Midkine Binds to Anaplastic Lymphoma Kinase (ALK) and Acts as a Growth Factor for Different Cell Types*

Midkine (MK) is a developmentally regulated, secreted growth factor homologous to pleiotrophin (PTN). To investigate the potential role of MK in tumor growth, we expressed MK in human SW-13 cells and studied receptor binding, signal transduction, and activity of MK. The MK protein stimulates soft agar colony formation in vitro and tumor growth of SW-13 cells in athymic nude mice, as well as proliferation of human endothelial cells from brain microvasculature and umbilical vein (HUVEC) in the low ng/ml range. MK binds to anaplastic lymphoma kinase (ALK), the receptor for PTN, with an apparent K d of 170 p M in intact cells, and this receptor binding of MK is competed by PTN with an apparent K d of (cid:1) 20 p M . Monoclonal antibodies raised against the extracellular ligand-binding domain of ALK inhibit ALK receptor binding of MK as well as MK-stimulated colony formation of SW-13 cells. Furthermore, MK stimulates ALK phosphorylation in WI-38 human fibroblasts and activates PI3-kinase and MAP kinase signal transduction in WI-38, HUVEC, neuroblastoma (SH SY-5Y) and glioblastoma (U87MG) cells that express the ALK protein. We conclude that MK can act as a growth, survival, and angiogenic factor during tumorigenesis and signals through the ALK receptor. Polypeptide growth factor expression is tightly regulated during development and in paraformaldehyde PBS room cell pellets resuspended PBS paraformaldehyde. A FACstar instrument (BD PharMingen) to measure the fluorescent Immunoprecipitations and Western Blotting— The detection of phos- pho-Akt, Akt, phospho-Erk1/2 and Erk1/2 was performed as described (51) using appropriate antibodies from Cell Signaling Inc. (Beverly, MA). The anti-ALK antibody used for Western blots was a rabbit polyclonal antibody from Sanbio Inc. (Uden, Netherlands). For immu- noprecipitation a mouse monoclonal antibody from Dako (Carpinteria, CA) was used. Both of these antibodies are raised against intracellular epitopes of ALK. The anti-phosphotyrosine antibody (No. 05-321) and the agarose-conjugated anti-phosphotyrosine antibody (4G10) were from U. S. Biochemicals Inc. (Lake Placid, NY). Lysates were prepared from cells grown in 6-well plates (BD PharMingen) for 24 h and then starved in serum-free medium for 16 h before treatment and lysis.

From the Lombardi Cancer Center, Georgetown University, Washington, D. C. 2007 Midkine (MK) is a developmentally regulated, secreted growth factor homologous to pleiotrophin (PTN). To investigate the potential role of MK in tumor growth, we expressed MK in human SW-13 cells and studied receptor binding, signal transduction, and activity of MK. The MK protein stimulates soft agar colony formation in vitro and tumor growth of SW-13 cells in athymic nude mice, as well as proliferation of human endothelial cells from brain microvasculature and umbilical vein (HUVEC) in the low ng/ml range. MK binds to anaplastic lymphoma kinase (ALK), the receptor for PTN, with an apparent K d of 170 pM in intact cells, and this receptor binding of MK is competed by PTN with an apparent K d of ϳ20 pM. Monoclonal antibodies raised against the extracellular ligand-binding domain of ALK inhibit ALK receptor binding of MK as well as MK-stimulated colony formation of SW-13 cells. Furthermore, MK stimulates ALK phosphorylation in WI-38 human fibroblasts and activates PI3-kinase and MAP kinase signal transduction in WI-38, HUVEC, neuroblastoma (SH SY-5Y) and glioblastoma (U87MG) cells that express the ALK protein. We conclude that MK can act as a growth, survival, and angiogenic factor during tumorigenesis and signals through the ALK receptor.
Polypeptide growth factor expression is tightly regulated during development and in the adult organism, but it appears to be deregulated during the course of malignant transformation (1). Here we report on the potential role of the growth factor midkine (MK) 1 in this process. MK is a member of the pleiotrophin (PTN)/MK family and was originally described as a retinoic acidinducible, developmentally regulated, heparin-binding, neurotrophic factor (2)(3)(4). MK is 50% homologous to PTN at the amino acid level and shares with PTN the genomic organization (5)(6)(7) and predicted protein structure (reviewed in Refs. 8 and 9).
Physiologically, MK is highly expressed during midgestation in the brain and numerous other organs and is down-regulated at birth. In the adult, MK shows a very restricted pattern of expression with the highest transcript levels in the intestine and low levels in the cerebellum, thyroid, kidney, bladder, lung alveoli, colon, stomach, and spleen (reviewed in Refs. 10 and 11). MK modulates epithelial-mesenchymal interactions during fetal development and organogenesis in the mouse (12) as well as in Xenopus (13). From in vitro cell biology studies and the embryonic expression pattern of MK in the central nervous system it was concluded that MK directs neurite interconnections during an early phase of brain development and may later on have a maintenance function in some restricted areas. In primary neuronal cultures isolated from mouse cerebral cortex MK inhibits the induction of apoptosis (14). Furthermore, MK promotes migration of various cells such as embryonic neurons (15), neutrophils (16), and macrophages (17). These biological activities suggest that MK may have a role in survival and invasion of tumor cells. Interestingly, during pathologic alterations in the adult brain such as Alzheimer's disease (18) or during cerebral infarction (19), the expression of midkine reemerges, and it has been suggested that this may reflect the activation of a repair mechanism (19). Several lines of evidence indicate that MK might also serve as an angiogenic factor. In the endometrial epithelium, MK is up-regulated by estradiol; this up-regulation correlates with enhanced angiogenic activity in this tissue (20). Furthermore, MK is known to enhance plasminogen activator and plasmin activity in bovine aortic endothelial cells (21,22), which suggests that MK may also have a role in tissue repair and angiogenesis.
The MK-related growth factor PTN is overexpressed in a number of tumors; we and others have show that PTN can function as a tumor growth factor (23,24) and can stimulate endothelial cell proliferation and angiogenesis (23,25,26). We used ribozyme-targeting of PTN mRNA to demonstrate that PTN can be the rate-limiting growth factor for melanoma growth, angiogenesis, and metastasis in vivo (27,28). Like PTN, MK mRNA expression is up-regulated in the majority of human tumors, including neuroblastomas (29), breast cancer (30) head and neck cancer, lung and esophageal cancer (31), Wilms' tumor, gastrointestinal cancers (10, 32), bladder cancer (11),and a variety of tumor-derived cell lines (10,30,31,33). It is worth emphasizing that midkine expression is very restricted in adult normal tissues, which makes all these findings more significant. To date, high levels of midkine expression have been correlated with a poor prognosis in neuroblastoma, glioblastoma, and bladder carcinoma (11,29,34). Recently, significantly increased midkine serum levels were detected in cancer patients; ten different types of cancer showed a similar profile of serum midkine level increases (35). Furthermore, in cases of gastric carcinoma and lung carcinoma, elevated serum midkine levels were found even at stage I.
Because this enhanced expression sharply contrasts with the very restricted expression pattern of MK mRNA in normal tissues, we decided to study directly whether MK expression could support tumor growth and perhaps serve as an angiogenesis factor. We subcloned the human MK cDNA into a eukaryotic expression vector downstream of the CMV immediate early gene promoter and generated transfected cells that express and secrete the full-length MK protein. This secreted MK stimulates the proliferation of human umbilical vein and human brain microvascular endothelial cells as well as the anchorageindependent growth of epithelial SW-13 cells in soft agar. Furthermore, stably transfected SW-13/MK cells form colonies in soft agar and tumors in athymic nude mice.
The mechanism of action of MK has been less clear because of the lack of a well defined MK receptor that can transduce the growth factor signal (15, 36 -42). Based on the recent discovery by our laboratory of anaplastic lymphoma kinase (ALK) as a receptor for pleiotrophin (43), we also tested the hypothesis that ALK may be a receptor for the PTN family member, MK.
Here we demonstrate that MK directly binds to ALK and show cross-competition between MK and PTN for binding to ALK. Furthermore, ALK receptor binding of MK as well as MKinduced colony formation of SW-13 cells are inhibited by monoclonal antibodies to the extracellular ligand-binding domain (LBD) of ALK (43). Finally, in different ALK-positive cell lines, we show signal transduction of MK via PI3K and MAPK pathways.

MATERIALS AND METHODS
Cell Lines and Tissues-Human brain endothelial cells (gift of Drs. P. Costello and R. L. Martuza, Dept. of Neurosurgery, Georgetown University) were isolated as primary cultures from the cerebellum and carried in 199E medium (Biofluids Inc.) with 10% fetal bovine serum. Human umbilical vein endothelial cells (HUVEC) were from Clonetics, Walkersville, MD, and SH SY-5Y cells were a gift of Dr. V. Movsesyan (Dept. of Neuroscience, Georgetown University). The other cell lines were obtained from American Type Culture Collection and were kept in improved minimal essential medium or Dulbecco's modified essential medium (Biofluids Inc., Rockville, MD) with 10% fetal bovine serum.
Construction of the MK Expression Vector-pMKHC4 contains the human MK cDNA clone as an EcoRI fragment and was a gift of Dr. A. Seddon (Lederle, Wayne, NJ) (44). The HindIII fragment of pMKHC4 containing the complete MK open reading frame and part (70 nucleotides) of the 5Ј-untranslated region of the MK1 transcript, was isolated from low melting point agarose and subcloned into the HindIII site of the pRc/CMV expression vector (Invitrogen) (23,27). The most 3Ј 143 nucleotides of the MK untranslated region, including the polyadenylation signal, were deleted during the cloning process. The resulting plasmid, pRc/MK, coding for an mRNA of about 0.9 kb, was used for transfections.
Stable Transfections-pRc/CMV (empty control vector) and pRc/MK plasmid DNAs were purified, and cells were transfected using the calcium phosphate/BES method as described previously (23). Transfected cells were cultivated in the presence of G418 (250 g/ml of medium; Invitrogen) to generate stably transfected cell lines. Clonal cell lines were obtained by diluting cell suspensions of transfected cells to ϳ1 cell/100 l of medium and plating 100-l aliquots in 96-well plates in the presence of G418. After 2 weeks, the wells showing growth of an individual clone were harvested and expanded. Clones where analyzed for MK protein expression by immunological detection using a protein slot blot assay. The transfection of ALK into 32D cells was described previously (43). The generation and characterization of U87MG cells that express ALK-targeted ribozymes were described very recently (54). In brief, ribozyme expression vectors that target the ALK mRNA were generated and shown to deplete stably transfected U87MG cells of their endogenous ALK. Using a series of cells, we reported that the reduction of ALK expression is paralleled by a reduction in the response of the cells to PTN. Furthermore, the reduction of ALK led to reduced xenograft tumor growth and increased apoptosis in the tumors (54). One of the paired U87MG/U87MG Rz cell lines from that study with a reduction of endogenous ALK mRNA by approximately 70% was also used in the present studies to assess the role of ALK for MK signal transduction.
Heparin Affinity Chromatography-One hundred ml of medium conditioned for 24 h by SW-13 cells (wild-type or transfected cells) was loaded onto 1-ml heparin-Sepharose columns (Amersham Biosciences) and washed, and the bound proteins were eluted with a step gradient of 3 ml each of 0.4, 0.9, and 2.0 M NaCl in 10 mM Tris buffer (pH 7.5) as described for the purification of PTN (45).
Soft Agar Growth of SW-13 Cells-Anchorage-independent growth assays of SW-13 cells in soft agar were carried out as described earlier, by growing 20,000 cells in 0.35% bactoagar on a bottom layer of solidified 0.6% bactoagar in 35-mm dishes (45,46). In some experiments the ability of transfected SW-13 cells to stimulate anchorage-independent growth of wild-type SW-13 was tested in a co-culture assay. A feeder layer of 2,000 cells was allowed to attach overnight to 35-mm dishes. After removal of the medium, this feeder layer of cells was covered with 1 ml of 0.6% agar, which was allowed to solidify. Finally, a top layer was added (0.8 ml of a 0.35% agar) containing 20,000 wild-type SW-13 cells as indicators of diffusible growth factors released from the feeder layer cells. After 2-3 weeks colonies in the top layer were counted using an inverted microscope equipped with a measuring grid. The size exclusion limit for positive colony counting was Ͼ60 m in diameter. The colony formation induced by control cells was used as background and was in the range of 5 to 10 colonies depending on the experiment.
Proliferation of Endothelial Cells-Human umbilical vein endothelial cells and human brain endothelial cells were plated in 24-or 96-well plates at 10,000 or 2,000 cells, respectively. In 1 or 0.2 ml, respectively, of growth medium, aliquots from heparin-affinity purified conditioned medium from transfected cells were included as indicated. FGF-2 added to the medium (10 ng/ml final concentration; R&D Systems) served as a positive control. After 3-5 days the cells were trypsinized and counted in a particle counter (24-well assay) or stained with Mosmann's tetrazolium dye (96-well assay; Promega, Madison WI), and the absorbance was read at 630 nm.
Tumor Formation in Athymic Nude Mice-Female athymic nude mice (Ncr nu/nu; Harlan Sprague-Dawley, Indianapolis, IN) were injected subcutaneously into each flank with 1 million cells in 100 l of medium and observed for tumor formation as described (23,47,48).
Receptor Binding Studies in Intact Cells-The assays were performed as described using either 35 S-PTN or 35 S-MK generated by metabolic labeling of PTN-or MK-transfected cells followed by purification of the radioligand from conditioned media (43). Murine hematopoietic 32D cells transfected with the human ALK cDNA (32D/ALK) or an empty vector (32D/control) served as receptor-containing and control cells, respectively (43).
Preparation of Anti-ALK LBD Murine Monoclonal Antibodies-Murine monoclonal antibodies to the ALK LBD were prepared by immunization of mice with the LBD as a bacterial glutathione S-transferase fusion protein. This antigen was described earlier and was used previously to generate rabbit polyclonal antibodies to the LBD (43). A different protein produced in mammalian cells was used for the screening of mouse sera and hybridoma supernatants. The complete extracellular domain (ECD) of ALK was expressed as a secreted protein in SW-13 cells by subcloning the ECD (49) into the eukaryotic expression vector pCDNA with a C-terminal Myc/His tag. This construct, including the original secretory signal sequence at the N terminus, was stably transfected and expressed in SW-13 cells, and the secreted ECD protein was isolated from the cell supernatants. To generate hybridoma cells producing antibodies of interest, spleen cells from an immunized mouse showing a positive enzyme-linked immunosorbent assay signal with the ECD protein was used for fusion with FOX-NY myeloma cells based on a protocol of Koehler and Milstein (50). The complete ALK ECD was then used to screen individual hybridoma clones by enzyme-linked immunosorbent assay for the presence of ␣LBD antibodies. The antibodies used in the present studies were derived from supernatants of two different hybridoma, 16G2-3 and 9C10-5, and are renamed here for easier identification as ␣LBD1 and ␣LBD2, respectively. The hybridoma supernatants were used at a 1:25 dilution for the experiments in Fig. 3.
Cell Surface ALK Staining by FACS-Subconfluent cell monolayers of COS-1 cells transiently transfected for 24 h with the ALK expression vector described previously (43) or with empty vector were detached from the flask with 0.02% sodium-EDTA in PBS. After two washes in cold PBS, 300,000 intact cells were incubated under gentle rotation in 0.4 ml of hybridoma supernatants for 15 min at 4°C. After three further washes with ice-cold PBS, cells were incubated in 0.4 ml of a 1:200 dilution of goat anti-mouse-fluorescein isothiocyanate (Jackson Immunoresearch Laboratories, West Grove, PA) for 15 min at 4°C. After three additional washes with ice-cold PBS, cell pellets were fixed in 3% paraformaldehyde in PBS for 10 min at room temperature. Fixed cell pellets were resuspended in PBS to a final concentration of 0.3% paraformaldehyde. A FACstar Plus instrument (BD PharMingen) was used to measure the fluorescent intensity (mean value of 10,000 cells).
Immunoprecipitations and Western Blotting-The detection of phospho-Akt, Akt, phospho-Erk1/2 and Erk1/2 was performed as described (51) using appropriate antibodies from Cell Signaling Inc. (Beverly, MA). The anti-ALK antibody used for Western blots was a rabbit polyclonal antibody from Sanbio Inc. (Uden, Netherlands). For immunoprecipitation a mouse monoclonal antibody from Dako (Carpinteria, CA) was used. Both of these antibodies are raised against intracellular epitopes of ALK. The anti-phosphotyrosine antibody (No. 05-321) and the agarose-conjugated anti-phosphotyrosine antibody (4G10) were from U. S. Biochemicals Inc. (Lake Placid, NY). Lysates were prepared from cells grown in 6-well plates (BD PharMingen) for 24 h and then starved in serum-free medium for 16 h before treatment and lysis.

Secretion of MK from Stably Transfected SW-13 Cells
SW-13 cells were chosen as a model cell line to study biological effects of MK because these cells express ALK and respond to the MK-related growth factor pleiotrophin (43). We transfected these cells with a human MK cDNA or with the empty expression vector and selected stably G418-resistant clonal cell lines for further studies. Conditioned media from these cells were concentrated and partially purified using heparin affinity chromatography with a step gradient similar to that employed earlier during the purification of PTN (45). After heparin affinity purification, only the 0.9 M NaCl eluate from conditioned media of MK-positive cells contained immunoreactive MK (not shown), and immunoblot analysis of supernatants of the different cell lines showed that only four of seven pRc/MK stably transfected cell lines secreted detectable amounts of MK (Fig.  1A, bottom). Using recombinant MK as a standard, we estimated that clones MK-4, MK-6, MK-7, and MK-8 secrete between 10 and 20 ng of immunoreactive MK/ml of conditioned medium. No immunoreactive MK was detected in the supernatants of wild-type SW-13 cells, G418-resistant controls transfected with the empty vector (pRc/CMV) or of three of the G418-resistant MK transfectants (Fig. 1A, bottom). In the lat- Wild-type SW-13 cells were used as indicator cells. They were added to a soft agar overlay on the feeder cells that had been attached to the bottom of the dishes. Indicator cell colonies were counted in triplicate dishes after 12 days of incubation (see "Materials and Methods" for details). The results from three experiments were pooled and graphed as the means Ϯ S.E. The left axis shows the absolute number of colonies Ͼ60 m, and the right axis indicates the fold increases over control. Bottom, conditioned media from the different SW-13 clones were analyzed by slot blot immunoassay for the presence of MK. B, activity of secreted MK on endothelial cells. Conditioned media from SW-13/MK-6 cells were passed over a heparin-Sepharose column and MK was eluted in the 0.9 M NaCl fraction of a step gradient. Aliquots from the 0.9 M NaCl fraction were incubated with HUVEC in 200 l of growth medium (left side of the graph) or with human brain microvascular endothelial cells (HBEC) in 1 ml of growth medium (right side of the graph); the cell number was determined after 3 or 5 days, respectively (see "Materials and Methods" for details). C, tumorigenicity of stably transfected SW-13 cells in athymic nude mice. Animals were injected subcutaneously with wild-type SW-13 cells, clone CMV-2 (empty vector), or clone MK-6 (1 million cells/site in 100 l of medium). The mean tumor volume Ϯ standard error for each group is shown (n ϭ 10 injection sites/group).
ter three clones, the expression unit for MK was most likely lost or disrupted during the stable integration of the expression vector into the genome, and only the G-418 resistance gene was expressed. This is expected for a fraction of stably transfected cells only selected for their drug resistance phenotype. Alternatively, MK mRNA might be expressed in these clones but not translated. We used all of the 10 different SW-13 derivative clonal cell lines in further in vitro studies to analyze the biological activities of MK.

Biological Activity of MK Secreted from Stably Transfected SW-13 Cells
Stimulation of Colony Formation of Wild-type SW-13 Cells-In the first series of assays, the stimulation of colony formation in soft agar of wild-type SW-13 cells was used as an indicator of growth factor activity. These cells do not form colonies in soft agar or tumors in athymic nude mice unless stimulated by growth factors such as FGFs (46,48) or PTN (23,45) that are added to their media or released from the cells. The different cell lines to be tested were plated as feeder cells on the bottom of 35-mm dishes. They were then overlaid with a soft agar layer that separates the indicator cells from the feeder cells on the bottom of the dishes. As described above, four of seven of the G418-resistant, pRc/MK-transfected clonal cell lines secreted immunoreactive MK into their media, whereas no MK was detected in three of the cell lines (Fig. 1A, bottom), and we found that only the MK-secreting cell lines stimulated colony formation of the wild-type SW-13 cells. For the SW-13/ MK-6 clone we obtained a 10-fold increase in the number of colonies over control and at least a 5-fold increase for the other MK-producing cells (Fig. 1A, top). The cell lines that did not secrete MK were also unable to stimulate soft agar colony formation of the indicator cells. An add-back experiment with MK from MK-6 cells confirmed the cross-feeding results and added MK stimulated SW-13 colony formation Ͼ 5-fold over control (not shown).
Endothelial Cell Proliferation-Our earlier studies had shown that the MK-related growth factor, PTN, can stimulate proliferation of endothelial cells in addition to SW-13 cells (23), and we tested MK in this regard. For this study we used conditioned medium from one of the MK-secreting clonal cell lines (SW-13/MK-6) as compared with a control transfectant. MK was concentrated and partly purified using heparin affinity chromatography. Because only the 0.9 M NaCl eluate contained MK (see above), we used this fraction in the endothelial cell proliferation assays. Fig. 1B shows a dose response of MK on HUVEC. The highest concentration of MK used in the assay was estimated as 8 ng/ml based on a slot blot analysis of the conditioned medium. This concentration of MK induced slightly more than 50% of the mitogenic effect of basic FGF (10 ng/ml) that stimulated proliferation of the cells 6.36 Ϯ 0.50fold. Human brain microvascular endothelial cells showed a similar proliferative response to MK although only one concentration of MK (4 ng/ml) was tested because of the limited supply of these primary cells. We conclude from these studies that MK is a mitogen for endothelial cells in addition to its activity on SW-13 cell colony formation. In this respect MK shows an activity profile that is qualitatively similar to PTN, although 5-10-fold higher concentrations of MK relative to PTN appear to be required for a half-maximal stimulation of endothelial cell proliferation (52). This could be a genuine difference in the potency between these two closely related growth factors, similar to well known differences in the reported potency, e.g. for different members of the FGF family (53). Alternatively, some of our preparation of MK may have been inactivated during the preparation, although separate preparations were similar with respect to their specific activity.
Tumorigenicity of SW-13 Cells Stably Transfected with pRc/ MK-We have shown previously that transfection of nontumorigenic SW-13 cells with a PTN expression vector makes these cells tumorigenic in athymic nude mice (23). Because MK showed a similar activity profile as PTN in vitro, we tested whether expression of MK also makes SW-13 cells tumorigenic in mice. We used one of the MK-secreting clones (SW-13/MK-6) as compared with a control transfectant and wild-type SW-13 cells for these studies. We observed exponentially growing tumors as early as 14 days in the SW-13/MK-6 group. After 36 days the tumor reached 1 cm 3 in volume, whereas mice that were inoculated with either SW-13 wild type or mock transfected SW-13 did not develop tumors (Fig. 1C)  with our earlier findings with PTN-secreting SW-13 cells (23). Northern analysis confirmed that MK mRNA was expressed in the tumors grown from the SW-13/MK-6 cells (data not shown), and we concluded from this study that MK expression can support tumor growth in vivo.

MK Binding to Anaplastic Lymphoma Kinase
Because MK and PTN belong to the same family, we tested whether MK binds to the tyrosine kinase receptor ALK, recently identified by us as a receptor for PTN (43). We used an experimental approach paralleling that used for PTN binding to ALK. We studied 35 S-labeled MK produced by metabolic labeling of SW-13/MK-6 cells for its binding to 32D cells transfected with the ALK cDNA (32D/ALK) or an empty vector. As shown in Fig. 2A, the MK radioligand binds to ALK expressed in 32D cells with an apparent K d of 170 Ϯ 90 pM, which is a somewhat lower affinity than observed for PTN (32 Ϯ 9 pM) (43). This Ͼ5-fold lower affinity of MK relative to PTN is also reflected in a 5-fold lower potency observed in bioassays. To test our hypothesis that MK binds to the same receptor as PTN, we performed a series of cross-competition assays. We used either 35 S-MK or 35 S-PTN and competed binding to 32D/ALK cells by unlabeled MK. As shown in Fig. 2B, MK inhibited the binding of both radioligands to ALK. Furthermore, a competition isotherm of unlabeled PTN for 35 S-MK ALK binding gave an apparent K d of ϳ20 pM for PTN (Fig. 2C), in good agreement with the K d determined for PTN radioligand binding to the ALK receptor, (32 Ϯ 9 pM) (43). From this series of experiments we conclude that MK and PTN bind to the same receptor, albeit with ϳ5-fold different affinities.

Inhibition of Receptor Binding and of the Biological Effects of MK by ␣LBD ALK Antibodies
Generation of ␣LBD Mouse Monoclonal Antibodies-To study the role of ligand/receptor interactions and potentially to generate an antibody that could block this interaction, we immunized mice with a glutathione S-transferase fusion protein that contains the LBD of ALK (43) and generated antibody-producing hybridoma cell lines from the spleen of immunoreactive animals. For the screening of the hybridoma supernatants for positive clones, the ALK ECD protein was generated as a secreted protein in SW-13 cells to assure that the selected monoclonal antibodies would ultimately recognize the eukaryotic LBD. In a first set of experiments we tested whether the selected antibodies detect the ALK receptor expressed on the cell surface. We used COS-1 cells that were transiently transfected with an ALK expression vector (or control vector) and FACS analysis of non-permeabilized cells to detect cell surface proteins. Fig. 3A shows two of the ␣LBD monoclonal antibodies that recognize the ALK protein on the cell surface of ALK-positive cells, whereas a negative control antibody was unable to distinguish between ALK-positive and ALK-negative cells.
␣LBD Effect on MK Binding to ALK-The ␣LBD antibodies inhibited binding of 35 S-MK to ALK-transfected 32D cells by 50 -60%, whereas the control antibody did not affect the MK binding to ALK (Fig. 3B). An anti-MK antibody inhibited 65% of the binding.
␣LBD Effect on Soft Agar Colony Formation of SW-13 Cells-As reported previously, SW-13 cells express ALK mRNA and protein at low levels (43). The expression of MK (SW-13/ MK-6 cells) or PTN (SW-13/PTN cells) does not appear to affect the expression of ALK in these cells (Fig. 3C). To assess the role of ALK for MK-stimulated colony formation of wild-type SW-13 cells in soft agar, MK was added without and with the ␣LBD antibodies. Inclusion of the ␣LBD antibodies inhibited the MK stimulation completely, whereas a control antibody had no effect (Fig. 3D, left). Stimulation of colony formation by FGF-2, on the other hand, was not affected by the ␣LBD antibodies (Fig. 3D, right), further supporting the specificity of the antibodies for MK signaling through ALK. Finally, the ␣LBD antibodies also inhibited colony formation of SW-13 cells stably expressing MK (SW13/MK-6 cells) by ϳ50%, whereas the control antibody had no effect (Fig. 3E). The incomplete effect in the latter assay is likely because of the continuous stimulus provided by MK expression in the cells in contrast to the single addition of MK in the experiment in Fig. 3D. Taken together, these results with the ␣LBD antibodies support the notion that MK binds to ALK and uses this receptor to exert its biological effects.

Signal Transduction of MK in Different Cell Lines
ALK Protein Expression-Several cell lines used here had been analyzed for ALK mRNA expression, and we had found a positive correlation between the response of cells to PTN and the detection of ALK mRNA (43,54). The Western blot in Fig.  4 complements these earlier mRNA expression studies that were carried out in the same cell lines. Wild-type MCF-7 cells as well as derivative MCF-7 cells (MCF-7/Adr) do not express detectable amounts of ALK mRNA (by reverse transcription PCR) or protein (by Western blot ; Fig. 4, lane 7). Expression of ALK protein, however, was found in two of the epithelial cell lines studied, ME-180 and 293 HEK, and at lower levels in endothelial cells (lanes 3-5). In WI-38 human fibroblasts the ALK protein migrates as a doublet (lane 6), a pattern that was also found by us in two different primary human breast fibroblast cultures (not shown). U87MG glioblastoma cells express ALK mRNA levels that are detectable by RNase protection (54) and by Western blot (Fig. 4, lane 1). Reduction of ALK mRNA in U87MG cells by ALK-targeted ribozymes reduced signal transduction of PTN by ϳ70% in these cells (54) and resulted in a similar reduction of ALK protein (Fig. 4, lane 2). The signal transduction of MK in these paired U87MG cells is described below (see Fig. 6). Finally, the human neuroblastoma cell line SH SY-5Y (Fig. 4, lane 8) showed the highest constitutive ALK protein expression among the cell lines surveyed.
MK Signal Transduction through PI3K and MAPK-Previous studies from our laboratory showed that PTN activates phosphatidylinositol 3-kinase (PI3K) and MAP kinase (MAPK) pathways and utilizes ALK as a receptor (43,51,54). Once we had established that MK binds to ALK, we tested the effects of MK on cells that express ALK (see Refs. 43 and 54). We first tested the WI-38 fibroblast cell line and found that Akt phosphorylation was increased Ͼ10-fold after MK treatment, whereas the increase in MAPK phosphorylation was ϳ2-fold (Fig. 5A). In endothelial cells (HUVEC), in which MK triggers proliferation (see Fig. 1B), MK induced a 2-fold increase in Akt phosphorylation and a 1.5-fold increase in MAPK phosphorylation. FGF-2 induced a higher signal in these cells (Fig. 5A). Because neuronal cells are known to be responsive to MK (55), we tested the effect of MK treatment on the ALK-expressing human neuroblastoma cell line SH SY-5Y and found a 4-fold increase in Akt phosphorylation and a 3-fold increase in MAPK phosphorylation in response to the MK treatment, (Fig. 5A). These results show that MK can activate PI3K and MAPK pathways in a manner very similar to the other family member, PTN (51).
MK Induction of Tyrosine Phosphorylation of ALK-We used the WI-38 human fibroblasts to assess ALK tyrosine phosphorylation in response to MK, because we had observed the highest stimulation in the signal transduction studies with these cells (see above). ALK was immunoprecipitated from control and from MK-stimulated cells (without and with preincubation of MK with an anti-MK or anti-PTN antibody), and the precipitates were blotted for phosphotyrosine. A comparison of the lanes in Fig. 5B shows that MK induces ALK phosphorylation and that this induction is reduced by the anti-MK but not by the anti-PTN antibody. This finding lends additional support to the notion that MK signals via the ALK receptor.

Rate-limiting Role of ALK for Akt Activation in U87MG Cells
To assess the significance of ALK for MK signal transduction, we performed a dose-response of MK on Akt phosphorylation in isogenic U87MG cells that had been transfected with a control vector or a ribozyme targeted against the ALK mRNA (Fig. 6). These cells have reduced endogenous ALK mRNA and protein levels (see Ref. 54 and Fig. 4, lanes 1 and 2), and PTN-induced Akt phosphorylation is diminished accordingly. Very similar to our earlier findings with PTN, MK is also unable to stimulate Akt phosphorylation upon reduction of ALK (Fig. 6). It is worth noting that in contrast with the diminished PTN and MK signals after reduction of ALK, Akt phosphorylation in the same cells via a different tyrosine kinase receptor, the platelet-derived growth factor receptor (PDGF-R), was not altered by the reduction of ALK levels (54). Interestingly, in the U87MG cells MAPK is activated constitutively and remains unaffected by the ALK reduction or by MK addition (not shown).
We conclude from our signal transduction studies that MK can activate both MAPK and PI3K pathways via ALK although at a different ratio and to a different extent in different cell types.

DISCUSSION
The widespread expression of MK in human tumors and tumor cell lines and its ability to stimulate endothelial cell proliferation in vitro and tumor growth in vivo support the notion that MK can play an important role in human cancer (10,11,29,30,32,34,56). Accordingly, our study shows that MK confers anchorage-independent growth to the SW-13 cell line both in a paracrine and autocrine manner; we link this to the ALK expression in these cells (43). In a study in NIH3T3 cells MK expression led to an increased number of colonies formed in soft agar (57), which is in agreement with the expression of ALK in NIH3T3 cells (43).
In the literature, the reports of the mitogenic capabilities of midkine are controversial. The native forms of mouse and chicken midkine are mitogenic for PC-12 cells (58, 59) but not for NIH3T3 fibroblasts and HUVEC and CC139 cells. The recombinant mouse MK was reported as a weak mitogen (60) on 10T1/2 fibroblasts and NIH3T3 cells but as a very potent mitogen for neuroectodermal precursor cells (61). Human recombinant MK expressed in bacteria was unable to induce mitogenesis in NIH3T3 cells (62). In our hands the chemically synthesized midkine exhibited only a very weak effect on SH SY-5Y and WI-38 cells (not shown), whereas both recombinant and synthetic MK had haptotactic effects on the UMR106 cell line (63). Our studies demonstrate that MK present in the conditioned media from MK-expressing cells stimulates soft agar colony formation of SW-13 cells and proliferation of human brain and umbilical vein endothelial cells. In addition, receptor binding and signal transduction were achieved with this MK protein.
Besides its effects in vitro, our studies provide evidence for the effect of MK on SW-13 cells in vivo. MK-secreting cells grew at an exponential rate in comparison with a control transfectant and wild-type SW-13 cells when inoculated in mice. This tumor growth effect was similar to our findings for PTN (23) and is corroborated by a report on similar effects of overexpression of MK or PTN in MCF-7 breast carcinoma cells (64). In a related study, MK also stimulated the in vitro and in vivo growth of NIH3T3 cells transfected with a MK cDNA (57). The angiogenic role of MK and PTN in tumor growth is underscored by an increased vascular density and endothelial proliferation (64), which matches our finding of endothelial cells proliferation. A potential mechanism for the angiogenic role of MK is represented by the up-regulation of urokinase-type plasminogen activator and down-regulation of plasminogen activator inhibitor-1 (21) as well as its mitogenic effect on different types of endothelial cells reported here.
In addition to its potential as an angiogenesis inducer, MK was described as an anti-apoptotic factor in primary neuronal cells (14). This paper described a dependence upon both MAPK and PI3K pathways, and as for PTN, MK has also been shown to cause transcriptional up-regulation of the anti-apoptotic protein Bcl-2 (65). In preliminary studies we found that MK can inhibit apoptosis induced by UV light in SW-13 cells; the pathway for these effects will be the subject of further studies.
Our previous finding that ALK is a receptor for PTN (43) made it likely that MK could also use this receptor for signal transduction. In the present paper we provide receptor binding and functional data for MK that also corroborate our previous studies on the PTN/ALK interactions for MK. Beyond those studies (43), we now introduce a new set of tools that were generated since, i.e. monoclonal antibodies to the LBD derived from the identification of ALK by phage display (43). Our goal was to generate antibodies that would recognize ALK expressed in mammalian cells as it presents itself to the ligand. For this purpose, we immunized mice with a bacterial fusion protein containing the LBD and screened the supernatants of hybridoma from the mice with the complete ALK ECD produced as a secreted protein in SW-13 cells. Indeed, two of the antibodies shown here recognize the cell surface receptor in intact cells, compete with MK ligand binding to ALK, and inhibit the biologic activity of MK in a colony formation assay. This is independent evidence of the crucial role of ALK and the significance of the LBD for PTN or MK signaling through this receptor. CONCLUSION MK shares its activity profile with PTN and appears to use similar signal transduction pathways. Furthermore, the lower potency of MK in comparison with PTN in different bioassays is also reflected in a lower affinity for their shared receptor, ALK. Inhibition of ligand binding and ligand-dependent soft agar colony formation by monoclonal antibodies raised against the LBD in ALK lend additional support to the proposed PTN or MK ligand/ALK receptor interaction and the biological significance of this interaction. Finally, the antibody-based approach should provide a basis for the development of clinically efficacious inhibitors that could be useful to treat pathologic alteration of the PTN or MK receptor pathway in benign diseases that appear to utilize MK, such as neurofibromatosis (66) and restenosis after coronary bypass (17). But this approach may be most useful in human cancers, including highly aggressive tumors of the lung, pancreas, brain, or gastrointestinal tract (67)(68)(69)(70)(71)(72), where currently only limited treatment options are available.