Characterization of a novel analogue of 1α,25(OH)2-vitamin D3 with two side chains: Interaction with its nuclear receptor and cellular actions

Article abstract of DOI:10.1021/jm0000160

The hormone 1α,25(OH)₂-vitamin D₃ (125D) binds to its nuclear receptor (VDR) to stimulate gene transcription activity. Inversion of configuration at C-20 of the side chain to generate 20-epi-1α,25(OH)₂D₃ (20E-12

Full text of DOI:10.1021/jm0000160

J . Med. Chem. 2000, 43, 2719-2730
2719
Ch a r a cter iza tion of a Novel An a logu e of 1r,25(OH)2-Vita m in D3 w ith Tw o Sid e
Ch a in s: In ter a ction w ith Its Nu clea r Recep tor a n d Cellu la r Action s
Anthony W. Norman,*,†,§ Percy S. Manchand,| Milan R. Uskokovic,| William H. Okamura,‡ J anet A. Takeuchi,‡
J une E. Bishop,† J un-Iichi Hisatake, H. Phillip Koeffler, and Sara Peleg#
Departments of Biochemistry and Chemistry and Division of Biomedical Sciences, University of California,
Riverside, California 92521, Hoffmann-La Roche Inc., Nutley, New J ersey 07110, Division of Hematology/ Oncology,
Cedars-Sinai Medical Center, UCLA School of Medicine, 8700 Beverly Boulevard, B-208, Los Angeles, California 90048, and
Department of Medical Specialties, The University of Texas M. D. Anderson Cancer Center, Houston, Texas 77030
Received J anuary 18, 2000
The hormone 1R,25(OH)2-vitamin D3 (125D) binds to its nuclear receptor (VDR) to stimulate
gene transcription activity. Inversion of configuration at C-20 of the side chain to generate
20-epi-1R,25(OH)2D3 (20E-125D) increases transcription 200-5000-fold over 125D with its 20-
normal (20N) side chain. This enhancement has been attributed to the VDR ligand-binding
domain (LBD) having different contact sites for 20N and 20E side chains that generate different
VDR conformations. We synthesized 1R,25-dihydroxy-21-(3-hydroxy-3-methylbutyl)vitamin D3
(Gemini) with two six-carbon side chains (both 20N and 20E orientations). Energy minimization
calculations indicate the Gemini side chain possesses significantly more energy minima than
either 125D or 20E-125D (2346, 207, and 127 minima, respectively). We compared activities of
125D, 20E-125D, and Gemini, respectively, in several assays: binding to wild-type (100%, 147%,
and 38%) and C-terminal-truncated mutant VDR; transcriptional activity (of the transfected
osteopontin promoter in ROS 17/2.8 cells: ED50 10, 0.005, and 1.0 nM); mediation of
conformational changes in VDR assessed by protease clipping (major trypsin-resistant fragment
of 34, 34, and 28 kDa). For inhibition of cellular clonal growth of human leukemia (HL-60)
and breast cancer (MCF7) cell lines, the ED50(125D)/ED50(Gem) was respectively 380 and 316.
We conclude that while Gemini readily binds to the VDR and generates unique conformational
changes, none of them is able to permit a superior gene transcription activity despite the
presence of a 20E side chain.
In tr od u ction
to employ libraries of analogues that display unique or
selective properties in comparison to 125D5-7 and (b)
to dissect and define how 125D and analogues interact
with the VDR to cause conformational changes in the
receptor protein6,7 that enhance VDR heterodimeriza-
tion with the retinoid X receptor (RXR). In turn, this
enhances heterodimer binding to other transcriptionally
required proteins, such as co-activators,8-10 and to DNA.
These structure-function studies have shown that it is
possible to design analogues of 125D that have both
favorable biological responses in selected target organs
and minimal calcemic side effects.1
An important focus of structure-function studies of
synthetic ligands for the VDR is the stereocenter at
carbon 20 of the side chain.11,12 Analogues with a 20-
epi (20E) [e.g. 20-epi-1R,25(OH)2-vitamin D3 (20E-
125D)] rather than the 20-normal (20N) orientation (see
Figure 1) have antiproliferative activities 200-5000-fold
greater than 125D when assayed in a variety of cell
types, including human leukemic HL-60 cells, rat os-
teoblastic ROS 17/2.8 cells, and human keratino-
cytes.5,13,14 It was originally postulated that the potency
of these analogues could be explained because: (i) their
affinity for the VDR was significantly higher than that
of 125D, (ii) their uptake into target cells was more
efficient, or (iii) the catabolism rate of the 20E analogues
was slower than that of 125D.11,12 However, studies from
our15,16 and other17,18 laboratories do not support these
possible explanations.
There are a wide range of biological effects which
are mediated by the hormonally active form of vita-
min D, namely 1R,25(OH)2-vitamin D3 (125D) through
the vitamin D endocrine system. This endocrine sys-
tem includes not only the classical arena of Ca2+
homeostasis (intestine, kidney, and bone) but also
actions on the immune system, the pancreatic B cell
and other endocrine cells, keratinocytes, the processes
of cell differentiation of normal cells, and inhibition of
proliferation of many types of neoplastic cells.1 Cen-
tral to the generation of these far-reaching biological
effects is the interaction of 125D with its nuclear
receptor (VDR), to produce a competent ligand-receptor
complex that is able to selectively interact with a VDR
response element present on the promoters of genes
which are transcriptionally regulated by the steroid
hormone.2,3 The VDR is present in over 30 different cell
types that participate in the vitamin D endocrine
system.4
The two principal approaches to studying 125D regu-
lation of gene transcription have been the following: (a)
* To whom correspondence should be addressed. Phone: (909) 787-
4777. Fax: (909) 787-4784. E-mail: Norman@ucrac1.ucr.edu.
† Department of Biochemistry, University of California.
‡ Department of Chemistry, University of California.
§ Division of Biomedical Sciences, University of California.
| Hoffmann-La Roche Inc.
UCLA School of Medicine.
The University of Texas M. D. Anderson Cancer Center.
#
10.1021/jm0000160 CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/21/2000

2720 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14
Norman et al.
F igu r e 1. Structural formulas of 1R,25(OH)2-vitamin D3 (125D), 20-epi-1R,25(OH)2-vitamin D3 (20E-125D), and the two-side-
chain analogue of 125D, 1R,25-dihydroxy-21-(3-hydroxy-3-methylbutyl)vitamin D3 (Gemini).
Recent studies from our laboratories suggest that the
primary basis by which 20E analogues achieve potency
superior to 125D in stimulating gene transcription lies
in their interaction with the VDR to generate unique
conformational changes of the receptor protein that
enhance its subsequent interactions with other tran-
scription factors, particularly the RXR.15,16 The principal
evidence supporting this conclusion was the observation
that three 20E analogues when bound in vitro to [35S]-
VDR resulted in trypsin sensitivity patterns different
from those generated by the equivalent 20N analogues
bound to the [35S]VDR.15 This was interpreted to be due
to different receptor conformations being induced by the
ligands. These same three 20E analogues, in comparison
to their 20N analogues, also conferred upon the VDR
an enhanced ability to form heterodimers with RXR, as
assessed by electromobility shift assays, and to activate
transcription via two different promoter constructs with
vitamin D-response elements (VDRE), which were
transfected into ROS 17/2.8 cells. Further support for
the differential interaction of 20E versus 20N analogues
with the VDR was obtained by preparation of several
mutations of the C-terminal region of the VDR. This
facilitated mapping of specific amino acids that interact
with the ligand and resulted in the conclusion that the
ligand-binding domain (LBD) of the VDR has multiple
and different contact sites for the two families of side-
chain-modified ligands.16
To further probe the role of the side chain of 125D in
its interactions with the VDR, we have synthesized a
unique analogue (Gemini) which has the same two six-
carbon side chains at C-20, one with a 20N orientation
and one with a 20E orientation (see Figure 1). Our
objective was to determine the various properties of this
two-side-chain analogue in interacting with the nuclear
VDR both in vitro and in host cells containing the
receptor. We report here that Gemini productively
interacts with the VDR to generate a conformational
shape different from that produced by either 125D or
20E-125D. Interestingly, despite this difference, the
biological potency of the Gemini-VDR ligand-receptor
complex is very similar to that of 125D in stimulating
gene transcription.
protected olefin 3 is easily accessible in high yield from
the Inhoffen-Lythgoe diol via 1 and then the iodide 2.
A double-ene reaction results in the conjugated diester
4. This process is a stepwise process wherein a mono-
ene intermediate (not shown) is rapidly formed at an
early stage of the reaction. Compound 4 undergoes a
smooth hydrogenation with palladium on carbon cata-
lyst, and the saturated analogue 5 is converted cleanly
to ditertiary alcohol 6. The following steps via 8 and 10
in the synthesis of the final product Gemini (Figure 1)
with the two side chains of 125D are reminiscent of the
protocol used for the synthesis of 125D.19 The structures
of Gemini and the various intermediates shown in
Figure 2 were fully confirmed by spectral and analytical
analyses, including NMR and mass spectrometry.
The eight-carbon side chains of both 125D and 20E-
125D are known to be highly conformationally flexible.20
The side chains of the seco-steroids 125D and 20E-125D
assume distinct orientations in their global or minimal
energy conformation; the 20N side chain is oriented to
the ‘northeast’ while the 20E side chain is oriented to
the ‘northwest’.21-23 Figure 3 provides a conformational
analysis using the previously described dot map formal-
ism21,24 of the individual side-chain arms of Gemini in
comparison to the side chain of 125D (C/C′ vs A/A′) and
20E-125D (D/D′ vs B/B′). Within a 3-kcal/mol window,
the double-side-chain analogue (C/C′ and D/D′) displays
2346 minima, whereas the parent 125D and 20E-125D
exhibit only 207 and 127 conformers, respectively. These
computations reveal the apparently greater flexibility
of the individual side-chain arms of the double-side-
chain analogue (compare the greater density of dots or
conformers in C/C′ and D/D′ to those in A/A′ and B/B′,
respectively).
The two hydroxyl groups of the double-side-chain
analogue (C/C′ and D/D′ in Figure 3) occupy a very
similar volume in space as their single 20N and 20E
side-chain counterparts despite the differences in their
global minimum orientations. The differences in their
global minima are particularly evident upon inspection
of the complementary side views (C′ vs A′ and D′ vs B′).
Both hydroxyl groups of the side-chain arms of Gemini
are localized above and below the plane of the CD ring.
Like the side chain of the hormone 125D, the 20N side-
chain portion of the double-side-chain analogue occupies
a volume largely “east to northeast” of the CD ring; and
like the side chain of the unnatural 20E-125D, the 20-
epimer arm of the double-side-chain analogue resides
in the “northwest” region. The two hydroxyl groups are
localized in different, but not exclusively different,
regions within 3 kcal/mol of the global minimum.
Resu lts
Figure 1 presents the structures of 125D, 20E-125D,
and 1R,25(OH)2-21-(3-hydroxy-3-methylbutyl)vitamin
D3 (Gemini) which are profiled in this report.
Figure 2 presents a summary of the overall synthetic
route for the chemical synthesis of Gemini, a novel
analogue of 125D with two six-carbon side chains. The

Two-Side-Chain 1R,25(OH)2D Interaction with VDR
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14 2721
F igu r e 2. Summary of the chemical synthesis of Gemini. The synthetic scheme shows the transformation of alcohol 1 to protected
ketone 9 followed by its Horner coupling to A-ring phosphine oxide 10, which is then transformed to the two-side-chain analogue
Gemini (structure shown in Figure 1). The chemical reagents, except where noted, are as follows: (a) Ph3P, I2 (94%); (b) KOtBu,
DMSO (85%); (c) ethyl propiolate, EtAlCl2 (32%); (d) H2, Pd/carbon (94%); (e) MeMgBr (93%); (f) H2SiF6, MeCN (85%); (g) PDC
(94%); (h) TMS-imidazole (92%).
Previous studies from these laboratories have estab-
lished that 20E analogues are 50-3000-fold more potent
than 125D in stimulating VDR-mediated gene tran-
scription;15,16 also the 20E analogues induce distinct
conformational changes in the VDR as judged by trypsin
protease sensitivity.15 Figure 4 compares the potency
of 125D, 20E-125D, and Gemini to transcriptionally
activate the VDRE on the osteopontin promoter in a
thymidine kinase promoter/growth hormone reporter
construct. The fusion gene containing the VDRE was
transfected into ROS 17/2.8 cells, and the growth
hormone reporter-protein production was measured
after treatment for 48 h with increasing concentrations
of the three ligands. The relative order of potency for
activating the osteocalcin promoter was 20E-125D >
Gemini > 125D.
Figure 5 presents results of ligand-exchange rate
studies in which recombinant hVDR transfected in
COS-1 cells were treated with ligand (125D, 20E-125D,
or Gemini) and then lysed. The cell lysates were
incubated in vitro with [3H]125D. It is apparent that
there is both a difference in the exchange properties and
stability of VDR occupied by 20E-125D as compared
with either 125D or the two-side-chain analogue Gemini.
Only 20% of the 20E-125D ligand was displaced by [3H]-
125D after a 60-min incubation, whereas approximately
55% of Gemini or 125D was displaced. Thus, the
exchange properties of Gemini, which has both 20N and
a 20E side chains, are similar to that of 125D, which
has only the 20N side chain.
terns when it is occupied with ligands with 20N or 20E
side chains.16,16 Figure 6 presents results of the protease
sensitivity studies on ligand-occupied VDR. In vitro
synthesized [35S]VDR was exposed to increasing con-
centrations of either 125D, 20E-125D, or Gemini fol-
lowed by treatment with trypsin and gel electrophesis
(SDS-PAGE) (see Figure 6). The proteolytic fragment
pattern resulting from trypsin treatment of the occupied
VDR was different for the three seco-steroids. Oc-
cupancy of the VDR with 125D or 20E-125D resulted
in a major fragment at 34 kDa, but in the presence of
20E-125D an additional fragment was seen at 32 kDa
(see Figure 6, panel A); this result is in agreement with
our previous report.15 However, when the VDR was
occupied with Gemini, a significant major fragment was
seen at 28 kDa with no evidence for the 34-kDa
fragment. Further, the concentration of ligand required
to achieve 50% of the maximal concentration of the
major proteolytic fragment (ED50) was different for each
seco-steroid (see Figure 6, panel B, and Table 1). The
ED50 for 125D was 45-fold higher than that for 20E-
125D, and the ED50 for Gemini was 30-fold higher than
that for 125D and 1400-fold higher than that for 20E-
125D. We conclude that analogue Gemini generates a
unique conformational change in the VDR different from
that achieved by either 125D or 20E-125D.
Figure 7 presents the consequences of two separate
mutations of the wild-type (WT) VDR on the ligand-
induced sensitivity of [35S]VDR to protease clipping by
trypsin under in vitro conditions. Our previous results
indicated that truncation of the AF-2 domain (at the
extreme carboxy terminus of the VDR) by introduction
Previous studies have established that the VDR
displays different trypsin-mediated fragmentation pat-

2722 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14
Norman et al.
F igu r e 3. Stereoscopic dot maps of the side chains of 125D, 20E-125D, and the two-side-chain Gemini for a 3-kcal/mol window.
Dot maps are used to depict the volume in space that the side chain can occupy. The dots locate possible occupation sites of the
side-chain hydroxyl group, which is considered to be an important factor in binding of the ligand to various host proteins. Using
the dot map representations, only a limited number of drawings are needed to display a large number of energy-minimized side-
chain conformers. Dot maps were constructed as described in Materials and Methods. Panels A and A′ pertain to the CD side-
chain fragment model system of 125D. B and B′ are a similar representation of the side chain of 20E-125D. Panels C, C′, D, and
D′ portray separately the individual arms of the double-side-chain analogue Gemini. Panels C and C′ reflect only the 20N side-
chain portion of Gemini, while panels D and D′ depict only the 20E side-chain portion of Gemini. Note that A-D are dot maps
viewing the natural CD fragment from the top face, while A′-D′ are views from the bottom edge of the approximate plane defined
by the CD ring.
of a stop codon at residue 390 or imposition of a double
mutation (V421M/F422A) in the AF-2 domain resulted
in a very significant loss of binding by the altered VDR
for 125D but not 20E-125D.16 Figure 7 presents a
comparison of the pattern of the major proteolytic
fragments which result from occupying, respectively, the
WT VDR or the two mutants (V421M/F422A and 390/
TGA) with 125D, 20E-125D, or Gemini. The major
proteolytic fragments were 34, 32, and 28 kDa (WT
VDR), 28, 32 + 28, and 28 kDa (V421M/F422A), and
26, 26, and 28 + 26 kDa (390/TGA), respectively, for
125D, 20E-125D, and Gemini. Thus, Gemini clearly
produces a unique 28-kDa fragment in the WT VDR
similar to that shown in Figure 6. This same 28-kDa
fragment is also significantly imposed by Gemini, but
not 125D, on the two mutants, indicating that it is likely
the 20E side chain of Gemini is important for the
stabilization of the 28-kDa fragment. For the V421M/
F422A mutant VDR, the pattern of proteolytic frag-
ments generated by occupancy of the VDR with 20E-
125D and Gemini were also different, again indicating
the unique response of this VDR construct to Gemini.
F422A mutant VDR, a result consistent with our earlier
results.16 Also, as noted earlier, 20E-125D can bind
virtually equivalently to the WT VDR and the V421M/
F422A mutant VDR. In a similar fashion, Gemini bound
to the V421M/F422A mutant only marginally less than
to the WT VDR. These results suggest that the 20E side
chain of Gemini is able to achieve the same general
affinity of binding as the 20E-125D, although as indi-
cated in Figure 7 the proteolytic fragment pattern is
markedly different from that achieved by 20E-125D
(only 1 fragment at 28 kDa vs fragments at 34, 32, and
28 kDa).
Figure 9 presents a comparison of the dose-dependent
efficacy of 125D and Gemini with respect to their
inhibition of the growth of human leukemic (HL-60),
breast (MCF7), and prostate (LNCaP) cancer cell lines.
The clonal growth of both the HL-60 and MCF7 cell
lines was inhibited via exposure to either 125D or
Gemini; however, the ED50 for Gemini was approxi-
mately 300 times lower than that of 125D (see Table
1). The prostate cell line only presented a blunted
inhibition of clonal growth response for both 125D and
Gemini.
Figure 8 presents results of competition of 125D and
its analogues 20E-125D and Gemini for binding to WT
VDR and its AF-2 mutant (V421M/F422A). In compari-
son to the WT VDR, 125D binds poorly to the V421M/
The relative affinities of the three seco-steroids for
binding to the chick intestinal VDR and human vitamin
D binding protein (DBP) were determined via standard

Two-Side-Chain 1R,25(OH)2D Interaction with VDR
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14 2723
only 2.5%. As reported previously, the VDR RCI for 20E-
125D is 147% or 1.5-fold greater than the reference
125D.1 However, for DBP the RCI for analogue 20E-
125D was only 2.6%. Thus the LBD of the VDR is
significantly more able than the LBD of DBP to accom-
modate either the 20E side chain or the two side chains.
Discu ssion
Over the past decade there have been many reports
describing structure-function relationships between
various structural functionalities of the molecule 125D
and a variety of biological responses (see review covering
278 analogues and 820 structural modifications26). This
information has been accumulated through synthesis
and biological and biochemical testing of a wide variety
of side-chain-modified analogues of 125D; these have
included lengthening or shortening the eight-carbon side
chain and introduction of double or triple bonds,27
allene,28 or aromatic ring29 functionalities along the
chain. While some of these analogues have displayed a
favorable profile of selective biological responses, none
have achieved the attention generated via inversion of
the C-20 chiral center to produce a family of 20E
analogues of 125D.11 Some of these 20E analogues
achieved separation of biological responses such that
they have antiproliferative12,18 or transcriptional activi-
ties17,30 200-5000-fold more potent than 125D but with
a reduced calcemic activity.
F igu r e 4. Transcriptional activity of 125D, 20E-125D, and
the two-side-chain analogue Gemini. ROS 17/2.8 cells were
transfected by the DEAE-dextran method with a thymidine
kinase/growth hormone (TK/ GH) fusion gene containing the
osteopontin VDRE (opVDRE). The seco-steroids were present
in the medium plus serum for 48 h prior to quantitation of
the growth hormone by radioimmunoassay. Each point of the
dose-response curve is the average of duplicate transfections.
The results shown are representative of 3-5 transfection
experiments. The ordinate values are normalized to the
maximum growth hormone generated by 1R,25(OH)2D3 at 100
nM: 0 ) 125D; b ) 20E-125D; 9 ) Gemini.
It is noteworthy that the unusual potency of the 20E
analogues is not directly explained by their affinity for
the VDR. For example, 20E-125D and 20-epi-22-oxa-
24a,26a,27a-trihomo-1R,25(OH)2D3 only bind to the
VDR ≈ 147% or 120%, respectively, as well as 125D,11
yet their antiproliferative activities are 200-5000-fold
greater than that of 125D. The LBD of the 50-kDa VDR
is a 37-kDa polypeptide encompassing two-thirds of the
receptor molecule.2 A three-dimensional molecular model
of the VDR LBD has been prepared using the atomic
coordinates of the thyroid receptor.31 Our studies have
shown that the C-terminal portion of the LBD, encoding
the transcription AF-2 domain of the VDR, has multiple
and different contact points for the 20N versus 20E
family of analogues.16 Thus, a 20E analogue when bound
to the VDR generates a different overall conformation
from that achieved by binding a 20N analogue. As a
consequence, the 20E analogue complex with VDR is
more stable, and the three-dimensional structure of the
receptor with a 20E ligand is apparently more effective
at forming heterodimers with RXR, interacting with
other transcriptionally important proteins, such as co-
activators,9,32 and generating a transcriptionally active
complex. The overall result is a significantly greater
transcriptional potency through the osteopontin or
osteocalcin VDRE.
F igu r e 5. Exchange assay of 125D, 20E-125D, and the two-
side-chain analogue Gemini with [3H]125D. The exchange
assay was conducted in COS-1 cells as described under
Materials and Methods. Each ligand at 10-8 M was incubated
with the cells for 1 h in a serum-free medium. Then the
medium was removed, and the cells were washed, homog-
enized, and incubated with 0.2 pmol of [3H]1,25(OH)2D3 with
or without a 100-fold excess of nonradioactive 125D. Each point
shown is the average of duplicate samples; the results shown
are representative of four assays: 0 ) 125D; b ) 20E-125D;
9 ) Gemini.
To further study the relative importance of the 20N
versus 20E side chain, we conceptualized synthesis of
a 125D analogue that simultaneously possessed two side
chains attached to C-20. We wished to learn whether
the VDR LBD can accommodate an analogue with two
side chains and whether the binding properties and the
biological properties of Gemini are more 20E-like or
20N-like. Here we report the chemical synthesis and
biological characterization of a novel analogue of 125D,
which has two six-carbon side chains, one in the 20N
steroid competition assays (see Table 1). In these assays
the relative competitive index (RCI) is calculated.25 By
definition, the RCI of 125D is set to 100%. The main
point of interest was to ascertain whether the two-side-
chain Gemini could bind to the VDR or DBP. For
Gemini, the VDR RCI was 38%, while the DBP RCI was

2724 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14
Norman et al.
A
B
F igu r e 6. Ligand-induced sensitivity of the WT VDR to in vitro treatment with trypsin. (A) In vitro translated WT VDR labeled
with [35S]methionine was incubated without or with increasing concentrations of the seco-steroids 125D, 20E-125D, and Gemini
before digestion with trypsin. The digestion products were analyzed by PAGE and autoradiography. The proteolytic products are
indicated by arrows. The fragment sizes are 34, 32, and 28 kDa for fragments 1-3, respectively. The large arrows indicate molecular
weight markers at 46, 30, and 21.5 kDa. Lanes marked R contained control [35S]VDR. Analogue treatment concentrations are
given above their respective lanes as -log M. (B) Each lane of the results shown in panel A was scanned with a densitometer,
and the results are presented. The calculated ED50 values ( SD for 125D, 20E-125D, and Gemini were 5.9 ( 0.6, 0.13 ( 0.09, and
180 ( 50 nM, respectively, for fragments 34, 34, and 28 kDa.
orientation and the other in the 20E orientation. The
side-chain analogue, 1R,25-dihydroxy-21-(3-hydroxy-3-
methylbutyl)vitamin D3 (designated as Gemini), was
synthesized in seven steps from a known starting
material; a preliminary report has appeared.33 Inde-
pendently, Calverley et al. also presented a brief report
of the chemical synthesis of the same analogue.34
The side chains of both 125D and 20E-125D are
known to be highly conformationally flexible and each
assumes its minimal energy characteristic orientations,
either to the ‘northwest’ for 20E or to the ‘northeast’
for 20N (see Figure 3).20 The stereoscopic dot maps of
the side chains of the two-side-chain Gemini provide
insight into the significant differences in the properties
of the single-side-chain analogues as compared with
Gemini. While the individual side chains of Gemini
assume the characteristic ‘northwest’ or ‘northeast’
orientation for their 20E or 20N chain, the two-side-
chain Gemini is much more conformationally flexible
as reflected by the 2346 minima, whereas the parent
125D and 20E-125D exhibit only 207 and 127 conform-
ers, respectively. This increased flexibility of the two

Two-Side-Chain 1R,25(OH)2D Interaction with VDR
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14 2725
Ta ble 1. Summary of Biological Properties of 1R,25(OH)2D3, 1R,25(OH)2-20-epi-D3, and a Two-Side-Chain Analogue, Geminia
ED50 (nM)
RCI (%)
colony formation, ED50 (nM)
transcription
activation
protease
sensitivity
seco-steroids
code
125D
20E-125D
Gemini
DBP
100
VDR
HL-60
130
MCF7
380
LNCaP
NR
1R,25(OH)2D3
20-epi-1R,25(OH)2D3
two-side-chain
100
10
0.005
1
5.9 ( 0.6
0.13 ( 0.09
180 ( 50
2.6 ( 1.5
2.5 ( 1.4
147 ( 12
38 ( 15
0.34
1.2
82
a
The human DBP RCI and the chick intestinal VDR RCI were determined from separate steroid competition assays as described
under Materials and Methods. The transactivation assay is a measure of the ability of the indicated seco-steroid to activate the VDRE
present in the promoter of the osteopontin gene coupled to growth hormone which was transfected into ROS 17/2.8 cells (Figure 4). The
protease sensitivity assay reflects ligand-induced changes in the conformation of [35S]VDR after treatment with trypsin followed by PAGE
separation (Figure 6). The ED50 for the protease sensitivity is tabulated for the concentration dependency of the major proteolytic fragment
present (see Figure 6) which is 34, 34, and 28 kDa, respectively, for 125D, 20E-125D, and Gemini. The procedures for the inhibition of
colony formation assays are described in Materials and Methods, and the dose-response results are in Figure 9. NR ) ED50 was not
reached at e10-6 M compound.
the A, seco B, and C/D rings identical to that of 125D.35
This is all the more impressive when consideration is
given to the additional molecular volume of Gemini (458
A3) as compared with either 125D or 20E-125D (375 A3)
as determined via molecular modeling. It is known that
some structural alterations of 125D that do not signifi-
cantly increase the molecular volume can, however,
result in a dramatic reduction in the VDR RCI; thus,
the RCI for 1â,25(OH)2D3 is 0.08% and for 18-acetoxy-
1R,25(OH)2D3 it is 0.04%.1 Thus, the additional molec-
ular volume associated with the presence of two side
chains on Gemini is not a serious liability with respect
to gaining access to the VDR LBD, suggesting that the
volume of the VDR LBD may be larger than necessary
to accommodate the natural hormone.
In contrast, for DBP, the RCI values of both the two-
side-chain Gemini and the single-side-chain 20E-125D
were dramatically reduced, with values of 2.5% and
2.6%, respectively. Clearly the presence of a 20E side
F igu r e 7. Effect of mutations in the WT VDR on ligand-
induced sensitivity to in vitro treatment with trypsin. In vitro
translated WT or mutated VDR was labeled with [35S]me-
thionine and incubated with the indicated concentrations of
125D or the analogues 20E-125D or Gemini (10-7-10-11 M)
before digestion with 20 µg/mL trypsin. The digestion products
were analyzed by PAGE and autoradiography. The arrows
indicate the position of the following ligand-dependent trypsin-
resistant fragments: (a) 34, (b) 32, (c) 28, and (d) 26 kDa; WT,
WT VDR; 421/422, AF-2-mutated VDR (V421M/F422A); 390/
TGA, VDR containing a stop codon at position 390.
chain is a severe liability for binding to the LBD of DBP.
Thus the VDR and DBP binding proteins display
significantly different ligand preferences for the three
seco-steroids employed in this communication, which is
consistent with the fact that there is no known struc-
tural similarity between these two proteins.
A further probing into the mode of interaction of 20E
and 20N analogues with VDR has shown that the 20E
analogues formed a significantly more stable complex
with the VDR than the 20N analogues, including the
natural hormone, 125D. This parameter of ligand-
receptor interaction was examined in the present study
by using an assay in which the rate of exchange of
analogue-occupied receptor with the natural hormone
is examined in vitro. Our experiments suggested that
the stability of VDR occupied by Gemini resembles the
stability of 125D-occupied VDR and not the 20E-
occupied VDR. Our earlier studies suggested that the
mechanism that supported the stability of VDR-20E
complex was their binding away from the AF-2 domain.
Therefore, we compared the binding requirements of the
three compounds to WT VDR and VDR that had the
AF-2 domain mutated (421/422) or deleted (390/TGA).
To our surprise, Gemini bound to the AF-2-mutated
VDR as well as it did to WT VDR. These results can be
interpreted to suggest that there is no cause-and-effect
relationship between the usage of the AF-2 domain for
binding and VDR ligand stability. Alternatively, it is
possible that Gemini uses the AF-2 residues in the
intact receptor to form a complex with similar stability
as the 125D-VDR complex, but once these residues are
side chains may be useful when the Gemini is gaining
initial access to the LBD of the VDR, since it effectively
allows the molecule to be more malleable and possibly
accommodate to the structural constraints imposed by
the side chains of the amino acids facing the interior of
the VDR LBD.
Two key components of the vitamin D endocrine
system are the plasma vitamin D-binding protein (DBP),
which transports all vitamin D metabolites throughout
the circulatory system, and the nuclear VDR. We have
compared for DBP and VDR the RCI, a measure of the
relative ligand affinity for the binding proteins, for the
two-side-chain Gemini with that of 20E-125D and 125D
with respect to their binding to VDR or DBP (see Table
1). The presence of the second side chain on Gemini did
not dramatically impair the ability of this analogue to
bind to the VDR as the RCI was 38% of that of 125D,
while the RCI for 20E-125D was 147%. Clearly a 20E
side chain alone or in combination with a 20N side chain
is competent to gain access to the LBD of the unoccupied
VDR; the molecular VDR LBD model can readily ac-
commodate Gemini as a ligand with an orientation of

2726 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14
Norman et al.
occupied with either 20E-125D or 125D has the same
major proteolytic fragment of 34 kDa, the 20E-125D
induced the appearance of an additional fragment at 32
kDa. In contrast, as reported in this communication, the
two-side-chain Gemini generates a major proteolytic
fragment of 28 kDa for both the WT VDR and the two
mutant VDRs (V421M/F422A and 390/TGA), suggesting
that it generates a conformational shape of the ligand-
occupied VDR which is different from those shapes
generated either by the 20N or 20E seco-steroids. We
speculate that because of the presence of two side
chains, Gemini may have a more difficult time than a
ligand with one side chain in gaining full entrance to
the LBD.
The conformational differences induced by the three
compounds should be quantitative or qualitative differ-
ences in their transcriptional potencies or efficacies. To
test this possibility, we examined the transcriptional
ED50 potency for activation of the thymidine kinase
promoter/growth hormone reporter construct containing
the opVDRE (see Figure 4 and Table 1) by 125D,
Gemini, and 20E-125D: 10, 1, and 0.005 nM, respec-
tively. Gemini was 10-fold more potent than 125D, but
20E-125D was 200- and 2000-fold more potent than
Gemini and 125D. Thus, the transcriptional potency of
the two-side-chain Gemini is modestly enhanced by the
presence of its 20E side chain present as compared to
125D. But, also the presence of the 20N side chain in
Gemini results in a 200-fold reduction in the transcrip-
tional activity achieved by 20E-125D. Although the
Gemini-occupied VDR does achieve a conformational
shape different from that of the 20E-125D-occupied or
125D-occupied VDR (28-kDa vs 34-kDa proteolytic
fragment), it is apparent that the shape of this Gemini-
occupied VDR is not one which is remarkably more
potent in transactivation of the osteopontin promoter.
Similar results were also noted for the osteocalcin
promoter (data not presented). In conclusion, each
receptor shape induced a different transcription profile.
Additional studies will reveal whether these differences
in transcription are due to different recruitment of
dimerization partners or of transcriptional co-activators.
The biological effectiveness of both Gemini and 125D
to inhibit the clonal growth of the HL-60, MCF7, and
LNCaP cells (Figure 9) is consistent with the observa-
tion that these cell lines possess the VDR.1,36-38 The
ED50(125D)/ED50(Gem) for HL-60 and MCF7 cells was
respectively 100 and 50. Thus for both cell lines, Gemini
was significantly more potent than 125D. Their relative
activities paralleled their relative gene transcriptional
activities rather than their binding affinity for the VDR
(see Table 1) which is consistent with the conclusions
derived from the protease sensitivity that binding of
Gemini by the VDR results in a different conformation
that is more productive with respect to both activation
of gene transcription and inhibition of clonal growth.
In contrast, in the LNCaP prostate cell lines (Figure 9)
neither Gemini nor 125D effected a full dose-response
inhibition of clonal growth. The reason for the refrac-
toriness of this cell line is unclear.
F igu r e 8. Competition of 125D and its analogues 20E-125D
and Gemini for binding to WT hVDR and its AF-2 mutant
(V421M/F422A). Competition assays were performed 3-4
times with each competitor 125D (A), 20E-125D (B), or Gemini
(C) by incubating homogenates from COS-1 cells transfected
with either WT VDR (WT) or the AF-2 mutant (V421M/F422A)
with [3H]125D and increasing amounts of the competitor
indicated above each graph. Each competition assay of the
mutant VDR was performed simultaneously with a competi-
tion assay of WT VDR. Linear regression coefficients (r) of the
plots shown were 0.92-0.96.
removed, it changes to the contact points to those used
by the 20E analogue.
A comparison of the ability of 125D, 20E-125D, and
Gemini to effect conformational changes in the VDR was
made by assessing the protease sensitivity of ligand
occupied [35S]VDRnuc to treatment with low concentra-
tions of trypsin as studied by subsequent PAGE (see
Figures 6 and 7). Our laboratories have previously
reported dramatic differences in protease sensitivity of
the ligand-occupied [35S]VDRnuc for a series of 20N
versus 20E analogues.15,16 Thus, while the [35S]VDRnuc
The presence of the second side chain on 125D
introduces several issues which may bear on determin-
ing the overall in vivo biological properties of analogue
Gemini. 125D is known to be inactivated principally by

Two-Side-Chain 1R,25(OH)2D Interaction with VDR
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14 2727
F igu r e 9. Dose-response of 1R,25(OH)2D3 and Gemini on clonal proliferation of (A) HL-60 human leukemic cells, (B) MCF7
breast cancer cells, and (C) LNCaP prostate cells. Results are expressed as percent of control plates containing no vitamin D
compounds. Each point represents the mean ( SD of at least three independent experiments with triplicate dishes: 1R,25-
(OH)2D3 (×-×) and Gemini (O-O).
Calcd for C19H37IOSi: C, 52.28; H, 8.54; I, 29.07; Si, 6.43.
Found: C, 52.27; H, 8.72; I, 28.99; Si, 6.45.
several metabolic pathways which hydroxylate the side
chain initiating side-chain cleavage.39 Thus the presence
of a second side chain may interfere with the orderly
metabolism process resulting in a longer metabolic half-
life and differences in potency between acute and
chronic dosing due to accumulation of higher circulating
levels of Gemini.
Our future studies with Gemini will further explore
a detailed determination of the kinetic ‘on’ and ‘off’ rates
for the ligand from the VDR and as well study the
metabolism and circulating concentrations of this two-
side-chain analogue in vivo. Also it is known that 125D,
in addition to generating biological responses via VDR-
mediated regulation of genomic events, can also produce
rapid, nongenomic responses40,41 though interaction
with a putative membrane receptor.42 Thus, it will also
be pertinent to determine the efficacy of the two-side-
chain Gemini to mediate rapid biological responses.
[1R-(1r,3a â,4r,7a r)]-(1,1-Dim eth yleth yl)d im eth yl[[oc-
t a h yd r o-7a -m et h yl-1-(1-m et h ylet h en yl)-1H -in d en -4-yl]-
oxy]sila n e (3). To a stirred solution of 40 g (92 mmol) of iodide
2 in 285 mL of dry THF and 125 mL of dry DMSO was added
20.5 g (183 mmol) of KOtBu. The mixture was stirred under
argon for 1.25 h and then poured into a mixture of 500 mL of
hexanes and 500 mL of water. Standard workup afforded 31
g of a pale yellow oil, which was chromatographed (silica gel,
hexanes) to afford 23.9 g (85%) of 3 as a colorless oil: [R]25
D
+36.95 (CHCl3, c ) 1.67); IR (CHCI3) 1639 cm-1
;
1H NMR
(CDCl3) δ 0.00 (6 H, s), 0.79 (3 H, s), 0.88 (9 H, s), 1.15 (1 H,
m), 1.40 (4 H, m), 1.60-1.70 (3 H, m), 1.73 (3 H, s), 1.74-1.90
(3 H, m), 1.97 (1 H, t, J ) 9 Hz), 4.01 (1 H, s), 4.68 (1 H, s),
4.85 (1 H, s); MS m/z 308 (M, 10). Anal. Calcd for C19H36OSi:
C, 73.95; H, 11.76; Si, 9.10. Found: C, 73.91; H, 11.90, Si, 9.16.
[1R-[1r(2E,4E,7E),3a â,4r,7a r]]-5-[4-[[(1,1-Dim eth yleth -
yl)d im et h ylsilyl]oxy]oct a h yd r o-7a -m et h yl-1H -in d en -1-
yl]-2,4,7-n on a tr ien ed ioic Acid Dieth yl Ester (4). To a
stirred solution of 3.08 g (10.0 mmol) of 3 and 3.92 g (40.0
mmol) of ethyl propiolate in 20 mL of CH2Cl2 was added 40.0
mL (40.0 mmol) of a 1.0 M solution of ethylaluminum dichlo-
ride in hexanes. The mixture was stirred under argon at room
temperature for 24 h, treated with an additional 981 mg (10
mmol) of ethyl propiolate and 7.5 mL (7.5 mmol) of a 1.0 M
solution of ethylaluminum dichloride in hexanes, and stirred
for an additional 18 h. The resultant orange-red solution was
added cautiously to a mixture of 200 mL of ethyl acetate and
100 mL of 50% brine, and after the fizzing had subsided, the
organic phase was collected and then worked up in the
standard fashion to give 5.76 g of a red gum. Fractional
chromatography (silica gel, 10% EtOAc in hexanes) afforded
2.2 g of crude 4, which upon HPLC purification (YMC silica
gel, 50-cm × 50-mm column) with 7.5% ethyl acetate in
Ma ter ia ls a n d Meth od s
Rea gen ts. [35H]Methionine and [23,24(n)-3H]1,25(OH)2D3
were obtained from Amersham Corp. A coupled transcription/
translation kit and a site-directed mutagenesis kit were
obtained from Promega. The two-side-chain analogue of 125D
was synthesized by Hoffmann-La Roche, Nutley, NJ . The
structural formulas for the key seco-steroids studied in this
communication are shown in Figure 1.
Ch em ica l Syn th esis of Tw o-Sid e-Ch a in An a logu e of
125D. The synthetic scheme employed for the chemical synthe-
sis of 1R,25-dihydroxy-21-(3-hydroxy-3-methylbutyl)vitamin D3
(designated as Gemini) is summarized in Figure 2.
Iod id e 2. To an ice-cold solution of 60 g (0.23 mmol) of
triphenylphosphine and 41 g (0.60 mol) of imidazole in 400
mL of CH2Cl2 was added 58 g (0.46 mol) of iodine. The cooled
mixture, which became heterogeneous after 5 min, was stirred
for 35 min and treated with a solution of 33 g (0.10 mol) of
alcohol 1 in 125 mL of CH2Cl2 during 15 min and then the
mixture was stirred at room temperature for 5 h. The mixture
was quenched with 400 mL of 2.5% sodium thiosulfate and
then worked up in the standard manner. The resulting
vacuum-dried residue was chromatographed (silica gel, 5%
ethyl acetate in hexanes) to afford a colorless oil, which
crystallized in the cold to afford the purified iodide 2: 41.3 g
(94%); mp 43-44 °C; [R]25D +50.58 (CHCl3, c ) 3.25); 1H NMR
(CDCl3) δ 0.00 (6 H, s), 0.89 (9 H, s), 0.95 (3 H, s), 0.99 (3 H,
d, J ) 5.6 Hz), 3.17 (1 H, dd, J ) 9.4, 5.0 Hz), 3.32 (1 H, dd,
J ) 9.4, 2.0 Hz), 4. 0 (1 H, s); MS (FAB) m/z 435 (M, 50). Anal.
hexanes afforded 1.62 g (32%) of 4 as a pale yellow gum: [R]25
D
+83.50 (c ) 0.98, EtOH); UV (MeOH) 284 (ꢀ ) 28200) nm; IR
1
(CHCl3) 1708, 1651, 1628 cm-1; H NMR (CDCl3) δ 0.006 (6
H, s), 0.80 (3 H, s), 0.88 (9 H, s), 1.16 (1 H, t, J ) 7.6 Hz), 1.28
(6 H, overlapping t, J ) 7 Hz), 2.16 (1 H, t, J ) 9 Hz), 3.00, (1
H, dd, J ) 6, 16 Hz), 3. 35 (1 H, dd, J ) 16, 4 Hz), 4. 02 (1 H,
s), 4.16 (4 H, overlapping q, J ) 7 Hz), 5.75 (1 H, d, J ) 16
Hz), 5.84 (1 H, d, J ) 15 Hz), 6.17(1 H, d, J ) 11 Hz), 6.88 (1
H, dt, J ) 16, 6 Hz), 7.50 (1 H, dd, J ) 15, 11 Hz); MS (EI)
m/z 504 (M, 23). Anal. Calcd for C29H48O5Si: C, 69.00; H, 9.58;
Si, 5.56. Found: C, 68.94; H, 9.69; Si, 5.67.
[1R -(1r,3a â,4r,7a r)]-5-[4-[[(1,1-D im e t h y le t h y l)d i-
m e t h ylsilyl]oxy]oct a h yd r o-7a -m e t h yl-1H -in d e n -1-yl]-
n on a n ed ioic Acid Dieth yl Ester (5). A stirred solution of
1.01 g (2.0 mmol) of triene ester 4 in 50 mL of ethyl acetate

2728 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14
Norman et al.
was hydrogenated over 200 mg of 10% palladium on charcoal
at room temperature and ambient pressure until hydrogen
absorption ceased (140 mL during 2.5 h). The mixture was
filtered with ethyl acetate washings, which upon concentration
gave 1.07 g of a colorless oil. This was purified by flash
chromatography (silica gel, 12% ethyl acetate in hexanes) to
mixture was stirred under argon at room temperature for 4
h. It was diluted with 7 mL of water, stirred for a further 15
min, and poured into a mixture of 75 mL of ethyl acetate and
50 mL of 50% brine. The organic phase was collected and the
aqueous phase was re-extracted with 3 × 50 mL of ethyl
acetate. The combined organic extracts after standard workup
gave a colorless oil, which was chromatographed (silica gel,
20% ethyl acetate in hexanes) to give 469 mg (92%) of 9 as a
give 964 mg (94%) of 5 as a colorless oil: [R]25 +32.1 (CHCl3,
D
c ) 1.04); IR (CHCl3) 1726 cm-1
;
1H NMR (CDCl3) δ 0.00
colorless oil: [R]25 -3.21 (CHCl3, c ) 0.87); IR (CHCl3); 1706
(3 H, s), 0.01 (3 H, s), 0.87 (9 H, s), 0.88 (3 H, s), 1.27 (6 H, t,
J ) 7 Hz), 2.25 (4 H, br t), 3.98 (1 H, s), 4.11 (4 H, q, J ) 7
Hz); MS (FAB) m/z 511 (M + 1, 100). Anal. Calcd for C29H54O5-
Si: C, 68.11; H, 10.66; Si, 5.50. Found: C, 68.21; H, 10.85; Si,
5.43.
D
cm-1; H NMR (CDCl3) δ 0.01 (18 H, s), 0.63 (3 H, s), 1.20 (6
1
H, s), 1.21 (6 H, s), 2.21-2.31 (2 H, m), 2.46 (1 H, dd, J ) 12,
11 Hz); MS (EI) m/z 495 (M - l5). Anal. Calcd for C29H58O3Si:
C, 68.17; H, 11.44; Si, 10.99. Found: C, 68.19; H, 11.41; Si,
11.07.
[1R -(1r,3a â,4r,7a r)]-6-[4-[[(1,1-D im e t h y le t h y l)d i-
m e t h ylsilyl]oxy]oct a h yd r o-7a -m e t h yl-1H -in d e n -1-yl]-
2,10-d im eth yl-2,10-u n d eca n ed iol (6). To a stirred ice-cooled
solution of 868 mg (1.7 mmol) of diester 5 in 12 mL of
anhydrous THF was added dropwise 5.0 mL (15 mmol) of a
3.0 M solution of methylmagnesium bromide in diethyl ether.
The mixture was stirred at room temperature for 45 min,
cooled to 5 °C, and quenched by the dropwise addition of 3
mL of saturated NH4Cl. After the fizzing had subsided, 15 mL
of ethyl acetate and 15 mL of saturated NH4Cl were added.
Stirring was continued for 20 min, and then the mixture was
poured into 100 mL of ethyl acetate and 50 mL of saturated
NH4Cl. Standard workup afforded 814 mg of a colorless gum,
which was purified by flash chromatography (silica gel, 50%
ethyl acetate in hexanes). High vacuum-drying (17 h) of the
resulting residue consisted of 763 mg (93%) of 6 as a colorless
foam: [R]25D +35.8 (EtOH, c ) 1.02); IR (CHCl3) 3608 cm-1;1H
NMR (CDCl3) δ 0.00 (6 H, s), 0.88 (9 H, s), 0.90 (3 H, s), 1.20
(12 H, s), 3.99 (1 H, s); MS (EI) m/z 482 (3, M). Anal. Calcd for
(1r,3â,5Z,7E)-21-(3-Hyd r oxy-3-m eth ylbu tyl)-9,10-seco-
ch olesta -5,7,10(19)-tr ien e-1,3,25-tr iol (Dou ble-Sid e-Ch a in
An a logu e, Gem in i). To a stirred, dry-ice-cooled solution of
466 mg (0.80 mmol) of the Horner reagent19 10 in 4.0 mL of
anhydrous THF was added 0.5 mL (0.80 mmol) of a 1.6 M
solution of butyllithium in hexanes. The resultant deep red
solution was stirred at -78 °C for 7 min, treated with 204 mg
(0.40 mmol) of ketone 9 in 3.0 mL of anhydrous THF, and
stirred at -78 °C for 3 h. The mixture was allowed to warm
to room temperature, quenched with 5 mL of a 1:1 mixture of
2 N Rochelle salt solution and 2 N KHCO3 solution, stirred
for an additional 15 min, and poured into a mixture of 80 mL
of ethyl acetate and 50 mL of a 1:1 mixture of 2 N Rochelle
salt solution and 2 N KHCO3 solution. The organic phase was
collected and the aqueous phase was re-extracted with 3 × 60
mL of ethyl acetate. Standard workup afforded a gum, which
was purified by flash chromatography (silica gel, 8% ethyl
acetate in hexanes) to give 208 mg of a colorless gum. The
latter was dissolved in 4.0 mL of THF and treated with 4.0
mL of a 1.0 M solution of TBAF in THF, and the solution was
stirred under argon for 17 h. It was diluted with 5 mL of water,
stirred for an additional 15 min, and poured into a mixture of
80 mL of 80% ethyl acetate in hexanes and 50 mL of water.
Standard workup gave 139 mg of a semisolid, which was
purified by flash chromatography (silica gel, 6% 2-propanol
in ethyl acetate) to give 108 mg of a colorless foam. This was
subjected to HPLC purification (silica gel, 50-cm × 50-mm
column, 3% 2-propanol in ethyl acetate) to give after vacuum-
C
29H58O3Si: C, 72.14; H, 12.11; Si, 5.82. Found: C, 72.18; H,
11.99; Si, 5.69.
[1S-(1r,3a â,4r,7a r)]-Oct a h yd r o-1-[5-h yd r oxy-1-(4-h y-
d r oxy-4-m eth ylp en tyl)-5-m eth ylh exyl]-7a -m eth yl-4H-in -
d en -4-ol (7). To a stirred solution of 700 mg (1.45 mmol) of 6
in 5 mL of THF and 15 mL of CH3CN contained in a Teflon
bottle was added 3.0 mL of a 30% aqueous solution of
fluorosilicilic acid and the mixture was stirred under argon
at room temperature for 1 h. Four 2.0-mL portions of the
fluorosilicilic acid solution were then added at hourly intervals,
for a total of 11 mL of reagent and a reaction time of 5 h. The
reaction mixture was diluted with 15 mL of water and poured
into a mixture of 60 mL of water and 125 mL of ethyl acetate.
After the fizzing had subsided, the organic phase was collected
and the aqueous phase was re-extracted with 3 × 75 mL of
ethyl acetate. Standard workup afforded 534 mg of a gum,
which was purified by flash chromatography (silica gel, 70%
ethyl acetate in hexanes) to give 458 mg (85%) of 7 as a
colorless foam: [R]25D +26.2 (CHCl3, c ) 0.76); IR (CHCl3) 3608
drying 82 mg of Gemini as a colorless, amorphous solid: [R]25
D
+13.8 (EtOH, c ) 0.5); UV (MeOH) 263 (ꢀ ) 17500) nm; IR
1
(CHCl3) 3608 cm-1; H NMR (CDCl3) δ 0.53 (3 H, s), 1.21 (12
H, s), 2.30 (1 H, dd, J ) 10, 7 Hz), 2.59 (1 H, d, J ) 11 Hz),
2.83 (1 H, d, J ) 13 Hz), 5.00 (1 H, s), 5.33 (1 H, s), 6.02 (1 H,
d, J ) 11 Hz), 6.37 (1 H, d, J ) 11 Hz); HRMS (El): calcd for
C32H54O2 m/z 502.4022; found m/z 502.4024.
En er gy Min im iza tion Ca lcu la tion s for 125D a n d An a -
logu e Sid e Ch a in s. Stereoscopic dot maps of the side chains
of 125D, 20E-125D, and Gemini were prepared by previously
described procedures.24,21 In brief, the dot maps are constructed
by overlaying all accessible low-energy conformations for the
side chain of the analogue under study by using a Monte Carlo
conformational search routine using the MM2 force field in
the BAKMDL and PC model programs (Serena Software,
Bloomington, IN). The resulting dot maps are used to depict
the volume in space that the side chain can occupy.
Deter m in a tion of Tr a n scr ip tion a l Activity. Rat os-
teosarcoma ROS 17/2.8 cells were plated in 35-mm dishes at
a density of 3 × 105 cells/dish. The cells were transfected with
2 µg of plasmid containing the VDRE from the human
osteopontin gene (opVDRE).43 This response element was
attached to the thymidine kinase promoter/growth hormone
fusion gene. All transfections were performed by the DEAE-
dextran method as previously described.15 Immediately after
1
cm-1; H NMR (CDCl3) δ 0.93 (3 H, s), 1.21 (12 H, s), 1.79-
1.95 (4 H, m); MS (FAB) m/z 369 (M + H).
[1S-(1r,3aâ,7ar)]-Octah ydr o-1-[5-h ydr oxy-1-(4-h ydr oxy-
4-m et h ylp en t yl)-5-m et h ylh exyl]-7a -m et h yl-4H -in d en -4-
on e (8). To a stirred solution of 400 mg (1.08 mmol) of 7 in
8.0 mL of CH2Cl2 was added 1.30 g (3.45 mmol) of pyridinium
dichromate and the mixture was stirred at room temperature
for 4.75 h. It was diluted with 20 mL of diisopropyl ether,
stirred for a further 15 min and filtered over a pad of Celite.
The Celite was washed with 4 × 40 mL of diisopropyl ether
and the combined filtrate and washings were evaporated to
give 405 mg of a pale yellow gum, which was purified by flash
chromatography (silica gel, 75% ethyl acetate in hexanes) to
give 372 mg (94% yield) of 8: [R]25 +0.45 (EtOH, c ) 0.92);
D
IR (CHCl3) 3608, 1706 cm-1; 1H NMR (CDCl3) δ 0.63 (3 H, s),
1.22 (12 H, s), 2.20-2.28 (2 H, m), 2.45 (1 H, dd, J ) 7.6,11
Hz); MS (EI) m/z 348.3 (M - H2O, 3). Anal. Calcd for
transfection, the ligands (125D, 20E-125D or Gemini) (10-7
-
10-13 M) were added to serum-free medium for 1 h and then
removed. The cells were washed twice with phosphate-buffered
saline (PBS), and DMEM with 10% fetal bovine serum was
added. Forty-eight hours after transfection, culture medium
was collected and the growth hormone levels were determined
with a radioimmunoassay. Maximal fold induction of tran-
C
23H42O3: C, 75.36; H, 11.55. Found: C, 75.13; H, 11.47.
[1S-(1r,3a â,7a r)]-Octa h yd r o-7a -m eth yl-1-[5-m eth yl-1-
[4-m eth yl-4-[(tr im eth ylsilyl)oxy]p en tyl]-5-[(tr im eth ylsi-
lyl)oxy]h exyl]-4H-in d en -4-on e (9). To a stirred solution of
367 mg (1.0 mmol) of ketone 8 in 10 mL of CH2Cl2 was added
1.25 mL (8.5 mmol) of 1-(trimethylsilyl)imidazole and the

Two-Side-Chain 1R,25(OH)2D Interaction with VDR
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14 2729
(2) Haussler, M. R.; Whitfield, G. K.; Haussler, C. A.; Hsieh, J . C.;
Thompson, P. D.; Selznick, S. H.; Dominguez, C. E.; J urutka, P.
W. The Nuclear Vitamin D Receptor: Biological and Molecular
Regulatory Properties Revealed. J . Bone Miner. Res. 1998, 13,
325-349.
(3) Freedman, L. P.; Lemon, B. D. Structural and Functional
Determinants of DNA Binding and Dimerization by the Vitamin
D Receptor. In Vitamin D; Feldman, D., Glorieux, F. H., Pike,
J . W., Eds.; Academic Press: San Diego, 1997; pp 127-148.
(4) Reichel, H.; Koeffler, H. P.; Norman, A. W. The Role of the
Vitamin D Endocrine System in Health and Disease. N. Engl.
J . Med. 1989, 320, 980-991.
(5) Elstner, E.; Lee, Y. Y.; Hashiya, M.; Pakkala, S.; Binderup, L.;
Norman, A. W.; Okamura, W. H.; Koeffler, H. P. 1R,25-Dihy-
droxy-20-Epi-Vitamin D3: An Extraordinarily Potent Inhib-
itor of Leukemic Cell Growth in Vitro. Blood 1994, 84, 1960-
1967.
(6) Peleg, S.; Liu, Y. Y.; Reddy, S.; Horst, R. L.; White, M. C.; Posner,
G. H. A 20-Epi Side Chain Restores Growth-Regulatory and
Transcriptional Activities of an A Ring-Modified Hybrid Ana-
logue of 1R,25-Dihydroxyvitamin D3 Without Increasing Its
Affinity to the Vitamin D Receptor. J . Cell. Biochem. 1996, 63,
149-161.
(7) Munker, R.; Kobayashi, T.; Elstner, E.; Norman, A. W.; Uskok-
ovic, M.; Zhang, W.; Andreeff, M.; Koeffler, H. P. A New Series
of Vitamin D Analogues Is Highly Active for Clonal Inhibition,
Differentiation, and Induction of WAF1 in Myeloid Leukemia.
Blood 1996, 88, 2201-2209.
(8) Zhao, X. Y.; Eccleshall, T. R.; Krishnan, A. V.; Gross, C.;
Feldman, D. Analysis of Vitamin D Analog-Induced Heterodimer-
ization of Vitamin D Receptor With Retinoid X Receptor Using
the Yeast Two-Hybrid System. Mol. Endocrinol. 1997, 11, 366-
378.
scription varied between experiments (3-5-fold), but maximal
transcriptional activity was the same for 125D (which was
included in each transfection experiment) and the analogues
in individual experiments.
Site-Dir ected Mu ta tion s in th e VDR. The WT VDR, the
residue 421/422,AF-2-mutated VDR (V421M/F422A), and the
C-terminal-truncated VDR 390/TGA were prepared as de-
scribed previously.16
Liga n d -Alter ed P r otea se Sen sitivity Assa y. Synthetic
WT and mutant VDR labeled with [35S]methionine (1000 Ci/
mmol) was prepared by in vitro coupled transcription/transla-
tion in reticulocyte lysates (Promega) with the human VDR
cDNA inserted into the pGEM-4 plasmid. The translated
receptor preparations were incubated with the indicated
concentrations of 125D or analogues for 20 min at room
temperature. Then 15 µg/mL trypsin (Calbiochem) was added,
and the mixtures were incubated for another 20 min. The
digestion products were analyzed by 14% sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and
the gels were dried and autoradiographed.
Liga n d Bin d in g Assa ys. To assess the relative affinity of
125D and the two analogues under study (20E-125D and
Gemini) for the chick intestinal VDR, human DBP, and the
WT or mutant VDR, a determination of the relative competi-
tive index (RCI) was carried out according to our standard
procedures.44 Both of these procedures are steroid competition
assays where increasing concentrations of the analogue or
standard nonradioactive 125D are incubated for 4 h at 4 °C
with a defined concentration of [3H]125D and the binding
protein. Then separation of bound and free ligand is achieved
by use of either hydroxyapatite (VDR) or charcoal-dextran
(DBP) and the RCI calculated.25 By definition the RCI of 125D
is set to 100%.
(9) Baudino, T. A.; Kraichely, D. M.; J efcoat, S. C., J r.; Winchester,
S. K.; Partridge, N. C.; MacDonald, P. N. Isolation and Char-
acterization of a Novel Coactivator Protein, NCoA-62, Involved
in Vitamin D-Mediated Transcription. J . Biol. Chem. 1998, 273,
16434-16441.
(10) Lemon, B. D.; Fondell, J . D.; Freedman, L. P. Retinoid
X
Exch a n ge Assa ys for Non r a d ioa ctive Liga n d s Bou n d
to th e VDR w ith [3H]125D. VDR-transfected COS-1 cells
were separately incubated with 10-8 M 125D, 20E-125D, or
Gemini in a serum-free medium for 1 h. Then the medium was
removed, and the cells were washed three times in ice-cold
PBS and homogenized. The number of unoccupied binding
sites was determined by incubating duplicate samples of cell
homogenate for the time indicated at 30 °C with 0.2 pmol of
[3H]125D with or without a 100-fold excess of nonradioactive
125D and then incubated for 3 h on ice. The free ligand was
separated from bound by hydroxyapatite, and the amount of
bound ligand was quantified by liquid scintillation spectrom-
etry. Each point is the average of two samples.
Colon y F or m a tion Assa y in Soft Aga r . Cells were
cultured in six-well culture dishes in a two-layer soft agar
system as described previously.45,46 The under layer contained
0.5% agar, the upper layer 0.3% agar (Difco Laboratories,
Detroit, MI). Compounds and cells to be tested were mixed
into the under layers and upper layers, respectively. Cell
concentrations were 1 × 103/plate for HL-60, MCF7 and
LNCaP cells. Cultures were placed in a humidified atmo-
sphere, 5% CO2, at 37 °C for 12 days. All experiments were
carried out using triplicate plates per experimental point, and
each experiment was performed at least three times. Colony
formation (>40 cells) were scored with an inverted microscope.
Receptor-Vitamin D3 Receptor Heterodimers Promote Stable
Preinitiation Complex Formation and Direct 1,25-Dihydroxyvi-
tamin D3-Dependent Cell-Free Transcription. Mol. Cell. Biol.
1997, 17, 1923-1937.
(11) Binderup, L.; Latini, S.; Binderup, E.; Bretting, C.; Calverley,
M.; Hansen, K. 20-Epi-Vitamin D3 Analogues: a Novel Class of
Potent Regulators of Cell Growth and Immune Responses.
Biochem. Pharmacol. 1991, 42, 1569-1575.
(12) Binderup, L. Vitamin D Analogues: New Regulators of Cancer
Cell Growth and Differentiation. Bioorg. Med. Chem. Lett. 1993,
3, 1891-1896.
(13) Elstner, E.; Linker-Israeli, M.; Le, J .; Umiel, T.; Michl, P.; Said,
J . W.; Binderup, L.; Reed, J . C.; Koeffler, H. P. Synergistic
Decrease of Clonal Proliferation, Induction of Differentiation,
and Apoptosis of Acute Promyelocytic Leukemia Cells After
Combined Treatment With Novel 20-Epi Vitamin D3 Analogues
and 9-Cis Retinoic Acid. J . Clin. Invest. 1997, 99, 349-360.
(14) Elstner, E.; Linker-Israeli, M.; Said, J .; Umiel, T.; De Vos, S.;
Shintaku, P.; Heber, D.; Binderup, L.; Uskokovic, M. R.; Koeffler,
H. P. 20-Epi-Vitamin D3 Analogues: A Novel Class of Potent
Inhibitors of Proliferation and Inducers of Differentiation of
Human Breast Cancer Cell Lines. Cancer Res. 1995, 55, 2822-
2830.
(15) Peleg, S.; Sastry, M.; Collins, E. D.; Bishop, J . E.; Norman, A.
W. Distinct Conformational Changes Induced by 20-Epi Ana-
logues of 1R,25-Dihydroxyvitamin D3 Are Associated With
Enhanced Activation of the Vitamin D Receptor. J . Biol. Chem.
1995, 270, 10551-10558.
(16) Liu, Y. Y.; Collins, E. D.; Norman, A. W.; Peleg, S. Differential
Interaction of 1R,25-Dihydroxyvitamin D3 Analogues and Their
20-Epi Homologues With the Vitamin D Receptor. J . Biol. Chem.
1997, 272, 3336-3345.
(17) Ryha¨nen, S.; Mahonen, A.; J a¨a¨skela¨inen, T.; Ma¨enpa¨a¨, P. H.
Synthetic 20-Epi Analogues of Calcitriol Are Potent Inducers of
Target-Gene Activation in Osteoblastic Cells. Eur. J . Biochem.
1996, 238, 97-103.
Ack n ow led gm en t. This study was supported by
NIH Grants DK09012-34 (A.W.N.) and DK-16595, UC
Intramural Research Fund and Solvay Pharmaceutical
Co. (W.H.O.), and Grant DK-50583 (S.P.).
(18) Elstner, E.; Heber, D.; Koeffler, H. P. 20-Epi-Vitamin D3
Analogues: Potent Modulators of Proliferation and Differentia-
tion of Breast Cancer Cell Lines in Vitro. Adv. Exp. Med. Biol.
1996, 399, 53-70.
(19) Baggiolini, E. G.; Iacobelli, J . A.; Hennessy, B. M.; Batcho, A.
D.; Sereno, J . F.; Uskokovic, M. R. Stereocontrolled Total
Synthesis of 1R,25-Dihydroxycholecalciferol and 1R,25-Dihy-
droxyergocalciferol. J . Org. Chem. 1986, 51, 3098-3108.
(20) Okamura, W. H.; Midland, M. M.; Hammond, M. W.; Rahman,
N. A.; Dormanen, M. C.; Nemere, I.; Norman, A. W. Chemistry
and Conformation of Vitamin D Molecules. J . Steroid Biochem.
Mol. Biol. 1995, 53, 603-613.
Su p p or tin g In for m a tion Ava ila ble: Details of spectro-
scopic methods, chromatographic profile, and additional NMR
data for the two-side-chain analogue Gemino. This material
is available, free of charge, via the Internet at http://pubs.
acs.org.
Refer en ces
(1) Bouillon, R.; Okamura, W. H.; Norman, A. W. Structure-
Function Relationships in the Vitamin D Endocrine System.
Endocr. Rev. 1995, 16, 200-257.

2730 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 14
Norman et al.
(21) Midland, M. M.; Plumet, J .; Okamura, W. H. Effect of C20
Stereochemistry on the Conformational Profile of the Side
Chains of Vitamin D Analogues. Bioorg. Med. Chem. Lett. 1993,
3, 1799-1804.
(22) Yamamoto, K.; Sun, W. Y.; Ohta, M.; Hamada, K.; DeLuca, H.
F.; Yamada, S. Conformationally Restricted Analogues of 1R,-
25-Dihydroxyvitamin D3 and Its 20-Epimer: Compounds for
Study of the Three-Dimensional Structure of Vitamin D Re-
sponsible for Binding to the Receptor. J . Med. Chem. 1996, 39,
2727-2737.
(23) Yamada, S.; Yamamoto, K.; Masuno, H.; Ohta, M. Conformation-
Function Relationship of Vitamin D: Conformational Analysis
Predicts Potential Side-Chain Structure. J . Med. Chem. 1998,
41, 1467-1475.
(24) Okamura, W. H.; Palenzuela, J . A.; Plumet, J .; Midland, M. M.
Vitamin D: Structure-Function Analyses and the Design of
Analogues. J . Cell. Biochem. 1992, 49, 10-18.
(25) Wecksler, W. R.; Norman, A. W. Structural Aspects of the
Binding of 1R,25-Dihydroxyvitamin D3 to its Receptor System
in Chick Intestine. In Methods in Enzymology: Vitamins and
Co-Enzymes; Academic Press: New York, 1980; pp 494-500.
(26) Spirichev, V. B.; Sergeev, I. N. Vitamin D: Experimental
Research and Its Practical Application. Wld. Rev. Nutr. Diet
1988, 173-216.
(27) Zhou, J . Y.; Norman, A. W.; Chen, D.; Sun, G.; Uskokovic, M.
R.; Koeffler, H. P. 1R,25-Dihydroxy-16-Ene-23-Yne-Vitamin D3
Prolongs Survival Time of Leukemic Mice. Proc. Natl. Acad. Sci.
U.S.A. 1990, 87, 3929-3932.
(28) Craig, A. S.; Norman, A. W.; Okamura, W. H. The Synthesis of
Two Novel Allenic Side Chain Analogues of 1R,25-Dihydroxyvi-
tamin D3. J . Org. Chem. 1992, 57, 4374-4380.
(29) Figade`re, B.; Norman, A. W.; Henry, H. L.; Koeffler, H. P.; Zhou,
J . Y.; Okamura, W. H. Arocalciferols: Synthesis and Biological
Evaluation of Aromatic Side-Chain Analogues of 1R,25-Dihy-
droxyvitamin D3. J . Med. Chem. 1991, 34, 2452-2463.
(30) Peleg, S.; Liu, Y.-Y.; Reddy, S.; Collins, E. D.; Norman, A. W.
Differential Interaction of 1R,25-Dihydroxyvitamin D3 and its
20-Epi-Analogue with the Transcription Activation Function
2(AF-2) Domain of the Vitamin D Receptor. In Vitamin D:
Chemistry, Biology and Clinical Applications of the Steroid
Hormone; Norman, A. W., Bouillon, R., Thomasset, M., Eds.;
University of California: Riverside, 1997; pp 203-209.
(31) Wagner, R. L.; Apriletti, J . W.; McGrath, M. E.; West, B. L.;
Baxter, J . D.; Fletterick, R. J . A Structural Role for Hormone in
the Thyroid Hormone Receptor. Nature 1995, 378, 690-697.
(32) Gill, R. K.; Atkins, L. M.; Hollis, B. W.; Bell, N. H. Mapping the
Domains of the Interaction of the Vitamin D Receptor and
Steroid Receptor Coactivator-1. Mol. Endocrinol. 1998, 12, 57-
65.
(33) Uskokovic, M. R.; Manchand, P. S.; Peleg, S.; Norman, A. W.
Synthesis and preliminary evaluation of the biological properties
of a 1R,25-dihydroxyvitamin D3 analogue with two side-chains.
In Vitamin D: Chemistry, Biology and Clinical Applications of
the Steroid Hormone; Norman, A. W., Bouillon, R., Thomasset,
M., Eds.; University of California: Riverside, 1997; pp 19-21.
(34) Kurek-Tyrlik, A.; Makaev, F. Z.; Wicha, J .; Zhabinskii, V.;
Calverley, M. J . Pivoting 2-Normal and 20-Epi Calcitriols:
Synthesis and Crystal Structure of a “Double Side Chain”
Analogue. In Vitamin D: Chemistry, Biology and Clinical
Applications of the Steroid Hormone; Norman, A. W., Bouillon,
R., Thomasset, M., Eds.; University of California: Riverside,
1997; pp 30-31.
(35) Norman, A. W.; Adams, D.; Collins, E. D.; Okamura, W. H.;
Fletterick, R. J . Three-Dimensional Model of the Ligand Binding
Domain of the Nuclear Receptor for 1R,25-Dihydroxy-Vitamin
D3
. J . Cell Biochem. 1999, 74, 323-333.
(36) Taoka, T.; Collins, E. D.; Irino, S.; Norman, A. W. 1,25(OH)2-
Vitamin D3 Mediated Changes in mRNA for c-Myc and 1,25-
(OH)2D3 Receptor in HL-60 Cells and Related Subclones. Mol.
Cell. Endocrinol. 1993, 95, 51-57.
(37) Nolan, E.; Donepudi, M.; Vanweelden, K.; Flanagan, L.; Welsh,
J . Dissociation of Vitamin D3 and Anti-Estrogen Mediated
Growth Regulation in MCF-7 Breast Cancer Cells. Mol. Cell.
Biochem. 1998, 188, 13-20.
(38) Miller, G. J .; Stapleton, G. E.; Ferrara, J . A.; Lucia, M. S.; Pfister,
S.; Hedlund, T. E.; Upadhya, P. The Human Prostatic Carcinoma
Cell Line LNCaP Expresses Biologically Active, Specific Recep-
tors for 1R,25-Dihydroxyvitamin D3. Cancer Res. 1992, 52, 515-
520.
(39) Henry, H. L.; Norman, A. W. Metabolism of vitamin D. In
Disorders of Bone and Mineral Metabolism; Coe, F. L., Favus,
M. J ., Eds.; Raven Press: New York, 1991; pp 149-162.
(40) Zanello, L. P.; Norman, A. W. Stimulation by 1R,25(OH)2-
Vitamin D3 of Whole Cell Chloride Currents in Osteoblastic ROS
17/2.8 Cells: A Structure-Function Study. J . Biol. Chem. 1997,
272, 22617-22622.
(41) Norman, A. W. Rapid Biological Responses Mediated by 1R,25-
(OH)2-Vitamin D3: A Case Study of Transcaltachia (the Rapid
Hormonal Stimulation of Intestinal Calcium Transport). In
Vitamin D; Feldman, D., Glorieux, F. H., Pike, J . W., Eds.;
Academic Press: San Diego, 1997; pp 233-256.
(42) Nemere, I.; Dormanen, M. C.; Hammond, M. W.; Okamura, W.
H.; Norman, A. W. Identification of a Specific Binding Protein
for 1R,25-Dihydroxyvitamin D3 in Basal-Lateral Membranes of
Chick Intestinal Epithelium and Relationship to Transcaltachia.
J . Biol. Chem. 1994, 269, 23750-23756.
(43) Ozono, K.; Liao, J .; Kerner, S. A.; Scott, R. A.; Pike, J . W. The
Vitamin D-Responsive Element in the Human Osteocalcin Gene.
Association With a Nuclear Proto-Oncogene Enhancer. J . Biol.
Chem. 1990, 265, 21881-21888.
(44) Norman, A. W.; Okamura, W. H.; Hammond, M. W.; Bishop, J .
E.; Dormanen, M. C.; Bouillon, R.; Van Baelen, H.; Ridal, A. L.;
Daane, E.; Khoury, R.; Farach-Carson, M. C. Comparison of
6-s-Cis and 6-s-Trans Locked Analogues of 1R,25(OH)2-Vitamin
D3 Indicates That the 6-s-Cis Conformation Is Preferred for
Rapid Nongenomic Biological Responses and That Neither 6-s-
Cis nor 6-s-Trans Locked Analogues Are Preferred for Genomic
Biological Responses. Mol. Endocrinol. 1997, 11, 1518-1531.
(45) Munker, R.; Norman, A. W.; Koeffler, H. P. Vitamin D Com-
pounds: Effect on Clonal Proliferation and Differentiation of
Human Myeloid Cells. J . Clin. Invest. 1986, 64, 424-430.
(46) Kubota, T.; Koshizuka, K.; Koike, M.; Uskokovic, M. R.; Miyoshi,
I.; Koeffler, H. P. 19-Nor-26,27-Bishomo-Vitamin D3 Ana-
logues: A Unique Class of Potent Inhibitors of Proliferation of
Prostate, Breast, and Hematopoietic Cancer Cells. Cancer Res.
1998, 58, 3370-3375.
J M0000160