Intestinal synthesis of 24-keto-1,25-dihydroxyvitamin D3. A metabolite formed in vivo with high affinity for the vitamin D cytosolic receptor.

24-Keto-1,25-dihydroxyvitamin D3 has been identified as an intestinal metabolite of 1,25-dihydroxyvitamin D3 by ultraviolet absorbance, mass spectroscopy, and chemical reactivity. The metabolite was produced from 1,25-dihydroxyvitamin D3 and 1,24R,25-trihydroxyvitamin D3 in rat intestinal mucosa homogenates. 24-Keto-1,25-dihydroxyvitamin D3 is present in vivo in the plasma and small intestinal mucosa of rats fed a stock diet, receiving no exogenous 1,25-dihydroxyvitamin D3, and in the plasma and small intestinal mucosa of rats dosed chronically with 1,25-dihydroxyvitamin D3. 24-Keto-1,25-dihydroxyvitamin D3 has affinity equivalent to 1,24R,25-trihydroxyvitamin D3 for the 3.7 S cytosolic receptor specific for 1,25-dihydroxyvitamin D3 in the intestine and thymus. In cytosolic preparations contaminated with the 5 S vitamin D-binding protein, both metabolites are about 7-fold less potent than 1,25-dihydroxyvitamin D3. In contrast, in cytosolic preparations largely free of the 5 S binding protein, both metabolites are equipotent with the parent compound. No evidence was obtained supporting a substantial presence of 23-keto-1,25-dihydroxyvitamin D3 in vivo; nor was the latter compound generated in detectable amounts from 1,25-dihydroxyvitamin D3 by intestinal homogenates. Thus, C-24 oxidation is a significant pathway of intestinal 1,25-dihydroxyvitamin D3 metabolism that produces metabolites with high affinity for the cytosolic receptor which mediates vitamin D action.

Production of 24-Keto-I,25-(OH)zD3 from 1,24R,25-(OH)3D3-The procedure described above was used except 1,24R,25- ( Sodium Borohydride Reduction-To 300 ng of 24-keto-1,25-(OH),D3 in methanol (30 pl) were added several small crystals of sodium borohydride. After 6 min, the reaction was quenched with dilute hydrochloric acid. The solvent was evaporated under a stream of nitrogen. The residue was extracted with chloroform (4 X 0.1 ml). The chloroform was evaporated and the residue was analyzed by HPLC. The material recovered from HPLC was analyzed by C1-NCI mass spectroscopy.
The calf thymus cytosolic receptor assay was also performed by a published procedure (41). Each incubation tube contained 1,25-(OH)z[26,27-3H]D3 (90 Ci/mmol, 0.06 pmol) in ethanol; unlabeled metabolite added in ethanol (25 pl); and -0.6 mg of protein in buffer (0.45 ml) consisting of 10 mM Tris-HC1, 500 mM KCl, 1 mM EDTA, 5 mM dithiothreitol, and 10 mM NaMo04, pH 7.5. The mixture was incubated for 18 h at 4 'C. Bound ligand was separated from free with hydroxyapatite. The bound steroid was extracted from hydroxyapatite with ethanol. The ethanol was evaporated and the radioactivity was measured.

RESULTS
Unlabeled 1,25-(OH)2D3 was incubated with intestinal homogenates prepared from 1,25-(OH)2D3-treated rats. The organic extract was partially purified by Sephadex LH-20 chromatography, and a metabolite that eluted in the 1,25-(OH)2D3 area was isolated through three different HPLC steps. The last HPLC system yielded a single major component (Fig.  Ut). The  I I the base peak in la-hydroxylated vitamin D compounds (42). In this case, however, the base peak was m/z 59, which is produced by C-24/C-25 bond schism. Although this peak is common in the mass spectra of 25-hydroxylated vitamin D compounds, it is usually not the base peak. The intensity of m/z 59 in this case suggests labilization of the C-24/C-25 bond. The molecular weight of 430, the side chain functionalization, and the labilization of the C-24/C-25 bond are consistent with a ketone at C-24, namely 24-keto-1,25-(OH)zD3. An electron impact mass spectrum of the silylated metabolite had a molecular ion at m/z 646, and therefore indicated that only three hydroxyl groups were present (Fig. 2C). The peaks at m/z 556 and 466 represent loss of one and two (CH3)&OH groups from m/z 646, respectively. Peaks at m/z 631,541, and 451 indicate loss of methyl groups from m/z 646, 556, and 466, respectively. The base peak at mlz 206 is the trimethylsilyl counterpart to m/z 134. The peak at m/z 73 represents (CH3)3Si+. Finally, the peaks at m/z 131 and 515  (19)-triene chromophore. Based on its UV absorbance, approximately 2 pg were isolated. C1-NCI' mass spectroscopy of the metabolite confirmed its molecular weight as 430. An electron impact mass spectrum also indicated a molecular weight of 430 ( Fig. 2A). Peaks at m/z 412 and 394 indicated loss of one and two molecules of water, respectively, from the molecular ion. The peak at m/z 371 was produced by loss of 59 atomic mass units from the molecular ion. The peak at m/z 269 resulted from loss of the side chain and one water molecule. Loss of water from m/z 269 produced m/z 251. The latter two peaks indicated that the functionalization of  -, relative intensity less than 0.1%. result from C-24/C-25 bond schism. In other words, m/z 131 is the silylated counterpart of mlz 59.
To distinguish 24-ket0-1,25-(OH)~D~ and another recently characterized vitamin D3 metabolite, 23-ket0-1,25-(OH)~D~ (36), their mass spectra were compared ( Table I). The 23keto compound had a molecular ion of very low intensity, most likely the result of a McLafferty rearrangement between the C-23-ketone and the C-25-hydroxyl group, which are able to align in a pseudo-six-member ring. The resultant proton transfer and loss of (CH3),C0 (m/z 58) produced peaks at m/ z 372, 354 (loss of H 2 0 from 372), and 336 (loss of H20 from 354). These were absent or of relatively low intensity in the mass spectrum of 24-keto-1,25-(OH),D3. Instead, the spectrum of the latter compound had a more intense molecular ion and peaks resulting from dehydration of the molecular ion. It also had a relatively intense peak (compared to the molecular ion) at m/z 371. Thus, these two vitamin D3 side chains are distinguishable by mass spectroscopy.
The relationship of the newly identified metabolite to the   HPLC system 1 (Fig. 4) system 1 (Fig.   4). The materials in fractions 29-34 were collected and re-analyzed in HPLC system 2 as described in the legend of Fig. 5. A, intestinal homogenates; B,  plasma (dashed line) and intestinal mucosa (solid line). The data are plotted as counts/min/fraction. The fractions were 1 rnl each. The compositions of the peaks from 1,25-(OH),D3-treated animals recovered from the first HPLC system (HPLC system 1; Fig. 4) were examined on a second HPLC system (HPLC system 2), which had different selectivity (2). Note that the elution order of standards in HPLC system 2 was different from HPLC system 1. Re-chromatography of the 24-keto-1,25-(OH)z[3H]D3 peak recovered from the intestinal homogenates showed a major peak that co-chromatographed with In contrast, re-analysis of the "23-ket0-1,25-(OH)~D~/ 1,23S,25-(OH)3D3" fractions of HPLC system 1 showed that the major product of intestinal homogenates and the major products in vivo were distinct from either of those two standards (Fig. 6). There may be a small amount of 1,23S,25-(OH)3[3H]D3 in each case (43); however, 23-keto-1,25-

TABLE I1
Competition of 1,25-(0H)2D3 and its deriuatiues /or the 3.7 S cytosolic receptors present in intestine and thymus Intestinal cytosol was prepared from chick mucosa (2). Thymus cytosol was prepared from calf (41). The thymus cytosol was treated with ammonium sulfate to effect separation of the 5 S binding protein from the 3.7 S receptor. Data were obtained from the curves in Fig.  7 .

Chick cytosol
Thymus cytosol Compound placement" affinityb placement" affinityb individually re-chromatographed in HPLC system 2. In each case, the radioactivity coeluted with authentic 1,25-(OH)zD3-26,23-lactone. The affinity of 24-ket0-1,25-(OH)~D~ was measured for the 1,25-(OH)2D3 cytosolic receptor in chick intestinal mucosa and in calf thymus (Fig. 7). In each case, 1,24R,25-(OH)3D3 and 24-keto-1,25-(OH),D3 were indistinguishable and were more active than the C-23-oxidized derivatives. In the bovine thymus cytosol preparation, which lacks substantial 5 S vitamin D-binding protein contamination, 1,25-(OH)zD3, 24-ket0-1,25-(OH)~D~, and 1,24R,25-(OH)3D3 were equipotent; in the chick cytosol preparation, which contains the 5 S vitamin-D binding protein, the two metabolites had 15% of the affinity of 1,25-(OH)*D3 (Table 11). One key difference between these two preparations is the diminished concentration in the latter preparation of a second protein, which sediments at 5 S (41). The 5 S protein, an artifact of the homogenation procedure, also binds vitamin D metabolites and probably discriminates differently than the 3.7 S protein. Therefore, data generated with cytosolic preparations substantially free of the 5 S con-taminant are likely to reflect affinity of ligand for receptor more accurately than those obtained with crude preparations. On the other hand, differences in receptor structure have not been ruled out.
In summary, this report describes the identification of a physiological 1,25-(OH)zD3 metabolite produced by intestine as 24-ket0-1,25-(OH)~D~. 24-Ket0-1,25-(OH)~D~ is rapidly generated from 1,25-(0H),D3 and is present in a vitamin D3 target tissue. It has high affinity for the cytosolic 1,25-(OH)2D3 receptor. Its physiological function, if any, is not certain, but it is intriguing to consider that the metabolite acts in situ as a mediator of calcium homeostasis. Alternatively, it may be part of an inactivation pathway. Further research will determine which of these is the more viable hypothesis.