WO1997019352A1

WO1997019352A1 - Therapeutic agents and screening method

Info

Publication number: WO1997019352A1
Application number: PCT/EP1996/004913
Authority: WO
Inventors: Mira Susa Spring
Original assignee: Novartis Ag
Priority date: 1995-11-22
Filing date: 1996-11-11
Publication date: 1997-05-29
Also published as: AU7622396A

Abstract

Although fluoride is a potent stimulator of bone formation, it remains of questionable use as a therapeutic principle in view of its narrow therapeutic window, toxicity and its tendency to induce poor quality bone formation. We provide a method for screening candidate compounds which act to increase bone formation comprising assessing the activity of the compounds as G protein stimulators.

Description

THERAPEUTIC AGENTS AND SCREENING METHOD.

The present invention relates to a method for screening compounds for activity as osteoblast stimulators to improve bone formation, particularly in cases of osteoporosis, and to fluoride mimics useful in treating osteoporosis.

Fluoride is one of the few agents known to cause an increase in bone formation. Fluoride therapy induces a dramatic and progressive increase in bone mass in -70 % of osteoporotic patients. Such effect on the bone mass is not produced with any other therapy yet known. However, the beneficial effects of fluoride on the bone mass were observed at concentrations close to ineffective and to toxic levels (5 - 10 μM serum concentration), which means that therapeutic window for fluoride is very narrow and hard to achieve. Over¬ dosed fluoride causes several side effects in « 25% ofthe patients (synovitis, anaemia, gastric irritation, pain syndrome in lower extremities), as well as poor quality of newly formed bone.

Bone formation is a multi-stage process, involving commitment of mesenchymal stem cells, proliferation of pre-osteoblasts and differentiation into osteoblasts which form the mineralised extracellular bone matrix. Fluoride apparently acts to increase bone formation by stimulating pre-osteoblast proliferation, but its mechanism of action thereon is not known. It is clear, however, that the activity of fluoride on bone formation is preferential, in that it appears to selectively activate bone formation since when applied at the right concentrations it is non-toxic and affects bone formation but not other tissues.

Heterotrimeric guanine nucleotide binding proteins (G proteins) constitute a family of signal- transducing proteins within the large superfamily of GTP-binding proteins. They consist of a heterotrimer of α, β, and 7 subunits, of which the α subunit binds guanine nucleotides and possesses a general GTPase activity. They function as general transducers of information across the cell membrane, coupling diverse receptors to effector molecules and pathways. The receptors are in turn regulated by extracellular signals, such as hormones, odorants, neurotransmitters and the like. Thus, G proteins form an essential link between many extracellular signalling molecules and the intracellular signalling pathways which they activate. The activity of G proteins on individual effectors can vary between G protein types. This difference is shown clearly by the effects of G proteins on adenylate cyclase (AC). Gs proteins, so named because they stimulate AC, induce an increase in cAMP levels. Gi proteins, on the other hand, decrease cAMP levels by inhibiting AC activity. The inhibitory effect of Gi proteins on AC is known to be inhibited by the toxin of Bordetella pertussis, PTX, which ADP-ribosylates most Gi proteins and thereby inactivates them.

The activation of G proteins essentially proceeds as a cycle. In the normal, inactive state, the α subunit is tightly bound to GDP and associated as a heterotrimer with the βγ complex. Interaction with the activated, intracellular portion of a receptor protein causes release of GDP and binding of GTP. The GTP-bound α subunit is then released from the receptor and the βγ complex, and interacts with and regulates the effector protein and/or pathway. In some cases, the βγ complex may also interact with the effector. The intrinsic GTPase activity of the α subunit eventually hydrolyses the GTP to GDP and the GDP-bound α subunit disassociates from the effector and returns to associate with the βγ complex.

One of the anomalous responses noted in G proteins is that fluoride ions, in the form of aluminium or beryllium tetrafluoride, can cause general and persistent G protein activation. It appears that the complex binds the α subunit adjacent to GDP and mimics the γ- phosphate group of GTP, leading to binding of the α subunit-fluoride complex to the effector and activation thereof.

As noted above, the mode of action of fluoride on bone formation is unknown. We now set forth that fluoride activates G proteins in osteoblasts, which is followed by osteoblast activation and increased bone formation. Moreover, because of as yet not defined reasons, fluoride-induced G protein activation in other tissues has no (or little) physiological consequences in comparison to the effects induced in bone. We demonstrate that fluoride induces G protein-mediated proliferation in osteoblastic cells, but not in fibroblastic cells. We provide evidence that fluoride activates Gi and Gs proteins in osteoblasts in a dose- dependent sequential manner. This information can be used to establish a screening method according to known principles to identify mimics of fluoride action in osteoblasts. Summary of the Invention

According to a first aspect of the invention, therefore, we provide a method for screening candidate compounds which act to increase bone formation comprising assessing the activity of the compounds as G protein stimulators.

Detailed Description of the Invention

Although fluoride is a potent stimulator of bone formation, it remains of questionable use as a therapeutic principle in view of its narrow therapeutic window, toxicity and its tendency to induce poor quality bone formation. The provision of a fluoride mimic which would possess the stimulatory properties of fluoride but lack at least one of the disadvantages associated therewith is therefore desirable. However, it has hitherto not been possible to search for such agents systematically, as the mode of action of fluoride in stimulating bone formation has not been known.

The stimulation of G proteins by fluoride is a general response which affects a multitude of pathways apparently indiscriminately in vitro. Therefore G protein stimulation has not so far been indicated as a candidate mechanism involved in the selective response of osteoblasts to fluoride ions.

By determining the mode of action of fluoride on osteoblasts surprisingly involves G protein activation, we hereby provide a method by which such fluoride mimics may be assayed for. Thus, the invention concerns a method for determining whether a compound is a potential fluoride mimic and potentiator of bone formation comprising the steps of:

(a) exposing a biological system comprising a G protein signal transduction pathway to the compound; and

(b) monitoring the extent of G protein activation in the system in response to the compound.

Preferably, the biological system is a cell or cell line which possesses a G protein signal transduction pathway. Most cell Iines will fulfil this requirement. Advantageously, the cell line is of osteoblastic origin. A cell or cell line may be chosen which enables a third parameter - mitogenic response to the fluoride mimic - to be monitored. For example, suitable cell Iines may be transformed or non-transformed and derived for example from new-born rodent calvaria. An example of such a cell line is the MC3T3-E1 cell line, a non- transformed line (Sudo, et al., (1983) J. Cell Biol. 96, 191-198). Alternatively, primary cells derived from new-born rodent, such as mouse, calvaria may be used. Preferably, the cell or cell line may be induced to overexpress a specific G-protein.

In an alternative embodiment, the biological system can be an in vitro system which is capable of monitoring G protein activation. For example, a recombinant His-tagged or GST- fused Gα subunit can be expressed in baculovirus system, purified, and assessed for activation by compounds. The read-out of the activation can be: (a) a fluorescence change induced by the change in protein conformation; or (b) competition between binding of radioactively labelled GTPγS and compounds to Gα subunit in the presence or absence of GDP. As reference compounds, GDP. AIF4^" and GTPγS can be used.

G protein activation is advantageously monitored by using G protein specific reagents to assess the effect of the fluoride mimic. For example, PTX, which has a specific inhibitory effect on Gi proteins, is found to block fluoride-induced DNA synthesis. Thus, agents which stimulated DNA synthesis but are blocked by PTX are candidate Gi protein stimulators.

Alternatively, it is known that cAMP levels are regulated by G proteins via their action on AC. AC activity is differentially modulated by G proteins, in that Gi proteins inhibit AC action and Gs proteins stimulate it. Fluoride inhibits direct forskolin-induced AC activation and the inhibitory effect can be reversed, for example by PTX, implying an involvement of Gi proteins in the mechanism of action of fluoride. Therefore, agents which inhibit forskolin- induced increases in cAMP levels in a PTX-dependent manner are candidate Gi protein activators.

As set out above, it has been found that, in contrast to their general non-specific activator properties in vitro, in osteoblastic cells G proteins show a selective activation of Gi over Gs proteins. Preferably, therefore, the method of the invention involves monitoring the extent of Gi protein activation in response to candidate fluoride mimics.

Monitoring of G protein activation may also be carried out by assessing the state of activity of effectors on pathways known to be activated by G proteins. Thus, for example, it is known that MAP kinase (MAPK) and S6 kinase (p70^S6K) are present on two separate signalling pathways which are at least in part regulated via G protein activation. Both MAPK and p70^S6K are activated by phosphorylation at key sites.

Fluoride induces general tyrosine phosphorylation and phosphorylation of both MAPK and p70^S6K PTX induces a partial reduction in specific MAPK and p70^S6K activation in response to fluoride, but does not affect general tyrosine phosphorylation. Thus, Gi proteins appear to act specifically to increase activity of both MAPK and p70^S6K in response to fluoride stimulation, but do not act to induce general tyrosine phosphorylation.

Accordingly, the invention preferably comprises monitoring G protein activation by assessing the activity or phosphorylation state of MAPK or p70^S6K. The activation state of p70^S6K may for example be monitored by assessing the level of phosphorylation of key residues, such as T₃₈₉, S₄₁₁, S₄ι₈, T₄₂₁ and S₄₂4 (see Pearson et al., (1995) EMBO J., 14, 5279-5287).

Using the methods provided by the invention, fluoride mimics useful for treatment of osteoporosis and other disorders of bone formation may be identified. Accordingly, the invention includes a fluoride mimic which possess the stimulatory properties of fluoride but lacks at least one of the disadvantages associated therewith, such as toxicity or the tendency to give rise to poor quality bone formation. For example, gastrointestinal complications of fluoride are probably due to the formation of corrosive hydrofluoric acid in the gastric lumen. Therefore, avoiding the use of fluoride based compounds is expected to lead to disappearance of this side effect. Further, poor quality bone that is formed after overdose of fluoride is at least partly due to the incorporation of fluoride to the bone mineral. Therefore, using non-fluoride-based compounds may help to avoid this undesirable effect. Finally, our data indicate that fluoride can sequentially activate Gi and at higher doses Gs proteins. It is likely that in vivo fluoride could activate one class of G proteins to produce bone-forming effects and another class of G proteins to produce side and toxic effects. Therefore, a class-specific G protein activator could produce preferentially bone-forming over side effects.

The invention is further described, for the purposes of illustration only, in the following examples. Example 1

Mitogenic response of osteoblastic MC3T3-E1 cells to activation bv Fluoride

(a) The ability of confluent, growth factor-deprived osteoblastic MC3T3-E1 and fibroblastic

NIH3T3 cells to respond to mitogenic stimuli by entering the S phase of the cell cycle and synthesising DNA is compared. We have chosen FCS as a general stimulator of cell growth,

PDGF as a defined growth factor that activated tyrosine phosphorylation signalling pathways, PMA, as a direct activator of the PKC pathway, forskolin and 8-Br-cAMP as direct activators of the PKA pathway, and fluoride as an agent whose action we want to determine. (Note that in our studies the name fluoride means NaF/AICI 3) The PKC and PKA pathways have been extensively studied in osteoblastic cells, since they are activated by

PTH, a hormone with known action on bone formation and resorption.

The cells (NIH3T3 and MC3T3-E1) are grown for 3 days to reach confluency, deprived of serum for 1 day, stimulated for 48h and labelled with ³H-thymidine for the last 24 h of stimulation. The results are quantified by liquid scintillation counting and expressed as a fold stimulation over non-stimulated control. The concentrations of stimulating agents are 30 ng/ml of PDGF, 1 mM/10 μM NaF/AICI₃, 25 μM forskolin, 500 nM PMA and 1 mM 8-Br- cAMP.

As previously established, NIH3T3 cells respond very strongly to FCS stimulation ( -60-fold increase over control), less to the growth factor PDGF and very weakly to the PKC stimulator PMA. Stimulators of the PKA pathway, forskolin and 8-Br-cAMP, inhibit the basal levels of DNA synthesis. In contrast, MC3T3-E1 cells respond much more weakly to mitogenic stimulation: maximal stimulation is achieved with PDGF (-4-fold), while FCS and PMA produce lower stimulation and forskolin induced an inhibition of DNA synthesis. We conclude that: a) osteoblastic MC3T3-E1 cells have a smaller capacity for mitogenic response than NIH3T3 fibroblasts; b) the mixture of growth factors present in FCS leads to synergistic effects in fibroblastic NIH3T3, but not in osteoblastic MC3T3-E1 cells; c) both cell Iines respond weakly to direct stimulation of PKC and are inhibited by direct stimulation of the PKA pathway. (b) Since confluent MC3T3-E1 cells in culture are progressively losing their proliferating capacity and starting to differentiate (Quarles, et al. (1992) J. Bone Min. Res. 7: 683-692), subconfluent MC3T3-E1 cells are investigated for their response to growth factors.

MC3T3-E1 cells are plated on tissue culture wells in α-MEM with 10% FCS and incubated for 2 - 3 h in order to adhere to the plastic. Subsequently, the medium is exchanged to α- MEM without serum and stimulating agents FCS, PDGF, IGF-I and NaF/AICl3 are added. The cells are incubated for the indicated times, labelled for the last 24 h with ³H-thymidine, extracted and the radioactivity quantified by liquid scintillation counting. The results are expressed as a fold increase over non-stimulated control and represent a mean of at least 3 separate experiments.

In subconfluent MC3T3-E1 cells, FCS is the strongest stimulator of DNA synthesis ( -5-fold), followed by PDGF, IGF-I and fluoride. The response to 1 mM NaF/10 μM AICI₃ (1.5-fold) is time dependent and is consistently observed at day 2 after plating. At 5 mM NaF/10 μM AICI_j, however, fluoride was toxic to the cells in this long-term treatment. The stimulation of DNA synthesis in MC3T3-E1 cells is rather small, which is in agreement with the data from other osteoblastic cells (Farley, et al. (1983) Science 222: 330-332; Reed, et al. (1993) J. Bone Min. Res. 8: 19-25). In order to check the reproducibility of this effect, the experiment is extensively repeated: in all cases we observe fluoride-induced stimulation of DNA synthesis that is significantly above the basal levels (Table 1). We have also found that fluoride can reproducibly stimulate confluent MC3T3-E1 cells in the presence of 1% FCS.

Table 1.

Reproducibility of the fluoride effect on DNA synthesis in subconfluent osteoblastic MC3T3-

E1 cells.

Experiment Fold Induction

Number mean S.D.

1 1.7 0.02

2 1.6 0.34

3 1.3 0.07

1.3 0.02

4 1.6 0.04

1.6 0.07

5 1.8 0.19

1.9 0.02

6 1.5 0.02

1.5 0.07

1.9 0.05

1.9 0.30

n= 12 mean: 1.63 S.D.: 0.21

DNA synthesis is measured by H-thymidine incorporation in the subconfluent cells. The cells are deprived from serum for 1 day and subsequently stimulated with the mixture of 1 mM NaF and 10 μM AICI3. The results are quantified by liquid scintillation counting and expressed as fold increase over non-stimulated cells. (c) In order to establish that the observed effect is due to formation of phosphate-group¬ like AIF4^" complex, the dose responses to both NaF and AICI 3 are tested in the constant presence of one of the agents. DNA synthesis measurements in day 2 subconfluent or confluent MC3T3-E1 cells are performed as described above. The concentration of either NaF in the presence of constant AICI3 or the concentration of AICI3 in the presence of constant NaF is varied. A dose-dependent response is observed in both subconfluent and confluent MC3T3-E1 cells in both cases, indicating that the formation of the AIF 4^" complex is responsible for the observed biological effect.

(d) The capacity of fluoride to increase the proliferation of MC3T3-E1 cells, an effect that usually follows stimulation of DNA synthesis, is also assessed. Sub-confluent cells are plated as described above, deprived of serum and stimulated. After indicated times, the cells are detached by 0.001 % pronase treatment and counted.

The results indicate that fluoride increases cell number at day 1 after plating. This effect is lost at day 2 and is insignificant at day 3, probably due to proliferation of control cells in the medium alone and insufficient amounts of the stimulating agent (the same loss of response is observed with fluoride, PDGF and IGF-I, but not with FCS). Together these data indicate that fluoride is mitogenic to osteoblastic MC3T3-E1 cells.

Example 2

Monitoring G-protein activation bv Fluoride

(a) Pertussis toxin (PTX) is a bacterial toxin that ADP-ribosylates most α subunits of the Gi class and blocks their function. PTX is used to examine whether fluoride-induced mitogenesis in osteoblasts is mediated by Gi proteins. As shown in Table 2, PTX completely inhibits fluoride-induced DNA synthesis in subconfluent cultures. PTX has a partial effect in DNA synthesis induced by F + 1% FCS, small effect on 10% FCS and no effect on PDGF- induced responses. It is not possible to perform similar experiments in subconfluent NIH3T3 fibroblasts, since PTX is toxic for them. The response to fluoride in these experiments varies, probably because of the additional 4 h treatment necessary for the PTX treatment. These results suggest that mitogenic response to fluoride in MC3T3-E1 cells is mediated by Gi proteins. Table 2.

Effect of pertussis toxin on DNA synthesis in subconfluent osteoblastic MC3T3-E1 and fibroblastic NIH3T3 cells.

Fold Induction

Stimulus MC3T3-E1 NIH3T3

mean±S.D . % inhibition with PTX

F 1 .6 ± 0.2 0.5 ± 0.2

1 % FCS 1 .5 ± 0.5 5.3 ± 0.8

1 % FCS + F 2.7 ± 0.4 6.5 ± 1 .4

10% FCS 4.4 + 1.2 42.7 ± 8.6

PDGF 3.2 ± 0.2 19.2 ± 7.3

PTX 1.0 ± 0.1

PTX + F 0.8 ± 0.1 133

PTX + F + 1 % FCS 2.0 ± 0.6 41

PTX + 10% FCS 3.7 ± 0.7 20

PTX + PDGF 3.1 ± 0.1 4

The cells are plated and incubated for 2.5 h until they adhered to the plastic. Subsequently, the cells are deprived from serum and the stimuli or inhibitors are added. After 2 days of stimulation and 1 day of metabolic labelling, ³H-thymidine incorporation is measured. Pertussis toxin (PTX) is added at 0.1 μg/ml for 4 h. The results are means from at least two independent experiments done in duplicates.

(b) The effect of fluoride on cAMP levels in MC3T3-E1 cells and primary calvarial osteoblasts is studied. Adenylate cyclase (AC), the enzyme that produces cAMP in the cell is regulated with two types of G proteins: Gs, that leads to stimulation of AC and increase in cAMP, and Gi, that leads to inhibition of AC and decrease in cAMP. Fluoride does not induce increase in cAMP levels (Tables 3 and 4), suggesting that it cannot activate Gs in the cells. However, fluoride can induce a decrease in cAMP levels induced by forskolin, a direct activator of AC. The fluoride-induced decrease is partially blocked by PTX, suggesting the involvement of Gi protein (Tables 3 and 4). These results indicate that, in the osteoblastic cells, fluoride is not a non-specific activator of G proteins as it is in vitro. In MC3T3-E1 cells and primary calvarial osteoblasts, therefore, we show the selective activation of Gi over Gs proteins by fluoride.

Table 3.

Effect of fluoride on forskolin-induced cAMP levels in MC3T3-E1 osteoblastic cells.

Treatment cAMP (pmol/ml) Fold % of mean ± S.P. forskolin

Basal 14 ± 4 n=8 1

Fluoride 7 ± 1 n=4 0.5

PTX 29 ± 36 n=4 2.0

Forskolin 829 ± 214 n=12 59.0 100%

Fluoride +

Forskolin 97 ± 80 n=12 13.8 12%

PTX +

Fluoride +

Forskolin 51 1 ± 174 n=5 36.5 62% Table 4.

Effect of fluoride on forskolin-induced cAMP levels in primary calvarial osteoblasts.

Treatment cAMP (pmol/ml) Fold % of mean ± S.P. forskolin

Basal 21 ± 6 n=2

Fluoride 23 ± 6 n=2

PTX 35 ± 8 n=2 1 .7

Forskolin 2437 ± 442 n=2 1 16.0 100%

Fluoride +

Forskolin 944 ± 362 n=2 44.9 39%

PTX +

Fluoride + Forskolin 2425 ± 70 n=2 1 15.5 100%

For the experiments reported in Tables 3 and 4, the cells are treated with PTX (0.1 μg/ml) for 4 h and with forskolin (5 μM) and fluoride (10 mM/10 μM NaF/AIC^) for 15 min. The cells are extracted with ethanol and cAMP is measured in extracts with a RIA kit (Amersham) following manufacturer's instructions. (c) We determine a dose-response for cAMP levels to fluoride in osteoblastic MC3T3-E1 cells and primary osteoblasts (Table 5). In agreement with data shown in Tables 3 and 4, fluoride induced a dose-dependent decrease in cAMP between 0.5 mM and 10 mM in MC3T3-E1 cells (IC₅₀=5 mM) and between 0.1 and 10 mM in primary osteoblasts (IC _S0=0.3 mM). However, surprisingly, this effect was reversed at high fluoride doses of 50 mM (Table 5). The cAMP levels not only returned to, but also increased over, the initial value. This result suggests activation of Gs protein with 50 mM fluoride that leads to an increase in cAMP over forskolin-induced levels. We conclude that fluoride sequentially activates Gi and Gs proteins in a dose-dependent manner.

(d) It is reported that fluoride can induce tyrosine phosphorylation and MAPK activation in MC3T3-E1 cells (J. Caverzasio et al., XXIVth European Symposium on Calcified Tissues, Aarhus, Denmark, 1995). The effect on fluoride on these two responses, as well as on phosphorylation of p70^S6K, that lies on a different signalling pathway than MAPK, is tested. We detect activation of tyrosine phosphorylation, MAPK and p70 ^sβκ phosphorylation (summary in Table 6). Fluoride has partly different effects in NIH3T3 cells: the pattern of tyrosine phosphorylation is different and the extent of MAPK phosphorylation is smaller. The phosphorylation of MAPK and p70^sβκ are partially sensitive to pertussis toxin, while total tyrosine phosphorylation is not. In agreement with the data in other cell types, the phosphorylation of p70^S6K is sensitive to wortmannin and rapamycin, the inhibitors acting upstream of p70^S6K. The tyrosine kinase inhibitor genistein inhibits only p70 ^S6K, but not MAPK. These results suggest involvement of Gi protein in activation of MAPK and p70 ^sβκ, but not in the general increase in tyrosine phosphorylation.

Table 5.

NaF ι cone. cAMP

(+ 10 _μM AICI_a1 (% of control)

MC3T3-E1 primary OB

0.01 n.d. 104

0.1 104 88

0.5 100 n.d.

1 82 8

5 52 n.d.

10 25 27

50 215 134

For Materials and Methods, see Table 3. OB - osteoblast; n.d. - not determined

Table 6.

Summary of the results showing fluoride-induced total tyrosine phosphorylation, MAPK and p7(f^eK phosphorylation in osteoblastic MC3T3-E1 cells and the effect of inhibitors on those responses.

A Time course

TyrP MAP K p70 S6 K starts at 5 min. 5 min. 10 min. maximal 10 min. - 1 h 10 min. - 1h 30 min. -1h

B Effect of pertussis toxin

TyrP MAP K p70 S6 K no partial inhibition partial inhibition

C Effect of wortmannin

TyrP MAP K p70 S6 K no no inhibition

D Effect of rapamycin

TyrP MAP K p70 S6 K no no inhibition

E Effect of αenistein*

TyrP MAP K p70 S6 K no no partial inhibition

The MC3T3-E1 cells are stimulated with fluoride for various times (time course) or for 10 min (experiments with inhibitors). Tyrosine phosphorylation is detected after Western blotting with 4G10 anti-phosphotyrosine antibodies. The phosphorylation of MAPK and p70^{S K} is estimated from "kinase band shifts" after Western blotting with anti-ERK (Santa Cruz Biotechnology) and anti-p70^S6K M1 antibodies (G. Thomas, FMI). The concentrations used are: 10 mM/10μM NaF/AIC , 0.1 μg/ml PTX, 100 nM wortmannin, 10 ng/ml rapamycin and 50 μM genistein.

Claims

1. A method for screening candidate compounds which act to increase bone mass comprising assessing the activity of the compounds as G protein stimulators.

2. A method according to claim 1 wherein the candidate compounds are fluoride mimics.

3. A method according to claim 1 or claim 2 comprising the steps of: a) exposing a biological system comprising a G protein signal transduction pathway to the compound; and

4. A method according to claim 3 wherein the extent of G protein activation is monitored by assessing the susceptibility of the response of the system to PTX.

5. A method according to claim 3 wherein the extent of G protein activation is monitored by assessing the modulation of cAMP levels in the system in response to the compound.

6. A method according to any preceding claim which is carried out in osteoblastic cells.

7. A method according to any preceding ciaim which is carried out in vitro.

8. A method according to claim 6 further comprising assessing the mitogenic response of osteoblastic cells to the compound.