Physicochemical prediction of a brain-blood distribution profile in polycyclic amines

Article abstract of DOI:10.1016/S0968-0896(03)00365-1

Recent investigation into the pharmacological character of the pentacyclo[5.4.0.0^2,6.0^3,10.0^5,9]undecyl and related polycyclic amines has revealed interesting facts regarding their possible use as neuroprotective agents. At this stage however, a clear shortcoming in the quest for further development of this novel class of compounds is the lack of concrete data on their ability to cross the blood-brain barrier (BBB). Working towards the aim of predicting BBB permeability, a series of related N-substituted 8-amino-8,11-oxapentacyclo[5.4.0.0^2,6.0^3,10.0 ^5,9]undecanes were synthesised. Compounds were characterised by both experimental and calculative methods, followed by biological assessment and statistical manipulation of the results obtained. In doing so, a simple biological model was established for the comparative evaluation of brain-blood distribution properties within the class. A highly sensitive ESI-MS.MS analytical procedure was developed for the detection of these compounds in biological tissues, indicating significant drug concentrations in the brain after intraperitoneal administration to C57Bl/6 mice. Stepwise multiple linear regression analysis of all data yielded two meaningful models (R²=0.9996 and R²=0.7749) depicting lipophilicity (log P_oct), solvent accessible molecular volume (SV), molar refractivity (MR) and system energy as the prime determinants of the brain-blood profile for these amines. The inherently high lipophilicity potential within the series is attributed to strong hydrophobic influences dominating hydrogen bonding effects. A possible conformational and energy dependent preference at the site of permeation is also suggested. The proposed estimations allow for the expedient and reliable prediction of brain partitioning behaviour for related polycyclic amines, facilitating the early rejection of unsuitable candidates and enabling research to focus on neuroprotective activity.

Full text of DOI:10.1016/S0968-0896(03)00365-1

Bioorganic & Medicinal Chemistry11 (2003) 3569–3578
Physicochemical Prediction of a Brain–Blood Distribution
Profile in Polycyclic Amines
Jaco Zah,a Gisella Terre’Blanche,a Elardus Erasmusb and Sarel F. Malana,*
aPharmaceutical Chemistry, Pothefstroom University for CHE, Private Bag X6001, Potchefstroom 2520, South Africa
bBiochemistry, Pothefstroom University for CHE, Private Bag X6001, Potchefstroom 2520, South Africa
Received 21 February2003; revised 27 May2003; accepted 2 June 2003
Abstract—Recent investigation into the pharmacological character of the pentacyclo[5.4.0.02,6.03,10.05,9]undecyl and related poly-
cyclic amines has revealed interesting facts regarding their possible use as neuroprotective agents. At this stage however, a clear
shortcoming in the quest for further development of this novel class of compounds is the lack of concrete data on their abilityto
cross the blood-brain barrier (BBB). Working towards the aim of predicting BBB permeability, a series of related N-substituted
8-amino-8,11-oxapentacyclo[5.4.0.02,6.03,10.05,9]undecanes were synthesised. Compounds were characterised by both experimental
and calculative methods, followed bybiological assessment and statistical manipulation of the results obtained. In doing so, a simple
biological model was established for the comparative evaluation of brain–blood distribution properties within the class. A highly
sensitive ESI-MS.MS analytical procedure was developed for the detection of these compounds in biological tissues, indicating sig-
nificant drug concentrations in the brain after intraperitoneal administration to C57Bl/6 mice. Stepwise multiple linear regression
analysis of all data yielded two meaningful models (R2=0.9996 and R2=0.7749) depicting lipophilicity(log Poct), solvent accessible
molecular volume (SV), molar refractivity(MR) and system energyas the prime determinants of the brain–blood profile for these
amines. The inherentlyhigh lipophilicitypotential within the series is attributed to strong hydrophobic influences dominating
hydrogen bonding effects. A possible conformational and energy dependent preference at the site of permeation is also suggested. The
proposed estimations allow for the expedient and reliable prediction of brain partitioning behaviour for related polycyclic amines,
facilitating the earlyrejection of unsuitable candidates and enabling research to focus on neuroprotective activity.
# 2003 Elsevier Ltd. All rights reserved.
Introduction
tial use of these compounds in preventing long-term
neurodegeneration (Fig. 1).
The chemistry of polycyclic amines has intrigued
manya scientist over the past five decades but only
recentlyhave researchers commenced probing their
pharmacological properties. Based on the premise of
structural analogyto the adamantanamines ( 1, 2),
investigation into the antiviral properties of two
Since the postulated target systems for these polycyclic
amines are within the brain and CNS, the need exists for
generating concrete data and guidelines on their BBB
permeability. Despite estimations as to the extent in
which these compounds are able to penetrate the brain,4
no specific BBB permeabilitydata has yet been pub-
lished. In essence, this project entailed the synthesis and
characterisation of a planned series of pentacyclo-
[5.4.0.02,6.03,10.05,9]undecyl amines in terms of lipophili-
city, minimum energy conformation, molecular size,
electronic character, hydrogen bonding behaviour and
brain–blood distribution data. Regression analysis was
applied in order to examine possible correlations between
brain–blood distribution data and physicochemical
descriptors. This was done in an attempt to establish a
quantitative structure–propertyrelationship (QSPR)
model with reliable predictive abilityto provide parameter
guidelines as to the potential degree of brain penetration
for premeditated compounds within the class.
4-amino-(D3)-trishomocubane
derivatives (3,
4)
revealed in vivo activityagainst Herpes simplex Type II
and Influenza A2.1 Shifting the emphasis to central
nervous system (CNS) activity, the potential of a series
of 8-alkylaminopentacyclo[5.4.0.02,6.03,10.05,9]undecanes
(5) as possible anti-Parkinsonian agents was also illu-
strated.2 Illuminating insights gained on the Ca2+
antagonistic effects3 of related pentacycloundecyl
amines (6–8) signified possible interaction with the
ionotropic NMDA receptor and therefore the poten-
*Corresponding author. Tel.: +27-18-299-2266; fax: +27-18-299-
4243; e-mail: fchsfm@puknet.puk.ac.za
0968-0896/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0968-0896(03)00365-1

3570
J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
Table 1. Comparison between input and optimised conformations
Results
generated byHyperchem 1, with designated energyminima in kcal
molꢀ1
Physicochemical characterisation
The synthesised compounds were characterised by both
experimental and computerised methods. Lipophilicity
was quantified at neutral pH as log Poct values. Deter-
minations were done experimentallyaccording to the
traditional shake-flask method.5 Predicted lipophilicity
[log P(ACD) and log P(Hyp)] values were calculated
algorithmicallyusing Advanced ChemistryDevel-
opment’s ACD/LogP1 program6,7 and Hyperchem1
modelling software,8 respectively. Hyperchem1 was
furthermore employed to calculate parameters provid-
ing quantum mechanical expression of minimised sys-
tem energy[Min.Energ(Hypy)], molar refractivity
Input
structure
Minimised
conformation
Min.Energy
(Hyp)
ꢀ4197.6
ꢀ4374.4
ꢀ4478.6
ꢀ4568.8
ꢀ4582.4
ꢀ4072.7
ꢀ4364.0
(MR), polarisability( a), solvent accessible molecular
volume (SV) and hydration energy (H) (Table 2).
As an experimental measure of hydrogen bonding
potential, a newlyexplored system to describe n-hexane/
ethylene glycol partitioning behaviour was employed.
This venture was rationalised bythe principle difference
between two immiscible solvents, one facilitating exclu-
sivelylipophilic interaction and the other capable of
hydrogen bonding. Equilibrated partitioning between
the two solvent phases would therefore allow a single,
direct measure of the solute’s hydrogen bonding (or
desolvation) potential.9 This parameter was subse-
quentlyexpressed as the logarithm of the implied
n-hexane/ethylene glycol partition coefficient and for the
purpose of this studydenoted as log PHB
.
Brain–blood distribution
Pilot studies were performed on the lead compound (6,
Table 1) to evaluate relative brain and blood analyte
concentrations (mg gꢀ1) at 30, 60, 90, and 120 min time
intervals after intraperitoneal administration to labora-
torymice, indicating optimum levels in both brain and
blood at approximately60 min. Although not indicative
of true steady-state levels, this observation provided a
meaningful centre for comparison within the congeneric
series. Brain partitioning data for each test compound
was consequentiallygenerated bydetermining the brain-
to-blood concentration ratio [(mg gꢀ1).(mg g
arbitrarysingle time-point (60 min) after admin-
istration, and denoted as log BBST that is,
)
ꢀ1 ꢀ1] at this
correlation values, hydrogen bonding capacity within
the ranges described for this series of compounds, did
not seem to significantlyinfluence BBB permeabilit.y
Furthermore, poor correspondence between log Poct and
the theoretical parameters log P(ACD) and log P(Hyp)
indicated that calculated log P values relate poorlyto the
lipophilic character of the polycyclic amines.
,
logð½brainŠ=½bloodŠÞ at a single time-point. Log BBST
values for the test compounds ranged between
ꢀ0.50ꢁ0.07 and 0.43ꢁ0.05 (Fig. 2), indicating that
these compounds significantlycrossed the BBB.
Stepwise multiple linear regression analysis of all data
yielded a four-component model (R2=0.9996; n=7, eq
1), depicting lipophilicity(log Poct), total system energy
[Min.Energy(Hyp)], solvent accessible molecular
volume (SV) and molar refractivity(MR) as the prime
determinants of BBB permeability for the polycyclic
amines n=7. Omitting the energyparameter Min.
Energy(Hyp) from the equation reduced linearity of the
subsequent three-component model to R2=0.7749
((Fig. 4) eq 2). No other significant multiple regression
correlations were evident (Table 2).
QSPR model
Individual correlation trends between brain–blood
distribution and physicochemical descriptors were
evaluated bymeans of linear regression analsyis as
well as polynomial least squares fit procedures. (The
nature of the curve was assessed bythe linear corre-
lation coefficient R2lin, and the non-linear correlation
coefficient R2non-lin). Statistical manipulation revealed a
parabolic relationship between permeabilityand log
Poct (R2non-lin=0.91, Fig. 3, Table 3). Based on poor

J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
3571
Figure 1. Adamantanamines and substituted polycyclic undecyl amines.
Figure 2. Log BBST values calculated as the average of six animal
experiments within each component group, with SD. All deviations
were within acceptable limits.23
Figure 3. Parabolic correlation trend for log BBST versus log Poct
Error bars indicate deviation for an average of six respective log BBst
assessments, and five log Poct measurements, for each compound.
.
Table 2. Summaryof experimental and computational descriptors
Log Log P Log P
Poct (ACD) (Hyp)
Log
PHB
MRa
ab
SAc
SVd
6
1.75
7p 1.47
1.28
1.68
1.41
1.71
2.22
0.98
0.18
0.4
2.57
2.53
2.83
2.70
2.54
1.26
1.79
0.57 75.13 29.80 478.66 789.68
ꢀ0.29 82.45 31.51 516.24 849.23
0.63 79.88 31.63 512.44 841.55
0.98 76.26 30.37 499.67 833.13
ꢀ1.67 81.57 32.27 524.03 862.55
9
10 1.59
11 1.44
12 2.22
13 0.68
ꢀ1.6
72.97 29.09 448.80 755.78
ꢀ2.04 79.83 31.15 496.47 822.71
aMolar refractivity(A 3).10,11
bPolarisability(A 3).12
cSolvent accessible surface area (A2).13,14
dSolvent accessible molecular volume (A3).13,14
Figure 4. Predicted log BBST from eq 1 versus experimental log BBST
Error bars represent variation in experimental values.
.
Table 3. Correlation values between log BBST and respective physico-
chemical descriptors
À
Á
log BBST ¼ À3:13289 À 0:20387 log Poct
Log
Poct
Log
P(ACD)
Log
PHB
Min.Energy
(Hyp)
SV
MR
þ 0:01133ðSVÞ À 0:07279ðMRÞ
ð2Þ
R2lin
R2non-lin
0.32
0.91
0.58
0.69
0.27
0.30
0.62
0.82
0.65
0.90
0.36
0.74
Discussion
This studyunambiguouslyproves that within this series,
the N-substituted 8-amino-8,11-oxapentacyclo-
À
Á
log BBST ¼ 4:56796 À 0:52226 log Poct
þ 0:00521½Min:EnergyðHypÞꢀ
[5.4.0.02,6.03,10.05,9]undecanes significantlypartition into
the brain after systemic administration—independent of
the mode of entryor whether specific transport systems
or efflux mechanisms are involved. Literature describes
log BB values ranging between ꢀ2.0 and +1.0, suggesting
þ 0:05559ðSVÞ À 0:33873ðMRÞ
ð1Þ

3572
J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
optimal brain partitioning around 0.3; larger values
usuallyexhibited byhighlypenetrable drugs, and values
<ꢀ1.0 indicating poor distribution to the brain.15 Values
for the polycyclic amines ranged between ꢀ0.50 and 0.43,
with optimal permeabilityexhibited bycompound 10 (log
BBST=0.43). Based on these values as crude guidelines, it
therefore implies that compound 12, despite displaying a
blood-favoured partitioning ratio, still reached meaningful
concentrations in brain parenchyma.
descriptors applied in this investigation. The models
establish the significance of lipophilicity, solvent acces-
sible molecular volume, molar refractivityand a energy
term with regard to the BBB permeabilityof the poly-
cyclic amines.
The decrease in linearityresulting from omitting the
energyterm in eq 1, points toward an energy-influenced
affinityat the site of permeation. However, as the
energyterm can be influenced bythe algorithm used
and might in some instances not represent the global
minimum, estimations of permeabilitycould also be
based on eq 2, especiallyfor larger and more flexible
congeners.
No linear relationships seemed to exist between BBB
permeabilityand the experimentallydetermined descriptors
log Poct and log PHB. Non-linear fit procedures however
revealed decidedlysignificant individual correlation
with regard to log Poct, log P(Hyp), Min.Energy(Hyp)
and SV. These observations suggest a mainlyparabolic
The negative contribution from log Poct probably
reflects on the fact that the measured range of values is
mainlydistributed around the plateau (optimum) of the
established parabolic curve, where slight increases/
decreases will reduce permeability. This phenomenon is
usuallyexplained in terms of diffusional transport - the
inclining leg of the curve simplyrepresenting improved
lipid-solubilitywith correspondinglyenhanced diffusion
across the barrier; the plateau indicative of diffusion
through the stagnant layers becoming rate-limiting,
while the declining leg represents a trend where diffusion
gets inhibited due to membrane association of highly
lipophilic compounds.
correlation between permeabilityand lipophilicity(log
16
Poct), following the trend described byHansch,
indicating a local maximum at log Poctꢁ1.5.
and
From the results obtained it seems riskyto assign spe-
cific correlation trends between permeabilityand the
respective semi-empiricallyderived parameters log
P(Hyp), Min.Energy(Hyp) and SV, because they reflect
both linear and non-linear tendencies. From this point
of view, log Poct seems to be the onlyreliable guideline
for BBB permeability, if one has to evaluate only one
physicochemical descriptor. In addition, the parameters
as calculated byACD 1, are unsatisfactorypredictors
for this class of compounds, considering yet again the
non-specificitythat emerged from such correlations [log
P(ACD)] or simplythe lack of any(log D7.4 and pKa).
Since ACD1 values are obtained from a fragmental
approach, prediction accuracymayprove inadequate in
novel classes of compounds.17 Atomic-based calculations
(Hyperchem1) clearlyproduce closer estimations to
QSPR properties, though still not satisfactory.
Unique to this model is the positive contribution to
permeabilityas exercised bythe solute size and energy
descriptors. Although Abraham et al.19 reported
increased brain penetration with increased solute size,
without providing mechanistic understanding of this
occurrence, this interesting observation should also be
mentioned here. If passive diffusion is regarded as the
main route of entry, one should expect the opposite
effect from solute size (according to the Stokes–Einstein
equation). Close inspection of the minimum energy
conformation (Hyperchem1) and volume (SV) of com-
pound 12 reveals that it is the smallest spatiallyarran-
ged molecule within the series and improved diffusion
compared to the rest of the series would thus be expec-
ted. Looking at its log Poct value (highest within the
series) this is exactlywhat happens with regard to octa-
nol–water partitioning; yet 12 displays the poorest dis-
tribution to the brain. This indicates that there
are probablycertain other mechanisms, like protein-
mediated extrusion transport, involved in the barrier
permeabilityof these compounds and therefore a con-
formational (energy-dependent) preference at the site of
permeation.
Hydrogen bonding (log PHB) does not seem to playa
pivotal role in the permeabilityof these compounds,
despite the fact that theydiffer significantlyin their
hydrogen bonding capacity. This suggests that BBB
permeabilityhere is driven mainlybyhdyrophobic
character. Strangelyenough, this reflects the exact
opposite situation from that described byChikhale et
al.,18 where poor correlation with lipophilicitywas
observed, but hydrogen bonding appeared the sig-
nificant determinant. A possible explanation might sim-
plylie in the number of hydrogen bonding groups and
that there also exists a ‘threshold’ from which hydrogen
bonding capacitystarts to elicit meaningful changes in
permeability. This seems rational since the current series
of polycyclic amines essentially differ by only one group
from compound to compound, or even onlyin the nat-
ure of such groups. More work is however required to
prove this hypothesis, and to establish whether the log
PHB parameter is reliable in its representation of
hydrogen bonding capacity. Polar surface area should
also be considered in such quantifications.
The positive y-intercept indicates the inherentlyhigh
potential of these compounds to reach meaningful con-
centrations in the brain.
Conclusion
The multivariate prediction models in eqs 1 and 2 verify
the conclusions drawn from evaluating the individual
correlation trends between permeabilityand the various
Both regression models establish the significance of
lipophilicity, solvent accessible molecular volume and
molar refractivitywith regard to the BBB permeability

J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
3573
of the polycyclic amines. The good linear regression
(R2=0.9996, eq 1) obtained byincluding an energyterm
also stresses the importance of conformational energyin
the transport of these compounds across the BBB. Pre-
dicting capabilities are however limited to this class of
compounds and within the respective physicochemical
descriptor limits investigated. Experimental log Poct
determination is required and the model is therefore not
purelycomputational in nature. This is necessaryas
calculated log P values do not correspond satisfactorily
to those determined experimentally. The BBB perme-
abilityassessment here is probablynot based on passive
diffusion alone, but is an integrated estimation of
transport mechanisms or preferences that mayexist
within the range of physicochemically different com-
pounds used to construct it. The experimental log Poct
furthermore remains the best lipophilicitydescriptor
with which these processes can be related, but proves
the complexityof BBB transport in that it cannot be
regarded as a static lipoid wall, but should rather be
recognised as a highlydynamic and selective interface.
The derived models do however provide valuable para-
meter guidelines as to those properties essential for cal-
culating the probabilityof these amines to significantly
reach the brain after systemic administration.
the synthesis of compounds 6, 9, 7p and 12, the methods
of Malan et al. were used.21
8-Benzylamino-8,11-oxapentacyclo[5.4.0.02,6.03,10.05,9]un-
decane (6). Pentacyclo[5.4.0.02,6.03,10.05,9]undecane-
8,11-dione (5.04 g, 28.98 mmol) was dissolved in tetra-
hydrofuran (THF, 50 mL) and cooled down to 5 ꢂC
while stirring on an external ice bath. Benzylamine (3.23
g, 30.21 mmol) was added slowly, with continued stir-
ring of the reaction mixture at lowered temperature.
The carbinolamine started precipitating after approxi-
mately10 min, but the reaction was allowed to reach
completion for an additional 20 min. This carbinol-
amine was isolated byfiltration. Water was removed
azeotropicallybyrefluxing this material in 60 mL
sodium-dried benzene (Dean–Stark dehydrating condi-
tions) for 1 h or until no more water was collected in the
trap. (The milkysuspension turned into a clear solution
as the reaction progressed.) The benzene was removed
under reduced pressure and the Schiff base (8-benzyli-
minopentacyclo[5.4.0.02,6.03,10.05,9]undecane-11-one,
a
yellow oil) was dissolved in a mixture of anhydrous
methanol (MeOH, 30 mL) and anhydrous THF (150
mL). Reduction was carried out byadding NaBH 4 (1.50
g, 39.46 mmol) in excess and stirring the mixture over-
night (18 h) at room temperature. The solvents were
removed in vacuo, the residue suspended in water (100
mL) and extracted with dichloromethane (DCM, 4ꢃ50
mL). The combined organic fractions were washed with
water (2ꢃ100 mL), dried over anhydrous MgSO4 and
evaporated in vacuo to yield a yellowish oil. Purification
of the product mixture was accomplished with column
chromatographyand the desired amine was isolated as a
colourless wax. Crystallisation from absolute ethanol
rendered the final product as a colourless microcrystal-
line solid. (Yield: 2.92 g, 11.01 mmol, 38.00%.)
C18H19NO; mp 78.7 ꢂC; HR-MS: calcd 265.1467, exp.
Since the experimental evaluation of permeability
potential is difficult, time-consuming and expensive, a
rapid yes/no decision concerning brain penetration can
be reached with confidence earlyin the drug develop-
ment process of these compounds, facilitating rejection
of unsuitable candidates.
Experimental
Materials and standard procedures
265.1466; IR (KBr) nmax: 3332, 2970, 1507, 1011 cmꢀ1
;
All reagents and analytical solvents were procured from
Sigma-Aldrich, Merck or Fluka and used without
intervention. Reaction and elution solvents were pur-
chased from commercial sources. Melting points were
determined bymeans of DSC utilising a Shimadzu
DSC-50 instrument while IR spectra were recorded on a
Nicolet 470 FT-IR spectrophotometer. Mass spectra
and HR-MS were recorded on a VG 7070E mass spec-
trometer and NMR spectra were obtained on a Varian
MS (EI, 70 eV) m/z: 266 [(M+)+1], 265 (M+), 237, 186,
131, 91, 65, 28; 1H NMR (300 MHz, CDCL3) dH:
7.34–7.15 (m, 5H, H-16,17,18,19,20), 4.62 (t, 1H, J=5.2
Hz, H-11), 3.98 (AB, 1H, JAB=13.5 Hz, H-14a), 3.93
(AB, 1H, JAB=13.5 Hz, H-14b), 2.83–2.47 (2ꢃm, 7H,
H-1, 2, 3, 5, 6, H-7/9, H-10), 2.38 (t, 1H, J=5.0 Hz, H-
7/9), 2.16 (bs, 1H, H-13), 1.86:1.51 (AB-q, 2H, J=10.4
Hz, H-4a, 4b); 13C NMR (75 MHz, CDCL3) dC: 140.96
(s, C-15), 128.40 (d, C-16, 20), 127.92 (d, C-17, 19),
126.85 (d, C-18), 109.62 (s, C-8), 82.48 (d, C-11), 55.22
(d, C-7/9), 54.73 (d, C-7/9), 47.73 (t, C-14), 44.83 (d,
1C), 44.79 (d, 1C), 44.50 (d, 1C), 43.18 (t, C-4), 43.07 (d,
1C), 41.93 (d, 1C), 41.47 (d, 1C).
1
Gemini 300. H spectra were recorded at a frequency
of 300.075 MHz and 13C spectra at 75.462 MHz, in a 7
Tesla magnetic field. Tetramethylsilane (TMS) was
used as internal standard and
a bandwidth of
1000 MHz at 24 kG applied for H–13C-decoupling. All
chemical shifts are reported in parts per million (ppm)
relative to TMS (d=0). The following abbreviations
are used to indicate multiplicities of respective signals:
s, singulet; bs, broad singulet; d, doublet; t, triplet; m,
multiplet.
1
2,6
8- (4 - Nitrobenzylamino) - 8,11 - oxapentacyclo[5.4.0.0
0
.
3,10.05,9]-undecane (7p). A solution of 4-nitrobenzyl-
.
amine HCl (5.0 g in 100 mL water) was alkalinised with
NaOH (10% w/v) and extracted with DCM (4ꢃ50 mL).
After in vacuo removal of the solvent at room
temperature, the free base (4.00 g, 26.32 mmol) was
slowlyadded to
a solution of the pentaclyo-
Synthetic procedures
[5.4.0.02,6.03,10.05,9]undecane-8,11-dione (4.40 g, 25.30
mmol) in THF (50 mL) and the stirring mixture cooled
down to 5 ꢂC. The carbinolamine started precipitating
after approximately40 min, but the reaction was
Pentacyclo[5.4.0.02,6.03,10.05,9]undecane-8,11-dione served
as the initial structure for all further syntheses and was
prepared in bulk according to literature methods.20 For

3574
J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
allowed to reach completion for an additional 20
min. This carbinolamine, a yellowish cream, was
quicklyisolated byfiltration and then suspended in 60
mL sodium-dried benzene. Molecular water was
removed byslowlyrefluxing this mixture under Dean-
Stark dehydrating conditions for 2 h. (The creamy sus-
pension developed into a clear solution as the reaction
progressed.) The benzene was removed under reduced
pressure and the Schiff base, 8-(4-nitrobenzylimino)-
pentacyclo[5.4.0.02,6.03,10.05,9]undecane-11-one, a yellow
oil) was dissolved in a mixture of anhydrous MeOH (30
mL) and anhydrous THF (150 mL). NaBH4 (1.52 g,
39.98 mmol) was added in excess and the mixture stirred
overnight at ambient temperature. (Decolouration indi-
cated successful reduction of the imine.) The solvents
were removed in vacuo, the residue suspended in water
(100 mL) and extracted with DCM (4ꢃ50 mL). The
combined organic fractions were washed with water
(2ꢃ100 mL), dried over anhydrous MgSO4 and evapo-
rated in vacuo to yield a deep amber-coloured wax with
signs of crystallisation. Purification of this product
mixture was accomplished with column chromato-
graphyand recrystallisation from absolute ethanol ren-
dered the final product as a yellowish microcrystalline
solid. (Yield: 3.56 g, 11.50 mmol, 45.44%.)
C18H18N2O3; mp 161.6 ꢂC; HR-MS: calcd 310.1317,
exp. 310.1322; IR (KBr) nmax: 3334, 2964, 1507, 1009,
852 cmꢀ1; MS (EI, 70 eV) m/z: 311 [(M+)+1], 310
h at 35 ꢂC. The solvents were removed in vacuo, the
residue suspended in water (100 mL) and extracted with
DCM (4ꢃ50 mL). The combined organic fractions were
washed with water (2ꢃ100 mL), dried over anhydrous
MgSO4 and evaporated under reduced pressure to yield
a yellowish oil. Partial resolution of the product mixture
was accomplished with column chromatographyand
the contaminated amine isolated as a yellow tainted
wax. This product was dissolved in acetone (50 mL) and
cooled down to ꢀ15 ꢂC while stirring on an external
ethanol bath. Forced evaporation of the acetone with a
mild stream of N2 gas caused oversaturation and con-
sequent precipitation of a colourless solid. The latter
was isolated bydecantation and allowed to reach room
temperature, affording the purified amine as a colourless
wax. (Yield: 2.54 g, 9.10 mmol, 31.52%.) C19H21NO;
mp 49.7 ꢂC; HR-MS: calcd 279.1534, exp. 279.1538; IR
(KBr) nmax: 3312, 2974, 1353, 846, 699 cmꢀ1; MS (EI, 70
eV) m/z: 279 (M+), 188, 159, 131, 105, 91, 77; 1H NMR
(300 MHz, CDCL3) dH: 7.29-7.13 (m, 5H, H-17, 18, 19,
20, 21), 4.59 (t, 1H, J=5.2 Hz, H-11), 3.11–2.99 (m, 2H,
H-14a, 14b), 2.82–2.44 (3ꢃm, 9H, H-1, 2, 3, 5, 6, 7/9,
10, 15a, 15b), 2.38 (t, 1H, J=4.9 Hz, H-7/9), 1.95 (bs,
1H, H-13), 1.87:1.51 (AB-q, 2H, J=10.4 Hz, H-4a, 4b);
13C NMR (75 MHz, CDCL3) dC: 139.99 (s, C-16),
128.76 (d, C-17, 21), 128.41 (d, C-18, 20), 126.10 (d, C-
19), 109.43 (s, C-8), 82.40 (d, C-11), 55.31 (d, C-7/9),
54.67 (d, C-7/9), 44.87 (t, C-14), 44.72 (d, 1C), 44.54 (d,
1C), 44.40 (d, 1C), 43.18 (t, C-4), 42.96 (d, 1C), 41.84 (d,
1C), 41.43 (d, 1C), 37.21 (t, C-15).
1
(M+), 231, 131, 91, 28; H NMR (300 MHz, CDCL3)
dH: 8.12 (d, 2H, J=8.9 Hz, H-17, 19), 7.50 (d, 2H,
J=8.9 Hz, H-16, 20), 4.62 (t, 1H, J=5.2 Hz, H-11),
4.10 (AB, 1H, JAB=15.4 Hz, H-14a), 4.04 (AB, 1H,
JAB=15.4 Hz, H-14b), 2.83-2.38 (3ꢃm, 8H, H-1, 2, 3, 5,
6, 7, 9, 10), 2.32 (bs, 1H, H-13), 1.86:1.51 (AB-q, 2H,
J=10.6 Hz, H-4a, 4b); 13C NMR (75 MHz, CDCL3) dC:
149.15 (s, C-18), 147.07 (s, C-15), 128.25 (d, C-16, 20),
123.58 (d, C-17, 19), 109.42 (s, C-8), 82.56 (d, C-11),
55.34 (d, C-7/9), 54.71 (d, C-7/9), 47.04 (t, C-14), 44.88
(d, 1C), 44.77 (d, 1C), 44.51 (d, 1C), 43.19 (t, C-4), 43.07
(d, 1C), 41.88 (d, 1C), 41.46 (d, 1C).
8-Cyclohexylmethylamino-8,11-oxapentacyclo[5.4.0.02,6
3,10.05,9]undecane (10). Pentacyclo[5.4.0.02,6.03,10.05,9]un-
.
0
decane-8,11-dione (5.01 g, 28.77 mmol) was dissolved in
THF (50 mL) and cooled down to 5 ꢂC. (Amino-
methyl)cyclohexane (3.31 g, 29.30 mmol) was added
slowly, continuously agitating the reaction mixture at
lowered temperature. The precipitated carbinolamine
was isolated byfiltration and washed with cold THF
(2ꢃ20 mL). Water was removed azeotropicallyby
refluxing this material in 60 mL sodium-dried benzene
(Dean–Stark conditions) for 90 min. The benzene was
removed under reduced pressure and the Schiff base
2,6
8- (2 - Phenylethylamino) - 8,11 - oxapentacyclo[5.4.0.0
0
.
3,10.05,9]undecane (9). Pentacyclo[5.4.0.02,6.03,10.05,9]un-
decane-8,11-dione (5.02 g, 28.86 mmol) was dissolved in
THF (50 mL) and cooled down to 5 ꢂC while stirring on
an external ice bath. 2-Phenylethylamine (3.52 g, 29.10
mmol) was added slowly, with continued stirring of the
reaction mixture at lowered temperature. The white
hydroxylamine started precipitating after approximately
15 min, but the reaction was allowed to reach comple-
tion for an additional 30 min. This hydroxylamine was
isolated byfiltration and washed with cold THF (2 ꢃ20
mL). Water was removed azeotropicallybyrefluxing
this material in 60 mL sodium-dried benzene under
dehydrating conditions (Dean–Stark) for 2 h. (The
milkysuspension developed into a clear solution as the
reaction progressed.) The benzene was removed under
reduced pressure and the Schiff base [8-(2-phenylethyl-
imino)pentacyclo[5.4.0.02,6.03,10.05,9]undecane-11-one, a
yellowish oil] was dissolved in a mixture of anhydrous
MeOH (30 mL) and anhydrous THF (150 mL). NaBH4
(1.62 g, 42.72 mmol) was added in excess and the imine
was subsequentlyreduced byrefluxing the mixture for 5
[8-(cyclohexylmethylimino)pentacyclo[5.4.0.02,6.03,10.05,9
]
undecane-11-one, an amber-coloured oil] was dissolved
in a mixture of anhydrous MeOH (30 mL) and anhy-
drous THF (150 mL). NaBH4 (1.52 g, 39.98 mmol) was
added in excess and the mixture refluxed for 5 h at
35 ꢂC. The solvents were removed and the residue sus-
pended in water (100 mL) and extracted with DCM
(4ꢃ50 mL). The combined organic fractions were
washed with water (2ꢃ100 mL), dried over anhydrous
MgSO4 and evaporated in vacuo. The residue was dis-
solved in acetone (50 mL) and cooled down to ꢀ15 ꢂC
on an external ethanol bath. Forced evaporation of the
acetone with a mild stream of N2 gas caused precipita-
tion of a colourless solid, which was quicklyisolated by
decantation and allowed to reach room temperature,
affording the purified amine as a colourless glass.
(Yield: 2.22 g, 8.18 mmol, 28.43%.) C18H25NO; mp
58.9 ꢂC; HR-MS: calcd 271.1936, exp. 271.1930; IR
(KBr) nmax: 3323, 2923, 1489, 1010, 851 cmꢀ1; MS (EI,
70 eV) m/z: 272 [(M+)+1], 271 (M+), 188, 176, 159,

J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
3575
1
131, 91, 55, 28; H NMR (300 MHz, CDCL3) dH: 4.57
temperature. The white carbinolamine started precipitat-
ing after approximately3 min, but the reaction was
allowed to reach completion for an additional 15 min. The
latter product was isolated byfiltration and washed with
cold THF (2ꢃ20 mL). Water was removed azeotropically
byrefluxing this material in 60 mL sodium-dried benzene
(Dean–Stark conditions) for 90 min. (The milkysuspen-
sion developed into a clear solution as the reaction pro-
gressed.) The benzene was removed under reduced
pressure and the Schiff base [8-(4-pyridylmethyl-
imino)pentacyclo[5.4.0.02,6.03,10.05,9]undecane - 11-one, a
yellow oil] was dissolved in a mixture of anhydrous
MeOH (30 mL) and anhydrous THF (150 mL). NaBH4
(1.29 g, 33.86 mmol) was added in excess and the mixture
stirred overnight at ambient temperature. (Decoloura-
tion indicated successful reduction of the imine.) The
solvents were removed under reduced pressure, the resi-
due suspended in water (100 mL) and extracted with
DCM (4ꢃ50 mL). The combined organic fractions were
washed with water (2ꢃ100 mL), dried over anhydrous
MgSO4 and evaporated in vacuo to yield a yellow-green
wax with signs of crystallisation. Purification of this
product mixture was accomplished with column chro-
matographyand slow recrystallisation from DCM–ace-
tone (1/1) at 5 ꢂC rendered a colourless microcrystalline
solid. At room temperature the latter transformed into
an amorphous white powder. (Yield: 4.12 g, 15.48
mmol, 52.94%.) C17H18N2O; mp 112.7 ꢂC; HR-MS:
calc. 266.1419, exp. 266.1415; IR (KBr) nmax: 3295,
2977, 1374, 1009, 797 cmꢀ1; MS (EI, 70 eV) m/z: 266
(M+), 238, 187, 174, 131, 92, 65, 39; 1H NMR
(300 MHz, CDCL3) dH: 8.46 (d, 2H, J=5.7 Hz, H-17,
19), 7.23 (d, 2H, J=5.7 Hz, H-16, 20), 4.59 (t, 1H,
J=5.3 Hz, H-11), 3.98 (AB, 1H, JAB=15.5 Hz, H-14a),
3.92 (AB, 1H, JAB=15.5 Hz, H-14b), 2.80–2.35 (3ꢃm,
8H, H-1, 2, 3, 5, 6, 7, 9, 10), 2.29 (bs, 1H, H-13),
1.83:1.48 (AB-q, 2H, J=10.5 Hz, H-4a, 4b); 13C NMR
(75 MHz, CDCL3) dC: 150.39 (s, C-15), 149.76 (d, C-
17,19), 122.51 (d, C-16, 20), 109.40 (s, C-8), 82.47 (d, C-
11), 55.26 (d, C-7/9), 54.64 (d, C-7/9), 46.44 (t, C-14),
44.78 (d, 1C), 44.70 (d, 1C), 44.44 (d, 1C), 43.12 (t, C-4),
43.01 (d, 1C), 41.82 (d, 1C), 41.39 (d, 1C).
(t, 1H, J=5.2 Hz, H-11), 2.79–2.41 (3ꢃm, 9H, H-1, 2, 3,
5, 6, 7/9,10,14a,14b), 2.36 (t, 1H, J=5.0 Hz, H-7/9),
1.88 (bs, 1H, H-13), 1.85:1.49 (AB-q, 2H, J=10.5 Hz,
H-4a,4b), 1.74–0.81 (4ꢃm, 11H, H-15, 16a, 16b, 17a,
17b, 18a, 18b, 19a, 19b, 20a, 20b); 13C NMR (75 MHz,
CDCL3) dC: 109.83 (s, C-8), 82.35 (d, C-11), 55.04 (d, C-
7/9), 54.69 (d, C-7/9), 50.19 (t, C-14), 44.78 (d, 2C),
44.43 (d, 1C), 43.16 (t, C-4), 42.97 (d, 1C), 41.81 (d, 1C),
41.44 (d, 1C), 38.77 (d, C-15), 31.20 (t, C-16,20), 26.55
(t, C-18), 25.89 (t, C-17,19).
8- [2 - (4 - Hydroxyphenyl)ethylamino] - 8,11 - oxapentacy-
clo[5.4.0.02,6.03,10.05,9]undecane (11). Pentacyclo[5.4.0.
2,6.03,10.05,9]undecane-8,11-dione (2.59 g, 14.87 mmol)
0
and tyramine (2.08 g, 15.15 mmol) were dissolved sepa-
ratelyin dryMeOH (30 mL each). These two solutions
were combined and continuallystirred at ambient tem-
perature for 50 h. The MeOH was removed in vacuo and
70 mL sodium-dried benzene added to the amber-
coloured residue. Water was removed byslowlyrefluxing
this mixture under Dean-Stark dehydrating conditions
for 2 h. The benzene was removed under reduced pressure
and the Schiff base, 8-[2-(4-hydroxyphenyl)ethylimino]-
pentacyclo[5.4.0.02,6.03,10.05,9]undecane-11-one, was dis-
solved in a mixture of anhydrous MeOH (30 mL) and
anhydrous THF (150 mL). Reduction was induced by
adding NaBH4 (0.90 g, 23.76 mmol) in excess and
slowlyrefluxing the amber solution for 6 h. The solvents
were removed, the residue suspended in water (100 mL)
and extracted with DCM (4ꢃ50 mL). The combined
organic fractions were washed with water (2ꢃ100 mL),
dried over anhydrous MgSO4 and evaporated to yield
an amber-coloured oil. Column chromatographypro-
duced an oilysubstance that was refluxed in acetone
with activated charcoal for 15 min. After filtration
through Celite1 the acetone was removed and the
desired amine crystallised from dry MeOH at 5 ꢂC to
render colourless grains. (Yield: 1.40 g, 4.75 mmol,
31.91%.) C19H21NO2; mp 160.9 ꢂC; HR-MS: calcd
295.1673, exp. 295.1676; IR (KBr) nmax: 3260, 2970,
1516, 1249, 1016, 827 cmꢀ1; MS (EI, 70 eV) m/z: 295
(M+), 188, 91, 66, 32, 28; 1H NMR (300 MHz, CDCL3)
dH: 6.97 (d, 2H, J=8.5 Hz, H-18,21), 6.62 (d, 2H,
J=8.5 Hz, H-17, 22), 4.62 (t, 1H, J=5.2 Hz, H-11),
4.01 (bs, 1H, H-20), 3.07–2.97 (m, 2H, H-14a, 14b),
2.83–2.49 (2ꢃm, 9H, H-1, 2, 3, 5, 6, 7/9, 10, 15a, 15b),
2.38 (t, 1H, J=4.9 Hz, H-7/9), 1.86–1.50 (AB-q, 2H,
J=10.4 Hz, H-4a, 4b), (H-13ꢄno signal); 13C NMR
(75 MHz, CDCL3) dC: 154.85 (s, C-19), 130.99 (s, C-16),
129.79 (d, C-17, 22), 115.47 (d, C-18, 21), 109.92 (s, C-
8), 82.47 (d, C-11), 55.28 (d, C-7/9), 54.69 (d, C-7/9),
45.00 (t, C-14), 44.76 (d, 1C), 44.50 (d, 1C), 44.42 (d,
1C), 43.16 (t, C-4), 43.02 (d, 1C), 41.86 (d, 1C), 41.44 (d,
1C), 35.81 (t, C-15).
8-(4-Aminobenzylamino)-8,11-oxapentacyclo[5.4.0.0 2,6
3,10.05,9] undecane (13). Tin powder (1.32 g, 11.05
mmol) and previouslyprepared 8-(4-nitrobenzylamino)-
.
0
8,11-oxapentacyclo[5.4.0.02,6.03,10.05,9]undecane (5, 2.56 g,
8.27 mmol) were weighed in a 50-mL round-bottomed
flask and set up under reflux conditions. HCl (10%, 20
mL) was added slowlydown the Liebig condenser, con-
tinuallystirring the mixture while elevating the reaction
temperature to reflux level. Ethanol (3 mL) was added
as a solubilisation agent. Further HCl additions (2ꢃ30
mL) were made consecutivelyat 10-min intervals,
allowing the reduction of the nitro group to reach com-
pletion over a period of 70 min. The clear supernatant
was then isolated bydecantation, cooled down to room
temperature and neutralised past the turning point with
NaOH (40%). Extraction was done with DCM (4ꢃ50
mL), the combined organic fractions washed with brine
(2ꢃ50 mL), dried over anhydrous MgSO4 and evapo-
rated in vacuo. Column chromatographyafforded the
desired 4-aminobenzylamino-derivative as a viscous,
8-(4-Pyridylmethylamino)-8,11-oxapentacyclo[5.4.0.02,6
.
0
3,10.05,9]undecane (12). The pentacyclo[5.4.0.02,6.03,10
.
0
5,9]undecane-8,11-dione (5.09 g, 29.23 mmol) was dis-
solved in THF (50 mL) and cooled down to 5 ꢂC while
stirring on an external ice bath. 4-(Aminomethyl)pyr-
idine (3.24 g, 30.02 mmol) was added slowly, with con-
tinued stirring of the reaction mixture at lowered

3576
J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
amber-coloured substance. (Yield: 1.02 g, 3.65 mmol,
44.11%.) C18H20N2O; HR-MS: calcd 280.1576, exp.
280.1583; IR (neat) nmax: 3343, 2964, 1616, 1517, 1275,
1009 cmꢀ1; MS (EI, 70 eV) m/z: 281 [(M+)+1], 280
ment of log Poct values. Standard solution preparations
and concentration determinations were however per-
formed within the phase of optimal solubilityat room
temperature (25 ꢂC). Determinations were done in five-
fold at lmax ꢀ208.4 nm (compound 6 in n-hexane),
277.0 nm (7p in ethylene glycol), 210.2 nm (9 in n-hex-
ane), 200.0 nm (10 in n-hexane), 224.2 nm (11 in ethyl-
ene glycol), 207.0 nm (12 in ethylene glycol) and 208.0
nm (13 in ethylene glycol).
1
(M+), 175, 131, 106, 91, 77, 28; H NMR (300 MHz,
CDCL3) dH: 7.12 (d, 2H, J=8.5 Hz, H-17, 20), 6.60 (d,
2H, J=8.5 Hz, H-16, 21), 4.63 (t, 1H, J=5.2 Hz, H-11),
3.87 (AB, 1H, JAB=12.7 Hz, H-14a), 3.82 (AB, 1H,
JAB=12.7 Hz, H-14b), 3.56 (bs, 2H, H-19a, 19b), 2.84–
2.48 (2ꢃm, 7H, H-1, 2, 3, 5, 6, 7/9, 10), 2.39 (t, 1H,
J=4.9 Hz, H-7/9), 2.11 (bs, 1H, H-13), 1.88–1.52 (AB-
q, 2H, J=10.5 Hz, H-4a, 4b); 13C NMR (75 MHz,
CDCL3) dC: 145.34 (s, C-18), 130.79 (s, C-15), 129.11 (d,
C-16, 21), 115.12 (d, C-17, 20), 109.61 (s, C-8), 82.40 (d,
C-11), 55.14 (d, C-7/9), 54.70 (d, C-7/9), 47.29 (t, C-14),
44.76 (d, 2C), 44.45 (d, 1C), 43.14 (t, C-4), 43.03 (d, 1C),
41.90 (d, 1C), 41.43 (d, 1C).
Acquisition of brain–blood distribution data
All animal trials were conducted on adult male C57Bl/6
laboratorymice (25–30 g), with approval and in accor-
dance with the guidelines as laid out bythe Medical
Ethics Committee of the Potchefstroom Universityfor
Christian Higher Education.
Computer modelling and minimisation
Administration and tissue harvesting
Modelling and structural optimisation were done using
the ChemplusTM extension of Hyperchem1 modelling
Test compounds were dissolved in a 80% (w/v) 2-
hydroxypropyl-b-cyclodextrin solution in water for
injection, bymeans of sonification and mild heating.
After notation of bodymass, this freshlyconstituted
preparation (37.5 mg mLꢀ1) was administered ꢀin1traper-
21
software.8 A previouslyminimised structure
of the
lead compound (6) served as the initial conformer geo-
metryto which functional group substitutions were
made to comprise all components in the series. MM+
and AM1 were used in sequence to carryout molecular
and electronic calculations. Ground state energies were
itoneallyat an arbitrarydosage of 300 mg kg
(ꢁ0.2
mL per mouse), after which animals were allowed free
movement and ad lib access to water and food. (Dosage
regimens of 200–400 mg kgꢀ1 for each compound were
tested beforehand to eliminate the possibilityof lethal
dosing, but no lethal effects were observed within 120
min post administration.) Animals were promptly
sacrificed bydecapitation 60 min after administration
and whole brain and blood samples collected in pre-
weighed open-topped 15 mL Pyrex1 vials and 4.5 mL
15% K3E anticoagulant-containing BD VacutainerTM
vials, respectively. Samples were put on ice for immedi-
ate further processing.
optimised using the Polak–Ribiere
´
procedure.22
Theoretical physicochemical parameters were computed
for all the minimised structures.
Partitioning measurements
Log Poct. Solvent phases of 1-octanol and double-dis-
tilled, de-ionised water were thoroughlycross-saturated
to equilibration in a separatoryfunnel for 24 h. Stan-
dard solution preparations and concentration determi-
nations were all performed within the 1-octanol phase at
room temperature (25 ꢂC). A stock solution of 50.0 mg
mLꢀ1 was prepared for each test compound and further
dilutions performed to render a standard concentration
series of 25.0, 20.0, 15.0, 10.0 and 5.0 mg mLꢀ1. 5-mL
aliquots of the stock solution were then pipetted into
equal volumes (5 mL) of the octanol-saturated water,
using Pyrex vials with Teflon-sealed screw caps. The
Standardisation and analyte recovery
Each component analysis comprised three groups of
animals, each group consisting of six subjects—an
experimental group receiving the compound under
investigation, a control group receiving dose-free, but
equal volumes (0.2 mL) of the injection medium, and
a standard group from which analyte-free brain and
blood tissues were harvested for constructing two
independent quantitative calibration (standard)
curves. Extraction efficiencywas not determined
beforehand, but rather incorporated into these linear
regression curves byintroducing both standard and
internal standard solutions into equal aliquots of the
respective brain and blood standard tissues and sub-
mitting both to the same extraction procedure as for the
samples and controls. The assumption was made that
the quantitative recoverypercentage remained constant
with regard to small experimental error-variance among
the six different standard tissues in each group. A stock
solution in CHCl3 (400 mg mLꢀ1) was prepared in each
instance and successivelydiluted into whole blood and
brain tissue to render a series of final working stan-
dards for ESI-MS.MS detection, ranging between 1.0
1
vials were vigorouslyshaken on a Labotec reciprocal
shaker for 60 min in order to achieve equilibrium parti-
tioning, followed bycentrifugation for 30 min at 2500
rpm to effect phase separation. UV absorbance readings
for stock, standard and sample solutions were obtained
in five-fold on
a Shimadzu UV120-02 spectro-
photometer at lmax: 211.0 nm (compound 6), 271.6 nm
(7p), 211.0 nm (9), 207.8 nm (10), 224.8 nm (11), 208.2
nm (12) and 209.6 nm (13). Concentrations were calcu-
lated from constructed linearlyregressed standard
curves. The corresponding water-phase concentrations
were obtained bythe difference between stock and
sample solution absorption values.
Log PHB. Equilibrated solvent phases of n-hexane and
ethylene glycol were used for log PHB determinations.
The same procedure was followed as for the establish-

J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
3577
and 34.0 mg mLꢀ1, covering the concentrations detected
within the sample assay. Compound 10 (35.0 mg mLꢀ1
was uniformlyemployed as internal standard, while
served this purpose in the constituent analysis of 10
itself.
ion spectra acquired for establishing the selectivity
required for direct sample injection. Based on selected
daughter ion peaks, conditions for each component’s
analysis were then optimised. Pertaining to the conditions
as for 8 - cyclohexylmethylamino - 8,11 - oxapentacyclo
[5.4.0.02,6.03,10.05,9]undecane (10), the internal standard,
the capillarytip was held at 3.50 kV and the cone vol-
tage maintained at 41 to produce the relevant [(M+)+1]
parent species (272+ m/z). Argon-induced fragmenta-
tion was instigated at a pressure of 1.4ꢃ10ꢀ3 mBar, with
the collision energysetting at 24.7 eV. The resolution of
MS2 was held at 13.0 and selective integration values
attained using parent ion (PAR) scans in the positive
ion mode. Dual function programming allowed for the
mutuallyexclusive quantification of the internal stan-
dard and the compound under investigation, both from
the same sample. After confirming the absence of ana-
lyte in the control samples, calibration curves were
employed and sample concentrations were related back
to the original tissue mass.
)
6
Brain tissue extraction
After careful notation of sample weight, the organs
were rinsed with ice-cold water (2ꢃ1 mL); HCl (10%; 2
mL) was added to each vial and the tissue homo-
genised using a Polytron1 homogeniser. (The rotator
head was thoroughlyrinsed with methanolic water
between applications, in order to prevent cross-con-
tamination.) The acidic homogenates were then trans-
ferred to polyethylene vials (Miniature Pony VialTM),
standard and internal standard solutions added and the
mixtures vortexed for 50 s. Following centrifugation
(15 min at 16,000 rpm; 4 ꢂC), the supernatants were
decanted and neutralised with ammonia solution (5%;
2 mL), establishing a final pH of 7–9. CHCl3 (3 mL)
was added and the deprotonated amine extracted by
vortexing the mixture for 60 s. Solvent phases were
separated bycentrifugation (5 min at 2800 rpm), the
aqueous layer discarded and the organic fraction
brought to dryness under a mild stream of nitrogen gas
at room temperature (25 ꢂC). The organic residue was
re-dissolved in mobile phase (1 mL) composed of 1%
formic acid in acetonitrile–water (3/2). A general factor
10 dilution was then performed to render the solutions
fit for ESI-MS.MS analysis without significant cross-
contamination.
Acknowledgements
Financial support from the Foundation for Pharma-
ceutical Education (Pharmaceutical Societyof South
Africa) is thankfullyrecognised.
References and Notes
1. Oliver, D. W.; Dekker, T. G.; Snyckers, F. O. Arzneim.-
Forsch. 1991, 41, 549.
2. Oliver, D. W.; Dekker, T. G.; Snyckers, F. O. Eur. J. Med.
Chem. 1991, 26, 375.
Blood extraction
Sample weight was carefullynoted, standard and inter-
nal standard solutions introduced and HCl (10%; 2 mL)
added to each vial. The mixtures were vortexed for 50 s.
Following centrifugation (15 min at 2800 rpm; 4 ꢂC), the
supernatants were decanted and neutralised with
ammonia solution (5%; 2 mL), establishing a final pH
of 7–9. Chloroform (3 mL) was added and the depro-
tonated amine extracted byvortexing the mixture for 60
s. Solvent phases were separated bycentrifugation (5
min at 2800 rpm), the aqueous layer discarded and the
organic fraction brought to dryness under a mild stream
of nitrogen gas at room temperature (25 ꢂC). The
organic residue was re-dissolved in mobile phase (1 mL)
composed of 1% formic acid in acetonitrile–water (3/2)
byvortexing the vials for 60 s. A general factor 10 dilu-
tion was then performed to render the solutions fit for
ESI-MS.MS analysis without significant cross-con-
tamination.
3. Malan, S. F.; Van den Heever, I.; Van der Schyf, C. J.
J. Pharm. Med. 1996, 6, 125.
4. Nguyen, V. H.; Mardon, K.; Kassiou, M.; Christie, M. J.
Nucl. Med. Biol. 1999, 26, 209.
5. Fujita, T.; Iwasa, J.; Hansch, C. J. Am. Chem. Soc. 1964,
86, 5175.
6. ACD/ChemSketch1 4.54; Advanced ChemistryDevelop-
ment Inc.: Toronto, Canada, 2000.
7. ACD/LogP1 4.54; Advanced ChemistryDevelopment Inc.:
Toronto, Canada, 2000.
8. Hyperchem for Windows1; Hypercube Inc.: Ontario,
Canada, 1994.
9. Burton, P. S.; Conradi, R. A.; Hilgers, A. R.; Ho, N. F. H.;
Maggiora, L. L. J. Control. Release 1992, 19, 87.
10. Ghose, A. K.; Crippen, G. M. J. Chem. Inf. Comput. Sci.
1987, 27, 21.
11. Viswanadhan, V. N.; Ghose, A. K.; Revankar, G. R.;
Robins, R. K. J. Chem. Inf. Comput. Sci. 1989, 29, 163.
12. Miller, K. J. J. Am. Chem. Soc. 1990, 112, 8533.
13. Bodor, N.; Gabanyi, Z.; Wong, C.-K. J. Am. Chem. Soc.
1989, 111, 3783.
14. Gavezzotti, A. J. Am. Chem. Soc. 1983, 105, 5220.
15. Clark, D. E. J. Pharm. Sci. 1999, 88, 815.
16. Hansch, C. Acc. Chem. Res. 1969, 2, 232.
17. Feher, M.; Sourial, E.; Schmidt, J. M. Int. J. Pharm. 2000,
201, 239.
Instrumentation and analysis
Electrosprayionisation tandem mass spectrometry(ESI-
MS.MS) analysis was performed using a VG Quattro II
tandem mass spectrometer (Micromass, Manchester,
UK) equipped with a HP1090 liquid chromatograph.
Standard samples of each of the seven synthesised com-
pounds were prepared in the mobile phase [1% formic
acid in acetonitrile–water (3/2)] and the relevant daughter
18. Chikhale, E. G.; Ng, K.-Y.; Burton, P. S.; Borchard, R. T.
Pharm. Res. 1994, 11, 412.
19. Abraham, M. H.; Chadha, H. S.; Mitchell, R. C. J. Pharm.
Sci. 1994, 83, 1257.

3578
J. Zah et al. / Bioorg. Med. Chem. 11 (2003) 3569–3578
20. Cookson, R. C.; Grundwell, E.; Hudec, J. Chem. Ind.
(London) 1958, 1003.
21. Malan, S. F.; Van der Walt, J. J.; Van der Schyf, C. J.
Arch. Pharm. Pharm. Med. Chem. 2000, 333, 10.
22. Polak, E. Computational Methods in Optimization, a
Unified Approach; Academic: New York, 1971.
23. Lombardo, F.; Blake, J. F.; Curatolo, W. J. J. Med.
Chem. 1996, 39, 4750.