Nucleoside Diphosphate Kinase A/nm23-H1 Promotes Metastasis of NB69-Derived Human Neuroblastoma¹

Tumor metastasis is a complex and dynamic process, involving several cellular events starting with the primary tumor site and ending with new tumor establishment in distal sites. These cellular events include detachment from the primary tumor, local tissue invasion, intravasation, survival in the circulatory system, extravasation, and colonization at target organs. Tumor metastasis remains a major cause of cancer mortality because detection markers and treatment options are currently limited for patients with metastatic tumors.

Nucleoside diphosphate kinase A (NDPK-A) is one of the proteins known to play a role in tumor metastasis. NDPK-A is encoded by the nm23-H1 gene and is located at human chromosome 17q21.3 (1-3). Its role in tumor metastasis was initially identified based on its reduced expression in murine melanoma cells with high metastasis potential (2). Subsequent studies have revealed that the expression level of NDPK-A is negatively correlated with the metastatic potential of certain human tumors, including melanoma and breast carcinoma (4), indicating a metastasis-suppressing role of NDPK-A. Transfection and animal studies indeed support the idea that NDPK-A acts as a metastasis suppressor in these types of tumors (4). Intriguingly, a positive correlation between NDPK-A level and metastatic potential has also been observed in other human tumors including neuroblastoma, osteosarcoma, and pancreatic carcinoma (5-9). However, it remains to be determined whether NDPK-A actually acts as a metastasis promoter in these tumor types.

Neuroblastoma arises from the neural crest of the sympathetic nervous system and accounts for 25% of infants and 7% of children with cancers in the United States. Most patients with limited stages (i.e., I and II) of neuroblastoma can be cured by the surgical removal of localized tumors (10, 11). However, the majority of patients with advanced stages (i.e., III and IV) of neuroblastoma die from metastatic disease despite intensive multimodal therapy (12). The gain of the chromosomal segment of 17q21-qter, within which nm23-H1 is located, occurs in 54% to 65% of neuroblastoma patients and is associated with poor clinical outcomes (13-15). Amplification and overexpression of nm23-H1 have been found in 14% to 30% of advanced neuroblastomas (5, 6, 16). Furthermore, we reported previously a unique point mutation of nm23-H1 that results in a Ser120→Gly substitution in NDPK-A (NDPK-A^S120G) in 21% of patients with advanced neuroblastomas (17). This mutation appears to be specific for advanced neuroblastoma because it has not been found in patients with limited stages of neuroblastoma or with leukemia or breast carcinoma (17). The association of these NDPK-A aberrations with the high metastatic potential of human neuroblastoma strongly suggests that, in this tumor type, NDPK-A is a metastasis promoter.

The human neuroblastoma NB69 cell line was established from a primary tumor in the adrenal gland (18), where ∼50% of human neuroblastomas originate (19). NB69 cells display low invasiveness and no N-myc amplification (20), the latter being associated with advanced stages of neuroblastoma (21). For these reasons, we therefore chose the NB69 line as our cell model to examine the role of NDPK-A independent of N-myc in neuroblastoma metastasis. We generated stable transfectants from NB69 cells to examine the effects of NDPK-A aberrations, associated with advanced neuroblastomas, on cellular events involved in the metastatic process. We also developed an orthotopic xenograft animal model with severe combined immunodeficient (SCID) mice to determine the incidence and colonization of neuroblastoma metastasis. Our results indicate that NDPK-A acts as a metastasis promoter rather than a metastasis suppressor in human neuroblastoma derived from NB69 cells.

We established from NB69 cells the stNB series of stable transfectants expressing NDPK-A aberrations found in patients with advanced neuroblastomas. The stNB-M and stNB-W transfectants constitutively overexpress NDPK-A^S120G and its wild-type, respectively. The vector-transfected stNB-V serves as a negative control. Based on Western blot analysis, all three randomly selected stNB-V clones expressed a very low level of NDPK-A protein (Fig. 1A and B), similar to the parental NB69 cells (data not shown). Compared with stNB-V, we observed a 10 to 25 times greater protein level of ectopic NDPK-A or NDPK-A^S120G in randomly selected stNB-W and stNB-M clones, respectively (the ∼21 kDa band in Fig. 1A and B). To our surprise, another form of NDPK-A with slower electrophoretic mobility was also detected in stNB-W and stNB-M transfectants with a monoclonal antibody specific for NDPK-A (Fig. 1A and B) but not for the NDPK-B isoform (data not shown). The protein levels of NDPK-A in stNB-W and stNB-M were comparable with 3 to 10 copies of the nm23-H1 gene detected in advanced stages of human neuroblastoma (17) as well as a 2- to 20-fold increase in NDPK activity in human tumors (22).

The stNB series coexpress a humanized Renilla green fluorescent protein (GFP) from the same transcript of ectopic NDPK-A in a nonfusion form (Fig. 1C), serving as a convenient cell marker (Fig. 1D). All randomly selected clones in the stNB series displayed different GFP intensity based on flow cytometry (data not shown). The stNB-W and stNB-M clones stably coexpressed NDPK-A with GFP in the presence of G418. In addition, this coexpression was faithfully maintained without G418 selection pressure for at least 1 month (data not shown), which is necessary for the in vivo studies.

Both protein and mRNA levels of NDPK-A are elevated during the S phase of the cell cycle (23), which suggests that a high level of NDPK-A may affect the proliferation rate. As determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay and cell counting, the doubling times of stNB-M, stNB-V, and stNB-W transfectants were 22.9 ± 2.9, 23.6 ± 2.3, and 21.2 ± 1.1 hours, respectively. Thus, NDPK-A overexpression did not significantly affect proliferation of neuroblastoma cells.

Retinoic acid (RA) induces neuroblastoma cells to undergo neuronal differentiation (i.e., neurite extension) while arresting the growth (24). Similar to the parental NB69 cells (25), ∼45% of stNB-V transfectants displayed neurite extension when treated with 30 μmol/L all-trans RA for 3 days (Fig. 2A). Under the same treatment condition, however, only 1% to 2% of stNB-W and stNB-M displayed neurite extension (Fig. 2A). Concomitant with neurite extension, RA inhibited stNB-V growth, whereas stNB-W and stNB-M escaped RA-induced growth arrest and accumulated cell numbers similar to the untreated counterparts (Fig. 2A). These results indicate that overexpression of NDPK-A or NDPK-A^S120G abrogates RA-induced neuronal differentiation and maintains neuroblastoma cells in a proliferating state.

Decreased cell adhesion is associated with the metastatic phenotype of tumors. Relative to stNB-W and stNB-V, we consistently observed that stNB-M transfectants displayed a subtle difference in cell adhesion. At 20% or 95% cell confluency, stNB-M transfectants took 20% to 30% less time than stNB-V or stNB-W transfectants to detach from mammalian cell culture dishes when treated with 0.25% trypsin. Mammalian cells adhere and grow well on the hydrophilic surface of tissue culture dishes and plates. The surface of bacterial culture dishes is relatively more hydrophobic, which might manifest the subtle difference observed in stNB-M adhesion. Indeed, we found that whereas stNB-W and stNB-V attached and spread onto bacterial Petri dishes within 2 days after plating, stNB-M remained rounded without spreading on the dish surface for at least 5 days (Fig. 2B). Rounded stNB-M cells were viable based on trypan blue exclusion (data not shown). In addition, stNB-V, but not stNB-W or stNB-M, displayed neurite extension 4 to 5 days after plating in Petri dishes (Fig. 2B), which further supports our earlier observation that NDPK-A aberrations abrogated neuronal differentiation.

To metastasize effectively to distal organs, tumor cells must improve their survival ability and cloning efficiency. We first examined whether NDPK-A affects serum-independent survival of neuroblastoma cells. The stNB series of transfectants were cultured at two different serum levels for 8 days without medium replenishment, and cell survival was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay on days 1, 2, 5, and 8. Compared with cells in the growth medium containing 10% serum, 90%, 46%, and 74% of stNB-M, stNB-V, and stNB-W, respectively, survived in medium containing 2% serum on day 8 (Fig. 2C). To determine the cloning efficiency, cells were singly sorted by flow cytometry into 96-well plates, three plates per transfection type. The cloning efficiency was calculated as 32 ± 3.2%, 26 ± 2.2%, and 15 ± 1.2% for stNB-W, stNB-M, and stNB-V, respectively. Taken together, our results indicate that NDPK-A aberrations significantly increased cloning efficiency (P < 0.02, t test) and serum-independent cell survival (P < 0.0001, t test) of human neuroblastoma cells by 1.6- to 2-fold.

Migration is essential for invading tumor cells to translocate from the primary tumor site to distal organs. Using a modified Boyden chamber system, GFP-labeled stNB-M, stNB-V, and stNB-W were allowed to migrate toward 5% serum in 15 hours. Migrated cells attached underneath the chamber membrane were detected with an inverted fluorescent microscope. After normalizing with the plating cell density, migrated stNB-M and stNB-W cells were 2.2- and 1.3-fold higher than migrated stNB-V, respectively (Fig. 3A), indicating that NDPK-A^S120G significantly (P ≤ 0.05, t test) enhanced neuroblastoma cell migration.

We further investigated whether NDPK-A aberrations affect neuroblastoma colonization on soft agar. Transfectants were resuspended in 0.3% agar before being overlaid onto 0.5% agar containing the growth medium. Colonies consisting of at least 50 cells were counted 14 days after plating. The stNB-M and stNB-W cells exhibited 3.6- and 2.8-fold more colonies on soft agar, respectively, than stNB-V (Fig. 3B), indicating that NDPK-A aberrations provide neuroblastoma cells with additional in vitro colonization ability (P ≤ 0.0007, t test).

The NB69 cell line was established from an adrenal neuroblastoma of a young patient (18). We therefore performed an intraadrenal (i.e., orthotopic) injection of the stNB-M, stNB-V, or stNB-W clones in SCID mice 7 to 8 weeks old using eight mice per clone. All animals developed the primary tumors near the injection site, which were visible as early as 6 days after injection with the tumor reaching ∼5 mm in diameter 14 to 18 days after injection. The primary tumor was allowed to grow ∼15 mm in diameter or until the animals showed discomfort, which ever occurred first, corresponding to 28 ± 2.5, 24 ± 0.6, or 23 ± 2.3 days after injection with stNB-M, stNB-V, or stNB-W, respectively (Table 1). During necropsy, the primary tumor was found exclusively in the injected adrenal gland (i.e., left) and extended to the left kidney in all mice examined. The average size of primary tumors derived from stNB-M, stNB-V, and stNB-W was 11 ± 1.7, 13 ± 1.7, and 12 ± 1.7 mm in diameter, respectively (Table 1). The latency and size of primary tumors in all animals did not differ significantly among the stNB-V, stNB-M, and stNB-W transfectants. This suggests that NDPK-A does not play a role in primary neuroblastoma development.

When the adrenal neuroblastoma was removed to establish the NB69 cell line, the patient displayed no detectable metastases. At the time of death despite 11 months of radiation and chemotherapy, however, the primary neuroblastoma had metastasized to the lung, liver, and sternum (18). We focused on the lung metastasis to avoid potential spillover into the abdominal cavity during intraadrenal injection. Macroscopic metastases were readily detectable in animal lungs by monitoring the GFP-labeled transfectants without microscopy (Fig. 4A). Macroscopic lung metastases (i.e., one or more fluorescent tumor nodules on the lungs) were found in 11, 5, and 8 of the animals injected with stNB-M, stNB-V, and stNB-W, respectively (Table 1). With light microscopy, microscopic lung metastases (i.e., one of more tumor foci containing at least 10 cells per cluster on the lungs) were detected in six, two, and five additional animals injected with stNB-M, stNB-V, and stNB-W respectively (Table 1). By adding the number of animals with macroscopic or microscopic metastases, the incidence was 71%, 29%, and 56% for animals injected with stNB-M, stNB-V, and stNB-W, respectively. Compared with stNB-V, stNB-M (P < 0.004) and stNB-W (P = 0.04) significantly promoted neuroblastoma metastasis in SCID mice based on Fisher's exact test.

Tumor growth at the secondary site (i.e., metastatic colonization) is a key regulatory step in metastasis. We therefore further examined the effect of NDPK-A aberrations on the ability of the stNB series of transfectants to proliferate and colonize in animal lungs. Counting directly from the fluorescent images taken from only one side of lungs, such as Fig. 4A, three animals injected with stNB-M, one with stNB-W, and none with stNB-V displayed 15 to 40 tumor nodules per mouse lung. The rest of animals with lung metastasis displayed 1 to 10 tumor nodules, and nodule sizes were predominantly larger in animals injected with stNB-M or stNB-W than that with stNB-V. Measuring only the largest microscopic tumor focus in the lungs of each animal, size distributions of tumor foci derived from stNB-M, stNB-V, and stNB-W were 0.3 to 2.8 mm (1.3 mm on average), 0.1 to 1.1 mm (0.4 mm on average), and 0.3 to 2.2 mm (1.0 mm on average) in diameter, respectively (Fig. 4B). These findings indicate that a high level of NDPK-A or NDPK-A^S120G promotes metastatic colonization of human neuroblastoma in animal lungs.

By G418 selection in culture, we reestablished cell lines from the primary and secondary tumor foci in animals injected with stNB-M, stNB-V, and stNB-W. We found that 100% of the cells in reestablished lines expressed GFP, as detected by fluorescent microscopy (data not shown). Based on Western blot analysis, protein levels of NDPK-A or NDPK-A^S120G in the reestablished cell lines were similar to that in the stNB series of transfectants used for animal injection (Fig. 4C). The findings indicate that ectopic NDPK-A was stably maintained and coexpressed with GFP during tumorigenesis and metastasis in our animal model of human neuroblastoma, supporting the role of NDPK-A in neuroblastoma metastasis.

In this study, we provide the first evidence that NDPK-A behaves as a metastasis promoter in human neuroblastoma at least in neuroblastoma derived from NB69 cells. This metastasis-promoting role of NDPK-A in neuroblastoma is independent of cell proliferation and primary tumor development, similar to the metastasis-suppressing role of NDPK-A in other tumor types (4). The dual role of a protein in tumor metastasis is not unprecedented because CD44 also plays opposite roles in the metastasis of neuroblastoma and breast carcinoma (26). Neuroblastoma is derived from the neural crest, which is destined to migrate and differentiate during embryonic development. NDPK-A expression is high in early development and decreases with advanced gestational age in the human placenta (27). It is plausible that different genetic backgrounds between embryonic tumors (e.g., neuroblastoma) and nonembryonic tumors (e.g., breast carcinoma) may allow NDPK-A to play different roles in metastasis. Alternatively, genetic alterations may accumulate differently among tumor types and/or between low invasive (e.g., neuroblastoma NB69) and high invasive (e.g., breast cancer MDA-MB435) cells (4) used in studying the role of NDPK-A in metastasis. Similarly, we cannot rule out that certain genetic alterations unique to NB69 and not occurring in other human neuroblastoma cell lines may cooperate with NDPK-A in promoting the metastasis. It is also likely that NDPK-A acts as a metastasis promoter in neuroblastoma because of the presence of an additional form of NDPK-A, which displays a slow electrophoretic mobility. This variant form may be a post-translationally modified NDPK-A, or nm23-H1B, a splice variant of NDPK-A that contains 25 additional residues at the NH₂ terminus (28). We are currently investigating the nature of this variant form of NDPK-A.

The metastasis-suppressing role of NDPK-A has been correlated with its motility-inhibitory effect on human breast cancer and murine melanoma cells (29, 30). In human neuroblastoma NB69 cells, however, NDPK-A did not exhibit any motility/migration-suppressing effect. In fact, NDPK-A^S120G significantly enhanced the migration of neuroblastoma cells by ∼2-fold. Similarly, the S120G mutation of NDPK-A also allows breast cancer cells to become more mobile by abrogating its motility-inhibitory effect (29). The S120G mutation of NDPK-A may increase cell motility/migration by reducing its phosphotransferase activity (31, 32) and/or the two-component kinase-like activity (33). Alternatively, the S120G mutation may increase cell motility/migration by altering its protein folding and protein-protein interactions (31, 34).

Tiam1 is one of proteins that interact with NDPK-A (35). NDPK-A inhibits Tiam1 activity specific for Rac1 and therefore may affect Rac1-mediated actin polymerization and lamellipodia formation, which are essential for cell adhesion and migration (35-37). Interestingly, NDPK-A protein has been detected in lamellipodia of human fibroblasts (38), which is likely due to an indirect interaction with integrin cytoplasmic domain-associated protein-1α via NDPK-B (1, 38) and/or via Tiam1 and Rac1 (39). It is known that integrin cytoplasmic domain-associated protein-1α associates with the cytoplasmic domain of β₁ integrin (40), and the latter is up-regulated by overexpression of DR-nm23 (another member of the NDPK family) in human SK-N-SH and murine N1E-115 neuroblastoma cells (41). It would be interesting to investigate whether Tiam1 and/or integrins are involved in increased migration and reduced adhesion of NB69 cells expressing NDPK-A^S120G.

RA induces neuronal differentiation and the growth arrest of neuroblastoma cells (24) and has been used for treating neuroblastoma patients (42). Neuroblastoma cells containing N-myc amplification have been shown to develop resistance to RA (43). NB69 cells display no N-myc amplification (18, 20) and undergo growth arrest and neuronal differentiation in responding to RA (25). However, overexpression of NDPK-A, either wild-type or S120G mutant, abrogated the effect of RA on NB69 cells. This abrogation may be due to the activity of certain interacting proteins altered by NDPK-A aberrations, such as Tiam1 and Rad (44). It is known that overexpression of Tiam1 or Rad stimulates cell spreading and neurite extension in the murine neuroblastoma N1E-115 cells (45, 46). Interestingly, NDPK-A overexpression in combination with N-myc amplification in human neuroblastoma IMR-32 cells enhances neuronal differentiation in the presence or absence of RA (47). When considering RA treatment, and until the regulation of neuronal differentiation is fully understood, it is important to determine whether neuroblastoma patients display NDPK-A aberrations and/or N-myc amplification because these genetic alterations have occurred individually or in combination (5, 6, 17).

In conclusion, NDPK-A does not behave as a metastasis suppressor in human neuroblastoma derived from NB69 cells. Overexpression and S120G mutation of NDPK-A promote neuroblastoma metastasis not only by preventing neuronal differentiation but also by increasing cell survival and colonization. Additionally, NDPK-A^S120G is able to reduce cell adhesion while increasing migration, rendering it a more potent metastasis promoter than its wild-type. Our findings suggest that NDPK-A is a potential marker for predicting the clinical outcome of neuroblastoma patients and the effectiveness of treatment with RA. Our findings also expand the role of NDPK-A from a metastasis suppressor to a metastasis promoter in certain tumor types, cautioning NDPK-A-based diagnostic and therapeutic development.

Using previously described pCRII-nm23H1-wt and pCRII-nm23H1-mut plasmids as templates (31), the respective cDNAs were amplified with a pair of primers: A-70 (GTT GGA TCC CAG CTG GAA GGA ACC ATG GC) and A-494 (GGT TCT CGA GCA TGG GAA GGA GGG GAA ATG G). The PCR products were cloned into the pIRES-hrGFP-1a vector (Stratagene, La Jolla, CA) followed by an insertion of the Neo-resistant gene via the pExchange module EC-Neo (Stratagene). The resulting expression plasmids were confirmed by DNA sequencing in the Cancer Center Core Facility of the University of California at San Diego (La Jolla, CA).

The human neuroblastoma NB69 cell line (18), generously provided by Dr. Jorge A. Colombo (Programa Unidad de Neurobiologica Aplicada, PRUNA, CEMIC-CONICET, Argentina), was cultured in DMEM-F12 medium (Invitrogen, Carlsbad, CA) containing 10% (v/v) fetal bovine serum (Invitrogen) at 37°C in a humidified atmosphere of 6% CO₂. Using the Effectene reagents (Qiagen, Valencia, CA), NB69 cells (2 × 10⁵ per well, six-well plates) were transfected with the above-described expression plasmids (1 μg) mixed with 8 μL Effectene enhancer and 10 μL Effectene reagent according to the manufacturer's instruction. Stable transfectants were selected with 400 μg/mL G418 before being cloned by Flow Cytometry at the Cancer Center Core Facility of University of California at San Diego.

To determine the proliferation rate, transfectants were plated at 2 × 10⁴ cells/mL in 6- or 96-well plates in triplicates for 1 week and fed every other day. Cell numbers in each well of the six-well plates were measured daily by a Coulter Counter (model Z, Beckman Coulter, Inc., Fullerton, CA) and plotted against hours to determine the doubling time during exponential growth. The proliferation rate was confirmed with a modified 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay in 96-well plates using a standard curve for cell numbers. At least three independent experiments were performed.

To induce neuronal differentiation, neuroblastoma cells were seeded at 1 × 10⁴ cells/mL in 24-well plates and treated in the dark with 30 μmol/L all-trans RA (Sigma Chemical Co., St. Louis, MO), whereas the negative control was treated with the vehicle (i.e., ethanol). Three days after treatment, cells at three different and random locations per well were photographed with a CCD camera attached to an inverted microscope (Leica Microsystems, Inc.,Bannockburn, IL, model DMIRB). Cells were scored positive for neuronal differentiation if neurites exceeded one cell body diameter in length. To determine the effect of RA on cell arrest, cells from different fields of microphotographs were manually counted and averaged for each treatment condition. Different treatment conditions were duplicated in each of three independent experiments.

The total proteins (50 to 100 μg) extracted from each clone of transfectants were resolved by 16% SDS-PAGE blotted onto the polyvinylidene difluoride membrane as described previously (31). NDPK-A on the membrane was probed with 1 μg/mL anti-NDPK-A monoclonal antibody (Kamiya Biomedical Co., Seattle, WA) or 0.67 μg/mL anti-NDPK-A polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) followed by a horseradish peroxidase–conjugated secondary antibody and chemiluminescence as described previously (31). To confirm the specificity of antibodies, recombinant NDPK-A and NDPK-B were generated as described previously (31).

Exponentially growing cells (2.5 × 10⁴ in 500 μL) in the DMEM-F12 medium containing 0.1% serum were added to a modified Boyden chamber in duplicates. Subsequently, each chamber containing cells was placed into a well of 24-well plates containing 500 μL DMEM-F12 medium supplemented with 5% serum, which serves as the source of chemoattractants. Cells were allowed to migrate toward 5% serum through the FluroBloc membrane (BD Biosciences, Lexington, KY) for 15 hours at 37°C and 6% CO₂. The FluroBloc membrane blocks 99% of fluorescence from nonmigrating cells; therefore, migrated GFP-expressing transfectants attached underneath the membrane were directly photographed in three different and random fields under an inverted fluorescent microscope (100× magnification). At least three independent experiments were performed.

Exponentially growing cells were resuspended in 0.3% agar, at a density of 2 × 10⁴ cells/mL, before being overlaid onto 0.5% agar in a well of six-well plates in duplicates. Once the agar solidified, cells were fed with the growth medium twice a week for 2 weeks. Under a light microscope with 40× magnification, photographs were taken at four different and random fields per well. At least three independent experiments were performed.

Six-week-old female Fox Chase SCID mice (Charles River Laboratories, Wilmington, MA) were maintained in a pathogen-free environment for at least 1 week prior to experimentation. Before surgery, the SCID mice were anesthetized with 75 mg/kg ketamine (Fort Dodge Animal Health, Fort Dodge, IA) and 1 mg/kg medetomidine (Pfizer Animal Health, Exton, PA) i.p. Additional anaesthetic was given when necessary. An incision was made over the left retroperitoneal space exposing the left adrenal gland into which 0.6 × 10⁶ neuroblastoma cells in 10 μL PBS were injected with a disposable 30 G needle attached to a 25 μL syringe (Hamilton, Reno, NV). The incision was closed with 4-0 dexon, and each animal was given 0.5 mL saline and 0.1 mg/kg buprenorphine (Rickitt and Coleman Pharmaceuticals Inc., Richmond, VA) s.c. During the first week after surgery, animals were given sulfamethoxazole trimethoprim antibiotics in the drinking water. Animal studies were preformed in accordance with protocols approved by the Animal Subjects Committee of the University of California at San Diego.

After intraadrenal injection, the general health of animals and tumor development were monitored daily for the first week followed by three times per week. Animals were sacrificed when tumors reached ∼ 1.5 cm in diameter when the animals showed a high level of discomfort or 1 month after injection, whichever came first. Animal lungs were examined and photographed under polarized and/or fluorescent light for macroscopic metastases with an imaging system consisting of an Epi lighting system with 470 nm excitation, 515 nm emission filters, a CCD camera (Illumatool Model LT-9500, Lightools Research, Encinitas, CA), and MagnaFire SP imaging software version 2.1 (Optronics, Goleta, CA). Microscopic metastases were examined by light microscopy in lung tissues after they had been fixed, paraffin embedded, serial sectioned, and stained with H&E.

Human neuroblastomas, ∼3 mm in diameter, were excised from the primary and secondary tumor foci in SCID mice. Excised tumors were washed in sterile PBS containing 50 units/mL penicillin and 50 μg/mL streptomycin, and tumor cells were dispersed by needles. Human neuroblastoma cells resistant to G418 were selected from animal cells by culturing with growth medium containing 400 μg/mL G418 at 37°C and 6% CO₂ for 10 to 14 days.

A two-tailed Student's t test was used to compare the mean values of cell survival, migration, or soft agar colonization between the control (stNB-V) and the transfectants that overexpress NDPK-A variants (i.e., stNB-W and stNB-M). Fisher's exact test was used to compare the incidence of lung metastasis between SCID mice injected with stNB-V and those with stNB-M or stNB-W. In both tests, P ≤ 0.05 was considered significant.

We thank Jorge A. Colombo for the kind gift of the NB69 cell line; Yamei Cheng, Jin Il Kim, Ralph Thurlow, and Kay Bates for their technical assistance; and Gordon Gill, Deching Chang, and Larry Paris for suggestions and critical reading of the manuscript.