(3R,5S,7as)-(3,5-bis(4-fluorophenyl)tetrahydro-1h-oxazolo[3,4-c] oxazol-7a-yl)methanol, a novel neuroprotective agent

Article abstract of DOI:10.1021/jm900254k

Compounds that interact with microtubules, such as paclitaxel, have been shown to possess protective properties against β-amyloid (Aβ) induced neurodegeneration associated with Alzheimer's disease. In this work, the novel agent (3R,5S,7as)-(3,5-bis(4-fluo

Full text of DOI:10.1021/jm900254k

J. Med. Chem. 2009, 52, 7537–7543 7537
DOI: 10.1021/jm900254k
(3R,5S,7as)-(3,5-Bis(4-fluorophenyl)tetrahydro-1H-oxazolo[3,4-c]oxazol-7a-yl)methanol,
a Novel Neuroprotective Agent^
Kelly E. Desino,† Sabah Ansar,§ Gunda I. Georg,*,z Richard H. Himes,‡ Mary Lou Michaelis,§ Douglas R. Powell,
Emily A. Reiff,z Hanumaiah Telikepalli,z and Kenneth L. Audus†
†Department of Pharmaceutical Chemistry, The University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047, §Department of
Pharmacology and Toxicology, The University of Kansas, 1251 Wescoe Hall Drive, Lawrence, Kansas 66045, zDepartment of Medicinal
Chemistry, The University of Kansas, 1251 Wescoe Hall Drive, Lawrence, Kansas 66045, ‡Department of Molecular Biosciences, The University
of Kansas, 1200 Sunnyside Avenue, Lawrence, Kansas 66045, and X-ray Crystallography Laboratory, The University of Kansas, 1251 Wescoe
Hall Drive Lawrence, Kansas 66045
Received February 28, 2009
Compounds that interact with microtubules, such as paclitaxel, have been shown to possess protective
properties against β-amyloid (Aβ) induced neurodegeneration associated with Alzheimer’s disease. In
this work, the novel agent (3R,5S,7as)-(3,5-bis(4-fluorophenyl)tetrahydro-1H-oxazolo[3,4-c]oxazol-
7a-yl)methanol was investigated for effectiveness in protecting neurons against several toxic stimuli and
its interaction with the microtubule network. Exposure of neuronal cultures to Aβ peptide in the
presence of 5 nM (3R,5S,7as)-(3,5-bis(4-fluorophenyl)tetrahydro-1H-oxazolo[3,4-c]oxazol-7a-yl)me-
thanol resulted in a 50% increase in survival. Neuronal cultures treated with other toxic stimuli such as
staurosporine, thapsigargin, paraquat, and H2O2 showed significantly enhanced survival in the presence
of (3R,5S,7as)-(3,5-bis(4-fluorophenyl)tetrahydro-1H-oxazolo[3,4-c]oxazol-7a-yl)methanol. Micro-
tubule binding and tubulin assembly studies revealed differences compared to paclitaxel but confirmed
the interaction of (3R,5S,7as)-(3,5-bis(4-fluorophenyl)tetrahydro-1H-oxazolo[3,4-c]oxazol-7a-yl)me-
thanol with microtubules. Furthermore, in vitro studies using bovine brain microvessel endothelial cells
experiments suggest that (3R,5S,7as)-(3,5-bis(4-fluorophenyl)tetrahydro-1H-oxazolo[3,4-c]oxazol-
7a-yl)methanol can readily cross the blood-brain barrier in a passive manner.
Introduction
There are a variety of barriers to brain penetration including
anatomical barriers, i.e., tight junctions, reduced pinocytosis,
and glycocalyx,3,4 metabolic barriers, i.e., peptidases, and
phase I and phase II enzymes,5 and an electrostatic barrier
due to sulfated glycoproteins at the cell surface. Another
major barrier to brain penetration is the existence of efflux
transporters expressed at the brain endothelium.6 If a drug is
recognized as a substrate for one of these efflux transporters,
the permeation across the blood-brain barrier can be sig-
nificantly compromised.7
In 1997, Shintani et al. reported that a small synthetic
compound named GS-164 stimulated microtubule assembly
in vitro by a mechanism similar to that of paclitaxel.8 Under
the conditions used by us to prepare a sample of GS-164, a
mixture of two diastereoisomers was obtained consisting of
meso compound (3R,5S,7as)-(3,5-bis(4-fluorophenyl)tetra-
hydro-1H-oxazolo[3,4-c]oxazol-7a-yl)methanol (1), racemate
(3R*,5R*)-(3,5-bis(4-fluorophenyl)tetrahydro-1H-oxazolo-
[3,4-c]oxazol-7a-yl)methanol (2), and (2-(4-fluorophenyl)ox-
azolidine-4,4-diyl)dimethanol (3), an incomplete condensa-
tion product (Figure 1). On the basis of modeling studies, the
authors had proposed that the R,R-enantiomer of compound
2 mimics parts of the paclitaxel structure and is responsible for
its paclitaxel-like activity. Our laboratory has now separated
the diastereoisomers present in GS-164 and has investi-
gated the properties of the (3R,5S,7as)-isomer 1 (Figure 1).
The compound exhibits neuroprotective properties when
It had previously been suggested that stabilization of
microtubules by antimitotic drugs could protect neurons from
the toxic effects of β-amyloid (Aβa) peptide, a peptide
believed to play a major role in the pathogenesis of Alzhei-
mer’s disease.1 Indeed, we reported a study in which we
demonstrated that microtubule stabilizing drugs such as
taxanes, epothilones, and discodermolide had a protective
effect against Aβ-induced degeneration of primary neurons in
culture.2 However, the success of a neuroprotective agent in
vivo as a pharmaceutical product depends on the ability of
the compound to reach the site of action within the brain.
^ This article is dedicated to the 100th anniversary of the American
Chemical Society Division of Medicinal Chemistry.
*To whom correspondence should be addressed. Current address:
Department of Medicinal Chemistry, University of Minnesota, 717
Delaware Street SE, Minneapolis, MN 55414. Phone: (612) 626-6320.
Fax: (612) 626-6318. E-mail: georg239@umn.edu.
a Abbreviations: Aβ, β-amyloid; BBMEC, bovine brain microvessel
endothelial cells; CsA, cyclosporine A; DAPI, 40,60-diamino-2-pheny-
lindole; DMEM/F12 Dulbecco’s modified Eagle’s medium F12,
DMSO, dimethyl sulfoxide; GTP, guanosine triphosphate; MAPs mi-
crotubule-associated proteins; PBS, phosphate buffered saline; PBSA
phosphate buffered saline with ascorbic acid; PIPES, piperazine-1,4-
bis(2-ethanesulfonic acid); EGTA, ethylene glycol bis(2-aminoethyl
ether)-N,N,N0,N0-tetraacetic acid; PEM, buffer containing piperazine-
1,4-bis(2-ethanesulfonic acid), ethylene glycol bis(2-aminoethyl ether)-
N,N,N0,N0-tetraacetic acid, and MgSO4; P-gp, P-glycoprotein; PI, pro-
pidium iodide; Tris/Cl, tris(hydroxymethyl)aminomethane HCl.
r
2009 American Chemical Society
Published on Web 09/03/2009
pubs.acs.org/jmc

7538 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
Desino et al.
Figure 1. Reaction scheme for the synthesis 1. Additional products are 2 and 3.
Figure 2. Structures and ellipsoid drawings of meso diastereoi-
somer (3R,5S,7as)-1 and (3R*,5R*)-2) from X-ray diffraction.
tested with cultured neurons and passively permeates the
blood-brain barrier. Interestingly, although this compound
binds weakly to microtubules, it does not show paclitaxel-like
microtubule stabilizing activity in vitro.
Results
Synthesis of 1. The reaction of p-fluorobenzaldehyde with
tris(hydroxymethyl)aminomethane led to three products
(Figure 1) that were separable by flash chromatography.
Compounds 1, 2, and 3 were produced in yields of 66%, 5%,
and 21%, respectively. Single crystal X-ray diffraction was used
to determine the stereochemistry of 1 and 2 (Figure 2). Thus,
the major product was the 3R,5S,7as meso compound 1. The
3R,5S,7ar meso compound, although a potential product, was
not formed. This is consistent with energy calculations.9
Microtubule Assembly and Stability. In the original article
on GS-164 the authors concluded that the compound acted
on tubulin in a manner similar to that of paclitaxel.8 In
addition, the authors used modeling to suggest that the
3R,5R isomer was responsible for the paclitaxel-like activity.
We have found that this isomer is a minor product in the
synthesis. We sought to determine whether the major pro-
duct, the 3R,5S,7as isomer 1, had paclitaxel-like activity. We
first tested the ability of 1 to stimulate tubulin assembly using
microtubule protein (tubulin containing microtubule asso-
ciated proteins) and pure tubulin. Microtubule assembly was
measured in two different ways, using a fluorescence assay
with microtubule protein without GTP (guanosine
triphosphate), and a centrifugation assay with pure tubulin
plus GTP and microtubule protein without GTP. Shown in
Figure 3. Stimulation of microtubule protein assembly by paclita-
xel (top) and 1 (bottom) as determined by monitoring the increase in
DAPI fluorescence. Microtubule protein (2 mg/mL) was incubated
in PEM buffer containing 10 μM DAPI and varying concentrations
of paclitaxel or 1 in 96-well plates for 30 min at 37 ꢀC. Maximum
fluorescence is that which is achieved by 25 μM paclitaxel.
Figure 3 are the results of the fluorescence assay. Compound
1 produced a small positive response (apparent microtubule
formation) at low concentrations, but the response did not
increase with increasing concentrations of 1. The maximum
increase in fluorescence was about 20% of that produced by
paclitaxel. In the centrifugation assays, the amount of
pelleted protein in the presence of 1 ranged from 10% to
20% of that formed in the presence of paclitaxel. Electron

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7539
microscopic examination indicated some fibrous material in
the presence of 1, indicating that the compound induces agg-
regation of tubulin but not microtubule formation. Similar
results were obtained with 2 and 3 (data not presented).
Paclitaxel is known to stabilize microtubules against cold-
induced depolymerization. As another means to compare the
taxane and 1, we studied the cold stability of microtubules
formed in vitro. Microtubules were formed from pure tubu-
lin at 37 ꢀC. Upon centrifugation, 81% of the protein was
found in the pellet. When the sample was placed on ice for
15 min before centrifugation, only 8% of the protein was in
the pellet fraction. The addition of 20 μM paclitaxel to the
microtubule suspension before the cold treatment stabilized
the microtubules; the pellet contained 79% of the protein.
On the other hand, the addition of 400 μM 1 had no effect on
cold stability; 9% of the protein was found in the pellet. The
results clearly demonstrate that 1 does not possess the
paclitaxel-like property of stabilizing microtubules against
cold-induced depolymerization.
8 nmol/mg protein. Since microtubule protein is about 70%
tubulin, 1 mg represents 0.7 mg of tubulin or 7 nmol. Thus,
both compounds bind in about a 1 to 1 molar ratio, but about
10-fold more 1 than paclitaxel was required to achieve
maximum binding. To determine whether the two com-
pounds bind to the same site in tubulin, competitive binding
experiments were performed. Microtubules were formed in
the presence of 100 μM [3H]1 and 0-40 μM paclitaxel and in
the presence of 20 μM [3H]paclitaxel and 0-400 μM 1.
Binding of 20 μM paclitaxel was reduced 40% by 400 μM
1. Binding of 100 μM 1 was reduced 42% by 40 μM
paclitaxel. Thus, the two compounds appear to be weakly
competitive with each other.
Protection of Primary Cortical Neurons against Aβ Peptide
and Cell-Death Initiators. We and others have shown that
exposure to Aβ induces beading, shortening, and eventual
disappearance of neuritic processes presumably due to dis-
ruption of the cytoskeletal network, and we have also found
that paclitaxel protects against Aβ toxicity.10-12 Although 1
did not have paclitaxel activity in vitro, it did protect primary
cortical neurons exposed to toxic agents. Protection against
Aβ25-35 peptide is shown in Figure 5. The EC50 for 1
(concentration that leads to a 50% increase in neuronal
survival in the presence of the Aβ peptide) was 5 nM. The
racemic mixture 2 and the incomplete condensation product
3 had EC50 values of 50 and 40 nM, respectively. Although
the small toxic Aβ25-35 peptide was used in the majority of
studies, the observations were confirmed with exposure to
Aβ1-42 peptide. Because 1 was the most effective of the
three compounds and was the major product in the synthesis,
further studies were restricted to this compound.
Binding of 1 to Microtubules. With the use of tritiated
compounds, the binding of 1 and paclitaxel to microtubules
formed from microtubule protein was measured (Figure 4).
Both compounds bound to microtubules in a ratio of 7 to
The effectiveness of 1 in protecting primary neurons
against diverse toxic stimuli that lead to cell death via various
pathways was also investigated. 1 was added at a concentra-
tion of 100 nM approximately 2 h before the toxic stimuli,
and cell viability was determined. As shown in Figure 6,
exposure of the neurons to the protein kinase C inhibitor
staurosporine, the sarcoplasmic-endoplasmic Ca2þ-AT-
Pase inhibitor thapsigargin, and the oxidative stress-indu-
cing agents paraquat and hydrogen peroxide all led to a
50% loss in neurons within 24 h. In cultures treated with 1
before addition of the toxic stimuli, neuronal survival was
Figure 5. Dose dependent effects of 1 on survival of primary
cortical neurons exposed to the toxic β-amyloid (Aβ) peptide for
48 h. The percentage of neurons surviving was determined by the
live/dead assay. Data are from six randomly selected fields from two
neuronal preparations (n ≈ 1500 cells/condition). Statistical sig-
nificance was determined by Students t test: (#) p < 0.001 for
controls versus Aβ only; (//) p < 0.001 for Aβ only versus Aβ þ
drug.
Figure 4. Binding of paclitaxel (top) and 1 (bottom) to microtu-
bules. Microtubules were formed from microtubule protein (2 mg/
mL) and incubated with different concentrations of [3H]paclitaxel
or [3H]1 for 10 min at 37 ꢀC. The microtubules were collected by
centrifugation and dissolved in 0.1 M NaOH. Protein concentra-
tions and radioactivity measurements were then determined.

7540 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
Desino et al.
Figure 7. Arrhenius plot of the apparent permeability coefficients
(Papp) of 1 in bovine brain microvessel endothelial cells at tempera-
tures ranging from 4 to 37 ꢀC over 1.5 h: data ( SD, n = 4.
[3H]1 was investigated in the luminal to abluminal direction
at temperatures ranging from 4 to 37 ꢀC over 1.5 h to
generate an Arrhenius plot (Figure 7). The activation energy
calculated from the slope of the Arrhenius plot was 26.3 kJ/
mol, which is suggestive of a passive process, since carrier-
mediated processes typically fall within a range of 29-105
kJ/mol.14
Figure 6. Effects of 1 on toxicity induced by diverse stimuli.
Primary neurons were exposed to 100 nM staurosporine (Str),
100 nM thapsigargin (Tg), 25 μM paraquat (PQ), or 25 μM H2O2,
( 100 nM 1. Mean percent survival was determined 24 h later, and
statistical significance was assessed using Student’s t test: (/) p <
0.05 and (//) p < 0.005 for each toxic stimulus alone versus the toxic
stimulus plus the drug. Data are from ∼1000 cells per condition.
significantly enhanced and, in most cases, the percentage of
live cells was nearly at control levels. Thus, the protective
effect of 1 is not selective for the cell death pathway induced
by Aβ peptides but seems to be due to a more generalized
cellular response that enables neurons to withstand stressful
conditions.
Discussion
Following the report by Shantani et al.8 that GS-164
promoted the assembly of tubulin and stabilized microtubules
in vitro, we sought to test the ability of the compound as a
neuroprotective agent. When we tested the three products of
GS-164 synthesis, we could not find evidence that any of them
promoted the assembly of tubulin. We did find that 1, the R,S
stereoisomer of GS-164, binds to microtubules in a molar
ratio of about one 1/tubulin molecule. However, it binds with
an apparent affinity that is about 10% that of the affinity of
paclitaxel and appears to be only partially competitive with
the taxane. Moreover, 1 did not stabilize microtubules in
vitro. In spite of these differences, 1, like paclitaxel, has
excellent neuroprotective properties against Aβ and other
toxic stimuli. Since 1 must be able to cross the blood-brain
barrier to exert its neuroprotective effects in neurodegenera-
tive diseases, the potential for 1 to penetrate the blood-brain
barrier was thoroughly investigated using in vitro methodol-
ogies. In vitro studies performed using BBMECs as a model of
the blood-brain barrier indicated that 1 had excellent poten-
tial for brain penetration. The physiochemical properties of 1
make it a potential candidate for use as a CNS drug. Com-
pared to paclitaxel, 1 has a much lower molecular weight,
possesses halogen groups that promote passive diffusion,15
and is not a substrate for any of the efflux transporters
expressed at the brain endothelium. All of these features are
important in determining whether a drug will be able to cross
the blood-brain barrier and reach a site of action within the
brain.7,15,16
Effects of 1 on Rhodamine 123 Uptake. Given the promis-
ing observations seen in the neuroprotective assays, addi-
tional studies were performed to investigate the ability of 1 to
cross the blood-brain barrier. The rhodamine 123 assay
assesses the interaction of a compound with the efflux
transporter P-glycoprotein, which is expressed at the brain
endothelium and can present a barrier to brain penetration.
Paclitaxel is a known substrate of P-glycoprotein (P-gp) and
consequently has very minimal brain penetration.13 Com-
pound 1 was tested in the rhodamine 123 uptake assay using
bovine brain microvessel endothelial cells (BBMECs) grown
in a 24-well plate. The uptake of rhodamine 123 was mon-
itored in the presence of 1 at concentrations ranging from
1 to 50 μM. No significant increase in rhodamine 123 was
observed except for the positive control, cyclosporine A
(CsA), a known substrate for P-gp (data not shown). These
data suggest that 1 is not recognized as a substrate for the
P-gp efflux transporter.
Permeability of 1. Bidirectional permeability studies were
used to predict the brain permeation of 1 and to determine
if 1 permeates via an active or passive process. Compound 1
(10 μM) with trace levels of [3H]1 was added to the luminal or
abluminal side of the BBMEC monolayer, and permeation
was monitored for 1.5 h. During this time, permeation across
the monolayer was linear (r2=0.99) and all experiments were
performed with n=4. In the luminal to abluminal direction,
Papp=2.86 ꢀ 10-4 cm/s. In the abluminal to luminal direc-
tion, Papp=2.44 ꢀ 10-4 cm/s. This indicates that there is no
statistical difference in permeability as a result of direction.
These data suggest that the permeability of 1 is a passive
process, since permeation via an active process would yield
directional differences in permeability.
The observation that 1 protects neurons against Aβ toxicity
is similar to what we have found with known microtubule-
stabilizing drugs such as paclitaxel.2,10,17 The in vitro studies
suggest that 1 interacts with microtubules differently than
paclitaxel does. Exposure to 1 led to the preservation of the
cytoskeletal integrity in primary neurons exposed to Aβ. Our
previously published work using calcein labeling demon-
strated that Aβ leads to the actual loss of neurites and not
just a loss of tubulin labeling.2,10,17 The neuritic dystrophy
attests to the complexity of the signaling events involved in the
To confirm that permeation of 1 is indeed a passive
process, the permeability of 1 (10 μM) with trace levels of

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7541
neuronal response to Aβ and suggests that 1 may be acting on
a different pathway than paclitaxel in leading to preservation
of the neurites. Furthermore, though it is reasonable to
assume that paclitaxel ultimately protects neurons by stabiliz-
ing microtubules, the cellular mechanism may be more com-
plex, possibly involving unique effects on specific signaling
pathways. For example, we have found that paclitaxel pre-
vents the Aβ-induced activation of Cdk5, one of the proline
kinases strongly implicated in the abnormal phosphorylation
ofthe τproteinfound intangles in Alzheimer’s disease brain.18
Diverse experimental approaches have provided strong
evidence that dysregulation of Ca2þ homeostasis is a driving
force for neuronal death in neurodegenerative diseases,19 and
paclitaxelwas showntoprotect neurons againstCa2þ influx,20
against glutamate-induced excitotoxicity,21 and against thap-
sigargin-induced endoplasmic reticulum stress and dysfunc-
tion,12 among other toxic insults. It is not yet known how this
occurs, but specific Ca2þ signaling pathways may be sensitive
to the state of the cytoskeletal network in neurons. Agents as
structurally different as paclitaxel and 1 could modulate very
different Ca2þ signaling cascades in intact cells. Although 1
does not mimic paclitaxel in vitro, this agent apparently
activates unique protein interactions and signaling cascades,
which ultimately leads to the preservation of neuritic pro-
cesses. As previously suggested,22 the state of the neuronal
cytoskeleton may serve as a surveillance system that trans-
duces multiple types of stress signals. Agents that invoke
mechanisms that slow or prevent the neuritic dystrophy
initiated by such signals give the neurons a chance to mobilize
other defenses and thereby enhance their survival. Experi-
mental strategies designed to delineate, at the molecular level,
the mechanisms through which agents such as 1 and paclitaxel
activate protective signaling cascades may well provide
opportunities for innovative therapeutic interventions that
capitalize on these targets.
as a colorless crystalline solid: mp 90 ꢀC; IR ν cm-1 3444, 2873,
1606, 1504, 1222, 1153, 1078, 1006, 837; 1H NMR (CDCl3) δ 3.50
(br s, 2H), 3.94(d, J=8.9 Hz, 2H), 4.05 (d, J=8.9 Hz, 2H), 5.55 (s,
2H), 7.07 (m, 4H), 7.44 (m, 4H); 13C NMR (CDCl3) δ 65.63,
72.66, 74.89, 96.65, 115.41, 115.58, 128.55, 128.62, 135.26, 135.29,
161.95, 163.91; HRMS (FABþ) m/z calculated for C18H18NO3F2
[M þ H] 334.1255, found 334.1230.
(3R*,5R*)-(3,5-Bis(4-fluorophenyl)tetrahydro-1H-oxazolo[3,
4-c]oxazol-7a-yl)methanol (2). was crystallized from hexanes/
CH2Cl2 as a colorless crystalline solid: mp 115 ꢀC; IR ν cm-1
1
3440, 2870, 1606, 1512, 1427, 1384, 1226, 1155, 1076, 837; H
NMR (CDCl3) δ 3.75 (m, 2H), 3.84 (d, J=8.8 Hz, 1H), 3.88 (d,
J=8.9 Hz, 1H), 4.10 (d, J=8.9 Hz, 1H), 4.20 (d, J=8.8 Hz, 1H),
5.16 (s, 1H), 5.51 (s, 1H), 6.89 (m, 6H), 7.27 (m, 2H); 13C NMR
(CDCl3) δ 65.60, 72.12, 74.87, 93.03, 93.91, 115.14, 115.31,
115.35, 115.52, 115.78, 129.21, 129.29, 129.34, 129.42, 130.21,
130.24, 135.87, 135.90, 161.89, 164.34; HRMS (FABþ) m/z
calculated for C18H18NO3F2 [M þ H] 334.1255, found 334.1237.
(2-(4-Fluorophenyl)oxazolidine-4,4-diyl)dimethanol (3). 3 was
crystallized from isopropanol as a colorless crystalline solid: mp
89 ꢀC; IR ν cm-1 3450, 2877, 1604, 1512, 1420, 1296, 1228, 1157,
1047, 837; 1H NMR (CDCl3) δ 3.58 (m, 3H), 3.71 (m, 2H), 3.85
(d, J=8.6 Hz, 1H), 5.41 (s, 1H), 7.06 (m, 2H), 7.44 (m, 2H); 13
C
NMR (CDCl3) δ 64.57, 64.98, 67.62, 70.73, 91.79, 115.81,
116.02, 128.29, 128.37, 134.91, 162.17, 164.63; HRMS
(FABþ) m/z calculated for C11H15NO3F [M þ H] 228.1036,
found 228.1036.
Microtubule Assembly Assay. Microtubule protein (tubulin
preparation containing microtubule-associated proteins (MAPs))
was obtained from bovine brain.23 Pure tubulin was iso-
lated from microtubule protein by a previously described pro-
cedure.24 Microtubule assembly in vitro was measured using
two different assays. One assay takes advantage of the fact that
40,6-diamidino-2-phenylindole (DAPI) binds to microtubules,
producing an increase in fluorescence.25 The reactions were
performed in 96-well plates in a volume of 120 μL per well.
The wells contained PEM buffer (0.1 M PIPES, 1 mM MgSO4,
1 mM EGTA, pH 6.9), 4% DMSO, 10 μM DAPI, 2 mg/mL
microtubule protein, and varying concentrations of paclitaxel or
1. The plates were incubated at 37 ꢀC for 30 min, after which the
fluorescence was measured in a multiplate reader. The readings
were corrected for a control lacking either compound. A cen-
trifugation assay was also employed in which microtubules are
pelleted after the assembly reaction (data not shown).26
Microtubule Binding Assay. Binding of 1 and paclitaxel to
microtubules was measured using tritiated compounds. [3H]Pac-
litaxel was purchased from Moraveck Biochemicals, Inc., Brea,
CA, and 1 was tritiated by ViTrax Co., Placentia, CA. Solu-
tions (100 μL) of 2 mg/mL microtubule protein in PEM
buffer containing 0.5 mM GTP and various concentrations of
[3H]1 (1 ꢀ 105 cpm/nmol) or [3H]paclitaxel (9 ꢀ 104 cpm/nmol)
in Beckman TLA-100 centrifuge tubes were incubated at 37 ꢀC
for 10 min. Microtubule pellets were collected by centrifugation
at 100000g for 4 min. The pellets were dissolved in 0.1 M NaOH,
and radioactivity and protein concentrations were determined.
Neuronal Cell Culture. To prepare the primary neurons,
dissociated cortical cell cultures were established from embryo-
nic day 18 rat fetuses recovered from pregnant Sprague-Daw-
ley rats (Harlan Sprague-Dawley, Inc., Indianapolis, IN).27
After the final precipitation step, neurons were suspended in
fresh DMEM/F12 (Sigma Chemical Co., St. Louis, MO) with
10% fetal bovine serum (Atlanta Biologicals, Atlanta, GA) and
plated at a density of 2.5 ꢀ 105 cells in 35 mm glass-bottom
microwell dishes (Mat-Tek Co., Ashland, MA), coated with
poly-D-lysine. Serum-containing medium was removed after
24 h, and the cells were maintained in serum-free DMEM/F12
containing the N2 supplements. Cultures were grown at 37 ꢀC in
5% CO2 and 97% humidity as described.27
Experimental Section
Chemistry. Analytical Characterization. 1H and 13C NMR
spectra were recorded on a 400 MHz NMR instrument (400 and
100 MHz respectively). High-resolution mass spectra (HRMS)
were obtained on a ZAB double-focusing mass spectrometer. IR
spectra were recorded on a FTIR instrument. Melting points are
uncorrected. Column chromatography was performed employ-
ing silica gel (230-400 mesh). The purities of 1 and 2 were
determined by HPLC and were found to be >95% pure. X-ray
diffraction data were collected on a CCD area detector using
˚
graphite-monochromated Mo KR radiation (λ = 0.710 73 A).
CCDC 706431 (1) and 706432 (2) contain the supplementary
crystallographic data for this paper. These data can be obtained
free of charge from The Cambridge Crystallographic Data
Centre via www.ccdc.cam.ac.uk/data_request/cif).
Synthesis of 1. A suspension of p-fluorobenzaldehyde (27 g,
0.22 mol)and tris(hydroxymethyl)aminomethane(13g, 0.11mol)
intoluene(350 mL) was heated at refluxtemperature for 12 h with
azeotropic removal of water. The reaction mixture was concen-
trated and stirred at room temperature, and the precipitated
unreacted p-fluorobenzaldehyde was removed by filtration. The
residue obtained after removal of solvent was subjected to flash
chromatography on silica gel using hexane/ethyl acetate (4:1 and
1:1). Crystallization of the respective fractions afforded pure
compounds (1, 2, and 3) in yields of 66%, 5%, and 21%,
respectively. The structures of the products were assigned by
spectral data, and the relative stereochemistry of 1 and 2 was
assigned from single crystal X-ray diffraction data.
(3R,5S,7as)-(3,5-Bis(4-fluorophenyl)tetrahydro-1H-oxazolo-
[3,4-c]oxazol-7a-yl)methanol (1). 1 was crystallized from hexane
Treatment with Aβ Peptide and Cell-Death Initiators. After
5 days in culture, the primary neurons were exposed to either

7542 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
Desino et al.
Aβ25-35 or Aβ1-42 in the presence or absence of concentra-
tions of 1 ranging from 0.5 to 60 nM. 1 was added approximately
2 h before the Aβ peptides. The Aβ25-35 was synthesized and
purified in the Biochemical Research Services Lab at The
University of Kansas. The reverse sequence peptide used as a
control, Aβ35-25, and the Aβ1-42 peptide used in confirma-
tory experiments were purchased from Bachem (Torrance, CA).
Prior to the addition of the peptides to the cultures, the Aβ
peptide stocks (1.3 mg/mL) were diluted into 10 mM Tris/Cl, pH
7.4, and maintained at 37 ꢀC for 24 h. Each batch of Aβ peptide
was analyzed for β-sheet formation by circular dichroism, but
no attempt was made to separate oligomers from fibrils. The
peptides were added directly to the culture medium, at a final
concentration of 10 μM. Control cultures received the DMSO
vehicle alone, and the final concentration of DMSO never
exceeded 0.04%. Assays were carried out 48 h following Aβ
peptide addition. The effects of neurotoxic stimuli other than Aβ
peptides were also used to assess the protective effect of 1 against
various cell-death initiators. Neurotoxic stimuli included staur-
osporine (100 nM) to induce apoptosis, thapsigargin (100 nM)
to cause ER stress, and paraquat (25 μM) or hydrogen peroxide
(25 μM) to induce oxidative stress. All reagents were obtained
from Sigma Chemical Co., St Louis, MO.
Measurement of Cell Viability. The effects of the Aβ peptides
and 1 were determined by monitoring neuronal cell survival
using the live/dead assay as previously described.2,10 Following
exposure to the peptides and other compounds, cells were
labeled with 20 μM propidium iodide (PI) and 150 nM calcein
acetoxy methyl ester (Molecular Probes, Eugene, OR) for
30 min at 37 ꢀC, rinsed with phosphate buffered saline (PBS),
and visualized under a Nikon inverted microscope (Nikon
Eclipse TE200, Japan) with filters for fluorescein isothiocyanate
and Texas red. Digital images were captured, and the number of
viable (green) and dead (red) neurons was determined by
counting the cells in 6-12 microscopic fields per culture dish
in duplicate dishes for each treatment. All experimental treat-
ments were carried out on at least two separate embryonic
neuronal preparations with approximately 1500 neurons scored
under each treatment condition. The raw data from each
experiment were combined, and the significance of differences
between cultures exposed to various treatments was determined
as described using Student’s t test.2 Neuronal survival in the
untreated control samples was considered to represent maximal
viability, and survival in the Aβ-only samples represents
minimal viability.
10 μL aliquots were removed from all wells to determine the
protein concentration using a Pierce BCA protein assay kit
(Pierce, Rockford, IL).
BBMEC Permeability Assays. BBMECs were grown on
0.4 μm Nucleopore polycarbonate membranes in a 100 mm
culture dish coated with rat tail collagen and fibronectin. Once
cells had formed a confluent monolayer as determined by light
microscopy, the membranes were transferred to Side-Bi-Side
diffusion chambers as previously described.31,32 Briefly, each
chamber was filled with 3 mL of PBSA and either the luminal
(blood) or abluminal (brain) donor chambers contained 10 μM
unlabeled 1 and a trace amount of [3H]1. The temperature was
maintained at 37 ꢀC within the chamber with an external
circulating water bath, and chamber contents were stirred with
Teflon coated magnetic stir bars driven by an external console.
At various time points up to 90 min, 200 μL aliquots were
removed from the receiver side and 20 μL aliquots of the donor
solution were taken at time zero. The integrity of the cell
monolayer was tested after experiment by monitoring the
permeability of [14C]sucrose, a low permeability paracellular
marker that should not readily cross the cell monolayer. All
samples were analyzed by liquid scintillation counting. For
temperature dependent permeability studies BBMECs were
grown on 0.4 μm Nucleopore polycarbonate membranes and
then transferred to Side-Bi-Side diffusion chambers as described
above. 1 (10 μM) with trace amounts of [3H]1 was added to the
luminal chambers, and 200 μL samples were taken from the
abluminal chamber at various time points up to 90 min. The
apparent permeability of 1 was calculated at a range of tem-
peratures from 4 to 37 ꢀC. An Arrhenius plot was generated to
determine the activation energy of transport. The integrity of the
cell monolayer was tested after experiment by monitoring the
permeability of [14C]sucrose, and samples were analyzed by
liquid scintillation counting.
Acknowledgment. We thank Jacquelyn Huff for excellent
technicalassistance. Supportfor thisresearchwas providedby
the Alzheimer’s Disease and Related Disorders Association,
the Institute for the Study of Aging, and NIH Grants HD
02528, CA 79641, and CA82801. E.A.R. acknowledges sup-
ported from the NIH Training Grant GM 07775, the Army
Breast Cancer Initiative Predoctoral Fellowship Program
(Grant DAMD 17-001-0303), and the American Association
for Pharmaceutical Education Predoctoral Fellowship Pro-
gram.
Isolation and Cell Culture of Brain Endothelial Cells. BBMECs
were isolated from the gray matter of bovine cerebral cortices by
enzymatic digestion followed by centrifugation as described.28
Isolated BBMECs were seeded at a density of approximately
50000 cells/cm2 onto 24-well culture plates or 100 mm culture
dishes as previously described.29
References
(1) Lee, V. M.; Daughenbaugh, R.; Trojanowski, J. Q. Microtubule
stabilizing drugs for the treatment of Alzheimer’s disease. Neuro-
biol. Aging 1994, 15 (Suppl. 2:S), 87–89.
Rhodamine 123 Uptake Assay. In this assay, rhodamine 123 is
used as a surrogate P-glycoprotein (P-gp) substrate.30 The effect
of the test compound on rhodamine 123 is determined by
monitoring intracellular fluorescence. If the test compound is
a substrate for P-gp, then addition of the compound will
increase rhodamine 123 uptake relative to the negative control.
BBMECs were seeded onto 24-well culture plates and used to
assay for P-gp interaction.29 Briefly, BBMECs were rinsed with
warm PBSA (phosphate buffered saline containing CaCl2,
MgSO4, glucose, and L-ascorbic acid) and allowed to acclimate
for 10 min. The PBSA was removed, and solutions of 1 (1 μM- 50
μM) were added to the wells and allowed to incubate for 30 min
at 37 ꢀC. Rhodamine 123 was then added to all wells at a
concentration of 5 μM, and cells were incubated for an addi-
tional 2 h. After the incubation, cells were quickly rinsed with ice
cold PBS several times. Lysis buffer (0.5% v/v Triton X-100 in
0.2 M NaOH) was then added to all wells and allowed to
solubilize cells for at least 30 min. Aliquots (200 μL) were taken
from all wells to determine rhodamine 123 concentration, and
(2) Michaelis, M. L.; Ansar, S.; Chen, Y.; Reiff, E. R.; Seyb, K. I.;
Himes, R. H.; Audus, K. L.; Georg, G. I. beta-Amyloid-induced
neurodegeneration and protection by structurally diverse micro-
tubule-stabilizing agents. J. Pharmacol. Exp. Ther. 2005, 312, 659–
668.
(3) Reese, T. S.; Karnovsky, M. J. Fine structural localization of a
blood-brain barrier to exogenous peroxidase. J. Cell. Biol. 1967,
34, 207–217.
(4) Brightman, M. W.; Reese, T. S. Junctions between intimately
apposed cell membranes in the vertebrate brain. J. Cell Biol.
1969, 40, 648–677.
(5) Spatz, M.; Mrsula, B. Progress in cerebral microvascular studies
related to the function of the blood-brain barrier. Adv. Cell.
Neurobiol. 1982, 3, 311–337.
(6) Pardridge, W. M. CNS drug design based on principles of
blood-brain barrier transport. J. Neurochem. 1998, 70, 1781–
1792.
(7) Hitchcock, S. A.; Pennington, L. D. Structure-brain exposure
relationships. J. Med. Chem. 2006, 49, 7559–7583.
(8) Shintani, Y.; Tanaka, T.; Nozaki, Y. GS-164, a small synthetic
compound, stimulates tubulin polymerization by a similar mechanism
to that of Taxol. Cancer. Chemother. Pharmacol. 1997, 40, 513–520.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7543
(9) Darabantu, M.; Maiereanu, C.; Silaghi-Dumitrescu, I.; Toupet, L.;
(21) Sponne, I.; Fifre, A.; Drouet, B.; Klein, C.; Koziel, V.; Pincon-
Raymond, M.; Olivier, J. L.; Chambaz, J.; Pillot, T. Apoptotic
neuronal cell death induced by the non-fibrillar amyloid-beta
peptide proceeds through an early reactive oxygen species-depen-
dent cytoskeleton perturbation. J. Biol. Chem. 2003, 278, 3437–
3445.
ꢀ
ꢀ
Condamine, E.; Ramondenc, Y.; Berghian, C.; Ple, G.; Ple, N. 3,7-
Dioxa-1-azabicyclo[3.3.0]octanes substituted at the C-5 position;
from local to global stereochemistry. Eur. J. Org. Chem. 2004,
2644–2661.
(10) Michaelis, M. L.; Ranciat, N.; Chen, Y.; Bechtel, M.; Ragan, R.;
Hepperle, M.; Liu, Y.; Georg, G. Protection against beta-amyloid
toxicity in primary neurons by paclitaxel (Taxol). J. Neurochem.
1998, 70, 1623–1627.
(11) Grace, E. A.; Rabiner, C. A.; Busciglio, J. Characterization of
neuronal dystrophy induced by fibrillar amyloid beta: implications
for Alzheimer’s disease. Neuroscience 2002, 114, 265–273.
(12) Seyb, K. I.; Ansar, S.; Bean, J.; Michaelis, M. L. beta-Amyloid and
endoplasmic reticulum stress responses in primary neurons: effects
of drugs that interact with the cytoskeleton. J. Mol. Neurosci. 2006,
28, 111–123.
(13) van Asperen, J.; Mayer, U.; van Tellingen, O.; Beijnen, J. H. The
functional role of Pglycoprotein in the blood-brain barrier.
J. Pharm. Sci. 1997, 86, 881–884.
(14) L’Hostis, C.; Geindre, M.; Deshusses, J. Active transport of
L-proline in the protozoan parasite Trypanosoma brucei brucei.
Biochem. J. 1993, 291 (Part 1), 297–301.
(15) Raub, T. J. P-Glycoprotein recognition of substrates and circum-
vention through rational drug design. Mol. Pharm. 2006, 3, 3–25.
(16) Mahar, D. K. M.; Humphreys, J. E.; Webster, L. O.; Wring, S. A.;
Shampine, L. J.; Serabjit-Singh, C. J.; Adkison, K. K.; Polli, J. W.
Passive permeability and P-glycoprotein-mediated efflux differ-
entiate central nervous system (CNS) and non-CNS marketed
drugs. J. Pharmacol. Exp. Ther. 2002, 303, 1029–1037.
(17) Michaelis, M. L.; Chen, Y.; Hill, S.; Reiff, E.; Georg, G.; Rice, A.;
Audus, K. Amyloid peptide toxicity and microtubule-stabilizing
drugs. J. Mol. Neurosci. 2002, 19, 101–105.
(18) Li, G.; Faibushevich, A.; Turunen, B. J.; Yoon, S. O.; Georg, G.;
Michaelis, M. L.; Dobrowsky, R. T. Stabilization of the cyclin-
dependent kinase 5 activator, p35, by paclitaxel decreases beta-
amyloid toxicity in cortical neurons. J. Neurochem. 2003, 84, 347–
362.
(22) Michaelis, M. L.; Seyb, K. I.; Ansar, S. Cytoskeletal integrity as a
drug target. Curr. Alzheimer Res. 2005, 2, 227–229.
(23) Tiwari, S. C.; Suprenant, K. A. A pH- and temperature-dependent
cycling method that doubles the yield of microtubule protein. Anal.
Biochem. 1993, 215, 96–103.
(24) Algaier, J.; Himes, R. H. The effects of dimethyl sulfoxide on the
kinetics of tubulin assembly. Biochim. Biophys. Acta 1988, 954, 235–243.
(25) Barron, D. M.; Chatterjee, S. K.; Ravindra, R.; Roof, R.; Baloglu, E.;
Kingston, D. G.; Bane, S. A fluorescence-based high-throughput assay
for antimicrotubule drugs. Anal. Biochem. 2003, 315, 49–56.
(26) Liu, Y.; Ali, S. M.; Boge, T. C.; Georg, G. I.; Victory, S.; Zygmunt,
J.; Marquez, R. T.; Himes, R. H. A systematic SAR study of C10
modified paclitaxel analogues using a combinatorial approach.
Comb. Chem. High Throughput Screening 2002, 5, 39–48.
(27) Michaelis, M. L.; Walsh, J. L.; Pal, R.; Hurlbert, M.; Hoel, G.;
Bland, K.; Foye, J.; Kwong, W.þH. Immunologic localization and
kinetic characterization of a Na /Ca2þ exchanger in neuronal and
non-neuronal cells. Brain Res. 1994, 661, 104–116.
(28) Audus, K. L.; Borchardt, R. T. Bovine brain microvessel endothe-
lial cell monolayers as a model system for the blood-brain barrier.
Ann. N.Y. Acad. Sci. 1987, 507, 9–18.
(29) Rose, J. M.; Peckham, S. L.; Scism, J. L.; Audus, K. L. Evaluation
of the role of Pglycoprotein in ivermectin uptake by primary
cultures of bovine brain microvessel endothelial cells. Neurochem.
Res. 1998, 23, 203–209.
(30) Fontaine, M.; Elmquist, W. F.; Miller, D. W. Use of rhodamine
123 to examine the functional activity of P-glycoprotein in primary
cultured brain microvessel endothelial cell monolayers. Life Sci.
1996, 59, 1521–1531.
(31) Audus, K. L.; Ng, L.; Wang, W.; Borchardt, R. T. Brain Micro-
vessel Endothelial Cell Culture Systems. In Model Systems for
Biopharmaceutical Assessment of Drug Absorption and Metabolism;
Borchardt, R. T., Smith, P. L., Wilson, G., Eds.; Plenum: New York,
1996; pp 239-258.
(19) Sherman, M. Y.; Goldberg, A. L. Cellular defenses against un-
folded proteins: a cell biologist thinks about neurodegenerative
diseases. Neuron 2001, 29, 15–32.
(20) Furukawa, K.; Wang, Y.; Yao, P. J.; Fu, W.; Mattson, M. P.;
Itoyama, Y.; Onodera, H.; D’Souza, I.; Poorkaj, P. H.; Bird, T. D.;
Schellenberg, G. D. Alteration in calcium channel properties is
responsible for the neurotoxic action of a familial frontotemporal
dementia tau mutation. J. Neurochem. 2003, 87, 427–436.
(32) Audus, K. L.; Rose, J. M.; Wang, W.; Borchardt, R. T. Brain
Microvessel Endothelial Cell Culture Systems. In An Introduc-
tion to the Blood-Brain Barrier: Methodology and Biology;
Pardridge, W., Ed.; Cambridge University: Cambridge, U.K., 1998;
pp 239-258.