1 ,Z&Dihydroxyvitamin D3 Increases Citrate Secretion from Osteosarcoma Cells*

Rat osteosarcoma cells respond to 1,25-dihydroxy- vitamin D3 with a 6- to lo-fold increase in the secretion of citric acid. The time required to attain a half-maxi- mal response is 12 h, a time course which is consistent with the postulated steroidal hormone action of this vitamin D metabolite. The citrate response is achieved by physiological concentrations of 1,25-dihydroxyvi- tamin Ds, with half of the maximal response at a vita- min concentration of 0.03 rig/ml. Both the time course and the dose dependence of the citrate response closely parallel the previously reported stimulation of bone Gla protein synthesis by 1,25-dihydroxyvitamin DS in these cells. of

The discovery of large quantities of citric acid in bone and its relative decrease in rickets raised the question of the relationship between citric acid and calcification (1). Subsequent studies established that both vitamin D and parathyroid hormone increase bone and serum citrate (2)(3)(4)(5)(6)(7)(8)(9)(10). The regulation of citrate secretion suggested by these observations led to the speculation that citrate may, by virtue of its interactions with calcium (11) and hydroxyapatite (12), play a role in the mobilization of calcium from bone in response to vitamin D and parathyroid hormone (1,2,13). Studies with bone in organ culture have shown that citrate is released from bone to media in quantities far greater than the original citrate content of the bone (14) and that the release of citrate from bone is stimulated by PTH' (15,16 To verify the identity of the putative citrate component, effluent corresponding to a retention time of 16.8 min was pooled, dried, and analyzed by liquid partition chromatography as described (27). The putative citrate component was also converted to its trimethylsilyl derivative and anaIyzed by gas chromatography-mass spectroscopy on a Finnigan 4021. An OV17 column (Supelco) was used for gas chromatography, and the mass spectrum was determined by electron impact ionization at 70 eV.
Citrate lyase treatment (26) was used to establish peak identity by selective destruction of citrate in media samples. In a typical experiment 1 ml of media was mixed with 50 p1 of 1 M Tris, pH 8.2,5 pl of 0.04 M ZnS04, 10 p1 of 0.1 M NADH, and 5 p1 of malate dehydrogenase (5 mg/ml in 3.2 M (NH4)zS04). Forty pl of citrate lyase (37.5 mg/ml in 0.15 M Tris-HC1, pH 8.2) was then added, and the reaction was incubated for 1.5 h at 25 "C before BaSO, adsorption and analysis. Control samples were treated in the same manner except 40 pl of 0.15 M Tris, pH 8.2, was substituted for citrate lyase. In some media samples from 1,25(OH)2D3-treated cells, citrate levels were high enough to permit direct determination by the citrate lyase method (26). In these samples, the average level of citrate determined by citrate lyase was in good agreement with the average level determined by the high pressure liquid chromatography method.

RESULTS
Media levels of citric acid were quantified by high pressure liquid chromatography on an organic acid column after adsorption on BaS04. As shown in Fig. 1 (Fig. 1). To further confirm identification, the presumptive citrate fraction from the high pressure liquid chromatography effluent was isolated and subjected to several further tests. Liquid partition chromatography of this component over silica (27) gave a single peak with a retention time identical to citrate. After trimethylsilation, gas-liquid chromatography of presumptive media citrate yielded a sindle component in the expected position of citrate which, on mass spectral analysis, proved to be identical to the trimethylsilyl derivative of citric acid.
One additional component resolved by high pressure liquid chromatography of media was also consistently stimulated by 1,25(OH)2D3 treatment, a component which eluted with a retention time of 14.9 min (Fig. 1). This component coelutes with transaconitate. If it is transaconitate, the level of transaconitate computed from the molar extinction coefficient is 0.5% of the citrate level in media from 1,25(0H)2D3-treated cells. Fig. 2 compares the time course of the increase in citrate and BGP levels in culture media following the addition of 1 ng/ml of 1,25(OH)2D3. The time required to attain a halfmaximal response to 1,25(OH)& is 12 h for citrate compared to 9.5 h for BGP. This time course for the BGP response is essentially identical to that reported earlier (18). Both the accumulation of citrate and of BGP in cell media cease after 24 h (Fig. 2), whether or not cells are treated with 1,25(OH)zD3. This effect has been noted previously for the accumulation of BGP in media (24). If the media is exchanged for fresh media at 24 h, BGP and citrate again accumulate in media and the rate of increase for each substance remains 6to 10-fold elevated in the cells treated continuously with 1,25(OH)zD3. The dependence of citrate secretion on the concentration of 1,25(OH)2D3 was determined after 24-h pretreatment with the vitamin at the test concentration to ensure that the response was maximal. As can be seen in Fig. 3, a 1,25(OH)2D3 concentration of 0.03 ng/ml is required for half-maximal stimulation of citrate secretion. This dose dependence is essentially identical to that previously seen for the stimulation of BGP secretion by these cells (18).
The hormonal specificity of the citrate response was evaluated using vitamin D3 and parathyroid hormone. Vitamin D3 gave the same magnitude of citrate secretion response as did 1,25(0H)*D3, but the concentration required for halfmaximal stimulation, 1 wg/ml, was over IO4 greater than the 1,25(OH)2D3 concentration needed for this level of response. This difference in dose dependence parallels the difference in the affinity of the cytosolic 1,25(OH),D3 receptor for these molecules. Intact bovine PTH and synthetic bovine P T H (1-34) both failed to stimulate citrate accumulation in media after 24 or 48 h of treatment at hormone concentrations of 4 X lo-' and

M.
Several clonal osteosarcoma cell lines were tested for a citrate secretion response to 1,25(OH)2D3 in order to better define the bone cell phenotype responsible for this effect. As may be seen in Table I, the only cell lines which increase citrate secretion in response to 1,25(OH)2D3 are ROS 2/3 and 17/2. These cell lines have a more osteoblastic phenotype than the 25/1 and 25/4 cell lines (28) and are the only cell lines tested which also synthesize BGP (24). The ability to increase citrate secretion in response to 1,25(OH)2D3, therefore, appears to be another marker for the osteoblastic phenotype. Two confluent 60-mm culture plates were pretreated for 24 h with each indicated concentration of 1,25(OH)zD3. The media was then exchanged for fresh media at the same 1,25(oH)*D3 concentration and, 12 h later, the media was removed and analyzed for citrate as described under "Experimental Procedures."

DISCUSSION
The present study demonstrates that physiological 1,25(OH)2D3 concentrations dramatically and specifically elevate media levels of citric acid in osteosarcoma cells. The kinetics of the citrate response is essentially identical to the BGP response of these cells (Fig. 1) and suggests that both the BGP and citrate effects may reflect the activation of specific genes by 1,25(0H)*D3 in the manner postulated for the cytosolic calcium-binding protein of intestine (29). To our knowledge, the present results are the first demonstration that 1,25(OH)2D3 stimulates the secretion of citrate from bone cells in culture.
In order to determine citrate at the levels present in cell media, it was necessary to develop an assay more sensitive than the citrate lyase procedure. The present assay employs an initial BaS04 adsorption step to extract citrate from most proteins and organic acids. Citric acid is then desorbed with acid and analyzed by high pressure liquid chromatography. The sensitivity limit of this assay is determined primarily by the photometer used to monitor the effluent. In our system, with a 0.02 absorbance unit full scale recorder, we could quantitatively measure 1 nmol of citrate in 5 ml of media. This is over 100 times less than the 110 nmol in 5 ml of media limit of the citrate lyase assay. Among the other advantages of this method are that labeled citrate can be used as an internal standard to evaluate recovery and that the presumptive citrate component can be recovered and subjected to mass spectral analysis for structural verification.
Previous studies with bone in organ culture and with isolated osteoblastic cells in primary culture have demonstrated that 1,25(OH)2D3 and PTH dramatically decrease the release of 14C02 from 1,5-14C-labeled citrate (17,30,31). These results have been interpreted as evidence for decreased intracellular decarboxylation of citrate in the tricarboxylic acid cycle of hormone-treated bone cells, a decrease which is presumed to 1,25-Dihydroxyvitamin D3 Increases Citrate Secretion increase intracellular citrate levels and so drive an increase in the passive diffusion of citrate from cells. There are several arguments against this interpretation of the 14C02 release studies. In order to explain the lo-to Xi-fold increase in citrate secretion observed in our investigations, it would be necessary to postulate that intracellular citrate levels could be increased to this extent without distorting cellular metabolism in the pathways now known to be allosterically regulated by citrate. In addition, decreased citrate decarboxylation in the tricarboxylic acid cycle implies that the aerobic metabolic rate in hormone-treated cells is correspondingly reduced, an effect which has not been observed (15,31). It seems to us more likely that the decreased release of '%O, from labeled citrate is due to the dilution of citrate label by secreted citrate prior to its entry into the cell. While we presently favor the alternative hypothesis that 1,25(OH)zD3 directly increases the rate of citrate transport from bone cells, it is clear that further studies will be needed to determine the relationship between hormone effects on 14C02 release from labeled citrate and citrate secretion.
The observation that 1,25(OH)2D3 stimulates a parallel secretion of citrate and BGP from osteoblastic bone cells suggests that these molecules may act in tandem in bone. Since both substances bind strongly to hydroxyapatite (11,20,21) and probably account for most of the molecules bound to the surface of bone mineral, the most likely site of their action is bone mineral. Although the mechanisms by which BGP and citrate affect bone mineral are not yet understood, recent studies have demonstrated that warfarin, a vitamin K antagonist, produces an excessive mineralization disorder characterized by complete closure of the growth plate in rats. It has been postulated that this disorder is caused by an inability to inhibit spontaneous hydroxyapatite growth (22) due to an absence of the vitamin K-dependent amino acid Gla in BGP. It is possible that citrate may also play a role in the inhibition of hydroxyapatite growth, perhaps by regulating the solubility of bone mineral (1,12,13). If the function of BGP and citrate is indeed the inhibition of bone mineralization, the increased serum 1,25(OH)2D3 levels in Ca"+-deficient animals may act to restore serum Ca2+ levels partly by stimulating BGP and citrate production and thereby reducing the extent to which newly formed bone is mineralized.