Synthesis and biological activities of vitamin D-like inhibitors of CYP24 hydroxylase

Article abstract of DOI:10.1016/j.steroids.2011.11.007

Selective inhibitors of CYP24A1 represent an important synthetic target in a search for novel vitamin D compounds of therapeutic value. In the present work, we show the synthesis and biological properties of two novel side chain modified 2-methylene-19-no

Full text of DOI:10.1016/j.steroids.2011.11.007

Steroids 77 (2012) 212–223
Contents lists available at SciVerse ScienceDirect
Steroids
journal homepage: www.elsevier.com/locate/steroids
Synthesis and biological activities of vitamin D-like inhibitors of CYP24 hydroxylase
Grazia Chiellini a,b, Simona Rapposelli c, Jinge Zhu a, Ilaria Massarelli d, Marilena Saraceno c,
Anna Maria Bianucci c,e, Lori A. Plum a, Margaret Clagett-Dame a, Hector F. DeLuca a,
⇑
a Department of Biochemistry, University of Wisconsin-Madison, 433 Babcock Drive, Madison, WI 53706-1544, USA
b Dipartimento di Scienze dell’Uomo e dell’Ambiente, Università di Pisa, via Roma 55, 56126 Pisa, Italy
c Dipartimento di Scienze Farmaceutiche, Università di Pisa, via Bonanno 6, 56126 Pisa, Italy
d Istituto Nazionale per la Scienza e la Tecnologia dei Materiali, via Giusti 9, 50121 Firenze, Italy
e International Centre for Studies and Research in Biomedicine (ICB), Luxembourg
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 10 August 2011
Received in revised form 14 November 2011
Accepted 15 November 2011
Available online 25 November 2011
Selective inhibitors of CYP24A1 represent an important synthetic target in a search for novel vitamin D
compounds of therapeutic value. In the present work, we show the synthesis and biological properties of
two novel side chain modified 2-methylene-19-nor-1,25(OH)2D3 analogs, the 22-imidazole-1-yl deriva-
tive 2 (VIMI) and the 25-N-cyclopropylamine compound 3 (CPA1), which were efficiently prepared in
convergent syntheses utilizing the Lythgoe type Horner–Wittig olefination reaction. When tested in a
cell-free assay, both compounds were found to be potent competitive inhibitors of CYP24A1, with the
cyclopropylamine analog 3 exhibiting an 80–1 selective inhibition of CYP24A1 over CYP27B1. Addition
of 3 to a mouse osteoblast culture sustained the level of 1,25(OH)2D3, further demonstrating its effective-
ness in CYP24A1 inhibition. Importantly, the in vitro effects on human promyeloid leukemia (HL-60) cell
differentiation by 3 were nearly identical to those of 1,25(OH)2D3 and in vivo the compound showed low
calcemic activity. Finally, the results of preliminary theoretical studies provide useful insights to rational-
ize the ability of analog 3 to selectively inhibit the cytochrome P450 isoform CYP24A1.
Keywords:
1,25-Dihydroxyvitamin D3
Cytochrome P450 inhibitors
CYP24A1
Cyclopropylamines
Cancer therapy
Molecular docking
Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction
is initiated in the liver to produce 25-hydroxyvitamin D3
(25(OH)D3), which is further converted to 1
D3 (1,25(OH)2D3), the hormonally active secosteroid, by 25(OH)D3
1a-hydroxylase (CYP27B1) mainly in the kidney. Metabolic inacti-
a,25-dihydroxyvitamin
The hormonal form of vitamin D3, 1a,25-dihydroxyvitamin D3
(1,25(OH)2D3, 1) (Fig. 1) has long been known for its central role
in controlling calcium and phosphate homeostasis. In addition, a
growing body of investigations is continuously unveiling the sig-
nificance of vitamin D in a diverse array of physiological functions,
including cell proliferation, differentiation and apoptosis, and im-
mune response. Consequently, vitamin D deficiency has been asso-
ciated with a number of diseases such as cancers, autoimmune
dysfunctions, cardiovascular diseases, and infections [1–3]. As the
therapeutic activity of exogenously administered 1,25(OH)2D3 is
limited by its capacity to induce hypercalcemia, numerous efforts
have been directed to develop therapeutically useful 1,25(OH)2D3
analogs. Although many of them have resulted in useful therapies,
in particular for the treatment of kidney, bone and skin diseases
[4,5], their use in other areas remains nonexistent primarily due
to the difficulty of achieving an effective concentration without
also raising serum calcium.
vation of 1,25(OH)2D3 in its target cells is initiated by side chain
hydroxylations at the C-23, C-24, and C-26 positions. Of these
hydroxylation sites, it is now accepted that the sequential oxidation
and cleavage of the side chain by mitochondrial 1,25(OH)2D3 24-
hydroxylase (CYP24A1) (Fig. 2) is the major pathway by which the
hormone is inactivated [6]. CYP24A1, as the main catabolizing en-
zyme of 1,25(OH)2D3, can also break down vitamin D analogs in a
similar catabolic process. Elevated CYP24A1 levels have been seen
in many types of human cancers [7], immune dysfunctions, and sec-
ondary hyperparathyroidism. In addition, intervention with vitamin
D analogs often stimulates the expression of CYP24A1, resulting in
rapid elimination of the active metabolites, that drastically reduces
the effectiveness of the treatment. Targeting CYP24 provides the
opportunity to increase endogenous levels of 1,25(OH)2D3, or reduce
the effective dose of exogenous 1,25(OH)2D3. Conventional drug de-
sign approaches have led to the development of a range of P450
inhibitors that are currently used in the clinic, primarily as anti-
fungal agents. Effective inhibitors can be found in a class of bifunc-
tional lipophilic compounds (entitled azoles) that directly link to
the heme iron by the azole nitrogen and, additionally, interact with
Vitamin D3, the natural form of vitamin D produced in the skin
through UV exposure, is biologically inert. Activation of vitamin D3
⇑
Corresponding author. Tel.: +1 608 262 1620; fax: +1 608 262 7122.
E-mail address: deluca@bioChem.wisc.edu (H.F. DeLuca).
0039-128X/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.steroids.2011.11.007

G. Chiellini et al. / Steroids 77 (2012) 212–223
213
Fig. 1. Chemical structures of 1
a
,25-dihydroxyvitamin-D3 and new designed CYP24 inhibitors.
OH
O
OH
H
OH
OH
H
H
H
H
H
HO
OH
HO
OH
HO
OH
1,24,25-(OH)3 D3
24-oxo-1,25-(OH)2 D3
1,25-(OH)2 D3
O
O
OH
H
OH
OH
H
H
OH
H
H
H
H
H
HO
OH
HO
OH
HO
OH
24-oxo-1,23,25-(OH)3 D3
Calcitroic acid
Fig. 2. CYP24A1 oxidation pathway.
other parts of the substrate site. Currently, the non-selective CYP
inhibitor ketoconazole (Fig. 3) has been shown to inhibit CYP24
and act synergistically with vitamin D analogs in cell cultures [8]
and is being tried in combination with Calcitriol in a phase I clinical
trial [9]. Liarazole (Fig. 3) has also been shownto inhibit 1,25(OH)2D3
hydroxylation and act synergistically with 1,25(OH)2D3 in androgen
independent DU-145 prostate cancer cells [10]. However, a major
problem of many azole inhibitors is their low selectivity.
Due to these limitations, selective CYP24 inhibitors offer a ther-
apeutic advantage, and several groups have developed compounds
to this end [11–16]. Aiming at increased/sustained hormone levels,
a project to develop azole-type inhibitors of CYP24A1 was started
in Novartis in the early 1990s. More than 400 structurally different
azole-type compounds were tested, using primary human kerati-
nocytes as model system. A compound termed VID 400 (Fig. 3)
showed the desired qualities as a strong selective CYP24A1 inhib-
itor, and was chosen for development in the indication of psoriasis
[11,12]. Simons’ group reported CYP24A1 inhibitory activity of
tetralone derivatives and N-[2-(1H-imidazol-1-yl)-2-phenylethyl]
arylamides, but their selectivity has not been assessed [13,14].
C-24 analogs of 1,25(OH)2D3 licensed to Cytochroma Inc. and
termed ‘‘vitamin D signal amplifiers’’ are both, potent inhibitors
of CYP24A1 and agonists of the vitamin D signaling pathway. The
lead compound CTA018, where C24 is replaced with a sulfoximine
functional group [15,16] (Fig. 3), is at the beginning of clinical
Phase II in the indication: topical treatment of mild to moderate
psoriasis.
Cyclopropylamine derivatives are also known to have interest-
ing and sometimes useful properties as enzyme inhibitors [17].
Since the initial reports that N-benzyl-N-cyclopropylamine

214
G. Chiellini et al. / Steroids 77 (2012) 212–223
Fig. 3. Chemical structures of cytochrome P450 inhibitors.
(Fig. 3) is a suicide substrate for cytochrome P450 enzymes, as well
as for monoamine oxidase, cyclopropylamines have been widely
applied as mechanistic probes of these and other oxidative en-
zymes. The cyclopropylamine substructure is indeed found in sev-
eral biologically active natural products and is increasingly found
within synthetic drug and drug candidate molecules [18–24].
Based on these findings, we decided to prepare two novel vita-
min D-like analogs having an azole group 2 (VIMI) or a cyclopro-
pylamine moiety 3 (CPA1) on the modified side chain (Fig. 1).
Interestingly, both these new compounds turned out to be potent
inhibitors of CYP24A1, with 3 showing specific inhibition of
CYP24A1 over CYP27B1. Although the azole analog 2 displayed a
very low potency as a vitamin D analog, the cyclopropylamine
compound 3 was found to induce HL-60 cells differentiation with
the same potency as 1,25(OH)2D3, while showing lower potency
in raising calcium tissue levels. Therefore this novel analog may
ultimately be useful as a new and safer therapeutic agent. A molec-
ular docking study provided useful insights to rationalize the abil-
ity of analog 3 to selectively inhibit the cytochrome P450 isoform
CYP24A1 and to optimize the lead structure to obtain compounds
suitable to enter the pre-clinical and clinical drug development
programs.
Instruments DMX-400 and DMX-500 Avance console spectrome-
ters. 13C NMR spectra were recorded in deuteriochloroform, at
100 and 125 MHz with the same Bruker Instruments. Chemical
shifts (d) in parts per million are quoted relative to internal Me4Si
(d 0.00). Electron impact (EI) mass spectra were obtained with a
Micromass AutoSpec (Beverly, MA) instrument. HPLC was per-
formed on a Waters Associates liquid chromatograph equipped
with a model 6000A solvent delivery system, model U6K Universal
injector, and model 486 tunable absorbance detector. THF was
freshly distilled before use from sodium benzophenone ketyl under
argon. A designation ‘‘(volume + volume)’’, which appears in gen-
eral procedures, refers to an original volume plus a rinse volume.
Both final vitamin D analogs synthesized by us gave single
sharp peaks on HPLC, and they were judged at least 99.5% pure.
The purity and identity of the synthesized vitamins were addition-
ally confirmed by inspection of their 1H NMR, 13C NMR, UV absorp-
tion, and high-resolution mass spectra.
2.2. Synthesis of compounds
2.2.1. (8S,20S)-des-A,B-20-(10-methylimidazolyl)-pregnan-8-ol (7)
A solution of diol 5 [25] (0.50 g, 2.35 mmol) in anhydrous pyri-
dine (5 mL) was cooled to ꢀ25 °C. A precooled solution of tosyl
chloride (0.55 g, 2.90 mmol) in anhydrous pyridine (1 mL) was
added dropwise to the diol solution via cannula. Upon stirring for
3.5 h at ꢀ25 °C, the reaction was warmed to 0 °C and allowed to stir
for an additional 20 h. The mixture was extracted with CH2Cl2,
washed with saturated CuSO4 aqueous solution, dried (MgSO4), fil-
tered, and concentrated to give a residue which was chromato-
graphed on a silica gel column with hexane/ethyl acetate (8:2) to
afford 0.60 g (1.68 mmol) of the corresponding tosylate 6 in 70%
yield. To a solution of imidazole (0.046 g, 0.67 mmol) in dry DMF
(5 mL) at 0 °C under argon was added NaH (60% dispersion in oil,
2. Experimental procedures
2.1. General
Optical rotations were measured in chloroform or methanol
using a Perkin–Elmer model 343 polarimeter at 22 °C. Ultraviolet
(UV) absorption spectra were recorded with a Perkin–Elmer Lamb-
da 3B UV–Vis spectrophotometer in ethanol. 1H nuclear magnetic
resonance (NMR) spectra were recorded in deuteriochloroform,
acetone-d6, or methanol-d4 at 400 and 500 MHz with Bruker

G. Chiellini et al. / Steroids 77 (2012) 212–223
215
0.057 g, 1.42 mmol), and the mixture was stirred at the same tem-
perature for 20 min. A solution of tosylate 6 (0.12 g, 0.34 mmol) in
dry DMF (3 mL) was added dropwise over 15 min. The reaction
mixture was warmed to room temperature, and after being stirred
for 24 h at room temperature, the mixture was quenched with
water and extracted with EtOAc. The organic phase was dried and
evaporated, and the residue was purified by flash chromatography.
Gradient elution (1–5% MeOH/CHCl3) afforded 0.085 g (0.32 mmol)
room temperature. After stirring of the reaction mixture for 3 h,
ice and 5% aq. HCl were added until pH 6. The solution was ex-
tracted with ethyl acetate (3 ꢁ 50 mL) and the combined organic
phases were washed with saturated aq. NaHCO3, dried (Na2SO4)
and concentrated under reduced pressure. The residue was chro-
matographed on silica gel with hexane/ethyl acetate (75:25) to
give the alcohol 9 (0.44 g, 84% yield) as a colorless oil.
[a
]D +41.1 (c 2.85, CHCl3); 1H NMR (400 MHz, CDCl3 + TMS) d 4.04
(1H, d, J = 2.4 Hz, 8 -H), 3.63 (1H, dd, J = 10.5, 3.2 Hz, 22-H), 3.38
of 7 in 95% yield as a white solid. [
a]
+28.7° (c 1.25, CH2Cl2); 1H
a
D
NMR (CDCl3, 400 MHz) d 7.42 (s, 1H), 7.05 (s, 1H), 6.87 (s, 1H),
4.11 (d, J = 2.3 Hz, 1H), 4.01 (dd, J = 13.7, 3.6 Hz, 1H), 3.54 (dd,
J = 13.7, 9.2 Hz, 1H), 0.98 (s, 3H), 0.84 (d, J = 6.6 Hz, 3H); 13C NMR
(CDCl3, 100.6 MHz) d, 137.7, 129.1, 119.4, 68.9, 54.2, 52.7, 52.3,
42.1, 40.1, 38.0, 33.6, 27.3, 22.6, 17.3, 16.9, 13.5; exact mass calcu-
lated for C16H26N2O (M+) 262.2040, found 262.2050.
(1H, dd, J = 10.5, 6.8 Hz, 22-H), 1.94 (1H, br. d, J = 12.4 Hz), 1.02
(3H, d, J = 6.6 Hz, 21-H3), 0.95 (9H, t, J = 7.9 Hz), 0.92 (3H, s, 18-
H3), 0.55 (6H, q, J = 7.9 Hz); 13C NMR (100 MHz) d, 69.2, 67.9, 53.1,
52.8, 42.1, 40.6, 38.3, 34.6, 26.8, 23.0, 17.6, 16.6, 13.5, 6.9, 4.9; exact
mass calculated for C19H38O2Si (M+) 326.2641, found 326.2626.
2.2.5. (8S,20S)-de-A,B-8-triethylsilyloxy-20-formylpregnane (10)
To a solution of oxalyl chloride (1.11 g, 8.74 mmol) in DMSO
(1.2 mL) at ꢀ60 °C, a solution of the primary alcohol 9 (0.22 g,
0.67 mmol) in anhydrous CH2Cl2 (2 mL) at ꢀ60 °C was added via
cannula. The resulting mixture was stirred at ꢀ60 °C for 2 h,
quenched with Et3N (4.9 mL), and warmed up to room tempera-
ture. Upon dilution with H2O, the mixture was extracted with
CH2Cl2, dried (MgSO4), filtered, concentrated, and purified by flash
column chromatography (95:5 hexane/ethyl acetate; Rf = 0.12) to
give the desired aldehyde 10 (0.120 mg, 0.37 mmol, 54% yield) as
an oil.
2.2.2. (20S)-des-A,B-8–20-(10-methylimidazolyl)-pregnan-8-one (4)
Molecular sieves A4 (0.4 nm, 4 Å) (0.15 g) were added to a solu-
tion of 4-methylmorpholine (0.074 g, 0.63 mmol) in dichlorometh-
ane (3 mL). The mixture was stirred at room temperature for
15 min and tetrapropylammoniumperruthenate (TPAP) (2 mg,
0.006 mmol) was added, followed by a solution of alcohol 7
(0.02 g, 0.076 mmol) in dichloromethane (1 mL). The resulting sus-
pension was stirred at room temperature for 1 h. The reaction mix-
ture was filtered through a Waters silica Sep-Pack cartridge (5 g)
that was further washed with ethyl acetate/2-propanol (10:1).
After removal of the solvent the ketone 4 (0.015 g, 77% yield)
was obtained as a colorless oil. 1H NMR (CDCl3, 400 MHz) d 7.44
(s, 1H), 7.07 (s, 1H), 6.88 (s, 1H), 4.03 (dd, J = 13.8, 3.6 Hz, 1H),
3.54 (dd, J = 13.8, 9.0 Hz, 1H), 0.90 (d, J = 6.6 Hz, 3H), 0.69 (s, 3H);
13C NMR (CDCl3, 100.6 MHz) d, 211.3, 137.7, 129.3, 119.4, 61.5,
54.1, 52.5, 49.8, 40.8, 38.7, 38.1, 27.6, 23.8, 19.2, 17.3, 12.5; exact
mass calculated for C16H24N2O (M)+ 260.1884, found 260.1885.
1H NMR (200 MHz, CDCl3) d 9.57 (1H, d, J = 3.0 Hz, CHO), 4.06
(1H, d, J = 2.4 Hz,
8a-H), 2.38 (1H, m, 20-H), 1.09 (3H, d,
J = 6.8 Hz, 21-H3), 0.95 (12H, m, Si(CH2CH3)3 + 18-H3), 0.55 (6H, q,
J = 7.8 Hz, Si(CH2CH3)3).
2.2.6. (8S,20S)-de-A,B-8-triethylsilyloxy-20-[2-(methoxycarbonyl)-
et-(1E)-en-yl] pregnane (11)
To a solution of the aldehyde 10 (0.120 g, 0.37 mmol) in abso-
lute EtOH (3 mL) at 0 °C was added methyl(triphenylphosphorany-
lidene)-acetate (0.307 g, 0.92 mmol) and Et3N (0.037 g,
0.37 mmol). The mixture was stirred at rt for 24 h and then the sol-
vent was evaporated. The residue was purified by flash column
chromatography (95:5 hexane/ethyl acetate, Rf = 0.45) to obtain
11 (0.105 g, 0.27 mmol, 75% yield) as an oil.
2.2.3. (8S,20S)-de-A,B-8-triethylsilyloxy-20-(acetyloxymethyl)
pregnane (8)
Acetic anhydride (0.41 g, 0.40 mL, 4.0 mmol) was added to a
solution of the diol
5 (0.5 g, 2.3 mmol) and Et3N (1.64 mL,
11.7 mmol) in anhydrous CH2Cl2 (20 mL) at room temperature
(rt). The reaction mixture was stirred at rt for 24 h, diluted with
methylene chloride (100 mL), washed with 5% aq. HCl, water, sat-
urated aq. NaHCO3, dried (Na2SO4) and concentrated under re-
duced pressure. The residue (0.68 g) was chromatographed on
silica gel with hexane/ethyl acetate (75:25) to give the desired
alcohol (0.53 g, 88% yield) as a colorless oil.
[a]
+58.5 (c 2.37, CH2Cl2); 1H NMR (400 MHz, CDCl3 + TMS) d
D
6.83 (1H, dd, J = 15.1, 9.0 Hz), 5.73 (1H, d, J = 15.6 Hz), 4.03 (1H,
d, J = 2.4 Hz, 8 -H), 3.76 (3H, s) 1.94 (1H, br. d, J = 12.4 Hz), 1.05
a
(3H, d, J = 6.6 Hz, 21-H3), 0.95 (12H, m), 0.55 (6H, q, J = 7.9 Hz);
13C NMR (100 MHz) d, 167.6, 155.3, 118.4, 69.2, 55.5, 52.9, 51.3,
42.4, 40.6, 39.4, 34.6, 27.3, 23.0, 19.1, 17.6, 13.8, 6.9, 4.9; exact
mass calculated for C20H36O3Si (M+–Et) 351.2350, found 351.2366.
To a stirred solution of the alcohol (0.53 g, 2.1 mmol) and 2,6-lu-
tidine (0.29 mL, 0.26 g, 2.5 mmol) in anhydrous methylene chloride
(5 mL) triethylsilyl trifluoromethane-sulfonate (0.54 mL, 2.5 mmol)
was added at 0 °C. The reaction mixture was allowed to warm to
room temperature (1 h), and stirring was continued for additional
30 min. Methylene chloride was added and the mixture was washed
with water, dried (Na2SO4) and concentrated under reduced pres-
sure. The residue was chromatographed on silica gel with hexane/
ethyl acetate (97:3) to afford the protected diol 8 (0.74 g, 95% yield)
2.2.7. (8S,20R)-de-A,B-8-triethylsilyloxy-20-(hydroxypropyl)
pregnane (12)
A solution of the compound 11 (0.105 g, 0.27 mmol) in absolute
EtOH (5 mL) was hydrogenated for 9 h in the presence of 10% pal-
ladium on powdered charcoal (15 mg). The reaction mixture was
filtered through a bed of Celite with several ethanol washes, the fil-
trate was concentrated and the residue was chromatographed on
silica gel with hexane/ethyl acetate (97:3, Rf = 0.47) to give the sat-
urated ester (0.092 g, 0.24 mmol, 89% yield), which after dissolving
in THF (1 mL) was immediately exposed to reduction with LiAlH4
1 M in THF (0.48 mL, 0.48 mmol) at ꢀ10 °C. The reaction mixture
was stirred at rt for 3 h, then water (0.2 mL) and NaOH 1 M
(0.05 mL) were added, and the resulting suspension was filtered.
The evaporation of the solvent afforded the alcohol 12 (0.070 g,
0.20 mmol, 82% yield) as a clear oil.
as a colorless oil: [a]
+40.77 (c 4.9, CHCl3); 1H NMR (400 MHz,
D
CDCl3) d 4.06 (2H, m), 3.77 (1H, dd, J = 10.6, 3.1 Hz, 22-H), 2.05
(3H, s), 1.93 (1H, br. d, J = 12.4 Hz), 0.98 (3H, d, J = 6.6 Hz, 21-H3),
0.96 (9H, t, J = 7.9 Hz), 0.92 (3H, s), 0.56 (6H, q, J = 7.9 Hz); 13C
NMR (100 MHz) d, 171.4, 69.6, 69.2, 53.4, 52.8, 42.3, 40.6, 35.4,
34.6, 26.8, 23.1, 21.0, 17.7, 17.1,13.6, 6.9, 5.0; exact mass calculated
for C19H35O3Si (M–C2H5) 339.2355, found 339.2347.
2.2.4. (8S,20S)-de-A,B-8-triethylsilyloxy-20-(hydroxymethyl)
pregnane (9)
1H NMR (400 MHz, CDCl3 + TMS) d 4.03 (1H, d, J = 2.4 Hz, 8
a-H),
The protected diol 8 (0.58 g, 1.6 mmol) was then treated with a
solution of NaOH (1 g, 25 mmol) in anhydrous ethanol (20 mL) at
3.62 (2H, m) 1.96 (1H, br. d, J = 12.4 Hz), 0.99–0.90 (15H, m), 0.55
(6H, q, J = 7.9 Hz); 13C NMR (100 MHz) d, 69.4, 63.7, 56.7 53.1, 42.1,

216
G. Chiellini et al. / Steroids 77 (2012) 212–223
40.8, 35.1, 34.6, 31.7, 29.7, 27.3, 23.0, 18.6, 17.7, 13.5, 6.9, 4.9; ex-
act mass calculated for C21H42O2Si (M+) 354.2949, found 354.2943.
1H NMR (400 MHz, CDCl3 + TMS) d 3.15 (2H, m), 2.45 (2H, m),
1.46 (9H, s), 0.96 (3H, d, J = 5.7 Hz), 0.73 (2H, m), 0.64 (3H, s), 0.58
(2H, m); 13C NMR (100 MHz) d, 212.1, 156.8, 79.2, 61.9, 56.6, 49.9,
40.9, 47.8, 38.9, 35.3, 32.8, 28.5, 27.5, 24.8, 24.0, 19.0, 18.7, 12.5,
8.0, exact mass calculated for C23H39NO3Na (M+Na+) 400.2823,
found 400.2821.
2.2.8. (8S,20R)-de-A,B-8-triethylsilyloxy-20-[3-(cyclopropylamine)
propyl] pregnane (14)
To a solution of oxalyl chloride (0.34 g, 2.91 mmol) in DMSO
(1.2 mL) at ꢀ60 °C a solution of the primary alcohol 12 (0.081 g,
0.23 mmol) in anhydrous CH2Cl2 (2 mL) was added via cannula.
The resulting mixture was stirred at ꢀ60 °C for 2 h, quenched with
Et3N (4.9 mL), and warmed up to room temperature. Upon dilution
with H2O, the mixture was extracted with CH2Cl2, dried (MgSO4), fil-
tered, concentrated, and purified by flash column chromatography
(95:5 hexane/ethyl acetate; Rf = 0.55) to give the desired aldehyde
13 (0.046 g, 0.13 mmol, 57% yield) which was immediately used
for the next step. 1H NMR (200 MHz, CDCl3) d 9.76 (1H, s, CHO),
2.2.11. (20S)-2-methylene-22-(imidazol-1-yl)-19,23,24,25,26,27-
esanor-1a-hydroxyvitamin D3 (2)
To a solution of phosphine oxide 17 [26] (0.107 g, 0.184 mmol)
in anhydrous THF (0.4 mL) at ꢀ20 °C was slowly added PhLi (1.8 M
in di-n-butylether, 0.129 mL, 0.232 mmol) under argon with stir-
ring. The solution turned deep orange. After 30 min the mixture
was cooled to ꢀ78 °C and a precooled (ꢀ78 °C) solution of ketone
4 (0.015 g, 0.058 mmol) in anhydrous THF (0.2 + 0.1 mL) was
slowly added. The mixture was stirred under argon at ꢀ78 °C for
3 h and at 0 °C for 18 h. Ethyl acetate was added, and the organic
phase was washed with brine, dried (Na2SO4) and evaporated.
The residue was dissolved in hexane and applied on a Waters silica
Sep-Pack cartridge (2 g). The cartridge was washed with hexane
and hexane/ethyl acetate (90:10) to give 19-norvitamin derivative
18 (0.023 g, 63% yield). Then the Sep-Pack was washed with ethyl
acetate to recover diphenylphosphine oxide 17 (0.06 g). For analyt-
ical purpose a sample of protected vitamin 18 was further purified
by HPLC (9.4 ꢁ 2.50 mm Zorbax Sil column, 4 mL/min, hexane/2-
propanol/TEA (94:6:0.5) solvent system, RT = 11.09 min).
4.02 (1H, br. signal, 8a-H), 2.40 (2H, m, 23-H2), 0.94 (15H, m,
Si(CH2CH3)3 + 18-H3 + 21-H3), 0.54 (6H, q, J = 8.0 Hz, Si(CH2CH3)3).
To a solution of the aldehyde 13 (0.046 g, 0.13 mmol) in anhy-
drous CH2Cl2 (1 mL) was added cyclopropylamine (0.0092 g,
0.16 mmol) and the mixture was cooled at 0 °C before adding so-
dium triacetoxyborohydride (0.047 g, 0.22 mmol). The reaction
mixture was stirred at rt for 2 h. Then sat. aq. NaHCO3 solution
(1 mL) was added and the mixture was extracted with EtOAc.
The organic phase was dried (MgSO4), filtered, and the solvent
was evaporated to give 14 (0.049 g, 0.12 mmol, 95% yield).
[a]
+44.5 (c 1.12, CH2Cl2); 1H NMR (400 MHz, CDCl3 + TMS) d
D
4.02 (1H, d, J = 2.4 Hz, 8a-H), 2.65 (2H, m), 2.13 (1H, m), 0.95
(9H, t, J = 7.9 Hz), 0.89 (6H, m), 0.55 (6H, q, J = 7.9 Hz), 0.43
(2H, m), 0.36 (2H, m); 13C NMR (100 MHz) d, 69.4, 56.7, 53.1,
50.2, 42.1, 40.8, 35.2, 34.6, 33.4, 30.3, 27.3, 26.4, 23.0 18.6,
17.7, 13.5, 6.9, 6.2, 6.1, 4.9; exact mass calculated for C24H48NOSi
(M+H+) 394.3505, found 394.3506.
1H NMR (CDCl3, 800 MHz) d 7.43 (s, 1H), 7.05 (s, 1H), 6.88 (s,
1H), 6.21 (d, J = 11.2 Hz, 1H), 5.86 (d, J = 11.2 Hz, 1H), 4.99 (s,
1H), 4.93 (s, 1H), 4.44 (m, 2H), 4.05 (dd, J = 13.6, 4.0 Hz, 1H), 3.57
(dd, J = 13.6, 8.8 Hz, 1H), 2.85 (dm, J = 12.8 Hz, 1H), 2.54 (dd,
J = 13.6, 5.6 Hz, 1H), 2.48 (dd, J = 12.8, 4.0 Hz, 1H), 2.35 (dd,
J = 13.6, 3.2 Hz, 1H), 2.20 (dd, J = 2.8, 8.0. Hz, 1H), 0.92 (s, 9H),
0.89 (s, 9H), 0.86 (d, J = 6.4 Hz, 3H), 0.61 (s, 3H), 0.10 (s, 3H)
0.096 (s, 3H), 0.073 (s, 3H), 0.055 (s, 3H); 13C NMR (CDCl3,
200.9 MHz) d 153.1, 140.5, 137.8, 133.5, 129.4, 122.4, 119.7,
116.8, 106.6, 72.7, 71.8, 56.2, 54.3, 53.0, 47.8, 46.0, 40.5, 39.1,
38.8, 28.8, 28.0, 26.0, 23.5, 22.6, 18.5, 18.4, 17.4, 14.4, 12.3, ꢀ4.6,
ꢀ4.9; exact mass calculated for C37H65N2O2Si2 (MH)+ 625.4580,
found 625.4570.
2.2.9. (8S,20R)-de-A,B-20-[3-(cyclopropyl-N-t-Boc-amine) propyl]
pregnan-8-ol (15)
To a solution of compound 14 (0.033 g, 0.084 mmol) in CH3CN
(3 mL) Boc2O (0.022 g, 0.10 mmol) and DMAP (0.001 g,
0.0084 mmol) were added under vigorous stirring. After stirring
at rt for 1 h, the mixture was diluted with EtOAc, washed with
water and brine then dried (MgSO4). Concentration gave the de-
sired protected amine (0.040 g, 0.081 mmol, 96% yield). 1H NMR
(400 MHz, CDCl3 + TMS) d 4.02 (1H, d, J = 2.4 Hz, 8
m), 2.49 (1H, m), 1.95 (1H, br. d, J = 12.4 Hz), 1.46 (9H, s), 0.94
(9H, t, J = 7.9 Hz), 0.89 (6H, m), 0.73 (2H, m), 0.56 (8H, m).
a
-H), 3.14 (2H,
The protected vitamin 18 (0.023 g, 0.037 mmol) was dissolved
in acetonitrile (2 mL). A solution of aq. 48% HF in acetonitrile
(1:9 ratio, 2 mL) was added at 0 °C and the resulting mixture was
stirred at room temperature for 8 h. Saturated aq. NaHCO3 solution
was added and the reaction mixture was extracted with ethyl ace-
tate. The combined organic phases were washed with brine, dried
(Na2SO4) and concentrated under reduced pressure. The residue
was diluted with 2 mL of hexane/ethyl acetate (3:1) and applied
on a Waters silica Sep-Pack cartridge (2 g). An elution with ethyl
acetate/hexane (1:1) and later with ethylacetate/2-propanol
(10:1) gave the crude product 2 (0.014 g). The vitamin 2 was fur-
ther purified by normal phase HPLC [9.4 ꢁ 250 mm Zorbax Sil col-
umn, 4 mL/min, hexane/2-propanol/TEA (74:25:1) solvent system,
RT = 7.74 min] to give a colorless oil (0.012 g, 80% yield).
Removal of the TES was achieved by treating the protected alco-
hol (0.032 g, 0.065 mmol) dissolved in anhydrous THF (5 mL) with
TBAF 1 M in THF (130 mL, 0.13 mmol). After 5 h of stirring at rt the
reaction mixture was diluted with EtOAc, washed with brine, dried
(MgSO4) and concentrated under reduced pressure. The residue
was purified by flash column chromatography on silica gel with
hexane/ethyl acetate (95:5) to give the alcohol 15 (0.016 g,
0.042 mmol, 65% yield). 1H NMR (400 MHz, CDCl3 + TMS) d 4.07
(1H, s, 8a-H), 3.14 (2H, m), 2.48 (1H, m), 1.98 (1H, br. d,
J = 12.4 Hz), 1.46 (9H, s), 0.93 (3H, s), 0.91 (6H, m), 0.73 (2H, m),
0.57 (2H, m).
UV (in EtOH) kmax 261.0, 251.5, 244.0 nm; 1H NMR (CDCl3,
800 MHz) d 7.42 (s, 1H), 7.03 (s, 1H), 6.86 (s, 1H), 6.33 (d,
J = 11.2 Hz, 1H), 5.89 (d, J = 11.2 Hz, 1H), 5.10 (s, 1H), 5.08 (s, 1H),
4.47 (m, 2H), 4.02 (dd, J = 13.6, 4.0 Hz, 1H), 3.56 (dd, J = 13.6,
8.8 Hz, 1H), 2.85 (dd, J = 13.6, 4.8 Hz, 1H), 2.82 (dm, J = 12.8,
4.8 Hz, 1H), 2.57 (dd, J = 13.6, 4.0 Hz, 1H), 2.33 (dd, J = 13.6, 6.4 Hz,
1H), 2.29 (dd, J = 12.8, 8.0. Hz, 1H), 0.86 (d, J = 6.4 Hz, 3H), 0.60
(s, 3H); 13C NMR (CDCl3, 200.9 MHz) d, 152.2, 142.5, 137.9, 131.4,
129.3, 124.1, 119.7, 116.0, 108.0 72.0, 70.8, 56.2, 54.3, 53.0, 46.1,
46.0, 40.4, 39.0, 38.4, 29.0, 28.0, 23.5, 22.7, 17.4, 12.4; exact mass
calculated for C25H37N2O2 [MH]+ 397.2850, found 397.2851.
2.2.10. (8S,20R)-de-A,B-20-[3-(cyclopropyl-N-t-Boc-amine)propyl]
pregnan-8-one (16)
Pyridinium dichromate (PDC) (0.079 g, 0.211 mmol) was added
to a solution of the alcohol 15 (0.016 g, 0.042 mmol) and pyridini-
um p-toluenesulfonate (PPTS) (0.007 g, 0.030 mmol) in anhydrous
CH2Cl2 (5 mL). The resulting suspension was stirred at rt for 3 h.
The reaction mixture was filtered through a Waters silica Sep-Pack
cartridge (5 g) that was further washed with hexane/ethyl acetate
(95:5). After removal of the solvent the ketone 16 (0.014 g,
0.037 mmol, 88% yield) was recovered as a colorless oil.

G. Chiellini et al. / Steroids 77 (2012) 212–223
217
2.2.12. N-cyclopropyl-(20R)-2-methylene-19,25,26,27-tetranor-25-
aza-1 -hydroxyvitamin D (3)
25(OH)D3 (1.25–2.50
The HPLC mobile phase was in 7% 2-propanol in hexane.
l
M) and inhibitors, and 1.0 mM NADPH.
a
To a solution of phosphine oxide 17 (0.086 g, 0.148 mmol) in
anhydrous THF (0.5 mL) at ꢀ20 °C was slowly added PhLi (1.7 M
in di-n-butylether, 0.087 mL, 0.148 mmol) under argon with stir-
ring. The solution turned deep orange. After 30 min the mixture
was cooled at ꢀ78 °C and a precooled (ꢀ78 °C) solution of ketone
16 (0.014 g, 0.037 mmol) in anhydrous THF (0.2 + 0.1 mL) was
slowly added. The mixture was stirred under argon at ꢀ78 °C for
3 h and at 0 °C for 18 h. Ethyl acetate was added, and the organic
phase was washed with brine, dried (MgSO4) and evaporated.
The residue was dissolved in hexane and applied on a Waters silica
Sep-Pack cartridge (2 g). The cartridge was washed with hexane
and hexane/ethyl acetate (99.5:0.5) to give the 19-norvitamin 19
(0.021 g, 0.028 mmol, 76% yield).
2.3.3. Determination of the 1,25(OH)2D3 half-life
Primary mouse osteoblasts were isolated from fetal calvaria and
cultured to 70% confluency in aMEM containing 10% FBS in 24-well
plates (1 mL/well). Two days later, the culture was treated with
2.5 nM or 2.5 pM 1,25(OH)2D3 and a trace amount of [3H]-
1,25(OH)2D3 in the presence and absence of 1
or 1 M 3. At various time points (0, 8, 24, and 48 h), 200
medium was withdrawn and extracted with 800 L of dichloro-
l
M ketoconazole
l
l
L of
l
methane twice. The organic phase was dried down and separated
via HPLC in 15% 2-propanol in hexane as described above. The
residual [3H]-1,25(OH)2D3 in the medium over a 48 h period was
quantitated by counting the tritium.
1H NMR (CDCl3, 900 MHz) d 6.20 (1H, d, J = 10.8 Hz), 5.82 (1H, d,
J = 10.8 Hz,), 4.96 (1H, s,), 4.91 (1H, s,), 4.41 (2H, m), 3.15 (2H, m),
2.81 (1H, dm, J = 12.6 Hz), 2.52 (1H dd, J = 13.5, 6.3 Hz), 2.45
(1H, dd, J = 12.6, 4.5 Hz), 2.32 (1H, dd, J = 13.5, 2.7 Hz), 2.17 (1H,
dd, J = 12.6, 8.1. Hz), 1.46 (9H, s), 0.94 (3H, d, J = 7.2 Hz), 0.89
(9H, s), 0.86 (9H, s), 0.73 (2H, m), 0.58 (2H, m), 0.54 (3H, s) 0.098
(3H, s) 0.096 (3H, s), 0.073 (3H, s), 0.055 (3H, s); 13C NMR (CDCl3,
200.9 MHz) d 157.0, 153.2, 141.4, 133.0, 122.6, 116.4, 106.5, 72.7,
71.8, 56.7, 56.5, 47.8, 45.9, 40.8, 38.8, 36.1, 33.2, 29.0, 28.7, 27.9,
26.1, 23.6, 22.4, 19.0, 18.5, 18.4, 12.3, 8.5, ꢀ4.6, ꢀ4.9; exact mass
calculated for C44H79NO4Si2Na (MNa)+ 764.5440, found 764.5469.
The protected vitamin 19 (0.021 g, 0.028 mmol) was dissolved
in THF (2 mL) and CH3CN (2 mL). A solution of aq. 48% HF in CH3CN
(1:9 ratio, 2 mL) was added at 0 °C and the resulting mixture was
stirred at rt for 8 h. Saturated aq. NaHCO3 solution was added
and the reaction mixture was extracted with ethyl acetate. The
combined organic phases were washed with brine, dried (MgSO4)
and concentrated under reduced pressure. The residue was chro-
matographed on silica gel ethyl acetate/hexane (80:20) to give
the desired final product 3 (0.008 g, 0.019 mmol, 69% yield). The
vitamin 3 was further purified by HPLC (9.4 ꢁ 250 mm Zorbax Sil
column, 4 mL/min, hexane/2-propanol (80:20) solvent system,
RT = 8.50 min). UV (in EtOH) kmax 261.0, 251.5, 244.0 nm; 1H
NMR (CDCl3, 400 MHz) d 6.35 (1H, d, J = 11.2 Hz), 5.88 (1H, d,
J = 11.2 Hz), 5.11 (1H, s), 5.09 (1H, s), 4.62 (2H, m), 2.87–2.80
(2H, m), 2.70–2.55 (3H, m), 2.35–2.26 (2H, m), 2.12 (1H, m), 0.93
(3H, d, J = 6.3 Hz), 0.55 (3H, s), 0.42 (2H, m), 0.34 (2H, m); 13C
NMR (CDCl3, 100 MHz) d, 152.0, 143.4, 130.4, 124.2, 115.3, 107.7,
71.8, 70.6, 56.4, 56.3, 50.2, 45.8, 40.4, 38.2, 36.0, 33.5, 30.4, 28.9,
27.7, 26.6, 24.7, 23.5, 22.3, 18.8, 12.1, 6.2; exact mass calculated
for C27H44NO2 (MH)+ 413.3289, found 413.3297.
2.3.4. In vitro studies
VDR binding, CYP24A1 transcription, and HL-60 differentiation
assays were performed as previously described [28].
2.3.5. In vivo studies
Bone calcium mobilization and intestinal calcium transport
were performed as previously described [28]. Briefly, weanling rats
were made vitamin D-deficient by housing under lighting condi-
tions that block vitamin D production in the skin. In addition, the
animals were fed a diet devoid of vitamin D and alternating levels
of calcium. Experimental compounds were administered intraper-
itoneally once per day for four consecutive days. Twenty-four
hours after the last dose was given, the blood was collected, and
everted gut sacs were prepared. Calcium was measured in the
blood and two different intestinal compartments using atomic
absorption spectrometry. Each study was comprised of at least
5–6 animals/experimental group and was controlled with a vehicle
group (5% ethanol:95% propylene glycol) and one or more positive
control groups [1,25(OH)2D3].
2.4. Docking studies
The 3D structure of the human CYP24A1 was built with
swisspdbviewer [29] thanks to its homology to the crystal struc-
ture of the rat isoform (ID: 3K9V), which was obtained from PDB
[30] and taken as template structure. The homology based model-
ing protocol relied on a sequence alignment, obtained from the
web server Blast [31] (as shown in Fig. 4). The alignment score re-
vealed a sequence similarity of 83%, which ensured that a reliable
model for the human isoform could be obtained.
The sequence of the human isoform was loaded in swisspdb-
viewer together with the 3D structure of the crystallized rat iso-
form (3K9V). The alignment, shown in Fig. 4, enabled automatic
building of a model for the human CYP24A1, which was then re-
fined within the swisspdbviewer software. A theoretical model
for the human CYP27B1 isoform was downloaded from the MOD-
BASE server [32].
Molecular docking was then accomplished by means of the
molecular docking software, AutoDock version 4.0 [33]. The polar
hydrogens and united atom Kollman charges were assigned to
the enzymes during the preparation of the protein input files, con-
taining fragmental volume and solvation parameters. For the
ligands and the heme group, partial atomic charges were deter-
mined by the Gasteiger method with modification to ensure unit
charge on each residue. Moreover, rotatable bonds were assigned
with AutoDock Tools, an accessory program that allows the user
to interact with AutoDock from a graphic user interface. Prior to
the AutoDock, AutoGrid was exploited for the preparation of the
grid map using a grid box with a npts (number of points in xyz)
of 50–60–50 Å which defines the simulation space. The box
2.3. Biology
2.3.1. CYP24A1 inhibition assay
Assay reactions were performed in a buffer containing 20 mM
Tris (pH 7.5), 125 mM NaCl, 0.1
adrenodoxin reductase (AdR), 0.075
trations of 1,25(OH)2D3 (1.25–2.50
NADPH. The reactions were conducted and analyzed by HPLC as
previously reported [27]. The Ki values were determined by fitting
the relative activity (V/V0) against the inhibitor concentration [I]
using the equation V/V0 = (KM + [S])/[KM(1 + [I]/Ki) + [S]], where V
and V0 are the reaction rates in the presence and absence of inhib-
itors, respectively.
l
M adrenodoxin (Adx), 0.1
M CYP24A1, varying concen-
M) and inhibitors, and 0.5 mM
lM
l
l
2.3.2. CYP27B1 inhibition assay
Inhibition of CYP27B1 [27] was assayed in a manner similar to
that of CYP24A1 as described above except that the buffer con-
tained 20 mM Tris (pH 7.5), 125 mM NaCl, 0.1% CHAPS, 2
lM
Adx, 0.2 AdR, 0.025 CYP27B1, various amounts of
l
M
lM

218
G. Chiellini et al. / Steroids 77 (2012) 212–223
Fig. 4. Alignment between rat and human CYP24A1 sequences. Query: CYP24A1 human, Sbjct: CYP24A1 rat.
spacing was 0.375 Å and the grid was set in order to cover the en-
tire space of the binding site. A distance-dependent function of the
dielectric constant was used for the calculation of the energetic
maps. AutoDock was run using the maximum number of energy
evaluation retries and generations, 10,000 and 27,000, respec-
tively. The Lamarckian genetic algorithm (LGA) with the pseudo-
Solis and Wets modification (LGA/pSW) method was used with
default parameters for calculation of the docking possibilities [34].
OH
OTs
a
b
OH
OH
6
5
3. Results
N
N
N
N
3.1. Chemistry
c
The synthetic strategy for new analogs 2 and 3 was based on the
Lythgoe type Horner–Wittig olefination reaction [26], which has
been extensively utilized by us for the preparation of a large num-
ber of vitamin D compounds [25,35]. This approach required using
two specific Grundmann-type ketones 4 (Scheme 1) and 16
(Scheme 2), which we prepared from the known diol 5, obtained
in our laboratory from commercial vitamin D2 [25]. Selective tosy-
lation of the primary hydroxy group of diol 5 provided the corre-
sponding tosylate 6 (Scheme 1), which after addition of the
sodium salt of imidazole in dry DMF, gave the imidazole alcohol
7 in 95% yields. Catalytic oxidation of the secondary alcohol with
tetrapropylammonium perruthenate [36] in the presence of 4-
methylmorpholine N-oxide afforded the expected ketone 4 in very
good yields. In order to obtain the CD-ring ketone 16 we prepared
the aldehyde 10 starting from the diol 5 (Scheme 2). The Wittig
reaction between aldehyde 10 and methyl (triphenylphosphorany-
OH
O
7
4
Scheme 1. Synthesis of ketone 4. Reagents and conditions: (a) TsCl/pyridine/ꢀ25 °C.
(b) Imidazole/NaH/DMF/0 °C. (c) TPAP/NMO/molecular sieves/rt.
lidene)-acetate provided the olefinic product 11 that after catalytic
hydrogenation followed by LiAlH4 reduction furnished the primary
alcohol 12. Then, Swern oxidation [37] provided the aldehyde 13
that after reaction with cyclopropylamine and reduction of the
resulting Schiff base with triacetoxyborohydride provided the
CD-ring cyclopropylamine side chain modified intermediate 14.
After protection of the secondary amino group as a t-butyl (BOC)

G. Chiellini et al. / Steroids 77 (2012) 212–223
219
Table 1
Inhibition constants of vitamin D analogs toward CYP24A1 and CYP27B1.
Inhibitor
Ki (lM)
CYP24A1
CYP27B1
Selectivity
2
3
0.021 0.007
0.042 0.014
0.032 0.005
0.090 0.017
3.34 0.38
0.053 0.010
4.3
80
1.6
Ketoconazole
800
600
400
200
0
[I] = 0.5 µM
[I] = 1 µM
[I] = 2 µM
[I] = 5 µM
-0.5
0.0
0.5
1.0
1/[1,25(OH)2D3] (µM-1)
Fig. 5. Lineweaver–Burke plot showing competitive inhibition of CYP24A1 by
analog 3 (CPA1).
Scheme 2. Synthesis of ketone 15. Reagents and conditions: (a) Ac2O/Et3N/rt. (b)
TESOTf/ 2,6-lutidine/CH2Cl2/0 °C. (c) NaOH/EtOH/rt. (d) C2O2Cl2/DMSO/CH2Cl2/
ꢀ60 °C. (e) Ph3P = CHCOOMe/EtOH/Et3N/0 °C. (f) H2/Pd–C/EtOH/rt. (g) LiAlH4/THF/
ꢀ10 °C. (h) C2O2Cl2/DMSO/CH2Cl2/ꢀ60 °C. (i) cyclopropylamine/CH2Cl2/rt. (j)
Na(OAc)3BH/0 °C. (k) Boc2O/DMAP/CH3CN/rt. (l) TBAF/THF/rt. (m) PDC/CH2Cl2/rt.
recombinant hamster cell line, or rat kidney mitochondria) and
consequently suffer from experimental variability in enzyme con-
tent and the presence of physiological partners and other interact-
ing factors, we recently developed a simple, clean, and sensitive
enzymatic assay [27]. In short, variants of human CYP24A1 pro-
teins, with a C-terminal His tag or an N-terminal fusion to maltose
binding protein (MBP), were overexpressed in Escherichia coli and
purified to apparent homogeneity. The hydroxylase activity was
reconstituted in vitro, and the resulting cell-free assay system
was applied in the screening of 2, 3 and other vitamin D analogs
synthesized in our laboratory. Additional assays were also per-
formed to measure their inhibitory activity towards CYP27B1
[27]. As shown in Table 1, 2 and 3 along with ketoconazole exhib-
ited potent inhibition of CYP24A1, with Ki values ranging from
carbamate followed by selective cleavage of the triethylsilyl
protecting group treating with tetrabutylammonium fluoride
(nBu4N+Fꢀ, TBAF) the secondary alcohol 15, was obtained. Oxida-
tion of alcohol 15 with pyridinium dichromate afforded the desired
CD-ring ketone 16 (Scheme 2). As shown in Scheme 3 the Horner–
Wittig reaction between the corresponding C,D-fragments 4 and 16
and the anion, generated from the phosphine oxide 17 by phenylli-
thium [38], produced the protected vitamin D compounds 18–19.
The silyl-protective groups as well as the t-butyl (BOC) carbamate
were cleaved in the presence of hydrofluoric acid and after the final
purification by HPLC the target vitamin D analogs 2 (VIMI) and 3
(CPA1) were obtained.
0.021 lM (VIMI) to 0.042 lM (CPA1). Ketoconazole, which had a
3.2. Biological evaluation
3.2.1. Inhibition of CYP24A1 and CYP27B1
Since current assays for CYP24A1 inhibition rely on the use of
different preparations (e.g., primary human keratinocyte culture,
Fig. 6. Time course of residual 1,25(OH)2D3 in the medium of the mouse osteoblast
Scheme 3. Synthesis of 2-methylene-19-nor-1,25(OH)2D3 analogs 2 and 3. Reagents
and conditions: (a) PhLi/THF/ꢀ78 °C. (b) Aqueous HF/THF/MeCN/rt.
culture in the presence and absence of CYP24A1 inhibitors. Each value is
mean SD of three determinations.
a

220
G. Chiellini et al. / Steroids 77 (2012) 212–223
Table 2
The ability of 3 to sustain 1,25(OH)2D3 in the mouse further dem-
onstrated its effectiveness in CYP24A1 inhibition.
VDR binding properties. Transcriptional activities, and HL-60 differentiation activities
of 2 (VIMI) and 3 (CPA1), in comparison to those of 1 (1,25(OH)2D3).
VDR binding
CYP24A1 transcription
HL-60 differentiation
3.2.2. In vitro and in vivo evaluations of 19-nor-1,25-(OH)2D3 analogs
Ki (M)
Ratio
EC50 (M)
Ratio
EC50 (M)
Ratio
2 and 3
The newly synthesized 19-nor-1a,25-dihydroxyvitamin D3 ana-
1
2
3
2 ꢁ 10ꢀ11
7 ꢁ 10ꢀ8
3 ꢁ 10ꢀ9
1
3500
150
2 ꢁ 10ꢀ10
1
2 ꢁ 10ꢀ9
1
>500
4
>10ꢀ6
>5000
15
>10ꢀ6
logs were also exposed to a standard set of in vitro assays [28] to
ascertain their effects on receptor binding, cell differentiation
and gene transcription. As shown in Table 2, both 2 and 3 appeared
to bind the VDR with lower affinity than the native hormone (1),
with 2 also being very inactive in inducing both HL60 cell differen-
tiation and transcription of 24-hydroxylase in rat osteosarcoma
cells. Interestingly, 3 causes the differentiation of HL-60 cells with
the same potency as 1,25(OH)2D3 (1), whereas its transcriptional
activity is markedly lower than 1,25(OH)2D3 (1). Importantly, this
cell type difference in biological potency was also observed
in vivo where 3 (CPA-1), showed very little activity in bone, but
was quite potent in stimulating calcium transport in the intestine
(Fig. 7). In accordance to its reduced in vitro activity, 2 was also
very inactive in vivo (data not shown).
3 ꢁ 10ꢀ9
8 ꢁ 10ꢀ9
Ki value of 0.032
pressing CYP27B1 activity, with a Ki value of 0.053
lM toward CYP24A1, was equally effective at sup-
l
M. This was
also true with 2, which showed only a moderate selectivity of 4.3
for inhibition of CYP24A1 over CYP27B1. 3, on the other hand,
exhibited specific inhibition of CYP24A1, with substantial selectiv-
ity of 80 over CYP27B1. Further analysis revealed that 3 acts as a
competitive inhibitor of both enzymes (Fig. 5) and that prolonged
incubation with the enzymes did not exert extra inhibitory activ-
ity, suggesting that inhibition occurred instantly without substan-
tial conformational changes in the proteins.
To confirm the inhibitory effect of 3 toward CYP24A1 in vitamin
D responsive cells, we determined the half life of 1,25(OH)2D3 in
mouse osteoblast culture with and without a supplement of inhib-
itors. At high doses of 1,25(OH)2D3 treatment (2.5 nM), the pool of
residual 1,25(OH)2D3 continuously decreased in the medium and
eventually reached near depletion in 48 h, while 1,25(OH)2D3 re-
mained stable in the presence of 3 over the entire period (Fig. 6).
3.3. Docking study
A molecular docking study was performed in order to rationalize
the selectivity profile of compound 3 towards the hCYP24A1 isoen-
zyme, by comparing its ability to discriminate between the two iso-
enzymes of interest (hCYP24A1 and hCYP27B1). Indeed an optimal
Fig. 7. Effects of 1,25(OH)2D3 and compound 3 (CPA1) on bone calcium mobilization and intestinal calcium transport. Values shown are average SEM. Asterisks indicate
values that are statistically different from Vehicle control (p < 0.05, Dunnett’s).

G. Chiellini et al. / Steroids 77 (2012) 212–223
221
Table 3
Relevant information coming from structural analysis of the natural ligands Calcitriol
and Calcifediol docked into the sites of hCYP24A1 and hCYP27B1 isoenzymes
respectively, and the compound 3 docked in the two isoenzymes.
Isoenzyme Ligand
Interactions
with the
H bonds Orientation
enzyme
HEM520
HOH4
MET246
ALA326
GLU329
THR330
Calcitriol
MET246 In agreement with the
one reported in Ref.
[39]
VAL391
PHE393
THR394
SER498
GLY499
GLU329
hCYP24A1
HEM520
HOH4
LEU148
MET246
ALA326
GLU329
PHE393
HIS497
SER498
GLY499
3
HIS497
Calcitriol-like
HEM520
SER111
THR114
GLU115
ARG117
ARG118
Calcifediol LEU127
ASN387 In agreement with the
one reported in Ref.
[40]
THR128
ALA224
VAL225
PHE229
LEU316
VAL384
ASN387
ARG117
Fig. 8. (A) A comparison of docking results for analog 3 (CPA1) and the natural
ligand Calcidiol (1,25(OH)2D3) in the model of hCYP24A1. Analog 3 is capable of
docking the hCPA24A1 active site in the same general configuration as Calcidiol,
with the side chain positioned over the heme. (B) A comparison of docking results
for analog 3 (CPA1) and the natural ligand Calcifediol (25(OH)D3) in the model of
hCYP27B1. Analog 3 docked the hCYP27B1 active site with an orientation opposite
to that of Calcifediol, thus preventing it’s A-ring from interacting with the heme.
hCYP27B1
HEM520
CYS108
THR114
ARG117
LEU127
THR128
PHE221
ALA224
VAL225
VAL228
LEU233
PHE299
LEU316
SER388
ARG453
VAL494
simulation, the best scoring pose was further relaxed and then ana-
lyzed to identify the most relevant interactions (Table 3).
We found that Calcitriol is engaged within the catalytic center
of hCYP24A1 with its 25-hydroxylated side chain positioned over
the heme. As shown in Fig. 8A, its conformation is stabilized by
multiple hydrophobic contacts, and at least two hydrogen bonds
–
Calcifediol Reverse
(i.e., between the
1a-OH and MET246, and the 25-OH and
GLU329). Notably, the residues involved in the interaction are
the same as those suggested by other authors to be important for
secosteroid binding and catalysis in rat models [39].
Calcifediol binds to the active site of hCYP27B1 in a configura-
tion that positions the C-1 of the A-ring moiety toward the heme
group [40]. Its conformation is also stabilized by two hydrogen
bonds (i.e., between the 25-OH group and ARG117, and the 3b-
OH with ASN387 (Fig. 8B).
Finally, compound 3 was docked within the binding site of the
two isoenzymes using the same protocol as for the natural ligands,
and the resulting theoretical models were analyzed to identify
those features that enable this compound to discriminate between
the two enzyme isoforms.
CYP24A1 inhibitor must inhibit CYP24A1 with minimal interfer-
ence with the activity of CYP27B1. Therefore the possibility to bind
according to orientations suitable for catalysis was investigated for
compound 3 both in the hCYP24A1 and hCYP27B1 active sites.
To validate our theoretical models of compound 3 bound to the
two isoenzymes and to help elucidate the molecular details that
are important for secosteroid recognition, Calcitriol [1,25(OH)2D3,
the natural ligand of CYP24A1] and Calcifediol [25(OH)D3, the nat-
ural ligand of CYP27B1] were first docked within the binding site of
their respective enzymes. Once the orientation and conformation
of each ligand within the binding site were optimized by docking
As depicted in Fig. 8A, compound 3 docks very well within the
hCYP24A1 binding site, in a manner very similar to Calcitriol. Both

222
G. Chiellini et al. / Steroids 77 (2012) 212–223
molecules (the substrate and the inhibitor) show contacts to the
same enzyme residues (see Table 3) and have a similar orientation
within the hCYP24A1 binding site, with the side chain positioned
over the heme, thus allowing for catalysis.
On the other hand, our docking studies show that, whereas Cal-
cifediol docks within the active site of hCYP27B1 in a configuration
that positions the C-1 of the A-ring moiety toward the heme group
of the hCYP27B1, thus allowing the catalysis, compound 3 takes an
orientation not suitable for catalysis, projecting it’s A-ring far away
from the heme group (See Fig. 8B). Furthermore, no hydrogen
bonds are observed between the ligand groups and the enzyme
residues (Table 3).
relative to the heme prosthetic group, thus reducing the ability of
compound 3 to compete with the natural ligand for the binding
pocket.
Taken together, these results are in agreement with our exper-
imental data that shows that 3 is approximately 80 times more
selective towards inhibiting CYP24A1 over CYP27B1.
Unlike what has been described for many cyclopropyamine-
based inhibitors, time-dependent and irreversible inactivation
was not observed for our cyclopropylamine analog 3, which in-
stead acted as a competitive inhibitor (Fig. 5). To this end, it is
interesting to note that when we docked 3 into the binding site
of the human CYP24A1, we observed that the distance between
These findings provide an explanation for the observed selectiv-
ity of compound 3 in inhibiting the human CYP24A1 versus human
CYP27B1.
the Fe3+ and the
7.3 Å. Such a great distance would not favor either the
a
-carbon of the cyclopropylamino group was
a
-hydroxyl-
ation or single electron transfer cyclopropane-ring opening, which
are considered the most accredited mechanisms for cyclopropyl-
amine suicidal inactivation of cytochrome P-450 enzymes [41].
Ongoing studies in our laboratory are directed toward optimiz-
ing the lead structure with respect to specificity, selectivity, effi-
cacy and absence of toxicity in appropriate model systems. If
successful these studies can result in compounds suitable to enter
pre-clinical and clinical drug development programs.
4. Discussion
CYP24A1 plays a key role in tuning the levels and function of
1,25(OH)2D3; inhibition of CYP24A1 opens up a very wide field of
possible applications ranging from basic research to the prevention
and treatment of diseases. There is general agreement that unbal-
anced high and/or long-lasting expression of CYP24A1 can contrib-
ute to the pathology of diseases that otherwise would respond to
endogenous or supplement vitamin D in a favorable way like
chronic kidney disease, bone disease, cancers, and psoriasis. In
these cases, inhibition of CYP24A1 could be the appropriate strat-
egy to increase the lifetime and thereby the function of vitamin D.
In this work, we showed that Azole-type (2) and cyclopropylamine
Acknowledgments
The work was supported in part by funds from the Wisconsin
Alumni ResearchFoundation. We gratefully acknowledge Jean Prahl,
and Jennifer Vaughan, for carrying out the in vitro studies, and Erin
Gudmundson for conducting the in vivo studies. We thank Dr. Mark
Anderson for his assistance in recording NMR spectra. This study
made use of the National Magnetic Resonance Facility at Madison,
which was supported by the NIH Grants P41RR02301 (BRTP/NCRR)
and P41GM66326 (NIGMS). Additional equipment was purchased
with funds from the University of Wisconsin, the NIH (RR02781
and RR08438), the NSF (DMB-8415048, OIA-9977486, and BIR-
9214394), and the US Department of Agriculture. The authors grate-
fully acknowledge and wish to extent their deep appreciation to
Prof. Aldo Balsamo for critically reviewing this manuscript.
(3) derivatives of 2-methylene-19-nor-1a,25-dihydroxyvitamin D3
are very potent CYP24A1 inhibitors. Consistent to what has been
reported in literature for others azole-type CYP inhibitors, the
imidazole side chain modified analog 2 displayed only a moderate
selectivity toward CYP24A1 over CYP27B1 while being basically
inactive as a 1,25(OH)2D3 analog. On the other hand, the introduc-
tion of a cyclopropylamino group on the side chain of 2-methy-
lene-19-nor-1a,25-dihydroxyvitamin D3 proved to be
a good
strategy at producing 1,25(OH)2D3 analogs with desirable physio-
logical characteristics. In fact, 3 combined the ability to specifically
inhibit CYP24A1, thus preventing 24-hydroxylation of 1,25(OH)2D3
and extending the hormone half life in vitro, with a potency to in-
duce differentiation of human promyelocytic leukemia HL-60 cells,
and showing lower potency in raising calcium tissue levels. There-
fore, this new cyclopropylamine entity 3, shows promise to be used
either alone or in combination with other chemotherapeutic
agents in the treatment or prevention of specific types of cancer.
A molecular docking study was also performed to rationalize the
selectivity profile of 3 towards the CYP24A1 isoenzyme. It was
found that 3 docked with high affinity to human CYP24A1 (Energy
ꢀ8.18 kcal/mol) with an orientation that is highly similar to Calci-
triol, with the side chain positioned in close proximity to the heme
group (Fig. 8A). In particular, the binding of 3 within the active site
is stabilized by a hydrogen bond between the 3b-OH and His497,
and multiple hydrophobic interactions with key conserved residues
(Table 3). However, compound 3 did not show noticeable turnover
under experimental conditions in the inhibition assay [27], suggest-
ing that a longer and bulkier side chain may impair CYP24A1’s abil-
ity to acquire the closed form needed for catalysis, and simply
enabling the compound to compete with the natural substrate for
the binding pocket. These insights correlate well with previous
observations that CYP24A1 requires a hydroxyl group at C25 to
introduce the C24 vicinal hydroxyl group [39].
References
[1] Deluca HF. Evolution of our understanding of vitamin D. Nutr Rev
2008;66:S73–87.
[2] Holick MF. The vitamin
nonskeletal health: mechanisms of action. Mol Aspects Med 2008;29:361–8.
[3] Bikle D. Nonclassic actions of vitamin D.
2009;94:26–34.
[4] Prosser DE, Jones G. Enzymes involved in the activation and inactivation of
vitamin D. Trends Biochem Sci 2004;29:664–73.
[5] Eeln G, Gysemans C, Verlinden L, Vanoibeek E, De Clercq P, Van Haver D, et al.
Mechanisms and potential of the growth inhibitory actions of vitaminD and
analogs. Curr Med Chem 2007;14:1893–910.
[6] Jones G. Vitamin D and analogues. In: Bilezikian JP, Raisz LG, Martin TJ, editors.
Principles of bone biology. Amsterdam: Elsevier Academic Press; 2008. p.
1777–99.
[7] Deeb KK, Trump DL, Johnson CS. Vitamin D signaling pathways in cancer:
potential for anticancer therapeutics. Nat Rev Cancer 2007;7:684–700.
[8] Reinhardt TA, Horst RL. Ketoconazole inhibits self-induced metabolism of 1,25-
dihydroxyvitamin D3 and amplifies 1,25-dihydroxyvitamin D3 receptor up-
regulation in rat osteosarcoma cells. Arch Biochem Biophys 1989;272:459–65.
[9] Peehl DM, Seto E, Hsu JY, Feldman D. Preclinical activity of ketoconazole in
combination with calcitriol or the vitamin D analogue EB 1089 in prostate
cancer cells. J Urol 2002;168:1583–8.
D deficiency pandemic and consequences for
J Clin Endocrinol Metab
[10] Ly LH, Zhao XY, Holloway L, Feldman D. Liarazole acts synergistically with
1a,25-dihydroxyvitamin D3 to inhibit growth of DU 145 human prostate
cancer cells by blocking 24-hydroxylase activity. Endocrinology
1999;140:2071–6.
[11] Schuster I, Egger H, Bikle D, Herzig G, Reddy GS, Stuetz A, et al. Selective
inhibition of vitamin
2001;66:409–22.
[12] Schuster I, Egger H, Nussbaumer P, Kroemer RT. Inhibitors of vitamin D
D hydroxylases in human keratinocytes. Steroids
On the other hand, when compound 3 is complex with
CYP27B1, its orientation within the active site (Fig. 8B) is not a
good mimic for Calcifediol with respect to the position of the A ring
hydroxylases:
2003;88:372–80.
structure–activity
relationships.
J
Cell
Biochem

G. Chiellini et al. / Steroids 77 (2012) 212–223
223
[13] Yee SW, Simons C. Synthesis and CYP24 inhibitory activity of 2-substituted-
3,4-dihydro-2H-naphthalen-1-one (tetralone) derivatives. Bioorg Med Chem
Lett 2004;14:5651–4.
[27] Zhu J, Barycki R, Chiellini G, Deluca HF. Screening of selective inhibitors of
,25-dihydroxyvitamin D3 24-hydroxylase using recombinant human
1a
enzyme expressed in Escherichia coli. Biochemistry 2010;49(49):10403–11.
[14] Aboraia AS, Yee SW, Gomaa MS, Shah N, Robotham AC, Makowski B, et al.
Synthesis and CYP24A1 inhibitory activity of N-(2-(1H-imidazol-1-yl)-2-
phenylethyl)arylamides. Bioorg Med Chem 2010;18:4939–46.
[15] Posner GH, Crawford KR, Yang HW, Kahraman M, Jeon HB, Li H, et al. Potent,
low-calcemic, selective inhibitors of CYP24 hydroxylase: 24-sulfone analogs of
[28] Glebocka A, Sicinski RR, Plum LA, Clagett-Dame M, Deluca HF. New 2-
alkylidene 1a,25-dihydroxy-19-norvitamin D3 analogues of high intestinal
activity: Synthesis and biological evaluation of 2-(30-alkoxypropylidene) and
2-(30-hydroxypropylidene) derivatives. J Med Chem 2006;49:2909–20.
[29] Guex N, Peitsch MC. SWISS-MODEL and the Swiss-PdbViewer: an environment
for comparative protein modeling. Electrophoresis 1997;18:2714–23.
[30] Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, et al. The
Protein Data Bank. Nucleic Acids Res 2000;28:235–42.
the hormone
1a,25-dihydroxyvitamin D3. J Steroid Biochem Mol Biol
2004;89–90:5–12.
[16] Kahraman M, Sinishtaj S, Dolan PM, Kensler TW, Peleg S, Saha U, et al. Potent,
selective and low-calcemic inhibitors of CYP24 hydroxylase: 24-sulfoximine
[31] blast.ncbi.nlm.nih.gov/.
analogues of the hormone
2004;47:6854–63.
[17] Hanzlik RP, Kishore V, Tullman R. Cyclopropylamines as suicide substrates for
cytochromes P-450. J Med Chem 1979;22:759–61.
1
a
,25-dihydroxyvitamin D3.
J
Med Chem
[32] Pieper U, Narayanan E, Davis FP, Braberg H, Madhusudhan MS, Rossi A, et al.
MODBASE: a database of annotated comparative protein structure models and
associated resources. Nucleic Acids Res 2006;34:291–5.
[33] Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew RK, et al.
Automated docking using Lamarckian genetic algorithm and an empirical
binding free energy function. J Comput Chem 1998;19:1639–62.
[34] Solis FJ, Wets RJB. Minimization by Random Search Techniques. Math Oper Res
1981;6:19–30.
[18] Angelastro MR, Laughlin ME, Schatzman GL, Bey P, Blohm TR. 17b-
(Cyclopropylamino)-androst-5-en-3b-ol,
inhibitor of cytochrome P45017 (steroid 17
Biochem Biophys Res Commun 1989;162:1571–7.
a
selective mechanism-based
a-hydroxylase/C17–20 lyase).
a
[19] Shah MA, Trackman PC, Gallop PM, Kagan HM. Reaction of lysyl oxidase with
trans-2-phenylcyclopropylamine. J Biol Chem 1993;268:11580–5.
[20] Njar VCO, Duerkop J, Hartmann RW. Novel 19-(cyclopropylamino)-androst-4-
en-3,17-dione: a mechanism-based inhibitor of aromatase. J Enzyme Inhib
1995;10:47–56.
[35] Grzywacz P, Chiellini G, Plum LA, Clagett-Dame M, Deluca HF. Removal of the
26-methyl group from 19-nor-1a,25-dihydroxyvitamin D3 markedly reduces
in vivo calcemic activity without altering in vitro VDR binding, HL-60 cell
differentiation, and transcription. J Med Chem 2010;53(24):8642–9.
[36] Ley SV, Norman J, Griffith WP, Marsden SP. Tetrapropylammonium
[21] Mitchell DJ, Nikolic D, Jang MH, van Breemen RB, Hille R, Silverman RB.
perruthenate, Pr4NRuO4, TPAP:
Synthesis 1994:639–66.
a catalytic oxidant for organic synthesis.
Inactivation of C30A trimethylamine dehydrogenase by N-cyclopropyl-a-
methylbenzylamine, 1-phenylcyclopropylamine, and phenylhydrazine.
Biochemistry 2001;40:8523–30.
[22] Totah RA, Hanzlik RP. Non-oxidative decarboxylation of glycine derivatives by
a peroxidase. J Am Chem Soc 2002;124:10000–1.
[37] Posner GH, Crawford KR, Peleg S, Welsh JE, Romu S, Gewirtz DA, et al. A non-
calcemic sulfone version of the vitamin D(3) analogue seocalcitol (EB 1089):
chemical synthesis, biological evaluation and potency enhancement of the
anticancer drug adriamycin. Bioorg Med Chem 2001;9:2365–71.
[38] Sicinski RR, Prahl JM, Smith CM, Deluca HF. New 1alpha, 25-dihydroxy-19-
norvitamin D3 compounds of high biological activity: synthesis and biological
evaluation of 2-hydroxymethyl, 2-methyl, and 2-methylene analogues. J Med
Chem 1998;41:4662–74.
[23] Vintem APB, Price NT, Silverman RB, Ramsay RR. Mutation of surface cysteine
374 to alanine in monoamine oxidase
A alters substrate turnover and
inactivation by cyclopropylamines. Bioorg Med Chem 2005;13:3487–95.
[24] Polasek TM, Elliot DJ, Somogyi AA, Gillam EMJ, Lewis BC, Miners JO. An
evaluation of potential mechanism-based inactivation of human drug
metabolizing cytochromes P450 by monoamine oxidase inhibitors, including
isoniazid. Br J Clin Pharmacol 2006;61:570–84.
[39] Annalora AJ, Goodin DB, Hong WX, Zhang Q, Johnson EF, Stout CD. Crystal
structure of CYP24A1, a mitochondrial cytochrome P450 involved in vitamin D
metabolism. J Mol Biol 2010;396:441–51.
[25] Chiellini G, Grzywacz P, Plum LA, Barycki R, Clagett-Dame M, DeLuca HF.
Synthesis and biological properties of 2-methylene-19-nor-25-dehydro-
1alpha-hydroxyvitamin D(3)-26,23-lactones–weak agonists. Bioorg Med
Chem 2008;16(18):8563–73.
[40] Yamamoto K, Uchida E, Urushino N, Sakaki T, Kagawa N, Sawada N, et al.
Identification of the amino acid residue of CYP27B1 responsible for binding of
25-hydroxyvitamin D3 whose mutation causes vitamin D-dependent rickets
type 1. J Biol Chem 2005;280:30511–6.
[26] Lythgoe B, Moran TA, Nambudiry MEN, Ruston S. Allylic phosphine oxides as
precursors of conjugated dienes of defined geometry. J Chem Soc Perkin Trans I
1976:2386–90.
[41] Lee HB, Sung MJ, Blackstock SC, Cha JK. Radical cation-mediated annulation.
Stereoselective construction of bicyclo[5.3.0]decan-3-ones by aerobic
oxidation of cyclopropylamines. J Am Chem Soc 2001;123:11322–4.