Nordihydroguaiaretic Acid Inhibits an Activated Fibroblast Growth Factor Receptor 3 Mutant and Blocks Downstream Signaling in Multiple Myeloma Cells

Receptor tyrosine kinases (RTKs) represent important signal transducers for the transmission of information across the cell membrane, including 58 distinct receptors in 16 different families of homologous proteins in humans (1, 2). The fibroblast growth factor receptor (FGFR) family includes four closely related receptors, designated FGFR1, FGFR2, FGFR3, and FGFR4. All FGFRs exhibit a cleaved NH₂-terminal signal sequence that directs the nascent protein into the endoplasmic reticulum (ER), followed by three extracellular immunoglobulin (Ig)-like domains, a membrane-spanning domain, and a split tyrosine kinase domain (3, 4), and are activated by FGFs, a large family of some 20 related growth factors (5–8). The ligand-binding specificity differs for each FGFR family member, with alternative splicing of FGFR transcripts resulting in further distinctions in ligand specificity accompanied by altered biological properties (4, 9). FGFR activation controls an array of biological processes, including cell growth, differentiation, migration, wound healing, and angiogenesis. Aberrant activation of these receptors, often through gain-of-function mutations, is associated with many developmental and skeletal disorders. Mutational activation of FGFR3 is responsible for the relatively common skeletal dwarfism, achondroplasia, due to the G380R mutation within the transmembrane domain, whereas strong activation by the mutation K650E within the kinase domain results in the neonatal lethal syndrome, thanatophoric dyspasia (TD) type II (10).

Aberrant signal transduction arising from expression of a constitutively activated FGFR3 is also an important event in a variety of human neoplasias, especially multiple myeloma and bladder carcinoma (11–15). Multiple myeloma is typified by the accumulation of secretory plasma cells in the bone marrow, which exhibit low proliferation but an extended life span. A frequent translocation observed in multiple myeloma, t(4;14)(p16.3;q32.3), involves the FGFR3 gene, and results in increased expression of FGFR3 alleles (11, 12). This translocation occurs with an incidence of ∼20% in multiple myeloma and places the FGFR3 gene located at 4p16.3 in proximity with the 3′ IgH enhancer at 14q32.3, leading to significant FGFR3 overexpression (11). Furthermore, FGFR3 overexpression is often accompanied by mutational activation, including the mutations Y373C and K650E, corresponding to germline FGFR3 mutations that cause the lethal skeletal syndromes TDI and TDII, respectively (10, 16). The aberrant signaling of overexpressed, mutated FGFR3, acting in concert with other dysregulated genes such as cyclin D1, c-maf, and MMSET, evidently contributes to the increased proliferative potential of B-cell myelomas (17). Mutational activation of FGFR3 has also been reported in human bladder and cervical carcinomas (13–15), including the mutations corresponding to R248C, S249C, G370C, and K650E, previously identified as causing TDI or TDII (10, 16).

Several recent reports show that the high level of kinase activity associated with the strongly activated FGFR3 mutants such as TDII (K650E) leads to accumulation of immature/mannose-rich, phosphorylated receptors in the ER. This can result in direct recruitment of Jak1 and activation of signal transducers and activators of transcription (STAT) 1 from the ER, as well as activation of Erk1/2 from the ER through a FRS2-independent and phospholipase C (PLC) γ-independent pathway (18–20). Given the ability of FGFR3 to signal from intracellular compartments within the secretory pathway, we investigated the possible inhibitory properties of a drug that specifically blocks ER/Golgi transport. Nordihydroguaiaretic acid (NDGA), a naturally occurring polyhydroxyphenolic compound isolated from the creosote bush, Larrea divaricatta, has been previously characterized as a lipoxygenase (LOX) inhibitor, as a strong antioxidant, and as an inhibitor of protein transport from the ER to Golgi, leading to a redistribution of Golgi proteins to the ER (21–24). NDGA has also been shown to have effects on cell proliferation, apoptosis, and differentiation. Recently, it was shown to have potentially valuable inhibitory effects against specific RTKs such as insulin-like growth factor receptor (IGF-IR), HER2/neu, and platelet-derived growth factor receptor (25–28).

The experiments presented here explore the ability of NDGA to inhibit signaling by activated FGFR3, including kinase activation and downstream signaling associated with an activated receptor. Given the ability of NDGA to inhibit other RTKs, and the ability of NDGA to collapse the secretory compartment from which activated FGFR3 is known to signal (18–20), we examined the possibility that NDGA would be particularly effective against the strongly activated FGFR3 mutant K650E. Indeed, NDGA was able to inhibit activation of the downstream signaling proteins STAT1, STAT3, and mitogen-activated protein kinase (MAPK). In addition, NDGA increased apoptosis in cells expressing FGFR3-TDII, as measured by poly(ADP)ribose polymerase (PARP) cleavage. NDGA was also able to increase apoptosis in the multiple myeloma cell line KMS-18 that expresses high levels of FGFR3 G384D. To examine possible therapeutic effects of NDGA, multiple myeloma–derived cell lines were treated with FGF2 to activate FGFR3, and the effects of NDGA treatment on downstream signaling pathways were observed. The results are discussed with respect to other therapeutic approaches for treatment of cancers presenting overexpressed and/or mutated FGFR3.

FGFR constructs. The full-length wild-type (WT) and kinase active (TDII:K650E) FGFR3 have been described previously (29). The membrane-localized kinase domain derivatives of FGFR3 have also been previously described (30).

Antibodies and reagents. Antibodies were obtained from the following sources: FGFR3 (C-15), STAT1 (E-23), STAT3 (C-20), β-tubulin (H-235) from Santa Cruz Biotechnology; 4G10 (antiphosphotyrosine) from Upstate Biotechnology; Phospho-STAT1 (Tyr701), Phospho-STAT3 (Tyr705; D3A7), Phospho-p44/42 MAPK (Thr202/Tyr204; E-10), cleaved PARP (Asp214) from Cell Signaling; MAPK (ERK1 + ERK2) from Zymed; and horseradish peroxidase (HRP) anti-mouse and HRP anti-rabbit from Amersham. Monoclonal antibody 10E6 was a gift from V. Malhotra Laboratory (University of California, San Diego, La Jolla, CA). Fluorescein-conjugated anti-mouse (Sigma), rhodamine-conjugated anti-rabbit (Boehringer-Mannheim), NDGA, and the poly (4:1 Glu, Tyr) peptide were obtained from Sigma. Zilueton was a gift from Dr. Edward Dennis (University of California, San Diego, La Jolla, CA).

Immunoprecipitation and immunoblot. HEK293 cells were grown in DMEM, supplemented with 10% fetal bovine serum (FBS) and incubated at 37°C in 10% CO₂. Cells (9 × 10⁵) plated on 10-cm dishes were transfected the next day with 2 μg of DNA by calcium phosphate precipitation at 3% CO₂, as previously described (31, 32). After 18 to 20 h, cultures were moved back into 10% CO₂ for 4 to 6 h before starving in 0% DMEM or treating with NDGA in 10% FBS DMEM overnight. Cultures that were starved overnight were treated with NDGA at varying concentrations or time, as indicated in figure legends. Cells were harvested, washed once in PBS, and lysed in 1% NP40 Lysis Buffer [20 mmol/L Tris-HCl (pH 7.5), 137 mmol/L NaCl, 1% Nonidet P-40, 5 mmol/L EDTA, 50 mmol/L NaF, 1 mmol/L sodium orthovanadate, 1 mmol/L phenylmethylsulfonyl fluoride (PMSF), and 10 μg/mL aprotinin] or radioimmunoprecipitation assay buffer [RIPA; 50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mmol/L NaF, 1 mmol/L sodium orthovanadate, 1 mmol/L PMSF, and 10 μg/mL aprotinin]. Total protein was measured by Bradford Assay or Lowry Assay. Lysates were immunoprecipitated with antibody for 2 h at 4°C. Protein A-Sepharose beads were added and incubated for 2 h at 4°C. The immunoprecipitated samples were washed thrice with Lysis Buffer, boiled 5 min in sample buffer, and separated by 10% SDS-PAGE. Proteins were transferred to Immobilon-P membranes (Millipore) and blocked in 3% milk/TBS/0.05% Tween 20 or 3% bovine serum albumin (BSA)/TBS/0.05% Tween 20 (for anti-phosphotyrosine, anti–phospho-STAT1, and anti–phospho-STAT3 blots). Membranes were immunoblotted with antibodies at room temperature for 2 h or overnight at 4°C. After primary incubation, membranes were washed with TBS/0.05% Tween 20 and incubated with HRP-conjugated secondary antibodies. Proteins were detected by enhanced chemiluminescence (ECL; Amersham) or (Millipore). To reprobe with other antibodies, membranes were stripped of bound antibodies in stripping buffer [100 mmol/L 2-mercaptoethanol, 2% SDS, and 62.5 mmol/L Tris-HCl (pH 6.8)] and incubated for 1 h at 55°C.

In vitro kinase assays. FGFR3-TDII was immunoprecipitated from 500 μg of total cell lysate per sample for 1.5 h at 4°C. Protein A-sepharose beads were added for 2 h. After 3 washes with Lysis Buffer, samples were washed once with Kinase Buffer [20 mmol/L Tris (pH 7.5), 10 mmol/L MnCl₂, and 5 mmol/L MgCl₂]. Immunoprecipitates were preincubated with NDGA in Kinase Buffer for 15 min on ice before the addition of ATP and the poly (4:1 Glu, Tyr) peptide and incubated at 37°C for 15 to 25 min. Reactions were washed thrice with cold Lysis Buffer before adding sample buffer. Proteins were separated by 7.5% or 10% SDS-PAGE and transferred to Immobilon-P membranes for immunoblotting, except for ³²P-labeled samples in which the gels were dried and exposed to film or a phosphor screen for use with a Phosphorimager.

Double label indirect immunofluorescence. HEK293 cells were seeded onto Collagen type I–coated glass coverslips (BD Biosciences) and transfected as above. After starving overnight, cells were treated with 30 μmol/L NDGA. At indicated time points, coverslips were fixed with 3% paraformaldehyde/PBS for 15 min, washed with PBS, and placed at −20°C in 50% glycerol/PBS. Before staining, cells were washed with PBS, permeabilized with 0.5% Triton X-100/PBS, and blocked with 3% BSA/PBS. The cis-Golgi was detected with monoclonal antibody 10E6 and fluorescein-conjugated anti-mouse secondary antibody. FGFR3 localization was detected with rabbit polyclonal antibody FGFR3 (C-15) and rhodamine-conjugated anti-rabbit secondary antibody. The nuclei were visualized by Hoechst dye added to the Rh secondary antibody mix. All antibodies were diluted in 3% BSA/PBS, and incubations were performed at room temperature. Coverslips were mounted on glass slides with 90% glycerol in 0.1 mol/L Tris (pH 8.5) plus phenylenediamine and photographed using a Nikon Microphot-FXA with a cooled CCD camera (Hamamatsu C5810).

Multiple myeloma cell lines. The human multiple myeloma cell lines, RPMI-8226, LP-1, KMS-18, and KMS-11, were a generous gift from Dr. Leslie Thompson (University of California, Irvine, Irvine, CA). They were grown in RPMI 1640 with l-glutamine medium, supplemented with 10% FBS and pen/strep, and incubated at 37°C in 5% CO₂. For FGF stimulation, 3 × 10⁶ cells per 35-mm well were starved overnight in serum-free RPMI 1640. NDGA was added to the cultures for 1 h at 37°C before simulating with 10 ng/mL FGF2 and 1 μg/mL Heparin for 10 min at 37°C. Plates were put on ice, and cells were collected, transferred to tubes, and spun down for 2 min at 1 krpm. Cell pellets were washed once with PBS before the addition of RIPA Lysis Buffer. Thirty micrograms of total lysate was separated by 10% SDS-PAGE and processed for immunoblotting as described above. For NDGA treatment of the KMS-18 and KMS-11 cell lines, 6 × 10⁶ cells per 5-cm plate were incubated at 37°C with various concentrations of NDGA for 1 h or overnight. Cells were collected by centrifugation as described above. Thirty micrograms of total lysate were separated by 10% SDS-PAGE and transferred to Immobilon-P membranes for immunoblotting.

In vitro inhibition of FGFR3-TDII autophosphorylation by NDGA. An activated FGFR3 clone containing the K650E mutation (FGFR3-TDII) was transfected into HEK293 cells. After an overnight starvation, the receptor was immunoprecipitated from the cell lysate and subjected to in vitro kinase assays. In Fig. 1A, γ-³²P-ATP was used in the kinase reactions after preincubating with NDGA for 15 minutes. The NDGA concentration of 0.5 μmol/L caused a dramatic reduction in the autophosphorylation of FGFR3-TDII. In Fig. 1B and C, the kinase reaction was performed in the absence of hot ATP and the phosphotyrosine on the receptor was detected by immunoblotting with a phospho-specific antibody. These results suggest a direct interaction of NDGA with FGFR3. To address if NDGA was inhibiting the ATP binding pocket of FGFR3-TDII, in vitro kinase reactions were performed over a range of ATP concentrations in the presence of the peptide substrate poly (4:1 Glu, Tyr). The incorporation of ³²P from [γ-³²P]-ATP into the peptide was quantified and plotted (Fig. 1D). Increasing concentrations of ATP did not affect the K_m at various concentrations of NDGA, indicating the inhibitor NDGA is noncompetitive with ATP. Additionally, the inhibition of FGFR3-TDII autophosphorylation was independent of ATP over the same concentration range (data not shown). This finding further supports the contention that NDGA is noncompetitive with ATP.

NDGA reduces in vivo autophosphorylation of FGFR3. To examine the in vivo effect of NDGA on the autophosphorylation of FGFR3-TDII and WT FGFR3, transfected HEK293 cells were treated with 40 μmol/L NDGA for 1 h. As seen in Fig. 2A, the NDGA is able to reduce the autophosphorylation of both the WT and activated receptor. In Fig. 2B, the time dependence of the NDGA incubations was determined. A concentration of 30 μmol/L was able to begin to inhibit the receptor autophosphorylation in only 5 min, and phosphorylation was almost completely gone by 60 min. Next, we tested a broad range of NDGA concentrations with 40 μmol/L being the highest amount to examine the optimal effect. Figure 2C shows ∼50% inhibition of FGFR3-TDII autophosphorylation with 10 μmol/L NDGA treatment. Also seen in Fig. 2C (bottom), there is a decrease in the amount of the immature form of the FGFR3-TDII receptor at 35 μmol/L NDGA. This is most likely due to the ability of NDGA to block ER/Golgi transport. Total cell lysate was examined for STAT1 activation in Fig. 2D. The phosphorylation on Y701 of STAT1 caused by FGFR3-TDII is attenuated at the 20 μmol/L concentration of NDGA.

FGFR3 localization and Golgi breakdown with NDGA treatment. Typically, the disassembly of the Golgi begins to occur rapidly after only a few minutes of NDGA treatment and is reversible (23). As seen in Fig. 3A, NDGA promoted the disassembly of the Golgi in ∼15 min, whereas the subcellular localization of FGFR3-TDII seemed to become more punctate in the cytoplasm with increasing time. This may suggest that inhibition of the autophosphorylation of FGFR3-TDII by NDGA observed in Fig. 2 is due to the change in the receptor localization and/or the altered interaction with downstream molecules. To examine this hypothesis, we expressed targeted kinase domains of activated FGFR3 (30) and treated the cells with NDGA for up to 1 hour. As seen in Fig. 3B, the myrR3-TDII derivative, which is targeted to the plasma membrane by a myristylation signal, exhibits membrane localization that is unaltered by NDGA treatment. Similarly, in Fig. 3C the localization of the cytoR3-TDII derivative, targeted to the cytoplasm by a nonfunctional myristylation signal, seems unaltered with NDGA treatment.

NDGA reduction of FGFR3 downstream signaling. Next, we examined the change in downstream signaling molecules that are activated in the FGFR3 pathway. Previous work from our laboratory has shown the activation of STAT1, STAT3, and STAT5 in response to FGFR3 activation (31, 32). Given the importance of STAT activation in a variety of human malignancies (33), we surveyed STAT1 and STAT3 for altered activation in response to NDGA treatment. The phosphorylation of MAPK was also determined. In Fig. 4A, HEK293-transfected lysates were probed with phospho-specific antibodies for STAT1, STAT3, and MAPK. The signaling from FGFR3-TDII was greatly attenuated in response to NDGA. The low level of activation caused by the overexpression of FGFR3 WT was also inhibited by 30 μmol/L NDGA in 5 minutes. As seen in Fig. 4, there is an increase in phospho-MAPK at the 60-minute time points. This is consistent with the previous report by Deshpande (34), which showed that the treatment of FL5.12 cells with NDGA led to an increase in activation of MAPKs. In Fig. 4B, as a control for the inhibition of autophosphorylation of FGFR3 by NDGA as seen above, FGFR3 was immunoprecipitated from the lysate and examined for phosphotyrosine. Next, in Fig. 4C, we examined the STAT1 activation in HEK293 cells transfected with localized activated kinase domains of FGFR3. The phosphorylation on STAT1 caused by the expression of both myrR3-TDII and cytoR3-TDII was inhibited by 5 minutes of NDGA treatment. This would suggest that NDGA is able to block the signaling from an activated receptor independent of its localization. Curiously, for unknown reasons, significantly enhanced STAT1 phosphorylation was observed in response to the cytoR3-TDII, although expression of this FGFR3 derivative was previously shown to lack transforming activity (30).

Zilueton, a LOX inhbitor, does not block FGFR3-TDII autophosphorylation. To determine if the LOX enzyme inhibiting activity of NDGA was important for its effect on FGFR signaling, we examined another LOX inhibitor, Zileuton. Zileuton is a specific 5-LOX inhibitor, whereas NDGA is nonspecific (35, 36). As seen in Fig. 5A, when Zilueton was added to HEK293 cells transfected with FGFR3-TDII, there was no decrease in receptor autohosphorylation. Even with 30 μmol/L of Zilueton added for 1 hour, no change was detected. As a control, a duplicated plate was treated with 10 μmol/L NDGA, which again shows a significant decrease in FGFR3-TDII autophosphorylation. This would imply that the ability of NDGA to inhibit LOX is not required to block FGFR signaling.

Overnight treatment with NDGA leads to increased apoptosis in FGFR3-TDII–expressing cells. The above experiments were performed with very short term treatments of NDGA, <1 hour. Next, we examined the effect of a 24-hour exposure of various concentrations of NDGA on HEK293 FGFR3-TDII–transfected cells. PARP is a zinc finger DNA-binding enzyme, which detects DNA strand breaks and is activated at an intermediate stage of apoptosis. During late-stage apoptosis, PARP is cleaved and inactivated by the apoptotic proteases caspase-3 and caspase-7 (37). We examined the PARP cleavage in cell lysates treated with NDGA to determine the extent of apoptosis. As seen in Fig. 5B, there is a significant increase in PARP cleavage in FGFR3-TDII–expressing cells as the concentration of NDGA is increased. Thus, in addition to the ability of NDGA to inhibit FGFR3 signaling, it is clearly able to increase cellular apoptosis.

NDGA increases apoptosis in a multiple myeloma cell line with activated FGFR3. To examine the possible therapeutic effects of NDGA for multiple myeloma, a lethal disease that is characterized by the slow proliferation of malignant plasma cells in the bone, we examined two multiple myeloma–derived cell lines that express high levels of mutated FGFR3 for an increase in apoptosis with NDGA treatment. As seen in Fig. 5C, there is an increase in PARP cleavage in the KMS-18 cell line (FGFR3 G384D) with overnight treatment of NDGA. In Fig. 5D, the same concentrations of NDGA did not seem to have a similar effect on the KMS-11 cell line (FGFR3 Y373C). This may indicate that NDGA is effective only on a subset of activating mutations in FGFR3, for example, kinase domain mutations but not extracellular domain mutations such as Y373C. Resolution of this question will require further detailed analysis.

NDGA blocks downstream signaling in multiple myeloma cells. First, we examined the effect of NDGA on multiple myeloma cell lines treated with FGF2 to activate the FGFR3 receptor. The RPMI-8226 cell line contains “normal” levels of expression of WT FGFR3, whereas the LP-1 cell line has higher levels of FGFR3 expression. The cells were starved overnight and NDGA was added to the medium for 1 hour before stimulating the cells for 10 minutes by the addition of FGF2. As seen in Fig. 6A, there is a significant decrease in MAPK activation with 30 μmol/L NDGA treatment before stimulation, which would indicate that NDGA is able to inhibit the activation of endogenous, WT FGFR3. Next, we examined the effect of NDGA on MAPK activation in multiple myeloma cell lines that express high levels of mutationally activated FGFR3. As the concentration of NDGA increased with the 1-hour treatment of the KMS-18 cell line, the phosphorylation on MAPK decreased in cell lysates (Fig. 6B). These data suggest that NDGA is able to block signaling from FGFR3 that is activated by ligand or mutation. However, there was no significant change in MAPK activation with the same concentrations of NDGA in the KMS-11 cell line as seen in Fig. 6C, which may be dependent on the particular mutation of FGFR3; in the case of KMS-11, a Y373C mutation that introduces an abnormal disulfide bond in the extracellular domain.

NDGA is a natural compound with an interesting history and broad spectrum of biological properties. Initially isolated from the creosote bush Larrea divaricatta, it was originally characterized as a potent antioxidant and shown to selectively inhibit arachidonic acid 5-LOX (38, 39), thereby leading to a reduction of inflammatory pathways through decreased leukotriene and prostaglandin synthesis. NDGA also shows profound effects on the secretory pathway, reflected in its ability to block protein transport from the ER to the Golgi apparatus, and to induce the redistribution of Golgi proteins into the ER (22–24, 40). As a nonspecific inhibitor of NADPH oxidase and protein kinase C, NDGA also disrupts the actin cytoskeleton and exerts effects on cell adhesion (41, 42). Although some reports indicate that NDGA treatment can inhibit apoptosis, as in inhibition of CD95L-induced apoptosis of glioma cells (43), other reports show NDGA-induced apoptosis in a variety of cells, including human breast cancer cells, pancreatic carincoma cells, and HL-60 cells (44–46). NDGA has been shown to induce death receptor 5/TRAIL-R2 expression, thereby sensitizing malignant tumor cells to TRAIL-induced apoptosis (47). More recently, NDGA has been shown to directly inhibit activation of two RTKs, the IGF-IR and the c-erbB2/HER2/neu receptor, resulting in decreased cellular proliferation (25, 27, 48, 49).

NDGA is a component of “Chaparral,” a natural product proposed several decades ago as a treatment for some cancers but was removed from the Food and Drug Administration (FDA) “generally recognized as safe” list in 1970. Since then, NDGA, also called masoprocol, received FDA approval for inclusion in a topical cream Actinex, under FDA Application No. (NDA) 019940, for treatment of actinic keratoses. Recent studies of its properties have led to rekindled interest in its biochemical and chemotherapeutic properties, as evidenced by an ongoing clinical trial, “Phase I Study of NDGA in Patients With Nonmetastatic, Biochemically Relapsed Prostate Cancer on Androgen Dependent Prostate Cancer (ADPC)” (UCSF-035510; ref. 48). Interestingly, different stereoisomers of NDGA occur, and whether these differ with regard to their biological properties has not been investigated. NDGA is found naturally as a mixture of chiral compounds, including the meso forms R,R-NDGA and S,S-NDGA, and the optically active form S,R-NDGA. To fully understand the spectrum of various biological activities exhibited by NDGA with regard to oncogenic RTKs such as IFG-IR and HER2 (25, 27, 49), or FGFR3 examined here, it may be desirable to synthesize chemically pure NDGA isomers for a thorough examination of their biological properties.

Given the importance of FGFR3 activation in several human cancers of clinical importance, it will be significant to determine whether the inhibitory properties of NDGA toward other RTKs, such as IGF-IR or HER2, apply to FGFR3. The activation of cellular signaling pathways by strongly activated mutants of FGFR3 such as the K650E mutant, responsible for TDII (10), or the mutation K650M, responsible for Severe Achondroplasia with Delayed Development and Acanthosis Nigricans (50), has been shown to occur within the secretory compartment, as evidenced by direct recruitment of Jak1 and STAT1 activation, and by Erk1/2 activation from the ER through FRS2α and PLCγ-independent pathways (18–20). Thus, the ability of NDGA to disrupt the secretory compartment, acting together noncompetitively with ATP and its previously shown inhibitory properties toward other RTKs, suggests that it might be particularly effective against FGFR3.

Indeed, in the experiments described here, we show that NDGA exhibits strong inhibitory effects against FGFR3 kinase activity and the activation of downstream signaling pathways. After expression in HEK293 cells and immunoprecipitation, the activated FGFR3 TDII mutant protein is directly inhibited when assayed for autophosphorylation activity. Furthermore, we show that although NDGA directly inhibits FGFR3-TDII kinase activity, it acts noncompetitively with ATP using this cell-free assay. This mode of inhibition resembles that of a structurally similar compound, AG 538, which was shown to inhibit IGF-IR and act noncompetitively with ATP (51). This indicates that NDGA may be a specific inhibitor toward FGFR3-TDII and other RTKs, rather than nonspecifically inhibiting a wide range of kinases. When cell lysates were examined for Tyr-phosphorylated FGFR3, a concentration of 30 μmol/L NDGA began to show inhibition of receptor phosphorylation in only 5 minutes, and receptor autophosphorylation was almost completely eliminated by 60 minutes. In this assay, the effects of NDGA were almost certainly the result of direct inhibition together with indirect inhibitory effects, such as disruption of protein secretion and collapse of the Golgi. This latter effect was clearly visible by immunofluorescent staining of the Golgi using the cis-Golgi marker 10E6. NDGA treatment also resulted in significantly diminished activation of STAT1 and STAT3 by the FGFR3 TDII mutant. Prior work shows that strongly activated mutants of FGFR3 actually recruit Janus-activated kinase 1 directly to the ER, leading to activation of STAT family members, which is not disrupted by brefeldin A (BFA; refs. 18, 19). In our results reported here, we found that NDGA, an agent previously shown to disrupt the Golgi by mechanisms distinct from BFA (52), rapidly blocked activation of STAT1 and STAT3. Although NDGA possesses strong antioxidant activity, the effects of NDGA on FGFR3 activation and signaling were not attributable solely to antioxidant activity, as shown by the inability of a chemically unrelated antioxidant, Zileuton, to affect FGFR3 activation. Using HEK293 cells expressing FGFR3, we showed that NDGA significantly increases apoptosis for cells expressing the activated TDII mutant and in the multiple myeloma cell line KMS-18, as shown by PARP cleavage. Lastly, using the multiple myeloma cell line LP-1 expressing FGFR3 WT, we show that FGF-dependent signaling, as reflected in the appearance of p-MAPK, is effectively blocked by NDGA treatment. NDGA also reduces p-MAPK in the KMS-18 multiple myeloma cell line expressing FGFR3 G384D.

Collectively, the results presented here indicate that NDGA, acting most likely by multiple mechanisms, including direct inhibition of kinase activity, disruption of the secretory pathway, inhibition of downstream signaling pathways, and increased apoptosis, possesses potentially beneficial attributes that merit further investigation. This is particularly true for those cancers where cellular proliferation depends upon continued stimulation of FGFR3 either by ligand activation or by mutational activation, as in bladder carcinoma and multiple myeloma.

No potential conflicts of interest were disclosed.

Grant support: NIH (R01-CA90900) and the Multiple Myeloma Research Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Prof. Leslie Thompson of the University of California, Irvine (Irvine, CA) for providing us with human multiple myeloma cells, Prof. Uli Muller for use of phosphorimaging equipment, Kristy Drafahl and Lisa Salazar for technical assistance, and Laura Castrejon for editorial assistance.