphoma Society of America.