Inhibition of colony-stimulating-factor-1 signaling in vivo with the orally bioavailable cFMS kinase inhibitor GW2580

Abstract

Colony-stimulating-factor-1 (CSF-1) signaling through cFMS receptor kinase is increased in several diseases. To help investigate the role of cFMS kinase in disease, we identified GW2580, an orally bioavailable inhibitor of cFMS kinase. GW2580 completely inhibited human cFMS kinase in vitro at 0.06 μM and was inactive against 26 other kinases. GW2580 at 1 μM completely inhibited CSF-1-induced growth of mouse M-NFS-60 myeloid cells and human monocytes and completely inhibited bone degradation in cultures of human osteoclasts, rat calvaria, and rat fetal long bone. In contrast, GW2580 did not affect the growth of mouse NS0 lymphoblastoid cells, human endothelial cells, human fibroblasts, or five human tumor cell lines. GW2580 also did not affect lipopolysaccharide (LPS)-induced TNF, IL-6, and prostaglandin E2 production in freshly isolated human monocytes and mouse macrophages. After oral administration, GW2580 blocked the ability of exogenous CSF-1 to increase LPS-induced IL-6 production in mice, inhibited the growth of CSF-1-dependent M-NFS-60 tumor cells in the peritoneal cavity, and diminished the accumulation of macrophages in the peritoneal cavity after thioglycolate injection. Unexpectedly, GW2580 inhibited LPS-induced TNF production in mice, in contrast to effects on monocytes and macrophages in vitro. In conclusion, GW2580's selective inhibition of monocyte growth and bone degradation is consistent with cFMS kinase inhibition. The ability of GW2580 to chronically inhibit CSF-1 signaling through cFMS kinase in normal and tumor cells in vivo makes GW2580 a useful tool in assessing the role of cFMS kinase in normal and disease processes.

Colony-stimulating factor 1 (CSF-1) promotes the survival, proliferation, and differentiation of mononuclear phagocyte lineages. CSF-1 exerts its activities by binding to cell-surface cFMS receptors, resulting in autophosphorylation by receptor cFMS kinase and a subsequent cascade of intracellular signals (reviewed in ref. 1). Receptor expression in macrophage lineages is consistent with the ability of exogenous CSF-1 to increase cytokine production in mice after lipopolysaccharide (LPS) challenge (2), increase the production of monocytes and macrophages in mice (3), and exacerbate arthritis in mice (4, 5) and rats (6). Mice with a nonfunctional CSF-1 ligand (7) or receptor (8) are osteopetrotic, deficient in several macrophage populations, and have diminished response to inflammatory challenge (9, 10), confirming the importance of the CSF-1-cFMS-receptor pathway in the regulation of macrophage lineages.

The CSF-1-cFMS-receptor pathway is up-regulated in a number of human pathologies that involve chronic activation of tissue macrophage populations and, thus, could be a target for drug therapy. CSF-1 promotes osteoclast development and bone degradation in vitro (11, 12) and, thus, could contribute to the excessive osteoclast activity in osteoporosis and at sites of orthopedic implant failure (13, 14). CSF-1 is elevated in the synovial fluid of rheumatoid arthritis patients (15), and synovial fibroblasts from rheumatoid arthritis patients produce high levels of CSF-1 (16), suggesting a role for CSF-1 in joint degradation. Increases in CSF-1 production are also associated with the accumulation of tissue macrophages seen in inflammatory bowel disease (17), glomerulonephritis (18), allograft rejection (19), and arteriosclerosis (20). In addition, the growth of several tumor types is associated with overexpression of CSF-1 and cFMS receptor in cancer cells and/or tumor stroma (21-24).

Antibodies, antisense, and genetic techniques have been used to evaluate the role of the CSF-1-cFMS pathway in animal disease models. Antibodies to CSF-1 inhibited arthritis-(4) and ovariectomy-induced bone loss in mice (25). Antibodies to the cFMS receptor inhibited the early stages of atherogenesis in mice (26) and decreased macrophage accumulation in mouse models of renal inflammation (27) and allograft rejection (19). Inhibition of CSF-1 signaling by antisense and small interfering RNAs inhibited tumor growth in mice (28), and tumors in mice with nonfunctional CSF-1 grew more slowly than tumors in control mice (29, 30).

To further investigate the importance of the CSF-1-cFMS-receptor pathway in normal and disease situations, we characterized the potency, selectivity, and bioavailablity of GW2580, a small-molecule inhibitor of cFMS kinase activity. The data show that GW2580 inhibits CSF-1 signaling in macrophage lineages and tumor cells in vitro and in vivo and is an inhibitor of TNF production in vivo.

Materials and Methods

Animals and Compound Dosing. Female C3H/HEN mice (Taconic Farms) or female CD-1 nude mice (Charles River Breeding Laboratories) weighing 22-26 g were used. The research complied with national legislation and company policy on the care and use of animals and related codes of practice.

GW2580 (31) was suspended in 0.5% hydroxypropylmethylcellulose and 0.1% Tween 80 by using multiple strokes with a Teflon-glass homogenizer and was dosed orally at 0.2 ml per mouse.

Kinase, Cell-Growth, and Cytokine-Production Assays. For detailed methods, see Supporting Methods, which is published as supporting information on the PNAS web site. For kinase assays, the catalytic domains of human kinases were expressed, isolated, and assayed with ATP and appropriate substrates. For cell-growth assays, the cells were cultured so that the addition of serum or growth factors resulted in linear growth. GW2580 was added to assays as a 10- to 12-point dose-response at a maximal concentration of 20 μM.

For detailed methods for assessing the effect of GW2580 on LPS-induced cytokine production from cells, see Supporting Methods (and see Figs. 6 and 7, which are published as supporting information on the PNAS web site). In one study, human peripheral-blood mononuclear cells (PBMCs), monocytes, and macrophages induced by a 1-week exposure to 35 ng/ml CSF-1 received GW2580 in dose-response, followed 0.5 h later by stimulation with LPS and measurement of TNF, IL-6, and prostaglandin (PG)E2 16 h later. A second study was similar, except i.p. macrophages from mice that received an i.p. injection of vehicle or 2.5 μg of CSF-1 per mouse 3.5 h earlier were used.

Human Osteoclast Bone-Degradation Assay. PBMCs were isolated from a healthy female volunteer, and 5 × 10⁵ cells in 1 ml of MEM with 10% FCS was added to the wells of a 24-well plate containing either a bone slice or a plastic coverslip. After the addition of receptor activator of NF-κB ligand (30 ng/ml) and CSF-1 (30 ng/ml) to all wells, the plates were incubated for 10 days. Six-hundred microliters of the medium was changed every 3 days. At day 10, the medium was changed, and GW2580 was added from a DMSO stock to yield 0.1% DMSO in all wells. Samples of bone slices were taken at day 10 as baseline controls for bone resorption. At day 13, the coverslips and bone slices were recovered, and total cells and 23C6-positive cells with three or more nuclei were counted on the coverslips by using immunocytochemistry. Resorption of bone was assessed by removing the cells with 10% NaOCl, coating the bone slices with gold, and quantifying the resorption area by incident-light microscopy. Actin-ring formation by osteoclasts was assessed after a 3-h exposure to GW2580 on day 13 (32).

Rat Calvaria and Fetal Long-Bone-Degradation Assays. For methods for the rat calvaria and fetal long bone assays, see Supporting Methods (and see Figs. 8 and 9, which are published as supporting information on the PNAS web site). Rat calvaria were incubated with parathyroid hormone (PTH) and GW2580 for two 48-h periods. The media from the second 48-h period were assayed for deoxypyridinoline crosslinks (DPPD). Long bones from fetal rats preloaded with ⁴⁵CaCl₂ were incubated for 24 h. GW2580 and PTH were then added, and the release of labeled calcium was assessed 48 h later.

GW2580 Plasma Concentrations and Plasma Protein Binding. Mice were dosed orally with GW2580 and killed with CO_2, and plasma was prepared. One-hundred microliters of 10% isopropyl alcohol/90% acetonitrile was added to 50 μl of plasma sample, followed by 1 min of vortexing, 2 min of sonication, and 2 min of centrifugation at 5,600 × g. The supernatant was analyzed by LC/MS by using atmospheric-pressure chemical ionization in the positive ion mode with multiple-reaction monitoring. GW2580 was quantified by using a standard curve of GW2580 prepared in plasma.

GW2580 was added to triplicate plasma samples at 3 μM and incubated 1 h at 37°C. The samples were centrifuged at 1,500 × g for 15 min at 4°C by using a centrifuge unit containing a 30,000-MW filter (Amicon). The percentage of GW2580 bound by plasma proteins was calculated from the GW2580 concentrations in the filtrate and retentate.

CSF-1 Priming of LPS-Induced Cytokine Production in Vivo. GW2580 was dosed before and after CSF-1 priming to evaluate effects on CSF-1 priming of LPS-induced cytokine production. Mouse CSF-1 (R & D Systems) was injected i.p. at 1.8 μg in 0.2 ml of PBS. LPS (L-2630, Sigma) was injected i.p. at 300 μg in 0.5 ml of PBS. Plasma was prepared 1.5 h after the LPS injection, and TNF and IL-6 were measured with mouse-specific ELISA kits (R & D Systems).

Thioglycolate-Induced Macrophage Influx into the Peritoneal Cavity. C3H/HEN mice received an i.p. injection of 1.5 ml of thioglycolate (VWR Scientific) on day 0, and, on day 4, the mice were killed with CO₂, and 5 ml of cold PBS containing 40 units/ml heparin was injected i.p. Fluid was withdrawn with an 18-gauge needle, and the cells were counted in a hemocytometer after dilution in 0.1% crystal violet in PBS.

Growth of M-NFS-60 Cells in the Peritoneal Cavities of Mice. Cells were suspended in PBS, and 1 × 10⁷ cells were injected i.p. into CD-1 nude mice. Four days later, the i.p. cells were counted as described in Materials and Methods. Almost all of the recovered cells were tumor cells, not endogenous macrophages.

Data Presentation and Statistical Analysis. All data are means (±SEM). All experiments were analyzed by using Dunnett's multiple-comparison test.

Results

Selective Effects on Kinase Activity, Cell Growth, and Cytokine Production in Vitro. GW2580 (Fig. 1) completely inhibited human cFMS kinase in vitro at 0.06 μM and was 150- to 500-fold selective compared to human (bRAF, CDK4, cKIT, cSRC, EGFR, ERBB2, ERBB4, ERK2, FLT-3, GSK3, ITK, JAK2, JNK3, MK2, P38, PDGFR-b, PDHK4, PKA, PLK1, PKCα, PKCβ1, PKCζ, SYK, TIE2, and VEGFR2) and mouse (LCK) kinases (see Table 5, which is published as supporting information on the PNAS web site). GW2580 completely inhibited the growth of CSF-1-dependent mouse myeloid M-NFS-60 (33) cells at 0.7 μM. In contrast, the growth of mouse myeloid NS0 cells, a CSF-1-independent cell line, was highly resistant to GW2580 (Table 1). In freshly isolated human monocytes, GW2580 at 1 μM maximally inhibited CSF-1-, granulocyte-macrophage-stimulating factor (GMCSF)- and LPS-induced growth by 100%, 80%, and 50%, respectively (Fig. 2). In contrast to monocytes, the growth of human foreskin fibroblasts, endothelial cells, and five tumor cell lines was highly resistant to GW2580 (Table 1).

Fig. 1.

Structure of GW2580

Table 1. Inhibition of cell growth by GW2580.

Cell type utilized	Stimulating growth factor(s)	Inhibitory IC50,* μM
Mouse M-NFS-60 myeloid tumor cells	CSF-1	0.33 ± 0.05 (15)
Mouse NSO myeloid tumor cells	Serum	13.5 ± 1.4 (15)
Freshly isolated human monocytes	CSF-1	0.47 ± 0.04 (7)
Human foreskin fibroblasts	Serum	>30
Human umbilical vein vascular endothelial cells	VEGF	12, 13
Human umbilical vein vascular endothelial cells	bFGF	>30, 10
Human MDA231 breast tumor cells	Serum	>30
Human A549 lung tumor cells	Serum	>30
Human BT474 breast tumor cells	Serum	21
Human HN5 head/neck tumor	Serum	29
Human N87 gastric tumor	Serum	>30

^*Data represent mean (±SEM) (n). Data without error terms are single determinations

Fig. 2.

Effects of GW2580 on CSF-1-, GMCSF-, and LPS-induced growth of human monocytes. (Upper) Monocytes from each of four donors were split, and effects on CSF-1- and GMCSF-induced growth were compared. (Lower) Monocytes from each of four donors were split, and effects on CSF-1- and LPS-induced growth were compared.

GW2580 did not affect LPS-induced TNF, IL-6, or PGE2 production from human PBMCs, monocytes, or CSF-1-generated macrophages in vitro (Fig. 6). GW2580 was also evaluated in macrophages isolated from the peritoneal cavities of mice injected i.p. with vehicle or CSF-1 3.5 h before being killed. As expected (2), the macrophages from the mice pretreated with CSF-1 produced about twice as much TNF and IL-6 after LPS challenge than macrophages from vehicle-treated mice. GW2580 did not affect LPS-induced TNF and IL-6 production in normal or CSF-1-primed macrophages (Fig. 7).

Inhibition of Bone Degradation in Vitro. Macrophages and osteoclasts were produced by culturing human monocytes for 10 days in the presence of 30 ng/ml CSF-1 and 30 ng/ml receptor activator of NF-κB ligand. Exposure of these cells to GW2580 from day 10 to day 13 decreased the number of cells attached to plastic dishes and the number of osteoclasts attached to bone slices in parallel cultures. GW2580 also inhibited bone resorption over the 3-day period and inhibited the actin rings formed over a 3-h period on day 13 (Table 2).

Table 2. Inhibition of human osteoclast degradation of bone in vitro with GW2580.

Day sampled*	Drug conc., μM	Cells per cm2 plastic	23C6(+) cells (>3 nuclei) per cm2 bone	Bone slice resorbed, %	Actin rings per cm2 of bone
10					3.3 ± 0.4†
13	0	9,723 ± 1,380‡	854 ± 122	26.6 ± 4.7	1,546 ± 232
13	1	2,984 ± 385§	244 ± 57§	3.3 ± 0.6§	479 ± 183§
13	3	3,369 ± 1,187	120 ± 38§	1.5 ± 0.4§	148 ± 69§
13	10	2,728 ± 931§	34 ± 9§	1.2 ± 0.6§	0 ± 0¶

Conc., concentration.

^*On day 10, GW2580 was added to the media, and cell numbers and bone resorption were assessed on day 13. Actin rings were counted after a 3-h exposure on day 13

^†P < 0.05

^‡Data represent mean (±SEM) for six cultures. Cultures with no GW2580 were used as controls in the statistical analysis

^§P < 0.01

^¶P < 0.001

PTH increased the concentration of the bone-degradation marker DPPD in the media of the rat calvaria from 17 to 27 nM. GW2580 inhibited the PTH response by 80% at 0.1 μM, with 1, 10, and 100 μM GW2580 decreasing DPPD to below the basal level of 17 nM (Fig. 8). PTH increased the release of labeled calcium in rat fetal long bone from 15% to 60%. GW2580 caused a 30-40% inhibition of PTH-induced calcium release at 0.1-0.3 μM, with higher concentrations of 1, 3, and 10 μM completely inhibiting the PTH response (Fig. 9).

Pharmacokinetics in Mice. Oral administration of 20 and 80 mg/kg GW2580 to mice gave maximal plasma concentrations of 1.4 and 5.6 μM, respectively (Fig. 3). At 3 μM in vitro, 24%, 93%, 95%, and 98% of the GW2580 was bound to protein in cell-culture media with 10% FCS, mouse plasma, rat plasma, and human plasma, respectively. Subtracting the fraction of compound bound to plasma proteins gives maximal plasma concentrations of 0.05 and 0.4 μM after oral dosing at 20 and 80 mg/kg, respectively. These maximal plasma concentrations of unbound GW2580 in vivo are near the concentrations of free compound needed for half-maximal inhibition of CSF-1-induced mouse M-NFS-60 cell growth in vitro (i.e., GW2580 = 0.25 μM).

Fig. 3.

Plasma concentrations of GW2580 after oral dosing in mice. Mice were dosed orally, and three were killed at different times for measurements of GW2580 concentrations in plasma.

Inhibition of CSF-1-Induced Priming of LPS-Induced Cytokine Production in Vivo. Exogenous CSF-1 given intraperitoneally to mice before an i.p. injection of LPS increases the plasma levels of IL-6 and TNF (2), thus providing a bioassay to assess the ability of GW2580 to inhibit CSF-1 signaling in vivo. GW2580 was dosed orally at 40 mg/kg 0.5 h before the CSF-1- or vehicle-priming dose (i.e., 4.5 h before the LPS injection or 0.5 h before the LPS injection). GW2580 inhibited TNF production at both times of injection with and without CSF-1 priming (Table 3). In contrast, GW2580 did not affect IL-6 production when given to CSF-1- or vehicle-primed mice 0.5 h before the LPS injection. When given to mice before the vehicle-priming injection, GW2580 caused a small increase in IL-6, but, when given to mice before CSF-1 priming, GW2580 completely blocked the ability of CSF-1 to prime the mouse for increased IL-6 production (Table 3). Dose-response studies using 20, 40, and 80 mg/kg GW2580 before a CSF-1- or vehicle-priming dose also showed that GW2580 inhibited both vehicle-primed and CSF-1-primed LPS-induced TNF production (Fig. 4). As in the first study with 40 mg/kg (Table 3), the dose-response study with GW2580 showed that GW2580 given before the vehicle-priming dose slightly increased LPS-induced IL-6 production, whereas GW2580 given before the CSF-1-priming dose completely blocked the ability of MCSF to increase LPS-induced IL-6 production (Fig. 4).

Table 3. Effects of GW2580 on CSF-1-priming of LPS-induced TNF and IL-6 production in mice.

t = –4.5 h* PO compound	t = –4 h i.p. Primer	t = –0.5 h PO compound	Plasma TNF,† ng/ml	Inhibition, %	Plasma IL-6, ng/ml	Inhibition, %
Vehicle	Vehicle		11.5 ± 1.0	Control	11.8 ± 1.5	Control
	Vehicle	GW2580	3.7 ± 0.4	68‡	9.5 ± 0.4	20
GW2580	Vehicle		6.7 ± 0.7	42	15.1 ± 2.4	–28
Vehicle	CSF-1		40.6 ± 4.1	Control	20.0 ± 1.6	Control
	CSF-1	GW2580	15.2 ± 2.6	63§	21.2 ± 2.9	–6
GW2580	CSF-1		17.1 ± 1.8	58§	10.2 ± 0.7	49‡

PO, orally administered.

^*Compound (40 mg/kg) and primer (CSF-1 at 1.7 μg per mouse) were dosed at various times (t) before an i.p. dose of LPS at t = 0

^†TNF and IL-6 were measured 1.5 h after LPS injection. Data represent mean (±SEM) with six to seven mice per group

^‡P < 0.01

^§P < 0.001

Fig. 4.

Effect of GW2580 on CSF-1-priming of LPS-induced TNF and IL-6 production in vivo in mice. GW2580 was dosed 45 min before an i.p. injection of vehicle (Upper) or CSF-1 (Lower)at1.7 μg per mouse. Three and a half hours later, all mice received an i.p. injection of LPS, followed by measurement of TNF (○) and IL-6 (▪) in the plasma 1.5 h after LPS. Each point represents data from six to seven mice. Statistical changes relative to oral-dosing vehicle are shown. *, P <0.01; †, P <0.001.

Inhibition of the Growth of M-NFS-60 Tumor Cells in Vivo in Mice. GW2580 was dosed orally at 20 and 80 mg/kg twice a day (b.i.d.), starting 1 h before the i.p. injection of M-NFS-60 cells, and the tumor cells in the peritoneal cavity were counted 4 days later. In both studies, GW2580 produced a dose-related decrease in the number of tumor cells, with the 80 mg/kg dose completely blocking tumor growth (Fig. 5). GW2580 showed no effect on body weights taken when the animals were killed (data not shown).

Fig. 5.

Effect of GW2580 on the growth of M-NFS-60 cells in vivo in mice. Mice were treated b.i.d. with vehicle or GW2580 starting 1 h before i.p. injection with 1 × 10⁷ M-NFS-60 cells. The number of i.p. cells was counted on day 4. Statistical changes relative to the tumor-only group are shown (*, P <0.05; †, P <0.001). In a second study, GW2580 also caused statistically significant inhibition of tumor load, with no-tumor, tumor-only, tumor-plus-20 mg/kg-GW2580, and tumor-plus-80 mg/kg-GW2580 groups yielding 5.6 ± 0.6 (†), 33.8 ± 2.8, 19.3 ± 4.5 (*), and 10.9 ± 1.7 (†) million i.p. cells per mouse, respectively.

Inhibition of Thioglycolate-Induced Macrophage Influx into the Peritoneal Cavity in Vivo. In the first study, GW2580 was dosed orally b.i.d. at 20 and 80 mg/kg, starting 1 h before thioglycolate injection, and i.p. macrophages were counted 4 days later. GW2580 at 20 and 80 mg/kg inhibited macrophage accumulation in the peritoneal cavity by 17% and 25%, respectively (Table 4). In the second study, GW2580 was dosed orally b.i.d. the week before thioglycolate injection and for the 4-day period after thioglycolate injection. With this extended treatment regimen, the 20 and 80 mg/kg doses of GW2580 inhibited macrophage accumulation by 8% and 45%, respectively (Table 4). GW2580 showed no effect on body weights taken when the animals were killed (data not shown).

Table 4. Effect of GW2580 on accumulation of i.p. macrophages 4 days after an i.p. injection of thioglycolate.

			Dosed days 0–4*		Dosed days –7 to 4
Compound	Dose, mg/kg	Thioglycolate on day 0	IP cells on day 4†	Inhibition, %	IP cells on day 4	Inhibition, %
Vehicle	–	Vehicle	1.2 ± 0.1		1.5 ± 0.1
Vehicle	–	Thioglycolate	9.4 ± 0.8	Control	8.9 ± 0.8	Control
GW2580	20	Thioglycolate	8.0 ± 0.9	17	8.3 ± 0.3	8
GW2580	80	Thioglycolate	7.5 ± 0.4	25	5.6 ± 0.7	45‡

^*Compounds dosed orally b.i.d. on days 0–4 starting 1 h before thioglycolate injection. The second group was dosed 7 days before and on days 0–4 after thioglycolate injection

^†Data represent mean (±SEM) of 10⁶ cells per ml of recovered peritoneal fluid. Six to seven mice per group were used

^‡P < 0.01

Discussion

Potency and Selectivity in Enzymatic and Cellular Assays. The utility of GW2580 as an investigative and potential therapeutic tool depends on its potency, selectivity, and bioavailability. GW2580 at 0.8-1 μM completely blocked the ability of CSF-1 to induce the growth of mouse M-NFS60 myeloid cells and human monocytes (Table 1). GW2580 at 1 μM also inhibited bone degradation in human osteoclasts, rat calvaria, and rat fetal long bone by 80-100% (Table 2) (see Figs. 8 and 9). These data show that GW2580 maximally inhibited the growth of monocytes and bone degradation at about 1 μM in vitro.

GW2580 acts as a competitive inhibitor of ATP binding to the cFMS kinase (31) and, thus, could show nonspecific kinase inhibition. To assess nonspecific effects, GW2580 was evaluated in assays of kinase activity, cell growth, and cytokine production. GW2580 was inactive against 26 kinases in vitro and did not inhibit the growth of mouse NS0 lymphoblastoid cells, human fibroblasts, human endothelial cells, and five human tumor cell lines (Table 1). GW2580's inactivity toward VEGF-induced endothelial cell growth (Table 1), an assay sensitive to VEGFR2 kinase inhibitors (34), correlated with its inactivity toward VEFGR2 kinase. GW2580's inactivity toward the growth of N87 gastric tumor, HN5 head/neck tumor, and BT474 breast tumor cells (Table 1), assays sensitive to EGFR and ERBB2 kinase inhibitors (35), correlated with its inactivity against EGFR and ERBB2 kinases. GW2580 also showed no effect on LPS-induced TNF, IL-6, and PGE2 production in human PBMCs, monocytes, and macrophages and in normal and CSF-1-primed mouse macrophages in vitro (Figs. 6 and 7). GW2580's inactivity against LPS-induced cytokine production in monocytes and macrophages, a response sensitive to P38 inhibitors (36), correlated with its inactivity against P38 kinase. These data show that GW2580 selectively inhibits monocyte growth compared to eight other cell types and is able to differentiate between monocyte growth and the complex pathways involved in LPS induction of TNF, IL-6, and PGE2 production in monocytes and macrophages in vitro.

GW2580 at 1 μM caused maximal 100%, 80%, and 50% inhibition of CSF-1-, GMCSF- and LPS-induced monocyte growth, respectively, over the 5 days (Fig. 2). GW2580 did not inhibit JAK2 kinase, a key component of GMCSF signaling (37). Inhibition of cFMS kinase could explain the inhibition of GMCSF- and LPS-induced growth over the 5-day assay. In a time-course study, GMCSF caused a maximal increase in CSF-1 production by monocytes by 3 h (38).In another study, both GMCSF and LPS increased CSF-1 production in monocytes at 18 h, the only time point reported (39). GMCSF-induced CSF-1 is likely active, given that antibodies to CSF-1 completely inhibited GMCSF-mediated monocyte survival over 8 days (40), and antibodies to the CSF-1 receptor inhibited GMCSF-mediated monocyte growth over 5 days by 40% (41). Defining the possible effects of GW2580 on GMCSF and LPS signaling will require analysis of signal-transduction events before the increase in CSF-1 production.

The ability of GW2580 to decrease the human osteoclast number, osteoclast actin-ring formation, and bone degradation in cultures containing CSF-1 (Table 2) is consistent with the role of CSF-1 in promoting osteoclast development and bone degradation (11, 12). The strong inhibition of PTH-induced bone degradation in fetal long bone and calvaria with GW2580 (Figs. 8 and 9) is consistent with an inhibition of cFMS kinase, given that PTH induces CSF-1 production in bone cultures (42). GW2580 did not inhibit PKA or PKC kinases known to mediate PTH induction of bone degradation in vitro (43).

Recently, the small-molecule kinase inhibitor SU11248 was found to inhibit CSF-1-receptor phosphorylation and osteoclast development and function in vitro (44). Unlike GW2580 (Table 1), SU11248 also inhibits VEGFR and FGFR kinases and VEGF- and FGF-induced proliferation of endothelial cells in vitro, showing that GW2580 is more selective for cFMS receptor kinase.

Inhibition of CSF-1 Signaling in Vivo. To directly assess inhibition of CSF-1 signaling in vivo, CSF-1 was used to “prime” mice to produce higher levels of plasma IL-6 and TNF after a LPS injection (2). Dosing GW2580 before the injection of CSF-1 decreased LPS-induced IL-6 production, whereas dosing after CSF-1 or dosing without CSF-1 did not affect IL-6 production, indicating that GW2580 specifically inhibits the ability of CSF-1 to prime the mouse for LPS-induced IL-6 production (Table 3; see also Fig. 4).

In contrast to the specific effects on CSF-1 priming of IL-6 production, GW2580 inhibited LPS-induced TNF production with or without CSF-1 priming (Table 3; see also Fig. 4). This inhibition of TNF production in vivo is unexpected, given that GW2580 did not inhibit TNF production from cells in vitro (Figs. 6 and 7). One might speculate that GW2580 is forming a unique metabolite in vivo that inhibits TNF production, but this seems unlikely, given that cFMS inhibitors of different structures inhibit TNF production in vivo (J.G.C. and S.D.C., unpublished data). Understanding why GW2580 inhibits TNF but not IL-6 production will require further research.

GW2580 is rapidly cleared in vivo (Fig. 3), thus it is possible that GW2580 does not inhibit cFMS kinase throughout the b.i.d. dosing schedule. To determine whether GW2580 can impact chronic CSF-1-dependent processes in vivo, we measured the growth of CSF-1-dependent M-NSF-60 cells in the peritoneal cavities of mice. GW2580 decreased tumor growth in a dose-dependent manner, showing that the degree and duration of cFMS kinase inhibition is sufficient to inhibit CSF-1-dependent growth in vivo (Fig. 5). It is possible that even intermittent inhibition of CSF-1 signaling can inhibit tumor growth in vivo.

One might expect GW2580 to diminish macrophage accumulation after injection of thioglycolate into the peritoneal cavity, given that antibodies to the cFMS receptor inhibited macrophage accumulation in the kidney during ureteric obstruction (27) and allograft rejection (19). GW2580 showed a modest inhibition of macrophage accumulation when dosed during the 4 days after the thioglycolate injection but showed more inhibition when also dosed 7 days before the injection of thioglycolate. This increase in effect is likely due to the time-dependent depletion of thioglycolate-responsive macrophage populations before the thioglycolate injection (Table 4).

In conclusion, GW2580 inhibited CSF-1-driven cell growth and bone degradation in vitro and had sufficient oral bioavailability to chronically inhibit cFMS kinase in vivo, thus providing a unique inhibitor for investigation of the role of cFMS kinase in normal and disease processes. The observation that GW2580 acutely inhibited LPS-induced TNF production in vivo suggests that cFMS kinase activity is involved in the mechanism by which LPS induces TNF production in vivo. Inhibition of TNF production by cFMS inhibitors, coupled with inhibition of macrophage production and differentiation, may prove of value in treating diseases such as rheumatoid arthritis, where macrophage lineages and TNF act in concert to promote inflammation and joint degradation. In addition, the ability of GW2580 to inhibit bone degradation in vitro and CSF-1-dependent tumor growth in vivo suggests that GW2580 could impact osteoporosis and tumor progression. Chronic dosing over a month or more will be needed to fully assess the impact of GW2580 on time-dependent tissue-macrophage turnover and resultant macrophage function in normal rats and in arthritis, osteoporosis, and cancer models.