Tartronates: A New Generation of Drugs Affecting Bone Metabolism

Abstract

In the search for a new class of bone-sparing agents for treating osteopenic disorders, we hypothesized that tartronic acid derivatives, sharing the chemical characteristics both of bisphosphonates and of Gla residues contained in matrix proteins such as osteocalcin, could positively affect bone metabolism. A series of tartronates was therefore tested for their ability to affect bone metabolism. In vitro resorption tests were performed examining pit formation by freshly isolated rat and rabbit osteoclasts plated onto bone slices and exposed to the drugs for 48 h. Tartronates bearing a linear side-chain (DF 1222 and DF 1363A) were the most effective in inhibiting pit excavation in the pM–nM range. Tartronates did not affect osteoclast viability, number, adhesion, or tartrate resistant acid phosphatase activity. Transient cell retraction was observed in osteoclasts plated onto glass and exposed to DF 1222. The maximal effect was seen in cells treated for 4 h at a concentration of 1 pM. DF 1222 accelerated mineralization in cultures of periosteal cells without affecting other osteoblast-like functions. This product was therefore tested in vivo in ovariectomized mice. Bone mass in femur was evaluated, by ash gravimetry, 21 days after ovariectomy. Unfortunately, DF 1222, the most active of tartronates in vitro, was inactive in this test because of its high hydrophilicity and the subsequent too short residence time. On the contrary, its tetrahydropyranyl ether derivative, DF 1363A, endowed with a significantly higher lipophilicity, showed a dose-dependent bone-sparing effect when administered subcutaneously at 10, 30, and 100 mg/kg/die, thus confirming the activity seen in in vitro tests. Because of their feasible parallel effect on both bone resorption and formation, tartronate derivatives may be tested to candidate this class of products for clinical studies.

INTRODUCTION

Osteopenic disorders, such as osteoporosis, Paget's disease, and hypercalcemia of malignancies, are extremely diffused, affecting a large number of individuals worldwide. The most extended disease is represented by osteoporosis, a typical elderly disorder, characterized by a reduced bone mass which is likely to be determined by unbalanced resorption/formation cycles. Among the individuals that may develop osteoporosis, women are considered at high risk, due to the estrogen deficiency that abruptly occurs at the menopause. Osteopenic disorders are treated with a restricted number of drugs, including estrogens, calcitonin, sodium fluoride, calcium supplementation, and bisphosphonates. The latter are carbon-substituted pyrophosphate analogs with potent bone resorption–inhibiting effects.⁽¹⁾ Even if the mechanism of action of bisphosphonates remains largely unclear, the use of alendronate for the treatment of osteoporosis was recently approved by the Food and Drug Administration (FDA). However, bisphosphonates are characterized by a very poor oral bioavailability—1% or less⁽²⁾ —that imposes to administer oral doses 100-fold higher than that theoretically active, doses that can produce epigastric complaints and irritation of the gastric mucosa in up to 50% of patients.⁽³⁾ Moreover, nitrogen-containing bisphosphonates can induce an acute phase response associated with flu-like symptoms (e.g., fever, malaise, and myalgias), and with increased levels of circulating IL-6.⁽⁴⁾ Besides alendronate, calcitonin and estrogens are the only drugs approved by the FDA for treatment of postmenopausal osteoporosis. The response to calcitonin is adequate in patients with increased bone turnover, but not in patients with low bone turnover.⁽⁵⁾ Treatment with estrogens may induce endometrial hyperplasia, breast tenderness, and vaginal bleeding, while its relationship with breast cancer has not yet been fully established.⁽⁶⁾ Calcium and fluoride supplements are not subjected to FDA regulation. The effect of calcium supplementation on bone mass in established osteoporotic syndromes is not well studied, whereas sodium fluoride is associated with a significant degree of gastrointestinal distress and painful lower extremity syndrome and, most importantly, with a fracture incidence even increased when compared with the placebo group.⁽⁷⁾ Taken together, these data indicate that the use of these drugs is limited by specific problems, concerning either the real efficacy, or the incidence of unwanted side effects.

Figure FIG. 1. Schematic representation of tartronates.

In the search of a new class of bone-sparing agents, we hypothesized that tartronic acid derivatives showing chemical characteristics (see Fig. 1) shared by bisphosphonates and of γ-carboxyglutamic (Gla) residues contained in the matrix proteins, such as osteocalcin (bone Gla protein, BGP),⁽⁸⁾ could positively affect bone metabolism. This hypothesis was supported by three main considerations:

(1) γ-carboxylated proteins, and in particular BGP, play a complex role in bone metabolism. A substantial body of evidence indicates that there is a negative correlation between the degree of γ-carboxylation of matrix proteins and bone fragility.^(9–12) However, more recent findings⁽¹³⁾ in transgenic mice seem to contradict this hypothesis. In fact, BGP-deficient mice have higher bone mass than control littermates and, interestingly, an even exaggerated bone resorption following ovariectomy.

(2) Malonates are a chemical class, comprising also Gla, which could affect cell metabolism. Malonate is a well-known inhibitor of succinic dehydrogenase, an enzyme in the Krebs cycle. Tartronic acid (i.e., hydroxymalonic acid) inhibits NADP-linked malic enzyme, an anaplerotic reaction of the Krebs cycle.⁽¹⁴⁾ The effects on respiratory enzymes could result in opposing effects in osteoblasts and osteoclasts.

(3) Bisphosphonates share a geminal diacidic group with malonates. This similarity has been used for designing bone-targeted drugs.⁽¹⁵⁾ However, it is not known whether bisphosphonates could mimic the action of Gla on bone cells, for instance by interacting with a hypothetical specific receptor.⁽¹³⁾

To test our hypothesis, a series of tartronic acid derivatives (tartronates) was investigated for their ability to affect bone metabolism. We observed that these compounds are endowed with a relevant bone resorption inhibitory activity and with a somehow parallel positive effect on in vitro mineralization. These molecules did not show remarkable effects on osteoclast viability, number, adhesion to glass or bone, and tartrate resistant acid phosphatase activity, thus indicating limited cytotoxic effects or cell metabolic injury.

MATERIALS AND METHODS

Materials

Dulbecco's modified minimum essential medium (DMEM), fatty acid-free bovine serum albumin (BSA), and the histochemical kits for tartrate-resistant acid phosphatase (TRAP) and alkaline phosphatase (ALP) activities were from Sigma Chemical Co. (St. Louis, MO, U.S.A.). The cresolphthalein complexone kit for calcium detection was from Boehringer Mannheim (Milan, Italy). Fetal bovine serum (FBS) was from Hyclone Laboratories (Logan, UT, U.S.A.). Penicillin, streptomycin, amphotericin B and mycostatin were from Eurobio (Paris, France). Culture dishes and sterile glassware were from Falcon Becton-Dickinson (Lincoln Park, NJ, U.S.A.) and from Costar Co. (Cambridge, MA, U.S.A.). Salmon calcitonin (sCT) was from Sandoz Pharma AG (Basel, Switzerland), 1α,25-dihydroxyvitamin D₃ (1,25(OH₂)D₃) was a kind gift of Dr. Domenico Criscuolo (Roche, Milan, Italy). All other reagents were of purest grade from Carlo Erba (Milan, Italy), from Merck (Damstadt, Germany), and from Aldrich Chimica (Milan, Italy).

Tartronates were synthesized at Dompé Research Center according to a procedure described earlier.⁽¹⁶⁾ Precoated silica gel plates (silica gels 60 F₂₅₄; Merck) with fluorescent indicator were used for thin layer chromatography (TLC). Silica gel 60 (70–230 mesh) was used for all column chromatography separations.

Cell isolation and culture

Osteoclasts:

Osteoclasts were isolated from newborn New Zealand rabbits or Wistar rats (Charles River, Calco, Italy) (4th–7th day) by a modification of the method described by Chambers et al.⁽¹⁷⁾ Briefly, animals were sacrificed by cervical dislocation, then long bones were removed, dissected free of soft tissues, placed in DMEM, and cut in small fragments. Osteoclasts were released by gently pipetting the fragments. The bone fragments were allowed to sediment for 30 s, then the cell suspension was transferred onto appropriate substrates (bone slices, Petri dishes, or glass coverslips) and incubated in DMEM, supplemented with 10% FBS, 100 IU/ml penicillin, 100 μg/ml streptomycin, 50 μg/ml amphotericin B, and 50 μg/ml mycostatin at 37°C in a humidified atmosphere of 95% air and 5% CO₂, for 90 minutes. Finally, cultures were vigorously rinsed to remove nonadhering cells, before adding fresh medium and before use for the experiments.

Osteoclasts were recognized by phase contrast microscopy as multinucleated cells, with variable shape and extensive lamellipodia. Nuclei and organelles were located in the central granular zone, while the peripheral area was homogeneous, irregularly undulated, and devoid of appreciable organelles. These cells were positive for the osteoclast marker TRAP, evaluated with Sigma histochemical kit number 181 -A, and retracted in response to 100 nM sCT.

Osteoblast-like cells:

Periosteal cells were obtained from the endocranial periosteum of newborn rat calvaria and characterized as described elsewhere.⁽¹⁸⁾ Briefly, the periosteum was removed and digested with 1 mg/ml neutral collagenase in 199 medium for 30 minutes 37°C. Released cells were extensively washed (three times), plated into 10-cm diameter Petri dishes and incubated in 199 medium supplemented with 20% FBS. Cells were grown until confluent, then trypsinized and transferred to appropriate dishes for characterization and experiments. These cells expressed alkaline phosphatase activity, and showed the ability to produce an extracellular matrix that underwent mineralization in the presence of 10 mM β-glycerophosphate, 50 μg/ml ascorbic acid, and dexamethasone.

Effect of tartronates on bone resorption in vitro

In vitro bone resorption was evaluated in cell cultures by a modification of the method described by Prallet et al.,⁽¹⁹⁾ which was rapid and required a light microscope without additional equipment. Transverse bone slices (4 × 4 mm) were cut from adult bovine cortical bone using a Buhler Isomet 2000 precision saw. Slices were cleaned by sonication (3 × 10 minutes in distilled water), briefly rinsed in acetone, and sterilized by overnight exposure to ultraviolet light. Osteoclasts were plated onto the bone slices and incubated for 48 h in the continuous presence of tartronates. At the end of the incubation, adherent cells were counted under light microscopy, then removed by sonication for 2 minutes in 0.01 N NaOCl. Bone slices were rinsed twice in distilled water (20 minutes) and stained for 4 minutes in 1% toluidine blue in 1% sodium borate, and observed by conventional light microscopy with a 16× objective. The resorption pits were divided in three visual categories, according to their diameter: category A, <10 μm; category B, 10–30 μm; category C, >30 μm. The diameter was measured under the microscope with the aid of a graduated eyepiece. The number of pits of each category in individual bone slice were scored by multiplying by a different factor according to their dimensions: for category A, 0.3; for category B, 1; and for category C, 3. The sum of the three scores gave the pit area index. Using this scoring method in six preliminary independent experiments performed in triplicate with rabbit osteoclasts (20 ± 4 cell/bone slice), we classified about 40% of pits in category A, 24% in category B, and 36% in category C, giving a pit area index of 32 ± 6 (mean ± SE). In a similar way, using rat osteoclasts (20 ± 1 cell/bone slice), we found 30% type A pits, 50 type B pits, and 20 type C pits, with a pit area index of 28 ± 5 (mean ± SE of six independent experiments, performed in triplicate). For each experiment, the mean control value of the pit area index was used as a base to calculate each value as a percentage of the mean control value. This allowed pooling of all the experiments for each condition tested. In some experiments, bone resorption was stimulated by the addition of 10 nM 1,25(OH₂)D₃.

Effect of tartronates on osteoclast morphology and TRAP activity

To evaluate osteoclast morphology, cells were fixed in 70% ethanol and observed by phase contrast microscopy. TRAP activity was evaluated by the Sigma histochemical kit number 181-A. For quantitative analysis, the number of multinucleated and mononuclear TRAP-positive cells was counted in 10 high-power microscopic fields and expressed as a percentage of the total number of cells present.

Effect of tartronates on rat periosteal cells

Periosteal cells (17th passage) were seeded in 24-well plates at a density of 10,000 cell/well in a final volume of 1 ml of DMEM F-12 supplemented with 15% FBS, 2.5 mM L-glutamine, and penicillin/streptomycin. At confluence (about 7 days), test compounds (tartronates or etidronate) were added in the presence or in the absence of 10⁻⁷ M dexamethasone. Three days later, medicated media were replaced with fresh mineralizing medium (DMEM F-12, supplemented with 10 mM β-glycerophosphate, 50 μg/ml ascorbic acid) containing the test substances. Three, 4, and 5 weeks later, mineralization was detected with Von Kossa staining, and alkaline phosphatase (ALP) histochemical semiquantitative assay was performed by Sigma Phosphatase Leukocyte Kit (procedure number 86) in parallel plates. For hydroxyapatite quantification, wells were incubated overnight with 300 μl of 1 N HCl, at room temperature, then the calcium content was evaluated in clarified supernatants using a cresolphtalein complexone kit (Boehringer Mannheim).

Effect of tartronates on bone metabolism in vivo

Female C3H/HeOUJ mice (Charles River) 5-weeks of age, were housed 10/cage and acclimated for 15 days at 21 ± 1°C room temperature and 55 ± 10% humidity. Food (containing 0.5% calcium and 0.4% phosphorus) and water were supplied ad libitum. The animals were ovariectomized (OVX) or sham-operated by the dorsal approach under ketamine-xylazine anesthesia. OVX mice were treated subcutaneously for 21 days with tartronates (10, 30, 100 mg/kg/day), alendronate (20 μg/kg/day), or vehicle. Twenty-one days after OVX—a time selected on the basis of recent results obtained by Kawaguchi et al. in 8-week-old CD-1 mice⁽²⁰⁾ and confirmed by our preliminary experiments—animals were sacrificed by pentobarbital hyperanesthesia, femurs and tibiae, freed from soft tissues, were ashed (700°C for 16 h in quartz fiber crucibles, CEM) and weighed. Femur ash weights were divided by the length of femurs, measured from the great trochanter to the external condyle, and expressed as milligrams per millimeter as described by Migliaccio et al.⁽²¹⁾

Statistics

Data are presented as mean ± standard error. Statistical differences were calculated by one-way analysis of variance (ANOVA), followed by Student's t-test or by Dunnett's test. A p < 0.05 was conventionally considered statistically significant.

RESULTS

Selection of active tartronates by screening their effect on unstimulated bone resorption

The activity of tartronates was screened at a fixed concentration of 10 nM against unstimulated bone resorption by rabbit osteoclasts to test the consistence of the work hypothesis. We selected the most promising term of the series: DF 1222 (2-hydroxy-2-[(N-tert-butyloxycarbonyl)-3-aminoprop-1-yl]propanedioic acid lithium salt), whose structure is depicted in Fig. 2. This compound, at 10 nM, strongly inhibited basal bone resorption by isolated rabbit osteoclasts (64 ± 11%, n = 7). In our screening conditions, two reference bisphosphonates, etidronate and alendronate (ALN), tested at 10 nM, gave inhibitions of 60 ± 23 and 79 ± 11%, respectively.

To test whether its action could be species specific, DF 1222 was then tested for its activity on rat osteoclasts, in comparison with ALN. The results, presented in Fig. 3B, demonstrate that DF 1222 was active also on this species. In particular, the activity of DF 1222 was statistically significant at 10⁻¹² M (55 ± 10% inhibition); however, a near complete inhibition (84 ± 6%) was achieved only by raising the concentration to 10⁻⁶ M. This shallow dose-response curve was shared also by ALN (Fig. 3A) and is in agreement with the dose-response curves obtained by Sahni et al.⁽²²⁾ using a similar assay protocol on rat osteoclast (i.e., bisphosphonates added concurrently with the cells to the medium).

Figure FIG. 2. Structures of DF 1222 and DF 1363A.

Characterization of DF 1222 activity

Effect on TRAP, adhesion, and morphology of isolated osteoclasts:

To investigate whether DF 1222 altered specific osteoclast functions, we performed a series of experiments to test its effect on TRAP activity, adhesion, and morphology of rabbit osteoclasts.

Histochemical detection of TRAP activity demonstrated that all the osteoclasts sitting on the glass coverslips and some mononuclear, irregularly shaped, macrophage-like cells were TRAP positive. The percentage of adhering TRAP-positive cells was modestly, but not significantly, reduced by 1 μM DF 1222, compared with control, (osteoclasts = 84 ± 33%, TRAP-positive mononuclear cells = 70 ± 10% of control, n = 4).

To address the question of whether DF 1222 could induce cell retraction, osteoclasts were observed by phase contrast microscopy, and the percentage of cells apparently retracted versus the total number of osteoclasts sitting on the glass coverslips was calculated. In control cultures, most osteoclasts were spread on the glass substrate, whereas 35 ± 6% of osteoclasts appeared retracted. DF 1222 stimulated osteoclast retraction with a biphasic effect. Maximal retraction was observed at the concentration of 10⁻¹² M, followed by a decline at higher concentrations (Fig. 4A). This behavior was different from that observed in osteoclasts treated with salmon calcitonin, which induced a dose-dependent osteoclast retraction with maximal effect in the micromolar range (Fig. 4B). The effect of DF 1222 was time dependent, with maximal stimulation observed within 4 hours of treatment (Fig. 4C).

Effect on bone resorption stimulated by 1,25(OH₂)D₃:

The effect of the tartronate DF 1222 was then studied on 1,25(OH₂)D₃-stimulated bone resorption in isolated rabbit osteoclasts. In this system, 10⁻⁸ M 1,25(OH₂)D₃ induced a significant increase of bone resorption, probably due to an indirect effect exerted by contaminating mononuclear cells. This effect is shown in Fig. 5. ALN was able to counteract the effect of 1,25(OH₂)D₃ in a dose-related manner, with an IC₅₀ of about 10 nM (Fig. 5A).

Figure FIG. 3. Effect of tartronates on bone resorption by rat osteoclasts, comparison with alendronate (ALN). Rat osteoclasts were isolated and allowed to attach to bone slices for 90 minutes. Cultures were then incubated for 48 h in the presence of (A) ALN, (B) DF 1222, or (C) DF 1363A at the doses indicated on abscissa. Mean ± SE (n = 7–8). Pit area index of control was (A) 9.2 ± 2.4, (B) 6.5 ± 1.5, and (C) 5.5 ± 1.7. *p < 0.05, **p < 0.01 versus control (Dunnett's t-test).

Similarly, DF 1222 was able to inhibit 1,25(OH₂)D₃-stimulated bone resorption in a dose-related manner. In particular, DF 1222 was active in the nanomolar range (Fig. 5B), with a calculated IC₅₀ of 4.24 ± 0.65 nM.

Effect on osteoblast-like cells:

DF 1222 was tested on periosteal osteoblast-like cells to detect its effect on proliferation and function. This tartronate was unable to affect cell proliferation, evaluated by [³H]thymidine incorporation, or ALP activity, evaluated by histochemical semiquantitative assay, up to a concentration of 10⁻⁴ M (data not shown). According to hydroxyapatite deposition, detected by Von Kossa stain (Fig. 6) and to calcium content determination after HCl solubilization (Table 1), these cells were able to mineralize in early culture (3–4 weeks) only in the presence of 10⁻⁷ M dexamethasone. Spontaneous hydroxyapatite deposition in the absence of dexamethasone was observed after 6 weeks from the addition of mineralization medium. Mineralization in culture was completely abrogated by etidronate at 10⁻⁵ M (data not shown). On the contrary, DF 1222, assayed at 10⁻⁷ and 10⁻⁵ M, was unable to affect dexamethasone-induced mineralization. However, in the absence of dexamethasone, DF 1222 was able to accelerate mineralization in an apparently dose-related manner (Fig. 6, Table 1). In fact, hydroxyapatite deposition was already evident in 3-week cultures treated with 10⁻⁵ M DF 1222 and in 4-week cultures treated with 10⁻⁷ M.

Effects in vivo:

DF 1222 was then tested subcutaneously for its activity in vivo on an OVX mouse model. Disappointingly, a substantial lack of activity was found (data not shown). To explain this in vivo/in vitro discrepancy, we have postulated that, owing to its high hydrophilicity, and the subsequent rapid urinary excretion, the activity of this compound could be prevented by a too short residence time in the body. This was confirmed by a preliminary pharmacokinetic study performed by evaluating urinary excretion of unmodified DF 1222 in three rats. DF 1222 was quantified by a selective high-performance liquid chromatography-mass spectrometric (HPLC-MS) method, with a triple stage quadrupole TSQ 700 instrument (Finnigan MAT) equipped with an Electrospray ion source (API) and operated in selected ion monitoring mode. Urinary excretion data were analyzed using the SIPHAR program (Version 4.0) applying the Powell minimization algorithm. We estimated the half-life of DF 1222 in 49 ± 7 minutes.

Therefore, we synthesized DF 1222 tetrahydropyranylether derivative, DF 1363A (2-(tetrahydropyran-2-yl)oxy-2-[(N-tert-butyloxycarbonyl)-3-aminoprop-1-yl] propanedioic acid lithium salt). This compound showed a significantly higher lipophilicity, as demonstrated by the different behavior on silica thin layer chromatography (TLC) (R_f DF 1222 = 0.3; R_f DF 1363A = 0.75). So we hypothesized that DF 1363A could overcome the problems we encountered for DF 1222.

Characterization of DF 1363A activity

Effect on unstimulated bone resorption:

To test this hypothesis, DF 1363A was tested in vitro against unstimulated bone resorption by rabbit osteoclasts. This compound, screened at 10 nM, showed an activity similar to that of its parent compound DF 1222, tested in the same experiments, giving an inhibition of 90 ± 4% (n = 9). Moreover, like DF 1222, DF 1363A was able to inhibit the bone-resorbing activity of rat osteoclasts. The results are presented in Fig. 3C. Also in this case the curve was too shallow to define a correct dose–response relationship.

Figure FIG. 4. Effect of DF 1222 on osteoclast morphology. Diagrams illustrating the effects of DF 1222 and salmon calcitonin (sCT) on osteoclast retraction. Rabbit osteoclasts were cultured as previously described, then treated with (A) DF 1222 and (B) sCT at the concentrations indicated on abscissa, or (C) with 10⁻¹²M DF 1222 for different time. A biphasic and a dose-dependent increase in the percentage of osteoclasts retracted were observed in cultures treated for 15 min, with DF 1222 and sCT, respectively. The effect of DF 1222 at the maximal active concentration (10⁻¹² M) was time-dependent and reached the plateau at 4 h. Mean ± SE (n = 3). *p < 0.05, **p < 0.01 versus control (Dunnett's t-test); a, p < 0.01 versus control at 48 h (open circle), Student's t-test.

Effect on bone resorption stimulated by (1,25-OH₂D₃):

Like its parent compound, DF 1363A was characterized for its effect on 1,25(OH₂)D₃-stimulated bone resorption in isolated rabbit osteoclasts. DF 1363A was able to inhibit 1,25(OH₂)D₃-stimulated bone resorption in a dose-related manner, being about 40 times less potent than its parent compound DF 1222, with an IC₅₀ of 166 ± 52 nM (Fig. 5C).

Effects in vivo:

DF 1363A was therefore studied in a model of estrogen-dependent osteopenia carried out in mice. The product was administered subcutaneously at three dosages (10, 30, and 100 mg/kg). Femur of OVX mice showed a significant decrease of ash weight normalized per bone length (−6.9 ± 0.7%, n = 6, p < 0.01). ALN given subcutaneously at the dose of 20 μg/kg/day was able to counteract OVX-induced osteopenia (Fig. 7). Similarly, DF 1363A was able to inhibit bone loss in femurs of OVX mice in a dose-related manner (Fig. 7), thus confirming its improved pharmacokinetic properties. The dose-related effect of DF 1363A was also evident in tibiae (data not shown) but, in this district, the loss of bone mass was less pronounced (−4.2 ± 0.7%, n = 6) than in femurs and did not reach statistical significance. Interestingly, however, the mean ash weight of tibiae from mice treated with 30 mg/kg of DF 1363A was even significantly higher than that of sham-operated animals (38.6 ± 0.3 vs. 36.3 ± 0.7 mg, n = 9, p < 0.01).

DISCUSSION

In this study, we have shown that a new class of molecules, belonging to the tartronate family, inhibits bone resorption and stimulate matrix mineralization in vitro. Maximal efficacy was observed in the tartronates presenting flexible side chains, such as DF 1222 and DF 1363A.

Figure FIG. 5. Effect of tartronates on 1,25(OH)₂D₃ stimulated bone resorption. Rabbit osteoclasts were cultured in the absence or in the presence of 10 nM 1,25(OH)₂D₃ (1,25D₃) for 96 h with or without different concentrations of (A) alendronate (ALN), (B) DF 1222, or (C) DF 1363A. The resorptive effect was expressed as mean ± SE. (n = 3–6). Pit area index of control was 4.6 ± 2.9 for (A) and (C), and 29.5 ± 5.2 for (B). *p < 0.05, **p < 0.01 versus 1,25D₃ (Dunnett's t-test); a, p < 0.05; b, p < 0.01 versus control (Student's t-test).

Inhibition of bone resorption by the selected tartronates was analogous to that induced by two reference bisphosphonates, etidronate and alendronate, whose bone resorption inhibiting effect is largely demonstrated by several studies.^(23–25) This finding validated the hypothesis that a molecule sharing the chemical characteristics both of bisphosphonates and of γ-carboxyglutamic (Gla) residues contained in the matrix proteins might affect bone resorption by isolated osteoclasts, since dose-dependent inhibition of in vitro pit excavation was observed. However, the curves of inhibition of tartronates against basal resorption were ill-defined due to a very flat profile. This is in agreement with the data obtained by Sahni et al.⁽²²⁾ by adding bisphosphonates concurrently with osteoclasts. On the contrary, active tartronates showed a clear-cut dose–response curve against bone resorption stimulated with the osteotropic factor 1,25(OH)₂D₃. This evidence could suggest, as in the case of bisphosphonates, a possible contribution of mononuclear cells (possibly osteoblasts or bone marrow stromal cells) in the mechanism of action of tartronates.⁽²⁶⁾

Tartronates were active both on rabbit and rat osteoclasts, indicating no differences between these two species. Furthermore, morphological and enzymatic analyses demonstrated that one of the most in vitro active tartronates, DF 1222, did not apparently affect cell viability. This tartronate modified osteoclast morphology, inducing cell retraction. Retraction was maximal at 10⁻¹² M and was reached within 4 h. However, this study does not provide insights into the cell signals and subsequent molecular events that cause arrest of the resorbing activity.

The increased bone loss that characterizes the pathogenesis of osteopenic syndromes as osteoporosis is due to abnormalities in the bone remodeling cycle, which occurs by resorption of old bone, recruitment of new osteoblasts, and deposition of new matrix that subsequently mineralizes.⁽²⁷⁾ It is generally accepted that with each cycle there is a modest, imperceptible deficit in bone formation, possibly due to progressive impairment of the signaling between bone resorption and bone formation, resulting in inefficient osteoblast recruitment.⁽²⁸⁾ Subsequently, osteopenia is caused not only by increased bone resorption, but also by a resultant substantial decrease of bone formation. From these considerations, a molecule that at the same time inhibits bone resorption and stimulates bone formation is an ideal candidate as an antiosteopenic drug. In this study, we observed that DF 1222 increased the velocity of matrix mineralization by rat periosteal cells in vitro, without affecting cell proliferation and ALP activity, indicating a possible role on the differentiating process rather than on the expansion of the population. The effect of DF 1222 on mineralization is similar to that of malonate, which is structurally related to tartronates (tartronic acid is hydroxymalonic acid), as recently reported by Klein et al.⁽²⁹⁾ These authors demonstrated that malonate is able to stimulate mineralization in rat marrow stromal cells. The mechanism that is the basis of this effect should be the inhibition of succinate dehydrogenase (SDH, complex II)³⁰ in agreement with the hypothesis that cells destined to mineralize rely less on respiration.⁽³¹⁾ However, the inhibition of Krebs enzymes in actively resorbing osteoclast could reduce their activity, as indirectly indicated by the stimulatory or inhibitory effects on osteoblastic SDH exerted by PTH⁽³²⁾ or calcitonin,⁽³³⁾ respectively.

We found the positive effect of DF 1222 on mineralization of particular interest because, to date, there is not convincing evidence of pharmacological molecules capable of increasing the bone mass through an anabolic effect on bone formation. Even the bisphosphonate etidronate, which is widely used in the therapy Paget's disease, completely abolished matrix mineralization in our experimental model, in agreement with several literature reports.^(34,35) Our results with tartronates are similar to those obtained with another bisphosphonate (i.e., alendronate), which, unlike etidronate, showed a favorable effect on mineralization.⁽³⁶⁾ However, our evidence is based on an in vitro model that may not fully reflect the in vivo situation.

Figure FIG. 6. Effect of DF 1222 on matrix calcification in rat periosteal cell cultures. Cells were grown for 3, 4, or 6 weeks in a mineralization medium (containing 10 mM β-glycerophosphate and 50 μg/ml ascorbic acid) in the presence of different factors. Mineralization was detected with von Kossa staining.

Table Table 1. Effect of DF 1222 on Calcium Deposition by Rat Periosteal Osteoblast-like Cells*

The positive results obtained in vitro prompted us to test the most effective tartronates in vivo. To accomplish this, we selected a mouse model in which the bone mass was previously reduced by OVX. The surgical removal of ovaries produces osteopenia due to the lack of the bone mass rescuing effects of estrogens.^(37–39) In our model, bone loss was clearly detectable 15 days after OVX and was highly statistically significant within 3 weeks, confirming previous reports.^(20,40)

Figure FIG. 7. Effect of chronic administration of Alendronate (ALN) and DF 1363A to ovariectomized (OVX) C3H/HeOUJ mice. DF 1363A (10, 30, 100 mg/kg/day) or ALN (20 μg/kg/day) were administered subcutaneously for 20 days. Sham-operated and OVX animals received saline subcutaneously daily. Twenty-one days after OVX, their femurs were ashed and weighed. The values are normalized per femur length and expressed as mean ± SE (mg/mm). a, p < 0.01 versus sham (Student's t-test); b, p < 0.05 versus OVX (Dunnett's test), n = 6–9.

The first tartronate tested in vivo was DF 1222. This was chosen because of its elevated efficacy in vitro. Unfortunately, the results obtained were quite disappointing, since no activity was found in the OVX mice treated with DF 1222. The reason why DF 1222 was unable to elicit any effect in vivo is probably related to its high hydrophilicity that may be responsible for its short half-life (about 50 minutes) demonstrated in pharmacokinetic study. This is suggested by the observation that DF 1363A, a tetrahydropyranyl ether derivative, almost equally active in vitro as DF 1222, but constructed with a side-chain conferring a significantly higher lipophilicity, protects in a dose-dependent manner the bone mass in OVX mice upon subcutaneous daily injections carried out for 20 days. This indicates that at least this component of the tartronate family retains its activity in vivo. Moreover, mice treated with high doses of DF 1363A showed higher bone mass in tibia with respect to the sham-operated animals. Although small (6.3 ± 0.8%), this increase was highly significant (p < 0.01), thus suggesting a potential anabolic effect.

Taken together, the results shown in this study demonstrate that tartronates are a promising new class of molecules that may be tested for their bone mass protective role. Of course, the in vitro approach, although allowing a rapid and reliable screening of these molecules for their bone resorption inhibiting effect and concurrent bone mineralization stimulating action, is not sufficient to establish whether they may be proposed as pharmacological molecules in humans. However, our in vivo results indicate that, at least a specific active tartronate plays a potential role as osteotropic drug, without inducing apparent toxic or side effects. Further studies in animal models are needed to establish whether one or more tartronates may be selected for clinical studies.