Synthesis and biological activity of 22-iodo- and (E)-20(22)-dehydro analogues of 1α,25-dihydroxyvitamin D3

Article abstract of DOI:10.1016/S0968-0896(99)00249-7

Construction of 25-hydroxy-steroidal side chain substituted with iodine at C-22 was elaborated on a model PTAD-protected steroidal 5,7-diene and applied to a synthesis of (22R)- and (22S)-22-iodo-1α,25-dihydroxyvitamin D₃. Configuration at C-22 in the iodinated vitamins, obtained by nucleophilic substitution of the corresponding 22S-tosylates with sodium iodide, was determined by comparison of their iodine-displacement processes and cyclizations leading to isomeric five-membered (22,25)-epoxy-1α-hydroxyvitamin D₃ compounds. Also, 20(22)-dehydrosteroids have been obtained and their structures established by ¹H NMR spectroscopy. When compared to the natural hormone, (E)-20(22)-dehydro-1α,25-dihydroxyvitamin D₃ was found 4 times less potent in binding to the porcine intestinal vitamin D receptor (VDR) and 2 times less effective in differentiation of HL-60 cells. 22-Iodinated vitamin D analogues showed somewhat lower in vitro activity, whereas (22,25)-epoxy analogues were inactive. Interestingly, it was established that (22S)-22-iodo-1α,25-dihydroxyvitamin D₃ was 3 times more potent than its (22R)-isomer in binding to VDR and four times more effective in HL-60 cell differentiation assay. The restricted mobility of the side chain of both 22-iodinated vitamin D compounds was analyzed by a systematic conformational search indicating different spatial regions occupied by their 25-oxygen atoms. Preliminary data on the in vivo calcemic activity of the synthesized vitamin D analogues indicate that (E)-20(22)-dehydro-1α,25-dihydroxyvitamin D₃ and 22-iodo-1α,25-dihydroxyvitamin D₃ isomers were ca. ten times less potent than the natural hormone 1α,25-(OH)₂D₃ both in intestinal calcium transport and bone calcium mobilization. Copyright (C) 1999 Elsevier Science Ltd.

Full text of DOI:10.1016/S0968-0896(99)00249-7

Bioorganic & Medicinal Chemistry 7 (1999) 2877±2889
Synthesis and Biological Activity of 22-Iodo- and (E)-20(22)-
Dehydro Analogues of 1ꢀ,25-Dihydroxyvitamin D3
Rafal R. Sicinskiy and Hector F. DeLuca*
Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin-Madison, Madison, WI 53706, USA
Received 17 May 1999; accepted 19 July 1999
AbstractÐConstruction of 25-hydroxy-steroidal side chain substituted with iodine at C-22 was elaborated on a model PTAD-pro-
tected steroidal 5,7-diene and applied to a synthesis of (22R)- and (22S)-22-iodo-1a,25-dihydroxyvitamin D3. Con®guration at C-22
in the iodinated vitamins, obtained by nucleophilic substitution of the corresponding 22S-tosylates with sodium iodide, was deter-
mined by comparison of their iodine-displacement processes and cyclizations leading to isomeric ®ve-membered (22,25)-epoxy-1a-
hydroxyvitamin D3 compounds. Also, 20(22)-dehydrosteroids have been obtained and their structures established by 1H NMR
spectroscopy. When compared to the natural hormone, (E)-20(22)-dehydro-1a,25-dihydroxyvitamin D3 was found 4 times less
potent in binding to the porcine intestinal vitamin D receptor (VDR) and 2 times less e€ective in di€erentiation of HL-60 cells. 22-
Iodinated vitamin D analogues showed somewhat lower in vitro activity, whereas (22,25)-epoxy analogues were inactive. Interest-
ingly, it was established that (22S)-22-iodo-1a,25-dihydroxyvitamin D3 was 3 times more potent than its (22R)-isomer in binding to
VDR and four times more e€ective in HL-60 cell di€erentiation assay. The restricted mobility of the side chain of both 22-iodinated
vitamin D compounds was analyzed by a systematic conformational search indicating di€erent spatial regions occupied by their 25-
oxygen atoms. Preliminary data on the in vivo calcemic activity of the synthesized vitamin D analogues indicate that (E)-20(22)-
dehydro-1a,25-dihydroxyvitamin D3 and 22-iodo-1a,25-dihydroxyvitamin D3 isomers were ca. ten times less potent than the nat-
ural hormone 1a,25-(OH)2D3 both in intestinal calcium transport and bone calcium mobilization. # 1999 Elsevier Science Ltd. All
rights reserved.
Introduction
by gamma spectrometry, namely, the radioiodinated
derivatives of the natural hormone, 1a,25-(OH)2D3.
Metabolism of vitamins D has been widely studied in
our and many other laboratories during the past
three decades.1,2 1a,25-Dihydroxyvitamin D3 (1a,25-
(OH)2D3, calcitriol) is the most potent known metabo-
lite in the vitamin D3 series for the regulation of calcium
and phosphate homeostasis.3±6 The extensive metabolic
studies have stimulated many e€orts to prepare radi-
olabeled vitamin D3 derivatives.7,8 The availability of
such radioactive compounds is of critical importance for
numerous biomedical experiments. Therefore, over the
past several years an e€ort has been made in our
laboratory to synthesize tritium labeled vitamin D2 and
D3 metabolites of still higher speci®c activity.9±15
Recently, we have turned our attention to the vitamin D
analogues containing radionuclids which can be detected
It was established that the nuclear receptor (VDR) for
1a,25-(OH)2D3 exists in more than 30 tissues and cancer
cell lines.16 It has become clear that important studies
could be made possible if hormone analogues contain-
ing a g-emitting isotope were available. Such vitamins
would have marked advantages over the presently used
tritium labeled ones. Since the speci®c activity of 125I
3
(ca. 2200 Ci/mmol) is much higher than that of H (ca.
29 Ci/mmol) it would allow receptors to be detected
with greater sensitivity. It was, therefore, of interest to
obtain such an iodo vitamin D compound which could
have similar tissue distribution properties as native
hormone and, in general, behave similarly to this nat-
ural steroid in most biochemical analyses. As a con-
sequence, our goal was to synthesize an iodinated
analogue possessing considerable ability to bind to the
intestinal nuclear 1a,25-(OH)2D3 receptor. It is well
known that the natural hormone is characterized by a
very high anity for the VDR. Broad structure-func-
tion studies clearly demonstrated that deletion of the
hydroxyl groups and almost all alterations of its carbon
Key words: Vitamin D3 analogues; receptor binding; molecular mod-
eling/mechanics; antiproliferative agents.
* Corresponding author. Tel.:+1-608-262-1620; fax: +1-608-262-
7122; e-mail: deluca@biochem.wisc.edu
y
On sabbatical leave from Department of Chemistry, University of
Warsaw, ul. Pasteura 1, 02-093 Warsaw, Poland.
0968-0896/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(99)00249-7

2878
R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
skeleton resulted in considerable diminished anities
of the synthesized analogues for the VDR binding
domains.8,17±19 Such analogues are usually character-
ized by low calcemic activity and decreased potency of
inducing cell di€erentiation.20 Therefore, we have deci-
ded to synthesize a compound being a simple iodo deri-
vative of the natural hormone. Thus, we attempted to
establish such a position in the seco-B-steroidal carbon
skeleton of the 1a,25-(OH)2D3 where the introduced
iodine atom would not considerably alter either the
basic hormone structure or its conformation. Addition-
ally, the position would have to be relatively easily
accessible during the synthesis and assure the resistance
of the formed iodo derivative to deiodination processes.
Taking into account the above requirements we exclu-
ded cyclohexane A-ring and hydrindane CD fragment
from our consideration and focused on the steroidal
side chain at C-17. Finally, we have decided to intro-
duce the iodine into the side chain at the 22-position.
with p-toluenesulphonyl chloride a€orded quantita-
tively a tosylate 3b which served as convenient model
compound for the nucleophilic substitution process.
The reaction of 3b with sodium iodide was carried out
in acetone:2-butanone (1:1) solvent system and required
prolonged heating at 45ꢀC for completion. After 48 h
the mixture of products was carefully separated by
HPLC and, in addition to the main 20(22)-ole®nic pro-
duct 5 (50%), two iodine-containing compounds 7 and
9 were also isolated in 13 and 26% yield, respectively.
The presence of both epimeric 22-iodo derivatives could
be easily explained taking into consideration the high
reactivity of the aliphatic iodo compounds for nucleo-
philic substitution, including the halogen±halogen
exchange processes. Classical examples of such `double
substitution' provide the reported formation of a race-
mic mixture of 2-iodooctanes on reaction of (R)-2-octyl
mesylate with potassium iodide30 and isolation of 3b-
iodocholestane as a product of the reaction of 3b-cho-
lestanyl tosylate with sodium iodide.31
Nucleophilic substitution at C-22 appears to be the most
popular method for construction of the 25-hydroxy-
cholesterol side chain whereas Wittig reaction of ster-
oidal 22-aldehydes is often used in the synthesis of
22,23-unsaturated analogues.21 C-22 Aldehydes, those
containing intact ABCD rings22 and those possessing
ring-B secosteroidal system,23 are readily available
compounds. Thus, introduction of iodine at C-22
seemed to be a reasonable choice for our studies, espe-
cially considering the literature data concerning simi-
larly modi®ed vitamin D compounds. Although the
presence of a secondary 22-hydroxy or methoxy group
resulted in considerably diminished biological activity of
vitamin D analogues,24±26 the corresponding derivatives
of 1a,25-(OH)2D3 possessing a bulky hydrophobic
22-methyl substituents showed (especially 22S-isomer)
signi®cant anity for intestinal VDR.27 It was, there-
fore, expected that introduction of the iodine at the
22-position of the side chain might also provide biolo-
gically active vitamins. This paper describes the synth-
esis, conformational analysis and biological potency of
epimeric 22-iodo-1a,25-(OH)2D3 compounds.
HPLC analysis of the reaction mixture of tosylate 3b
with NaI indicated that a ratio of 7:9 was changed from
approx. 1:1 to 1:2 after 24 and 48 h, respectively. That
the interconversion of both isomers occurred in the
process was further con®rmed when individual iodo
compounds 7 and 9 were subjected to the reaction con-
ditions with NaI, analogous to those applied to the
tosylate 3b. HPLC analysis revealed that 9 was partially
(20%) converted into isomeric 7 after 24 h and the
composition of the products did not change con-
siderably upon further heating, reaching 21% of 7 after
100 h. In the case of reaction of iodo compound 7 with
NaI, however, the ratio 1:1.9 of 7 and 9 was established
after 24 h, whereas 23% of 7 remained in the equili-
brium mixture after 100 h. These observations clearly
indicated the primary and secondary reaction products,
and therefore, allowed us to ascribe the 22R- and 22S-
con®gurations for the formed iodo compounds 7 and 9,
respectively. The assignment of the side-chain double
bond con®guration in the product 5 was based on its 1H
NMR spectrum, especially on the chemical shifts of
the 18- and 21-methyl signals typical of (E)-20(22)-
dehydrosteroids.32±35 Compound
5 could also be
obtained, albeit in a low yield (35%), by prolonged
heating of the tosylate 3b in pyridine.
Results
Synthesis
An analogous reaction sequence was performed starting
from the known23 vitamin D 22-aldehyde 2. Thus, the
aldehyde was converted to the respective (22S)-hydroxy
compound 4a in 72% yield. Its tosyl derivative 4b was
subjected to the nucleophilic substitution with NaI,
similar as in the case of the PTAD-adduct 3b. In view of
the known susceptibility of 5Z-vitamin D compounds to
undergo an iodine-catalyzed isomerization to their 5E-
isomers,36,37 the reaction was carried out in the presence
of mercury and in the dark. Monitoring of the reaction
course by HPLC indicated that a primary substitution
product 8a was slowly transformed into the isomeric
iodide 10a. After 80 h at 45ꢀC, preparative HPLC led to
isolation of the main reaction product, namely, 20(22)-
dehydrovitamin 6a (32%) and iodo compounds 8a (8%)
and 10a (20%). Both 22-iodovitamins, when protected
We decided to test the synthetic pathway leading to 22-
iodo vitamin D derivatives 8c and 10c (Scheme 1) using
much less expensive model compounds derived from
ergosterol. Thus, we used as a starting material C-22
aldehyde 1, obtained in high yield by ozonolysis of a
TBDMS-ether of PTAD-diene protected ergosterol,28
and subjected it to a reaction with Grignard reagent,
prepared from bromo compound A.29 Since it is well
known that steroidal (20S)-aldehydes, combined with
organometallic reagents, form predominantly products
which can be predicted by Cram's rule,21 it was not
surprising that we isolated the corresponding (22S)-hy-
droxy compound 3a in 83% yield. Even careful HPLC
analysis did not indicate the presence of the epimeric
(22R)-alcohol in the reaction mixture. Treatment of 3a

R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
2879
Scheme 1.
from light, can be stored for a prolonged time in a free-
zer (ca. 20ꢀC) with neither 5,6-E/Z isomerization nor
epimerization at C-22. However, both processes, if
desired, can be performed easily. Thus, the isomeriza-
tion of 8a and 10a to the respective 5E-compounds 8b
and 10b, can be accomplished by the well-known pro-
cedure using iodine as a catalyst.36 In order to achieve
the interconversion between both C-22 epimers, each
compound 8a and 10a can be subjected to the analogous
nucleophilic substitution reaction with sodium iodide as
described above for the tosylate 4b. In both cases the
similar equilibrium mixture can be reached consisting of
epimeric (22S)- and (22R)-iodo compounds 10a and 8a
in a ratio of about 3±4:1.
out the hydrolysis in the dark and shorten the reaction
time (ca. 10 h). Elongation of the time of the hydrolysis
of vitamins 8a and 10a resulted in diminished yields of
the desired vitamins 8c and 10c, and a steady built-up of
less polar products (i.e. cyclic ethers 12 and 11, respec-
tively). These conversions, although undesired, were
useful for assignment of con®guration in the respective
C(22)-epimeric iodo compounds by their correlation
with the starting tosylate 4b. Since the tosylate 4b, when
subjected to the analogous hydrolysis conditions,
formed the tetrahydrofuran derivative 11 as the only
isolable product, it was reasonable to ascribe to this
compound a structure of (22R)-22,25-epoxy-1a-hydro-
xyvitamin D3. Consequently, con®gurations at C-22 in
the iodo vitamins 10 and the tosylate 4b must be the
same (22S).
Deprotection of hydroxyl groups in the synthesized
compounds 6a, 8a and 10a, providing the corresponding
1a,25-dihydroxyvitamins 6b, 8c and 10c, can be e€ec-
tively accomplished using AG 50W-X4 ion exchange
resin in methanol. In the case of iodo vitamins it was
necessary to add mercury to the reaction mixture, carry
Biological evaluation and conformational analysis
An initial examination of biological potency of the syn-
thesized vitamins was undertaken by determining their

2880
R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
ability to displace tritiated 1a,25-(OH)2D3 from the
porcine intestinal vitamin D receptor. The results shown
in Table 1 demonstrate that (E)-20(22)-dehydrocalcitriol
(6b) is 4 times less e€ective in binding to the receptor
than the parent hormone 1a,25-(OH)2D3 (13a, Fig. 1),
whereas iodo vitamins 8c and 10c show, respectively, 1/
60 and 1/20 of the activity of the natural hormone.
Cyclic ethers 11 and 12 show very little activity in this
regard (data not shown). Cellular activity of the
synthesized vitamins was established by studying the
di€erentiation of HL-60 cells into monocytes. Also in
this system side-chain ole®n 6b was the most active,
followed by 22S- and 22R-vitamins 10c and 8c, being,
respectively, 3 and 12 times less active than 1a,25-
(OH)2D3.
Table 1. VDR Binding anitiesa and HL-60 di€erentiating activi-
tiesb of 1a,25-(OH)2D3 and its 22-substituted analogues
Side chain analogue of 1a,25-(OH)2D3
VDR
1
HL-60
1
1/60e
1/80f
1/3e
1f
1/60c
1/12d
It was interesting to compare biological activities of the
synthesized iodo vitamins 10c and 8c with recent results
of similar studies performed by Yamada's group on
closely related 22-methylated vitamin D analogues
13b,c.27,38 These 22S- and 22R-methyl derivatives were
3 and 60 times, respectively, less active than the parent
hormone 13a in the porcine VDR assay. Measurements
of cellular di€erentiation showed that 22S-isomer had
similar potency as 1a,25-(OH)2D3, whereas its 22R-epi-
mer was 80 times less potent. The observed di€erences
in the receptor binding between 22-methylated analo-
gues were related, through interesting conformational
and molecular mechanics studies,38,39 to the corre-
sponding spatial regions A (for 22S-isomer) and G (for
22R-isomer) occupied by their 25-oxygen atoms.
1/20c
1/3d
1/4c
1/2d
1/95g
1/170h
Comparison of activity data listed above indicated that
the binding potency of iodo- and methyl-derivatives was
identical among 22R-substituted compounds, whereas
in the case of 22S-isomers the methyl-substituted ana-
logue showed signi®cantly increased activity. In the lat-
ter case, the observed discrepancy can be the result of an
increased electronegativity of an iodine atom (2.66 ver-
sus 2.55 for carbon). However, considering also sub-
stantial di€erence between size of iodine atom and
methyl group, in terms of space-®lling capacity (van der
Waals radii 2.15 and 2.0, respectively), it is obvious that
such bulky substituent can severely restrict the rotation
around the C(20)±C(22) bond and, therefore, in¯uence
mobility and conformation of the C(17)-side chain.
These considerations prompted us to perform con-
formational analysis of model 8-methylene compounds
14a,b,c with 25-hydroxy- and 25-hydroxy-22-iodo-sub-
stituted side chains and compare the results with those
of the corresponding 22-methyl counterparts.
1/35g
1/4h
<1/400h
1/13h
<1/1000c,i
a
Competitive binding of 1a,25-(OH)2D3 and the synthesized vitamin
D analogues to the porcine intestinal vitamin D receptor (VDR). The
experiments were carried out in triplicate on two di€erent occasions.
The values of VDR binding anities are related to that of 1a,25-
(OH)2D3 for which the value is set at 1.0.
b
Induction of di€erentiation of HL-60 promyelocytes to monocytes
by 1a,25-(OH)2D3 and the synthesized vitamin D analogues. Di€er-
entiation state was determined by measuring the percentage of cells
reducing nitro blue tetrazolium (NBT). The experiment was repeated
three times. The values of HL-60 di€erentiation activity are related to
that produced by 1a,25-(OH)2D3 for which the value is set at 1.0.
Relative binding anities to the porcine intestinal VDR determined
in our laboratory.
d
We carried out MM+ force ®eld calculations of com-
pounds 14a±c by using the HyperChem molecular
modeling program and performed a conformational
search with its extension modules from ChemPlus
program. These molecular mechanics studies were
carried out according to protocol described in detail in
Experimental. The distribution of the side-chain con-
formers resulted from this conformational study is pre-
sented in Table 2 together with steric energy (Es) values
of the corresponding global minimum structures. These
most stable conformers and the respective positions of the
c
Determined in our laboratory.
Relative binding anities to the porcine intestinal cytosol receptor.39
e
Literature data.38
f
Relative binding anities to the chick intestinal cytosol receptor.25
Literature data.25
Compound prepared by deprotection of hydroxyl groups in 3a with
g
h
i
AG 50W-X4 resin in methanol (80% yield).

R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
2881
cated the relation between the receptor (VDR) anity
of the vitamin D analogue and the spatial region occu-
pied by its side-chain 25-hydroxy group and showed
that region A is more important than G. Thus, the
observed di€erence in receptor binding properties of
the iodo vitamins 8c and 10c can be explained, at least
to some extent, by Yamada's ``active space group con-
cept''.
As might be expected from X-ray crystallography of
several steroids,40 for the low energy conformers gener-
ated during our conformational search, only the
gauche(+) conformation in relation to the dihedral
angle C(16±17±20±22) was found. Thus, the largest
substituent at C-20 (i.e. the remaining side chain frag-
ment) is directed towards ring D and `outside the mole-
cule'. However, the conformation around the next two
C±C bonds of the side chain is in¯uenced by the pre-
sence of C(22)-substituents. X-ray data of the 25-OH-
Figure 1. Chemical structure of 1a,25-dihydroxyvitamin D3 (calcitriol)
and its 22-substituted analogues.
41
D3 revealed that only the anti isomer, with respect to
C(17±20±22±23) torsion angle, exists in the crystalline
form. Our calculations, however, indicate that an alter-
native gauche(+) form is a minimum energy con-
formation for 25-hydroxy-8-methylene derivative 14a;
its most stable conformer was found to have ca.
0.2 kcal/mol lower steric energy than the lowest energy
structure with the anti conformation.42 In the remaining
9 least energy side-chain conformers of 14a, falling into
1 kcal/mol energy window, both gauche(+) and anti
conformations were present. However, the correspond-
ing least energy side-chain conformers of the 22R-iodo
compound 14b adopted only one conformation in
respect to the C(17±20±22±23) and C(20±22±23±24) tor-
sion angles, gauche(+) and anti, respectively. In the
case of the lowest energy conformers (0±1 kcal/mol
energy window) of the 22S-iodo isomer 14c, the anti
conformation was the only preferred one in respect to
the C(17±20±22±23) bond, whereas anti and gauche(+)
25-oxygen in the other lowest energy structures are
shown in Figure 2. Figure 3 shows the result of super-
imposition of all conformers of 14a±c, falling into 1 kcal/
mol energy window, on the carbon skeleton of the glo-
bal minimum structure of 14a. It is immediately appar-
ent from the above ®gures that location in space of the
25-oxygen, crucial for anchoring of the vitamin D side
chain to the receptor, is di€erent for the compounds
studied. Thus, in the case of compound 14a, the 25-hy-
droxy group, situated in the side chain that is char-
acteristic of the natural hormone, occupies two spatial
regions A and G (as de®ned by Yamada).39 Due to
restricted conformational mobility, the location in space
of the 25-oxygens belonging to the side chain of 22R-
and 22S-iodo compounds 14b and 14c is limited
to single regions G and A, respectively. Detailed con-
formation-activity studies of Yamada et al.38,39 indi-
Table 2. Distribution of the side-chain conformers of model compounds 14a,b,c resulted from conformational search
Model 8-methylene compound
Number of conformers (energy window in kcal/mol)
2nd step 3rd step
The lowest energy (Es)
conformer (kcal/mol)
1st step
(0±6)
11
(0±1)
11
(0±4)
78
(0±1)
11
(0±3)
99
33.09a/33.31b
10
8
41
49
10
15
48
41
33.81a
9
15
33.65b
a
gauche(+) Conformation in respect to the C(17±20±22±23) torsion angle.
anti Conformation in respect to the C(17±20±22±23) torsion angle.
b

2882
R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
Figure 2. Stereoplot of the global energy minimum conformations (wireframe drawings) of the model 8-methylene compound 14a with 25-
hydroxyvitamin D3 side chain (a), its 22R-iodo substituted analogue 14b (b), and isomeric compound 14c (c). The circles show the location of 25-
oxygen atoms in the other higher energy conformers (energy window 1 kcal/mol). All hydrogen atoms are omitted.
forms were found for the neighboring C(20±22±23±24)
bond.
above allow us to hypothesize that side chain of 22R-
substituted vitamin D analogues does not ®t well to the
hydrophobic cavity of VDR, and it is likely that in the
formed complex, the 22R-group attached to the ligand
does not interact with the receptor. On the contrary, the
side chain modi®ed with 22S-substituent ®ts better to
the receptor, the 22S-group being probably in the
proximity of some hydrophobic center. Therefore,
increasing electronegativity of the atom bonded to C-22
should result in enhancing repulsion between ligand and
binding domain. In agreement with this postulate are
literature data on 22R- and 22S-methoxy analogues of
1a,25-(OH)2D3 (13d and 13e),25 reported to be 95 and
35 times less potent, respectively, in binding to the chick
intestinal vitamin D receptor than the parent hormone
(Table 1). Not surprisingly, an analogue bearing a more
polar substituent at C-22, that is, 1a,22S,25-(OH)3D3
(13g) showed even more signi®cantly (by three orders of
magnitude!) reduced anity for VDR. The analogous
relationship between potency of 22S-functionalized
vitamins and electronegativity of 22-substituent can be
observed in the HL-60 cell di€erentiation assay; in the
case of 22R-compounds such clear-cut tendency does
not exist. Interestingly, the literature data indicate that
22-hydroxy and 22-methoxy analogues of 1a,25-
(OH)3D3 were essentially devoid of calcemic activity,25
whereas 22-iodovitamins synthesized by us, especially
22R-isomer 8c, were e€ective both in intestinal calcium
Discussion
The side chain of the vitamin D3 is a highly ¯exible
fragment of this molecule. That the tertiary hydroxyl
group, situated at its terminal carbon atom, plays an
important role in anchoring the whole vitamin D com-
pound to VDR is evident: removal of this substituent
reduces the binding anity by three orders of magni-
tude.8 Therefore, vitamin D side chain has been a pre-
ferred target of many synthetic e€orts and structure±
activity studies.7,17±19 During last few years interesting
conformational analyses of 25-hydroxylated side chain
have been performed by Okamura's43,44 and Yama-
da's38,39 groups. They introduced a concept of three-
dimensional `dot maps' showing the possible locations
in space of 25-oxygen in the native hormone and its
di€erent side-chain analogues. They also attempted to
correlate the receptor (VDR) anity and cell di€er-
entiation potency of such analogues with the corre-
sponding spatial regions that can be reached by their
terminal 25-hydroxy groups. Stimulated by these ®nd-
ings we envisaged the conformational analysis of the
22-iodovitamins, whose synthesis is presented in this
study. Results of our conformational search described

R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
2883
Figure 3. Two di€erent stereoviews ilustrating results of conformational analysis of the side chain of model CD-ring compounds 14a±c. Energy-
minimized conformations of these compounds (energy window 1 kcal/mol) were overlaid on the global minimum conformer of 14a as described in
Experimental. The circles show the location of 25-oxygen atoms in the corresponding conformers. They are indicated two distinct spatial regions A
and G de®ned by Yamada.39 Side-chain carbons and all hydrogen atoms are omitted for clarity.
Table 3. Support of intestinal calcium transport and bone calcium mobilization by (E)-20(22)-dehydro- and 22-iodo-analogues of 1a,25-(OH)2D3
in vitamin D-de®cient rats on a low-calcium dieta
Compound
Compound no.
Amount (mg/d)
Ca transport S/M (mean‹SEM)
Serum Ca (mean‹SEM)
None (control)
1a,25-(OH)2D3
(E)-20(22)-dehydro-1a,25-(OH)2D3
None (control)
1a,25-(OH)2D3
(22R)-22-iodo-1a,25-(OH)2D3
(22S)-22-iodo-1a,25-(OH)2D3
0
4.0‹0.2b
11.2‹1.2c
12.4‹0.7d
3.8‹0.4b
10.5‹1.1c
10.7‹0.7d
13.2‹1.5e
4.1‹0.1b
5.9‹0.3c
5.2‹0.2d
4.0‹0.1b
5.9‹0.2c
6.2‹0.2d
4.5‹0.1e
13a
6b
0.1
1.0
0
0.1
1.1
1.1
13a
8c
10c
a
Weanling male rats were maintained on a 0.47% Ca diet for 1 week and then switched to a low-calcium diet containing 0.02% Ca for an addi-
tional 3 weeks. During the last week, they were dosed daily with the appropriate vitamin D compound for 7 consecutive days. All doses were
administered intraperitoneally in 0.1 mL propylene glycol:ethanol (95:5). Controls received the vehicle. Determinations were made 24 h after the last
dose. There were 5±6 rats per group. Statistical analysis was done by Student's t-test. Statistical data: serosal/mucosal (S/M), panel 1, b from c and
d, p<0.001; panel 2, b from c,d, and e, p<0.001; serum calcium, panel 1, b from c, p<0.001, b from d, p=0.05; panel 2, b from c and d, p<0.001, b
from e, NS.
transport and bone calcium mobilization. Preliminary
biological in vivo tests performed in our laboratory
showed that both 22-iodovitamins, as well as E-20(22)-
dehydro analogue do not di€er considerably in their
calcemic potency (Table 3). Thus, all these analogues
have about one-tenth the activity of 1a,25-(OH)3D3.
Conclusion
A convenient method of the synthesis of 22-iodinated
steroids from their easily available C(20)-aldehyde pre-
cursors have been elaborated. The introduction of iodine
atom can be achieved by nucleophilic substitution of the

2884
R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
corresponding 22S-tosylates with sodium iodide. This
reaction provides a convenient means for preparing
radioiodinated derivatives if isotopically labeled iodides
(e.g. NH4 123I, Na 125I, Na 131I, etc.) are employed. Due
to susceptibility of aliphatic iodo derivatives to nucleo-
philic substitution, such radioiodinated compounds can
also be prepared by iodine±iodine isotope exchange
process. The synthesized vitamins, especially (22S)-22-
iodo-1a,25-(OH)2D3 (10c), retain substantial binding
anity for the vitamin D receptor and have the ability
to di€erentiate HL-60 cells. Also, the isomer 10c shows
high in vivo calcium transport activity with little or no
bone calcium mobilization activity. Therefore, the iso-
topically labeled compound 10c has good potential for
development as a new tracer in various known binding
assays and for other experimental research studies. The
observed di€erences in VDR binding anity of both
isomeric 22-iodovitamins 8c and 10c were attributed,
after molecular modeling studies and conformational
analysis, to the di€erent spatial regions occupied by
their 25-oxygens.
Reaction of C-22 aldehydes 1 and 2 with Grignard re-
agent derived from bromo compound A. (a) A solution
of 4-bromo-2-methyl-2-(triethylsilyloxy)butane A (281 mg,
1 mmol) in anhydrous ether (0.5 mL, containing a cata-
lytic quantity of iodine) was added dropwise to a stirred
mixture of magnesium powder (29 mg, 1.2 mmol; ꢁ50
mesh, Aldrich) in anhydrous ether (0.5 mL) under argon
at room temperature with occasional warming it up to
35ꢀC. After addition was complete the mixture was
stirred for 1 h at room temperature and for 30 min at
40ꢀC. Then it was cooled to 0ꢀC and a solution of a
PTAD-protected diene (20S)-C-22 aldehyde 1 (123 mg,
0.2 mmol) in anhydrous THF (0.5 mL, cooled to 0ꢀC)
was added dropwise. After the mixture was stirred for
20 min at 0ꢀC and 1 h at room temperature it was
quenched with aqueous solution of NH4Cl (2 mL) and
diluted with benzene (20 mL). The organic layer was
separated, washed with brine and diluted NaHCO3,
dried (Na2SO4) and evaporated. Flash chromatography
of the residue using 20% ethyl acetate in hexane as an
eluent yielded pure (22S)-alcohol 3a (135 mg, 83%) as a
1
foam: H NMR d 0.084 and 0.106 (3H and 3H, each s,
2ÂSiCH3), 0.578 (6H, q, J=8.0 Hz, 3ÂSiCH2), 0.813
(3H, s, 18-H3), 0.885 (9H, s, Si-t-Bu), 0.936 (3H, d,
J=7.7 Hz, 21-H3), 0.944 (9H, t, J=8.0 Hz, 3ÂSiCH2
CH3), 0.968 (3H, s, 19-H3), 1.222 (6H, br s, 26- and 27-
H3), 3.12 (1H, dd, J=14.2, 5.1 Hz, 9a-H), 3.65 (1H, m,
22-H), 4.40 (1H, br m, 3a-H), 6.20 and 6.38 (1H and
1H, each d, J=8.2 Hz, 6- and 7-H), 7.3±7.5 (5H, br m,
Ar-H); MS m/z (rel intensity) 819 (M+, 19), 762 (48),
644 (M+±RDA, 74), 497 (61), 119 (PhNCO, 100); exact
mass calcd for C47H77N3O5Si2 819.5402, found
819.5407. Anal. calcd for C47H77N3O5Si2: C, 68.82; H,
9.47; N, 5.13. Found: C, 69.08; H, 9.51; N, 5.10.
Experimental
Chemistry. Ultraviolet (UV) absorption spectra were
recorded with a Hitachi Model 60-100 UV±vis spectro-
1
meter in the solvent noted. H nuclear magnetic reso-
nance (NMR) spectra were recorded at 500 MHz with a
Bruker AM-500 FT spectrometer in deuteriochloro-
form. Chemical shifts, (d) are reported down®eld from
internal Me4Si (d 0.00). Low- and high-resolution mass
spectra were recorded at 70 eV on a Kratos DS-50 TC
instrument equipped with a Kratos DS-55 data system.
High-resolution data were obtained by peak matching.
Samples were introduced into the ion source maintained
at 120±250ꢀC via a direct insertion probe. High-perfor-
mance liquid chromatography (HPLC) was performed
on a Waters Associates liquid chromatograph equipped
with a Model 6000A solvent delivery system, a Model 6
UK Universal injector, and a Model 486 tunable absor-
bance detector. Analytical and preparative separations
were done using the same Zorbax-Sil (Dupont, Wil-
mington, PA) 6.2 mmÂ25 cm column (4 mL/min,
1500 psi).
(b) A solution of the Grignard reagent was prepared
from bromo compound A (94 mg, 0.33 mmol) and mag-
nesium powder (9.7 mg, 0.4 mmol) in anhydrous ether
(0.5 mL) as described in the preceding procedure. Then
it was cooled to 0ꢀC and a solution of (20S)-C-22 vita-
min D aldehyde 2 (32 mg, 0.056 mmol) in anhydrous
ether (0.3 mL, cooled to 0ꢀC) was added dropwise. After
the mixture was stirred for 20 min at 0ꢀC and 75 min at
room temperature it was quenched with aqueous solu-
tion of NH4Cl (2 mL) and diluted with 4:1 (v/v) ben-
zene:ether (20 mL). The organic layer was separated,
washed with brine and diluted NaHCO3, dried
(Na2SO4) and evaporated. TLC and HPLC control
indicated formation of only one (22S) of the two possi-
ble isomers. Pure (22S)-hydroxyvitamin D derivative 4a
was obtained as a colorless oil (31 mg, 72%) by pre-
parative HPLC using 3.5% ethyl acetate in hexane as an
eluent; peak at 30 mL was collected. 4a: UV (EtOH)
Microanalyses of crystalline compounds were within
‹0.4% of the theoretical values. THF was freshly dis-
tilled before use from sodium benzophenone ketyl under
argon.
The starting (20S)-22-aldehyde 122 was obtained by
ozonolysis of PTAD adduct of 7-dehydrocholesterol
TBDMS ether.28 The synthesis of starting vitamin D
(20S)-22-aldehyde 2 was described by us previously.23
l
max 264 nm, lmin 225 nm, A264/A225=1.6; 1H NMR d
0.062 (12H, br s, 4ÂSiCH3), 0.546 (3H, s, 18-H3), 0.583
(6H, q, J=8 Hz, 3ÂSiCH2), 0.887 (18H, s, 2ÂSi-t-Bu),
0.920 (3H, d, J=6.4 Hz, 21-H3), 0.947 (9H, t, J=8 Hz,
3ÂSiCH2CH3), 1.221 and 1.225 (3H and 3H, each s, 26-
and 27-H3), 2.83 (1H, br d, J=12.5 Hz, 9b-H), 3.63 (1H,
m, 22-H), 4.19 (1H, m, 3a-H), 4.37 (1 H, m, 1b-H), 4.87
and 5.18 (1H and 1H, each s, 19-H2), 6.02 (1H, d,
J=11.2 Hz, 7-H), 6.24 (1H, d, J=11.2 Hz, 6-H); MS
m/z (rel intensity) 774 (M+, 15), 642 (43), 75 (100);
1
1. H NMR d 0.084 and 0.105 (3H and 3H, each s,
2ÂSiCH3), 0.842 (3H, s, 18-H3), 0.885 (9H, s, Si-t-Bu),
0.973 (3H, s, 19-H3), 1.146 (3H, d, J=6.8 Hz, 21-H3),
3.13 (1H, dd, J=14.4, 4.9 Hz, 9a-H), 4.39 (1H, m, 3a-
H), 6.22 and 6.36 (1H and 1H, each d, J=8.3 Hz, 6- and
7-H), 7.3±7.5 (5H, br m, Ar-H), 9.58 (1H, d, J=3.3 Hz,
22-H).

R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
2885
exact mass calcd for C45H86O4Si3 774.5834, found
774.5850.
The 20(22)-dehydro compound 5 was collected at Rv
18 mL and repuri®ed by HPLC using 3% ethyl acetate
in hexane to give an analytically pure material (2.0 mg,
50%) as a foam. Pure isomeric 22-iodo compounds 7
(0.6 mg, 13%) and 9 (1.2 mg, 26%) were collected at Rv
23 mL and 28 mL, respectively. Some unreacted tosylate
(0.6 mg, 12%) was also isolated at Rv 90 mL.
Reaction of 22-hydroxy compounds 3a and 4a with p-to-
luenesulfonyl chloride. (a) To a solution of alcohol 3a
(75 mg, 0.093 mmol) in dry pyridine (200 mL) was added
freshly recrystallized p-toluenesulfonyl chloride (49 mg,
0.26 mmol) and the reaction was allowed to proceed for
64 h at 4ꢀC. The reaction mixture was poured into ice/
saturated NaHCO3 with stirring. After 40 min of stir-
ring the aqueous suspension was extracted with 1:1 (v/v)
benzene:ether (3Â10 mL). The combined extracts were
washed with saturated NaHCO3, water, saturated
CuSO4, water again, dried (Na2SO4) and evaporated.
The oily yellowish residue (quantitative yield) was pure
enough to be used for the subsequent synthetic step. An
analytical sample of the tosylate 3b was obtained after
HPLC puri®cation using 10% ethyl acetate in hexane
1
5. H NMR d 0.087 and 0.109 (3H and 3H, each s,
2ÂSiCH3), 0.571 (6H, q, J=8 Hz, 3ÂSiCH2), 0.658 (3H,
s, 18-H3), 0.888 (9H, s, Si-t-Bu), 0.950 (9H, t, J=8 Hz,
3ÂSiCH2CH3), 0.964 (3H, s, 19-H3), 1.212 (6H, s, 26-
and 27-H3), 1.639 (3H, s, 21-H3), 3.13 (1H, dd, J=14.1,
5.1 Hz, 9a-H), 4.40 (1H, br m, 3a-H), 5.23 (1H, t,
J=7.0 Hz, 22-H), 6.19 and 6.36 (1H and 1H, each d,
J=8.3 Hz, 6- and 7-H), 7.3±7.5 (5H, br m, Ar-H); MS
m/z (rel intensity) 801 (M+, <1), 626 (M+±RDA, 100),
479 (77), 119 (PhNCO, 58); exact mass calcd for
C39H70O2Si2 (M+±RDA) 626.4915, found 626.4909.
1
(Rv 16 mL): H NMR d 0.089 and 0.108 (3H and 3H,
each s, 2ÂSiCH3), 0.537 (6H, q, J=7.9 Hz, 3ÂSiCH2),
0.723 (3H, s, 18-H3), 0.887 (9H, s, Si-t-Bu), 0.921 (9H, t,
J=7.9 Hz, 3ÂSiCH2CH3, 0.94 (3H, d, J ꢁ7 Hz, 21-H3),
0.943 (3H, s, 19-H3), 1.123 and 1.157 (3H and 3H, each
s, 26- and 27-H3), 2.43 (3H, s, Ar-CH3), 3.13 (1H, dd,
J=14.2, 5.0 Hz, 9a-H), 4.39 (1H, br m, 3a-H), 4.49
(1H, t, J=7.0 Hz, 22-H), 6.18 and 6.30 (2H, each d,
J=8.3 Hz, 6- and 7-H), 7.3±7.5 (7H, br m, Ar-H), 7.79
(2H, d, J=8.2 Hz, Ar-H).
1
7. H NMR d 0.082 and 0.104 (3H and 3H, each s,
2ÂSiCH3), 0.568 (6H, q, J=7.6 Hz, 3ÂSiCH2, 0.789
(3H, s, 18-H3), 0.885 (9H, s, Si-t-Bu), 0.947 (9H, t,
J=8 Hz, 3ÂSiCH2CH3), 0.964 (3H, s, 19-H3), 1.167
(3H, d, J=7.0 Hz, 21-H3), 1.189 and 1.194 (3H and 3H,
each s, 26- and 27-H3), 3.12 (1H, dd, J=14.0, 5.0 Hz,
9a-H), 4.39 (1H, br m, 3a-H), 4.46 (1H, d, J=11.5 Hz,
22-H), 6.21 and 6.36 (1H and 1H, each d, J=8.2 Hz, 6-
and 7-H), 7.3±7.5 (5H, br m, Ar-H); MS m/z (rel inten-
sity) 929 (M+, <1), 872 (M+±t-Bu, 1), 754 (M+±RDA,
46), 607 (40), 173 (100), 119 (PhNCO, 46); exact mass
calcd for C39H71IO2Si2 (M+±RDA) 754.4037, found
754.4036.
(b) 22-Hydroxyvitamin D (4a) (28 mg, 0.036 mmol) was
treated with p-toluenesulfonyl chloride (20 mg,
0.10 mmol) in dry pyridine (100 mL) as described above
for compound 3a. The crude oily product was puri®ed
by preparative HPLC using 2% ethyl acetate in hexane
as an eluent. Pure tosylate 4b (Rv 20 mL) was obtained
as a colorless oil. 4b: UV (hexane) lmax 264 and 223 nm,
1
9. H NMR d 0.083 and 0.106 (3H and 3H, each s,
2ÂSiCH3), 0.568 (6H, q, J=7.6 Hz, 3ÂSiCH2), 0.870
(3H, s, 18-H3), 0.885 (9H, s, Si-t-Bu), 0.943 (9H, t,
J=7.9 Hz, 3ÂSiCH2CH3), 0.969 (3H, s, 19-H3), 0.975
(3H, d, J=6.9 Hz, 21-H3), 1.191 and 1.212 (3H and 3H,
each s, 26- and 27-H3), 3.12 (1H, dd, J=14.2, 5.1 Hz,
9a-H), 4.16 (1H, m, w/2=18 Hz, 22-H), 4.40 (1H, br m,
3a-H), 6.21 and 6.37 (1H and 1H, each d, J=8.2 Hz, 6-
and 7-H), 7.3±7.5 (5H, br m, Ar-H); MS m/z (rel inten-
sity) 929 (M+, 1), 872 (M+±t-Bu, 2), 754 (M+±RDA,
80), 607 (58), 173 (100), 119 (PhNCO, 38); exact mass
calcd for C39H71IO2Si2 (M+±RDA) 754.4037, found
754.4029.
1
lmin 238 nm; H NMR d 0.059 and 0.067 (6H and 6H,
each s, 4ÂSiCH3), 0.479 (3H, s, 18-H3), 0.528 (6H, q,
J=8 Hz, 3ÂSiCH2), 0.877 (18H, s, 2ÂSi-t-Bu), 0.915
(9H, t, J=8 Hz, 3ÂSiCH2CH3), 0.929 (3H, d, J=
6.0 Hz, 21-H3), 1.103 and 1.141 (3H and 3H, each s, 26-
and 27-H3), 2.43 (3H, s, Ar-CH3), 2.80 (1H, br d,
J=12.3 Hz, 9b-H), 4.19 (1H, m, 3a-H), 4.38 (1H, m, 1b-
H), 4.58 (1H, t, J=7.1 Hz, 22-H), 4.86 and 5.19 (1H and
1H, each s, 19-H2), 5.99 and 6.22 (1H and 1H, each d,
J=11.2 Hz, 7- and 6-H), 7.32 and 7.80 (2H and 2H,
each d, J=8 Hz, Ar-H); MS m/z (rel intensity) 928 (M+,
1), 796 (2), 756 (3), 664 (3), 624 (47), 492 (27), 173 (100);
exact mass calcd for C52H92O6Si3S 928.5922, found
928.5894.
(b) To a stirred solution of vitamin D tosylate (4b)
(4.6 mg, 5 mmol) in 1:1 (v/v) acetone:2-butanone
(200 mL) was added a drop of mercury (220 mg) and
calcium carbonate (1 mg, 10 mmol) followed by sodium
iodide (3.7 mg, 25 mmol). The resultant mixture was
stirred and heated for 80 h at 45ꢀC in the dark under
argon, by which time no starting material remained.
The reaction mixture was poured into water (10 mL)
and extracted with ethyl acetate (2Â10 mL). The com-
bined organic layers were washed with 1% Na2S2O3 and
water, dried (Na2SO4) and evaporated. The workup of
the reaction mixture and following chromatographic
separations were done using subdued light in the
laboratory. The mixture of products was separated by
preparative HPLC using 0.1% ethyl acetate in hexane
Reaction of 22-tosyloxy compounds 3b and 4b with so-
dium iodide. (a) To a stirred solution of tosylate 3b
(4.9 mg, 5 mmol) in 1:1 (v/v) acetone:2-butanone
(200 mL) was added calcium carbonate (1 mg, 10 mmol)
followed by sodium iodide (3.7 mg, 25 mmol). The
resultant mixture was stirred and heated for 48 h at
45ꢀC under argon. The reaction mixture was poured
into water (10 mL) and extracted with ethyl acetate
(2Â10 mL). The combined organic layers were washed
with 1% Na2S2O3 and water, dried (Na2SO4) and eva-
porated. The residue was separated by preparative
HPLC using 5% ethyl acetate in hexane as an eluent.

2886
R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
as an eluent. The 20(22)-dehydrovitamin (6a) was col-
lected at Rv 37 mL and repuri®ed by HPLC in the same
solvent system to give an analytically pure material
(1.2 mg, 32%). The peaks of 22-iodo vitamins 10a and
8a partially overlapped (Rv 48 and 53 mL, respectively)
but rechromatography (or recycling of both peaks)
a€orded pure 22R-iodocompound 8a (0.4 mg, 8%) and
its 22S-isomer 10a (0.9 mg, 20%).
and after HPLC separation using 0.1% ethyl acetate in
hexane the 5,6-trans isomer 8b was obtained (Rv 70 mL).
Similarly, compound 10a, upon treatment with iodine
under the above conditions, was isomerized to 10b,
which can be obtained in pure form after HPLC (con-
ditions as above) separation (Rv 60 mL).
1
8b. H NMR d 0.532 (3H, s, 18-H3), 1.167 (3H, d, J=6
Hz, 21-H3), 1.185 (6H, br s, 26- and 27-H3), 2.88 (1H, br
d, J ꢁ13 Hz, 9b-H), 4.22 (1H, m, 3a-H), 4.46 (1H, d,
J=11 Hz, 22-H), 4.54 (1H, m, 1b-H), 4.95 and 4.99 (1H
and 1H, each s, 19-H2), 5.83 and 6.46 (1H and 1H, each
d, J=11 Hz, 7- and 6-H).
6a. UV (hexane) lmax 265.0 nm, lmin 228.5 nm (A265/
A228=1.7); H NMR d 0.063 and 0.072 (6H and 6H,
1
each s, 4ÂSiCH3), 0.403 (3H, s, 18-H3), 0.567 (6H, q,
J=8 Hz, 3ÂSiCH2), 0.884 (18H, s, 2ÂSi-t-Bu), 0.946
(9H, t, J=8 Hz, 3ÂSiCH2CH3), 1.264 (6H, br s, 26- and
27-H3), 1.630 (3H, s, 21-H3), 2.84 (1H, br d, J=12.8 Hz,
9b-H), 4.19 (1H, m, 3a-H), 4.38 (1H, m, 1b-H), 4.87
(1H, s, one of 19-H2), 5.19 (2H, m, one of 19-H2 and 22-
H), 6.02 and 6.24 (1H and 1H, each d, J=11.2 Hz, 7-
and 6-H); MS m/z (rel intensity) 756 (M+, 12), 624 (31),
248 (47), 117 (100); exact mass calcd for C45H84O3Si3
756.5728, found 756.5707.
1
10b. H NMR d 0.609 (3H, s, 18-H3), 0.967 (3H, d,
J=6 Hz, 21-H3), 1.201 and 1.216 (3H and 3H, each s,
26- and 27-H3), 2.88 (1H, br d, J=13 Hz, 9b-H), ca. 4.2
(2H, m, 3a- and 22-H), 4.57 (1H, m, 1b-H), 4.95 and
4.99 (1H and 1H, each s, 19-H2), 5.83 and 6.47 (1H and
1H, each d, J=11 Hz, 7- and 6-H).
Hydrolysis of tosylate 4b and 20(22)-dehydrovitamin 6a.
(a) To a solution of vitamin D tosylate (4b) (300 mg) in
benzene (20 mL) was added AG 50W-X4 ion exchange
resin (5 mg, prewashed with methanol) in anhydrous
methanol (50 mL). The mixture was vigorously stirred at
room temperature for 13 h under argon, and it was
diluted with 1:1 (v/v) ether:ethyl acetate (1 mL). The
solution was decanted and transferred to a separatory
funnel and the resin was washed with 1:1 ether:ethyl
acetate (2Â2 mL). The combined organic phase was
washed with 5 mL portions of brine and saturated
NaHCO3, dried (Na2SO4) and evaporated. Puri®cation
of the product by HPLC using 1:1 (v/v) ethyl acet-
ate:hexane as an eluent provided an oily (22R,25)-epoxy
vitamin 11 (84 mg, 63%) collected at Rv 34 mL. 11: UV
(EtOH) lmax 264.5 nm, lmin 227.5 nm (A264/A227=
8a. UV (hexane) lmax 264.0 nm, lmin 227.0 nm (A264/
1
A227=1.8); H NMR d 0.064 (12H, br s, 4ÂSiCH3),
0.517 (3H, s, 18-H3), 0.568 (6H, q, J=8 Hz, 3ÂSiCH2),
0.875 (18H, br s, 2ÂSi-t-Bu), 0.950 (9H, t, J=8 Hz,
3Â SiCH2CH3), 1.158 (3H, d, J=6.8 Hz, 21-H3), 1.179
(6H, br s, 26- and 27-H3), 2.83 (1H, br d, J ꢁ13 Hz, 9b-
H), 4.19 (1H, m, 3a-H), 4.37 (1H, m, 1b-H), 4.45 (1H, d,
J=11 Hz, 22-H), 4.86 and 5.18 (1H and 1H, each s, 19-
H2), 6.02 and 6.23 (1H and 1H, each d, J=11.2 Hz, 7-
and 6-H); MS m/z (rel intensity) 884 (M+, 20), 752 (54),
624 (24), 248 (100); exact mass calcd for C45H85O3Si3I
884.4851, found 884.4899.
10a. UV (hexane) lmax 264.0 nm, lmin 227.0 nm (A264/
1
A227=1.6); H NMR d 0.058 and 0.064 (6H and 6H,
1
each s, 4ÂSiCH3), 0.569 (6H, q, J=8 Hz, 3ÂSiCH2),
0.597 (3H, s, 18-H3), 0.875 (18H, br s, 2ÂSi-t-Bu), 0.944
(9H, t, J=8 Hz, 3ÂSiCH2CH3),0.964 (3H, d, J=6 Hz,
21-H3), 1.194 and 1.211 (3H and 3H, each s, 26- and 27-
H3), 2.82 (1H, br d, J ꢁ13 Hz, 9b-H), 4.19 (2H, m, 3a-
and 22-H), 4.37 (1H, m, 1b-H), 4.86 and 5.18 (1H and
1H, each s, 19-H2), 6.02 and 6.23 (1H and 1H, each d,
J=11.2 Hz, 7- and 6-H); MS m/z (rel intensity) 884
(M+, 21), 752 (62), 624 (22), 248 (100); exact mass calcd
for C45H85O3Si3I 884.4851, found 884.4875.
1.8); H NMR d 0.553 (3H, s, 18-H3), 0.915 (3H, d,
J=6.3 Hz, 21-H3), 1.230 and 1.240 (3H and 3H, each s,
26- and 27-H3), 2.84 (1H, br d, J=12.9 Hz, 9b-H), 4.05
(1H, m, w/2=16 Hz, 22-H), 4.23 (1H, m, 3a-H), 4.43
(1H, m, 1b-H), 5.00 and 5.32 (1H and 1H, each s, 19-H2),
6.02 and 6.38 (1H and 1H, each d, J=11.0 Hz, 7- and 6-
H); MS m/z (rel intensity) 414 (M+, 8), 396 (M+±H2O,
11), 378 (M+±2H2O, 4), 152 (13), 134 (42), 99 (100); exact
mass calcd for C27H42O3 414.3134, found 414.3132.
(b) Deprotection of hydroxyl groups in (E)-20(22)-
dehydrocompound 6a (0.93 mg) was performed under
conditions identical to those used for the conversion of
4b into 11. Crude product was puri®ed by HPLC using
1:1 (v/v) ethyl acetate:hexane as an eluent. Crystalline
20(22)-dehydrocalcitriol 6b (337 mg, 66%) was eluted at
RV 59 mL. 6b: UV (EtOH) lmax 265.0 nm, lmin 228.5 nm
(A265/A228=1.8); 1H NMR d 0.424 (3H, s, 18-H3),
1.231 (6H, br s, 26- and 27-H3), 1.651 (3H, s, 21-H3),
2.84 (1H, br d, J=12.2 Hz, 9b-H), 4.23 (1H, m, 3a-H),
4.44 (1H, m, 1b-H), 5.00 (1H, s, one of 19-H2), 5.23
(1H, t, J=7.1 Hz, 22-H), 5.33 (1H, s, one of 19-H2), 6.03
and 6.38 (1H and 1H, each d, J=11.3 Hz, 7- and 6-H);
MS m/z (rel intensity) 414 (M+, 8), 396 (M+±H2O, 15),
378 (M+±2H2O, 10), 152 (37), 134 (100); exact mass
calcd for C27H42O3 414.3134, found 414.3138.
Elimination of p-toluenesulfonate ester 3b in pyridine. A
solution of tosylate (3b) (1 mg, 1 mmol) in a dry pyridine
(200 mL) was heated under argon for 48 h at 70ꢀC. Sol-
vent was evaporated, the residue taken up in ethyl ace-
tate and the solution was washed with saturated CuSO4,
water and saturated NaHCO3, dried (Na2SO4) and
evaporated. HPLC separation of the residue using 3%
ethyl acetate in hexane yielded the ole®n 5 (0.29 mg,
35%; collected at 52 mL).
5,6-Double bond isomerization of 22-iodo vitamin D
compounds 8a and 10a. Treatment of compound 8a in
ether with a catalytic amount of iodine (2% of the
amount of 8a), while keeping the solution under di€use
daylight for 1 h, resulted in cis to trans isomerization,

R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
2887
Deprotection of hydroxyl groups in 22-iodovitamins 8a
and 10a. (a) To a solution of protected compound 8a
(1 mg) in anhydrous benzene (50 mL) was added AG
50W-X4 ion exchange resin (20 mg, prewashed with
methanol) as a slurry in anhydrous methanol (200 mL).
A drop of mercury (400 mg) was added and the resul-
tant mixture was vigorously stirred at room temperature
for 10 h in the dark under argon. The mixture was dilu-
ted with 1:1 (v/v) ether:ethyl acetate (1 mL), the solution
was decanted and transferred to a separatory funnel and
the resin was washed with 1:1 ether:ethyl acetate
(2Â2 mL). The combined organic phase was washed
with 5 mL portions of brine, 1% Na2S2O3, saturated
NaHCO3, and brine again, dried (Na2SO4) and evapo-
rated (temp. below 35ꢀC). The work up of the reaction
mixture and following chromatographic separations
were done using subdued light in the laboratory. Pre-
parative HPLC using 1:1 (v/v) ethyl acetate:hexane as
an eluent provided small amount (<50 mg) of (22S,25)-
epoxy-1a-hydroxyvitamin D3 (12) and 0.33 mg (52%) of
(22R)-iodo-1a,25-dihydroxyvitamin D3 (8c) eluted at Rv
28 and 63 mL, respectively.
(M+±HI±2 H2O, 21), 99 (100); exact mass calcd for
C27H41IO2 (M+±H2O) 524.2151, found 524.2145.
Molecular modeling: conformational search. The calcu-
lation of optimized geometries and steric energies (Es)
were carried out using the algorithm from the MM+
HyperChem (release 4.0) software package (Autodesk,
Inc., 1994). MM+ is an all-atom force ®eld based on the
MM2 functional form. Molecular mechanics studies on
the three model 8-methylene compounds 14a±c were
performed according to the protocol described below.
1. Initial geometry of each compound was obtained
by steric energy minimization performed by
HyperChem program.
2. This energy-minimized structure was used for the
generation of other conformers with the aid of
Conformational Search module of ChemPlus
(release 1.5) program (Hypercube, Inc., 1995) and
the following three-step procedure. At the ®rst
step, each of the 17,20- and 20,22-bonds in the
examined structure was subjected to rotations with
15ꢀ intervals. All resulting conformers were
energy-minimized and compared with the pre-
viously generated ones. Carbons 20, 23, 24, as well
as iodine and oxygen atoms were chosen for com-
parison in an RMS ®t. Structures were considered
to be duplicates when: (a) their steric energy dif-
ference was smaller than 0.05 kcal/mol, (b) their
rotated torsions di€ered less than 5ꢀ, and (c) RMS
error was found within 0.25 A. The least energy
structures (energy window 6 kcal/mol) obtained
after 50 iterations, were subjected to the second
step of conformational search, where rotations
(15ꢀ) of the 22,23-, 23,24-, and 24,25-bonds were
applied. After 300 iterations the lowest energy
conformers (energy window 4 kcal/mol) were sub-
jected to the third step, consisting of rotations (15ꢀ)
of the 17,20-, 20,22-, 22,23-, 23,24- and 24,25-bonds.
3. Conformers, resulting from the additional 650
iterations and falling into the energy window
1.0 kcal/mol, were chosen for a construction of the
three-dimensional dot maps indicating the position
of their oxygen atom at C-25. Since the CD-
hydrindane part of the molecules is very rigid and
does not undergo any signi®cant conformational
changes it was possible to superimpose all con-
formers on the global minimum structure using the
carbons 13, 14, 17, and 20. For this purpose RMS
Fit module of ChemPlus was used and a special
PC computer program that deleted all atoms from
the conformer being overlaid, except its oxygen at
C-25, which was then attached with a single O±O
bond to its counterpart belonging to the basic,
lowest energy conformer. Where all conformers of
each compound 14a±c were overlaid in such man-
ner, oxygen±oxygen bonds were deleted from the
resulted `polioxide' structure. The global minimum
conformers of all model compounds and positions
of the 25-oxygen in the other lowest energy con-
formers are shown in Figure 2.
8c. UV (EtOH) lmax 264 nm, lmin 228 nm, A264/
A228=1.7; 1H NMR d 0.535 (3H, s, 18-H3), 1.172 (3H,
d, J=6.5 Hz, 21-H3), 1.224 (6H, br s, 26- and 27-H3),
2.82 (1H, br d, J=13 Hz, 9b-H), 4.24 (1H, m, 3a-H),
4.45 (1H, m, 1b-H), 4.47 (1H, d, J=11.6 Hz, 22-H), 5.00
and 5.33 (1H and 1H, each s, 19-H2), 6.02 and 6.37 (1H
and 1H, each d, J=11.2 Hz, 7- and 6-H); MS m/z (rel
intensity) 542 (M+, 8), 524 (M+±H2O, 64), 506 (M+±2
H2O, 19), 414 (M+±HI, 10), 396 (M+±HI±H2O, 62),
378 (M+±HI±2 H2O, 9), 99 (100); exact mass calcd for
C27H43IO3 542.2257, found 542.2248; calcd for
C27H41IO2 (M+±H2O) 524.2151, found 524.2136.
12. 1H NMR d 0.533 (3H, s, 18-H3), 0.910 (3H, d,
J=6.8 Hz, 21-H3), 1.191 and 1.236 (3H and 3H, each s,
26- and 27-H3), 2.82 (1H, br d, J=13.3 Hz, 9b-H), 4.08
(1H, t, J=7.0 Hz, 22-H), 4.22 (1H, m, 3a-H), 4.42 (1H,
m, 1b-H), 5.00 and 5.32 (1H and 1H, each s, 19-H2),
6.01 and 6.38 (1H and 1H, each d, J=11.2 Hz, 7- and 6-
H); MS m/z (rel intensity) 414 (M+, 33), 396 (M+±H2O,
49), 378 (M+±2 H2O, 21), 152 (18), 134 (46), 99 (100);
exact mass calcd for C27H42O3 414.3134, found
414.3129.
(b) The isomeric 22-iodo vitamin 10a (1 mg) subjected to
hydrolysis and HPLC separation as described above for
compound 8a gave small amount (ca. 50 mg) of
(22R,25)-epoxy-1a-hydroxyvitamin D3 (11) and 0.3 mg
(49%) of (22S)-iodo-1a,25-dihydroxyvitamin D3 (10c),
eluted at Rv 34 and 79 mL, respectively.
10c. UV (EtOH) lmax 264 nm, lmin 227 nm, A264/
A227=2.0; 1H NMR d 0.607 (3H, s, 18-H3), 0.975 (3H,
d, J=5.9 Hz, 21-H3), 1.23 and 1.24 (3H and 3H, each s,
26- and 27-H3), 2.83 (1H, br d, J=13 Hz, 9b-H), 4.21
(2H, m, 3a- and 22-H), 4.44 (1H, m, 1b-H), 5.00 and
5.33 (1H and 1H, each s, 19-H2), 6.02 and 6.38 (1H and
1H, each d, J=11.3 Hz, 7- and 6-H); MS m/z (rel
intensity) 542 (M+, 2), 524 (M+±H2O, 31), 506 (M+±2
H2O, 17), 414 (M+±HI, 8), 396 (M+±HI±H2O, 54), 378
4. In the manner analogous to that described above, all
conformers of the model compounds 14a±c, falling

2888
R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
into 1 kcal/mol energy window, were overlaid on
1,2-propanediol:ethanol by intraperitoneal injection
each day for 7 days and they were sacri®ced 24 h after
the last dose. The rats were sacri®ced under ether anes-
thesia by decapitation; their blood and intestines were
collected and used immediately to determine intestinal
calcium transport and serum calcium concentration.
Calcium was determined using the Calcette automatic
calcium titrator (Precision Systems, Inc., Natick, MA)
and intestinal calcium transport by the everted intestinal
sac method using the proximal 10 cm of intestine as
described earlier.47 Statistical analysis was by the Stu-
dent's t-test.49 Intestinal calcium transport is expressed
as serosal:mucosal ratio of calcium in the sac to the
calcium in the ®nal incubation medium, or S/M. Bone
calcium mobilization represents the rise in serum cal-
cium of the rats maintained on a very low calcium diet.
In that measurement, the rise in serum calcium must
arise from bone and hence is a determination of bone
calcium mobilization.
the carbon skeleton of the global minimum con-
former of 14a (steric energy 33.09 kcal/mol). The
result of such superimposition is shown in Figure 3.
Biological studies 1. Measurement of binding to the por-
cine intestinal vitamin D receptor. Porcine intestinal
nuclear extract was prepared as described earlier.46 It
was diluted, and 0.1 mg of protein (200 fmol of binding
activity) in 100 mL was used in each tube; 10,000 cpm of
1a,25-(OH)2[26,27-3H]D3 was added in 2.0 ml of etha-
nol. To this was added either standard radioinert 1a,25-
(OH)2D3 at various concentrations or the indicated
analogue at various concentrations in 5 mL of ethanol.
The mixture was incubated at room temperature for 4 h
on a shaker and then 100 mL of hydroxyapatite (50%
slurry) added. The sample was vortexed at 5 min inter-
vals for 15 min on ice. The hydroxyapatite was then
washed 3 times by adding 0.5 mL TE 5% TritonÂ100,
centrifuging at 200 g for 5 min, and aspirating the
supernatant. The radioactivity bound to the hydro-
xyapatite was determined by liquid scintillation count-
ing in Bio-Safe II scintillation ¯uid. The displacement
experiments were carried out in triplicate on two di€er-
ent occasions with identical results. The relative VDR
binding anities of the analogues were determined by
comparison of the compound concentrations required
for 50% displacement of the radiolabeled 1a,25-
(OH)2D3 from the receptor protein.
Acknowledgements
We acknowledge the assistance of Rowland Randall in
recording the mass spectra. The work was supported by
a grant from the Wisconsin Alumni Research Founda-
tion and a grant from the National Foundation for
Cancer Research.
References and Notes
Measurement of cellular di€erentiation. Human leuke-
mia HL-60 cells, originally obtained from ATTC, were
plated at 2Â105 cells/plate, incubated in Eagle's mod-
i®ed medium as described previously.45 The compounds
tested were added in the indicated concentrations in
0.05 mL of ethanol so that the ethanol concentration
never exceeded 1%. The incubation was carried out for
4 days and at the end of the fourth day, superoxide
production was measured by nitro blue tetrazolium
(NBT) reduction. The number of cells containing intra-
cellular black-blue formazan deposits was determined
by light microscopy using a hemacytometer. At least
200 cells were counted in duplicate per determination.
Percentage di€erentiation represents percentage cells
providing NBT reduction appearance. The experiment
was repeated 3 times. The relative di€erentiation activ-
ities of the analogues were determined by comparison of
the compound concentrations capable of inducing 50%
maturation according to the assay. This method is
described in detail elsewhere.46
1. Kodicek, E. Lancet 1974, 1, 325.
2. DeLuca, H. F. Fed. Proc. 1974, 33, 2211.
3. DeLuca, H. F. FASEB J. 1988, 2, 224.
4. Jones, G.; Vriezen, D.; Lohnes, D.; Palda, V.; Edwards, N.
S. Steroids 1987, 49, 29.
5. Ikekawa, N. Med. Res. Rev. 1987, 7, 333.
6. DeLuca, H. F.; Schnoes, H. K. Annu. Rep. Med. Chem.
1984, 19, 179.
7. Zhu, G.-D.; Okamura, W. H. Chem. Rev. 1995, 95, 1877.
8. DeLuca, H. F.; Paaren, H. E.; Schnoes, H. K. Top. Curr.
Chem. 1979, 83, 1.
9. Sicinski, R. R.; Tanaka, Y.; Phelps, M.; Schnoes, H. K.;
DeLuca, H. F. Anal. Biochem. 1987, 161, 96.
10. Jones, G.; Schnoes, H. K.; DeLuca, H. F. Biochemistry
1975, 14, 1250.
11. Suda, T.; DeLuca, H. F.; Schnoes, H. K.; Blunt, J. W.
Biochemistry 1969, 8, 3515.
12. Paaren, H. E.; Fivizzani, M. A.; Schnoes, H. K.; DeLuca,
H. F. Proc. Natl. Acad. Sci. USA 1981, 78, 6173.
13. Napoli, J. L.; Mellon, W. S.; Fivizzani, M. A.; Schnoes, H.
K.; DeLuca, H. F. Biochemistry 1980, 19, 2515.
14. Napoli, J. L.; Fivizzani, M. A.; Hamstra, A. J.; Schnoes,
H. K.; DeLuca, H. F. Anal. Biochem. 1979, 96, 481.
15. Frolik, C. A.; DeLuca, H. F. J. Clin. Invest. 1973, 52, 543.
16. Minnghetti, P. P.; Norman, A. W. FASEB J. 1988, 2, 3043.
17. Bouillon, R.; Okamura, W. H.; Norman, A. W. Endocr.
Rev. 1995, 16, 200.
Measurement of intestinal calcium transport and bone
calcium mobilization. Weanling male rats from the low
vitamin D colony were purchased from the Sprague±
Dawley Co. (Indianapolis, IN) and fed the vitamin D-
de®cient diet,47 containing 0.47% calcium and 0.3%
phosphorus, for 1 week. They were then switched to the
reduced calcium diet (0.02% Ca) for an additional 2
weeks. These animals have no detectable levels of 25-
OH-D3 or 1a,25-(OH)2D3 in their plasma as measured
by methods described previously.48 The rats were given
the indicated doses of compounds in 0.1 mL of (95:5)
18. Ikekawa, N. Med. Res. Rev. 1987, 7, 333.
19. Paaren, H. E.; Mellon, W. S.; Schnoes, H. K.; DeLuca, H.
F. Bioorg. Chem. 1985, 13, 62.
20. Some recent ®ndings indicate that diminished VDR bind-
ing by a vitamin D analogue does not mean that it will neces-
sarily have diminished antiproliferative/prodi€erentiation
potency, see Peleg, S.; Nguen, C.; Woodard, B. T.; Lee, J.-K.;
Posner, G. H. Mol. Endocrinol. 1998, 12, 525.

R. R. Sicinski, H. F. DeLuca / Bioorg. Med. Chem. 7 (1999) 2877±2889
2889
21. Piatak, D. M.; Wicha, J. Chem. Rev. 1978, 78, 199.
22. Yamada, S.; Nakayama, H.; Takayama, H. Tetrahedron
Lett. 1981, 22, 2591.
23. Kutner, A.; Perlman, K. L.; Lago, A.; Sicinski, R. R.;
Schnoes, H. K.; DeLuca, H. F. J. Org. Chem. 1988, 53, 3450.
24. Ikekawa, N. Bioorg. Med. Chem. Lett. 1993, 3, 1789.
25. Eguchi, T.; Yoshida, M.; Ikekawa, N. Bioorg. Chem. 1989,
17, 294.
38. Yamada, S.; Yamamoto, K.; Masuno, H.; Ohta, M. J.
Med. Chem. 1998, 41, 1467.
39. Yamamoto, K.; Sun, W. Y.; Ohta, M.; Hamada, K.;
DeLuca, H. F.; Yamada, S. J. Med. Chem. 1996, 39, 2727.
40. Duax, W. L.; Wawrzak, Z.; Grin, J. F.; Cheer, C. In
Biology of Cholesterol, Yeagle Ed., CRC: Boca Raton, 1988,
p. 1.
41. Toan, T.; Ryan, R. C.; Simon, G. L.; Calabrese, J. C.;
Dahl, L. F.; DeLuca, H. F. J. Chem. Soc., Perkin Trans. 2
1977, 393.
26. Hirano, Y.; Ikekawa, N.; Tanaka, Y.; DeLuca, H. F.
Chem. Pharm. Bull. 1981, 29, 2254.
27. Yamamoto, K.; Takahashi, J.; Hamano, K.; Yamada, S.;
Yamaguchi, K.; DeLuca, H. F. J. Org. Chem. 1993, 58, 2530.
28. Barton, D. H. R.; Shioiri, T.; Widdowson, D. A. J. Chem.
Soc. (C) 1971, 1968.
29. Fall, Y.; Torneiro, M.; Castedo, L.; Mourino, A. Tetra-
hedron Lett. 1992, 33, 6683.
30. Landini, D.; Quici, S; Rolla, F. Synthesis 1975, 430.
31. Corey, E. J.; Howell, M. G.; Boston, A.; Young, R. L.;
Sneen, R. A. J. Am. Chem. Soc. 1956, 78, 5036.
32. Schow, S. R.; McMorris, T. C. J. Org. Chem. 1979, 44, 3760.
33. Byon, C. Y.; Gut, M. J. Org. Chem. 1976, 41, 3716.
34. Schmit, J. P.; Piraux, M.; Pilette, J. F. J. Org. Chem. 1975,
40, 1586.
35. Narwid, T. A.; Cooney, K. E.; Uskokovic, M. R. Helv.
Chim. Acta 1974, 57, 771.
36. Verloop, A.; Koevoet, A. L.; Van Moorselaar, R.;
Havinga, E. Rec. Trav. Chim. 1959, 78, 1004.
37. Verloop, A.; Koevoet, A. L.; Havinga, E. Rec. Trav. Chim.
1955, 74, 1125.
42. These ®ndings are in good agreement with the published
results of MM2PP calculations of the same 25-hydroxy-8-
methylene derivative 14a, see ref. 27.
43. Midland, M. M.; Plumet, J.; Okamura, W. H. Bioorg.
Med. Chem. Lett. 1993, 3, 1799.
44. Okamura, W. H.; Palenzuela, J. A.; Plumet, J.; Midland,
M. M. J. Cell. Biochem. 1992, 49, 10.
45. Dame, M. C.; Pierce, E. A.; Prahl, J. M.; Hayes, C. E.;
DeLuca, H. F. Biochemistry 1986, 25, 4523.
46. Ostrem, V. K.; Lau, W. F.; Lee, S. H.; Perlman, K.; Prahl,
J.; Schnoes, H. K.; DeLuca, H. F.; Ikekawa, N. J. Biol. Chem.
1987, 262, 14164.
47. Suda, T.; DeLuca, H. F.; Tanaka, Y. J. Nutr. 1970, 100,
1049.
48. Uhland-Smith, A.; DeLuca, H. F. J. Nutr. 1993, 123,
1777.
49. Snedecor, G. W.; Cochran, W. G. In Statistical Methods,
6th ed.; Iowa State University Press: Ames, IA, 1967.