Structure Function Analysis of Vitamin D Analogs with C-ring Modifications*

Analogs of la,26-dihydroxyvitamin D3 (la,26-(OH)2D3) with substitutions on C-11 were synthesized. Small apolar substitutions (1 la-methyl, 1 la-fluoro-methyl) did not markedly decrease the affinity for the vitamin D receptor

Vitamin D is metabolized into a large number of metabolites, but lc~,25-(0H)~D~' is the major hormonal form of vitamin D that binds to specific receptors with high affinity (1)(2)(3).These receptors are not only present in the classical target tissues for vitamin D (intestine and bone) but also in many other cells and organs, especially those with a calcium transport function (e.g.placenta) or with a calcium-dependent hormonal secretion (e.g.endocrine pancreas).Moreover, the vitamin D hormone modifies growth and differentiation of both normal and malignant cells (4,5).This new knowledge opened the search for vitamin D analogs with more specific actions on one of these target tissues with the aim of specific therapeutic intervention (e.g. for immune disorders, cancer control, psoriasis, or osteoporosis).Many other steroid hormones have natural hydroxylations in the C-ring (e.g.cortisol), or C-ring analogs have important agonist or antagonist effects (e.g.RU 38486 with antiprogesterone and antiglucocorticoid effects (6)).The role of the C-ring of the secosteroid structure of vitamin D in the binding to or activation of the natural vitamin D-binding proteins was not yet fully explored (7).We therefore synthesized a series of analogs of la,25-(OH)zD3 with modifications of the C-ring at (2-11.These analogs retain or increase their binding properties for the serum vitamin D-binding protein (DBP), whereas large or polar substitutions decrease the receptor binding and biological activity in vitro and in vivo.

Methods Synthesis of C-11-substituted 1a,25-(0H)a3
Analogs-All analogs of la,25-(OH)*D3 were obtained by total synthesis.The method rests on the Horner coupling (10) of the A-ring with the appropriate CDring of the vitamin D3 skeleton, both in the required enantiomeric form (Fig. 1).In practice, ketones lb-lk, possessing the various substituents at C-11, under protected form if necessary (see Id, lh, and li), were condensed with the lithium salt of the protected phosphine oxide 2, yielding the corresponding trienes 3. Consistently lower yields (-20-40%) were observed when coupling derivatives that possess the 0-oriented substituent at C-11, as in IC, li, and lk.At the stage of the coupling reaction, all hydroxyl groups (te. at C-1, C-3, and C-25) needed to be protected.Removal of these protective groups in 3 under acidic conditions was the final step in the synthesis and led to the analogs listed in Fig. 1.The protected phosphine oxide 2 was obtained following a known method (11).Depending on the nature of the C-11 substituent, the various derivatives lb-lk were synthesized using different methods.These are summarized in Fig. 2 and briefly described below.Full synthetic details will be described elsewhere.
The synthesis of derivatives lb, lf, lg, and l j is shown in Fig. 2 4 .The appropriately functionalized CD-ring skeleton was obtained by oxidative fission of the triene system and of the 22,23-double bond in vitamin Dz, followed by reduction to Inhoffen-Lythgoe diol 4 (12).After selective activation of the primary alcohol in 4, the side chain at C-22 was completed, bearing the 25-hydroxyl under protected form (13).After oxidation of the alcohol in 5 to the corresponding ketone, a conjugated double bond was introduced following standard methodology (14, 15).The resulting enone 6 then served as a substrate for the introduction at C-11 of the methyl (lb), ethyl (If), vinyl (lg), and phenyl (lj) substituents using the appropriate organocopper reagents (16)(17)(18).
As shown in Fig. 2B, the vinyl-substituted derivative l g was the precursor for Id and l e and may also serve to obtain lh.The oxidative fission of the vinyl bond, followed by reduction, leads to diol 7.After selective protection of the primary alcohol, further oxidation led to ketone Id.On the other hand, selective activation of the primary alcohol allowed for the further introduction of a fluorine substituent (19); analogous oxidation here resulted in ketone le.A nondestructive oxidation of the vinyl double bond in l g using 9borabicyclononane/hydrogen peroxide opened the way to the 2-hydroxyethyl-substituted derivative l h (20).
Derivatives possessing a @oriented substituent at (2-11, i.e.IC, li, and l k , were obtained as shown in Fig. 2C.Starting from the already described lb, lh, and lj, a double bond was first introduced as in 8. Subsequent catalytic hydrogenation then afforded the required derivatives with epimeric configuration at C-11. Cells and Animals-The human promyelocytic leukemia cell line (HL-60) and UMR 106 lines were obtained from the American Tissue Culture Company (Rockville, MD) and T. J. Martin (Melbourne, Australia), respectively.Peripheral blood mixed mononuclear cells (PBMC) were prepared from buffy coats of human blood obtained from the Belgian Red Cross.Vitamin D-deficient chicks were reared on a vitamin D-deficient diet (1% calcium, 1% phosphorus, Hope farms, Woerden, The Netherlands) and housed in a windowless room.After 3 weeks, they received a daily intramuscular injection for 10 days with la,25-(OH)*D3 or its analogs solubilized in polyethylene glycol.Serum, bone, and duodenal mucosa were then collected under Nembutal anesthesia.Calcium was measured by atomic absorptiometry in serum and ashed bone.Serum osteocalcin and duodenal calbindin D-28K were measured by radioimmunoassay using specific anti-chick antisera raised against these chick proteins.

Vitamin D Analogs with C-ring Modifications
Binding Studies-Binding of vitamin D metabolites and analogs to hDBP was performed at 4 "C essentially as described previously (21).[3H]la,25-(OH)zD3 and la,25-(OH)zDa or its analogs were added in 5 p1 of ethanol into glass tubes and incubated with hDBP (0.18 pM) in a final volume of 1 ml (0.01 M Tris-HC1, 0.154 M NaCl, pH 7.4) for 3 h at 4 "C.Phase separation was then obtained by the addition of 0.5 ml of cold dextran-coated charcoal.The binding to the vitamin D receptor was estimated in either rat duodenal mucosa or in rat osteoblastic cells (UMR 106 cells).Duodenal mucosa from normal Wistar rats was sonicated in 4 volumes of buffer (0.05 M Tris-HCl, 0.5 M KC1, 5 mM dithiothreitol, 10 mM Na2MoO4, and 1.5 mM EDTA, pH 7.5).The high speed supernatant was then incubated with 0.2 nM [3H]1a,25-(OH)2D3 and increasing concentrations of la,25-(OH)2D3 or its analogs in a final volume of 0.3 ml for 15 min at 25 "C followed by 5 min at 0 "C.Phase separation was then obtained by the addition of cold dextran-coated charcoal.The osteoblast homogenate was obtained from UMR 106 cells grown from the last 24 h in a- minimum Eagle's medium supplemented with 2% charcoal-treated fetal calf serum.The cells were then trypsinized, and cytosolic extracts were prepared.The cell pellet was extracted on ice in a hypertonic buffer as was done for the duodenal mucosa, and the receptor binding was measured (22).
Culture Conditions for Mitogen-induced Proliferation of Peripheral Blood Mononuclear Cells-PBMC were obtained from the buffy coat of blood donors by centrifugation of heparinized venous blood on Ficoll-Paque (Pharmacia, Uppsala, Sweden).The culture conditions were identical with those previously described (9), except that the culture medium in the present study was RPMI 1640.
Culture Conditions for HL-60 Cells-HL-60 cells were seeded at 1.2 X 10' cells/ml, and la,25-(0H),D3 or its analogs were added in ethanol (final concentration, <0.2%) in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum (GIBCO), 100 units/ml penicillin, and 100 pg/ml streptomycin (Boehringer).After 4 days of culture in a humified atmosphere of 5% COz in air at 37 "C, the dishes were shaken to lose any adherent cells, and all cells were then assayed for maturation by NBT reduction assay and for proliferation by [3H] thymidine incorporation.
NBT Reduction Assay-Superoxide production was assayed by NBT-reducing activity as described previously (23).HL-60 cells at 1.5 X IO6 cells/ml were mixed with an equal volume of freshly prepared solution of phorbol 12-myristate 13-acetate (200 ng/ml) and NBT (2 mg/ml) and incubated for 30 min at 37 "C.The percentage of cells containing black formazan deposits was determined using a hemacytometer.
Bone Resorption Assay-Bone resorption was measured as the release of previously incorporated "Ca from 17-day-old fetal radii and ulnae after 3 and 6 days of culture as described in the fetal rat limb explant assay of Raisz (24).The radii and ulnae were cultured in BGJb medium with 2% charcoal-treated fetal calf serum at 37 "C in 40% 02, 55% Nz, and 5% CO,.To reduce free exchangeable calcium, the limb explants were precultured for 24 h.The medium was then replaced by control medium or medium containing 1a,25-(OH)& or one of the analogs.After 3 days, the medium was replaced by fresh medium of the same composition.The 4'Ca content of the medium at 3 and 6 days and of the 5% trichloroacetic acid extracts of the bones were measured by liquid scintillation counting and used to calculate the total "Ca content and the cumulative percentage of 46Ca released.
Measurement of 25-OH-D3-24-hydroxyyle Actiuity-UMR 106 cells were cultured as described above.After the second culture period of 24 h, the medium was changed to serum-free a-minimum Eagle's medium containing 0.1% bovine serum albumin, and the cells were incubated for 1 h with 1a,25-(OH)*D3 or one of the analogs.The cells were then washed for 2 h with a-minimum Eagle's medium supplemented with 2% bovine serum albumin, followed by an incubation for 1 h with [3H]25-OH-D3 in the presence of 0.1% bovine serum albumin.Subsequently, the formation of 24,25-(0H)zD3 was assessed as described previously (25).I and shown in Fig. 3.A methyl or fluoromethyl group on C-11 did not markedly alter the binding properties, whereas Ila- ethyl or lla-vinyl groups caused a 5-lo-fold decrease in receptor binding.A still larger (phenyl) or more polar (Ila- OHCH, or lla-(2-hydroxyethyl)) group further decreased the receptor affinity (Table I).The stereochemistry of the substitution also influenced the affinity, but the effect was different for 11-methyl and 11-phenyl groups (Table I).

Binding
The affinity of la,25-(OH)zD3 to hDBP at 4 "C was 1.5 f 0.2 X lo' M" (mean f S.D., n = 6).Apolar substitutions with one or two carbons in position C-11 increased the affinity of la,25-(OH)zD3 analogs for hDBP, whereas even large polar substitutions did not markedly decrease the binding to hDBP (Fig. 4).The analogs with the highest affinity for hDBP were lla=vinyl-la, and its analogs between 10-and 100fold, but the rank order remained essentially unchanged (Fig. 5B).The antiproliferative effect of the hemisuccinate of lla- (2-hydroxyethyl)-la,25-(OH)zD3 was markedly lower than its nonesterified analog in contrast to their similar receptorbinding properties (data not shown).
The differentiation of HL-60 cells was markedly enhanced by la,%-(OH)z& and all the 11a-or 11/3-substituted analogs (Table I).Using NBT as the end point for vitamin D effects, lla-methyl, Ila-fluoromethyl, Ila-vinyl, and lla-ethylla,%-(OH)zD3 were all very effective, whereas the larger or more polar analogs were less active (Table I).The effect of the vitamin D analogs on the induction of nonspecific esterase activity differed only slightly from their effect in the NBT reaction (data not shown).The differentiating effect of la,25-(OH)zD3 on HL-60 cells was reciprocal to the effect of this steroid on HL-60 cell proliferation as measured by [3H]thymidine incorporation (Fig. 6) so that an excellent negative correlation between both parameters was observed ( r = 0.9; p < 0.001).A similar response was observed for three of the C- Biological Activity of 1a,25-(0f&D3 Analogs in the Rachitic Chick in Vivo-The concentrations of calcium in serum and bone, osteocalcin in serum, and calbindin D-28K in duodenal mucosa were measured after 10 days of treatment of vitamin D-deficient chicks with la,25-(OH)2D3 or its C-11-substituted analogs.Among the analogs tested, lla-methyl-la,25-(OH),D3 was the most potent on all parameters but nevertheless reached only 50% of the potency of la,25-(OH)2D3, whereas its &isomer was about 10 times less active.The lla- fluoromethyl-and lla-vinyl analogs also showed mild agonist activity (Table I and Fig. 8).Larger and/or more polar substitutions on C-11 nearly totally abolished the biological activity in uivo.

DISCUSSION
Analogs of la,25-(OH)2D3 with substitutions in the C-ring at C-11 are of special interest, and only limited chemical or biological data on such analogs are available (7, 25).The synthesis of such analogs is best realized by total synthesis involving classical coupling of A and CD fragments of the vitamin D skeleton, as indicated in Fig. 1.
The binding of C-11-substituted la,%-(OH)2& analogs to the rat vitamin D receptor was markedly affected by the size, polarity, and stereochemistry of the substitution (Table I).Analogs with a single carbon substitution retained the highest affinity.lla-Methyl-and lla-fluoromethyl-1a,25-(OH)~D3 retained nearly the affinity of la,25-(OH)2D3 for the rat osteoblast, calf thymus, or rat duodenal receptor (Table I).ll@-Methyl and lla-chloromethyl substitutions already reduced the affinity, which was further lost by introduction of a polar substitution (Ila-hydro~ymethyl-la,25-(OH)~D~) (Fig. 3).Analogs with two or more carbons on C-11 substitutions of la,25-(OH)zD3 all had decreased receptor affinity, but again, the more polar analogs had the lowest affinity.The largest substitution (phenyl) also decreased the receptor affinity, but of the two stereoisomers, the j3 analog retained the highest affinity, whereas for 11-methyl isomers, the a analog had a higher affinity than the @ analog (Table I).The receptorvitamin D interaction is thus hindered by polar or large substitutions on the C-ring.This must be due to the interaction at the hormone-binding site of the receptor, since it occurs as well in direct receptor binding experiments (Fig. 3) and in experiments that evaluate the further gene activation by the receptor-hormone complex (see below).The proximity of C-11 to the hormone-receptor binding site could probably be exploited for photoaffinity labeling.
The affinity of la,%-(OH)2D3 for hDBP was similar to our previous observations (20) and to the results obtained by others (27, 28).In contrast to the effect on receptor binding, substitutions on C-11 did not decrease but instead increased the affinity for hDBP (Table I and Fig. 4).Indeed, no C-11 analog (except 1lj3-(2-hydro~yethyl)-la,25-(OH)~D~) showed a decreased binding to hDBP, whereas several small substitutions increased the affinity severalfold.The stable binding to hDBP was also confirmed in in vitro studies on the effect of vitamin D analogs on lymphocyte proliferation, since the presence of hDBP decreased the activity of all the analogs (Fig. 5).The increased affinity of C-11-substituted la,25-(0H)ZDS analogs was also observed for rat and chick DBP (data not shown).When C-11 analogs of 25-OHD3 were analyzed, a similarly high affinity for human or rat DBP was observed (29).This could even be exploited for the production The affinity of further metabolites or analogs of la,25-(OH)zD3 for DBP has not been extensively studied before, but very few have increased affinity for this plasma protein.
The two stereoisomers of la,24,25-trihydroxyvitarnin D3 lose about half of their affinity for rat DBP (30).The natural and unnatural stereoisomers of la,25-(OH)2D3-26,23-lactone have a nearly %fold increase in binding to rat DBP, but their binding to the chick intestinal receptor is only 0.6 or 3%, respectively (31) A more recent analog, 2j3-(3-hydroxypropyl)la,25-(OH)& has a 2-fold increase in DBP binding but an %fold decrease in chick intestinal receptor binding (32).Our present data thus demonstrate that C-11 substitutions of 1a,25-(OH)zD3 have variable effects on receptor binding but do not sterically hinder the binding and, on the contrary, can even enhance up to 4-fold the binding affinity towards DBP.This probably implies that C-11 extensions stick outside the binding site of the vitamin D-DBP complex but nevertheless create new interactions that stabilize the complex.
The effect of C-11 substitutions on the proliferation of normal leucocytes differed from the results obtained on proliferation and differentiation of promyelocytic tumor cells.As described before (9,23), inhibition of proliferation of leukemic cells required an about 100-fold higher concentration of la,25-(OH)zD3 than necessary for normal mononuclear cells.Of all the C-11-substituted analogs, only lla-methyl-la,25-(0H)ZDa markedly inhibited PBMC proliferation, doing so slightly better in the absence than the presence of DBP (Fig. 5).However, several C-11 analogs markedly inhibited proliferation and induced differentiation of HL-60 cells.The induction of differentiation (NBT reduction) was a mirror image of the effect of vitamin D analogs on proliferation ([3H] thymidine incorporation) (Fig. 6).Most C-11-substituted analogs thus have higher effects on differentiation and proliferation of promyeloid leukemia cells than on normal mixed blood mononuclear cells.The reason for this discrepancy is not clear, but differences in receptor structure, regulation, or accessibility, postreceptor differences, or simple differences in the metabolism of the vitamin analogs are all plausible.
The bone-resorbing effect of the new 1a,25-(OH)zD3 analogs were in line with their affinity for the vitamin D receptor and their effect on the differentiation of HL-60 cells, whereas the concordance with inhibition of PBMC proliferation was rather poor (Table I).This concordance in differentiating effects on HL-60 cells and bone resorption is not unexpected R. Bouillon and E. Van Herck, unpublished results.since bone marrow monocyte cells, with precursors similar to those of HL-60 cells, are supposed to be at the origin of osteoclasts (33, 34) and la,25-(OH)2D3 can stimulate the differentiation of these precursors into more mature osteoclasts.For unexplained reasons, however, no such correlation was found for other vitamin D analogs with side chain modifications (e.g.A22-24,24,24-trihomo-la,25-(OH)~D~ (35)).
The biological effects of C-11-substitutedvitamin D analogs were tested both for serum and bone calcium responses and for the effect on serum osteocakin and duodenal calbindin D-28K concentrations.Similar results were obtained for all these parameters, but the dose-response curve for the vitamin Dresponsive proteins was much steeper and allowed a better evaluation of the relative potency.The difference between the stereoisomers of ll-methyl-la,25-(OH)2D3 on receptor binding and HL-60 cell proliferation i n vitro was confirmed in uiuo for effects on calcium metabolism (Table I and Fig. 8).
The activity of all C-11-substituted analogs i n vivo was 2-10fold lower than the effects on bone resorption in vitro and lower than could be expected from their receptor-binding properties (Table I).This difference was most obvious for as expected, also demonstrated a low biological effect in vivo (Fig. 8).
Previous data on vitamin D analogs with C ring modifications are limited to the synthesis of a few Ila-substituted-9(11)-dehydr0-la,25-(OH)~D~ analogs (7) with decreased receptor affinity (36) and to the synthesis of lla-hydroxyvitamin D3 (26).This latter compound bound slightly better to DBP than vitamin D3 itself but displayed, as expected, a low affinity for the vitamin D receptor.No direct studies of its further hydroxylated metabolites were performed, but 11ahydroxyvitamin DS showed low biological activity in vivo (26).
Several of the C-11-substituted analogs have a higher affinity for hDBP than the natural hormone and still retain good receptor-binding properties (e.g.lla-CHa-, lla-CHzF-, or lla-~inyl-la,25-(0H)~D~).No natural metabolite has such properties, and only one other analog approached this combination (32).Increased DBP binding is expected to decrease the biological effectiveness, as has been demonstrated repeatedly in vitro (36), but can otherwise prolong the circulating half-life.One recent analog, 2j3-(3-hydroxypropyl)-la,25-(OH)&, with only a %fold increase in DBP binding and an 8-fold decrease in receptor binding, increased the bone mass of ovariectomized rats when given intermittently twice a week (32).Several of the C-11-substituted analogs also demonstrated increased DBP binding but, at least when given once daily, this did not increase, but rather decreased, their biological efficacy.Giving it more intermittently, as in the study by Okano et al. (32), might be more advantageous.
for psoriatic lesions (41)) or induce greater effects on lympho- Several C-11-substituted analogs were more potent on leukemic cell differentiation and proliferation than on normal leucocytes i n uitro or calcium metabolism i n uiuo (Table I).
This difference was, however, smaller than that observed for several analogs with side chain modifications.Possible therapeutic effects in vivo on cancer growth or differentiation therefore will depend on further specific testing.

FIG. 4 .
FIG. 4. Binding of la,26-(OH)ZDs ( C -. ) and its C-11-substituted analogs to purified human DBP measured at 4 "C and pH 7.4.For details, see "Experimental Procedures."Left panel, binding of analogs with a single carbon substitution on (2-11; right panel, binding of analogs with more than one C substitution on C-11.The symbols are the same as in Fig. 3. B/Bo, [3H]1,25-(OH)zDs, boundbound in the absence of unlabeled steroid.

FIG. 5 .
FIG.5.Effect of la,2S-(OH)2Ds and its C-11-substituted analogs on the proliferation of phytohemagglutin-stimulated mixed peripheral human blood lymphocytes, as measured by [SH]thymidine incorporation.All experiments were done in replicates of six in microcultures of PBMC from the same donor.A, studies in the presence of hDBP-free serum; B, studies in the presence of hDBP (7.5 NM) by using DBP-free human serum reconstituted with purified hDBP.The symbols are the same as in Fig.3.

3
FIG.5.Effect of la,2S-(OH)2Ds and its C-11-substituted analogs on the proliferation of phytohemagglutin-stimulated mixed peripheral human blood lymphocytes, as measured by [SH]thymidine incorporation.All experiments were done in replicates of six in microcultures of PBMC from the same donor.A, studies in the presence of hDBP-free serum; B, studies in the presence of hDBP (7.5 NM) by using DBP-free human serum reconstituted with purified hDBP.The symbols are the same as in Fig.3.
FIG.5.Effect of la,2S-(OH)2Ds and its C-11-substituted analogs on the proliferation of phytohemagglutin-stimulated mixed peripheral human blood lymphocytes, as measured by [SH]thymidine incorporation.All experiments were done in replicates of six in microcultures of PBMC from the same donor.A, studies in the presence of hDBP-free serum; B, studies in the presence of hDBP (7.5 NM) by using DBP-free human serum reconstituted with purified hDBP.The symbols are the same as in Fig.3.
Vitamin D Analogs with C-ring Modifications llp-methyl-and lla-~inyl-la,25-(OH)~D3. Increased DBP binding could have decreased the immediate availability of vitamin D analogs in vivo, but this would normally be offset by their longer half-lives.Analogs with poor receptor binding,