Screening solution-phase combinatorial libraries using pulsed ultrafiltration/electrospray mass spectrometry

Article abstract of DOI:10.1021/jm960729b

A method is described whereby a family of homologues is synthesized in a one-pot reaction, without isolation or purification, and the reaction mixture is screened using a competitive binding assay based on pulsed ultrafiltration/electrospray mass spectrom

Full text of DOI:10.1021/jm960729b

4
006
J . Med. Chem. 1997, 40, 4006-4012
Scr een in g Solu tion -P h a se Com bin a tor ia l Libr a r ies Usin g P u lsed Ultr a filtr a tion /
Electr osp r a y Ma ss Sp ectr om etr y
†
†
†
†
†
Yong-Zhong Zhao, Richard B. van Breemen, Dejan Nikolic, Chao-Ran Huang, Charles P. Woodbury,
‡
,†
Alexander Schilling, and Duane L. Venton*
Department of Medicinal Chemistry and Pharmacognosy, College of Pharmacy, University of Illinois at Chicago,
8
33 South Wood Street, M/ C 781, Chicago, Illinois 60612-7231, and Hewlett-Packard Company, 1200 East Diehl Road,
Naperville, Illinois 60566
X
Received October 17, 1996
A method is described whereby a family of homologues is synthesized in a one-pot reaction,
without isolation or purification, and the reaction mixture is screened using a competitive
binding assay based on pulsed ultrafiltration/electrospray mass spectrometry (PUF/ESMS) to
tentatively identify those derivatives having the highest affinity for a target receptor. As a
model system to test this approach, a synthetic scheme designed to prepare a series of analogues
of the adenosine deaminase inhibitor erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA), as dia-
stereomeric mixtures, was carried out. Pulsed ultrafiltration screening of the crude reaction
mixture against controls without protein detected protonated molecules corresponding to EHNA-
type derivatives and three of its linear, alkyl homologues but did not show protonated molecules
for an isobutyl or benzylic EHNA derivative, suggesting the latter was inactive. To verify this
conclusion, we prepared E/THNA, the linear homologues, and the benzylic derivative (each as
a diastereomeric mixture) and bioassayed them for their adenosine deaminase inhibition index
([I]/[S]0.5). The bioassay results for the individually synthesized analogues were in good
agreement with that predicted by the observed relative ion enhancement in the PUF
experiments. Thus, the PUF protocol might be used as a general method to quickly provide
direction to the chemist in search of drug candidates.
A most promising new approach to drug discovery
concerns the synthesis and screening of combinatorial
libraries in order to identify new compounds that
express high affinity and specificity for a pharmacologi-
cally relevant, biomolecular target. Advances in mo-
lecular biology, automated chemical synthesis, and
robotics have facilitated the formulation of vast libraries
We are developing techniques with these goals in
mind, using a combination of pulsed ultrafiltration
(PUF) and electrospray mass spectrometry.7 Pulsed
ultrafiltration is a technique under development in our
laboratory for the quantitative measurement of macro-
molecule/ligand interactions. During pulsed ultrafil-
tration, a macromolecular target is trapped in a flow-
through cell by an ultrafiltration membrane; then an
aliquot of a ligand is pumped through the ultrafiltration
cell. As ligands bind and dissociate from the solution-
phase receptors (macromolecules or complexes thereof),
their elution profile from the ultrafiltration cell is
altered relative to control. We have developed solutions
to the differential equations describing this interaction
which in individual experiments have allowed us to
conveniently measure classical binding parameters, i.e.,
Kd, Ki (competitive binding), n (number of binding sites),
∆H, ∆S of binding, and in the case of enzymes, Michae-
1
of structurally related molecules.
An essential aspect of screening large combinatorial
libraries is the ability to identify the active components
in these complex mixtures, which is usually based on
the strength of binding to a selected target macromol-
ecule. A common approach to this problem has utilized
immobilization of either the target receptor or library
molecules on a solid-phase support in order to facilitate
2
,3
identification of the active compounds in the library.
Technically this is also true of the phage-related meth-
4
odologies. Another approach is the iterative resynthe-
8
sis of components of the library with progressively lower
complexity until a single most active compound is
lis-Menten constants, Km, are projected.
By coupling the pulsed ultrafiltration cell to an
electrospray mass spectrometer detector, ligand/biomol-
ecule interactions can be analyzed using extraordinarily
small amounts of material while concurrently providing
structural information on the ligand eluate. During
these studies, we realized that this system might
address the aforementioned problems associated with
identification of ligands of biological interest in a
combinatorial library and that this analysis could be
carried out with both ligand and macromolecule free in
solution. Thus, a combinatorial library pulse passed
through an ultrafiltration cell containing the target
macromolecule would show elution profile perturbation,
as a function of affinity, of that eluate ligand which
interacted with the macromolecule. In preliminary
work we have shown that we can adjust this system to
5
,6
identified.
However, immobilization may change the
affinity characteristics of the binding ligand or the
receptor from its native, solution-phase form, and the
resynthesis approach is still labor-intensive and time-
consuming. A method which would both facilitate the
screening of solution-phase libraries and target mol-
ecules with a minimum amount of sample handling and
also permit a rank ordering of binding constants of
target and library molecules in solution would be of
value.
*
Corresponding author. Tel: (312) 996-5233. Fax: (312) 996-7107.
E-mail: venton@uicvm.uic.edu.
†
University of Illinois at Chicago.
Hewlett-Packard Co.
Abstract published in Advance ACS Abstracts, November 15, 1997.
‡
X
S0022-2623(96)00729-7 CCC: $14.00 © 1997 American Chemical Society

Screening Combinatorial Libraries via PUF/ ESMS
J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 25 4007
F igu r e 1. Schematic representation of the PUF molecular diversity screening method.
completely separate ligands which have high affinity
and specificity for the target macromolecule compared
to those which are nonbound or weakly bound, i.e., to
ligand bound, [Lb], will be governed by the law of mass
action:
7
identify the “needle in the haystack”. In the present
[L ][M ]K
0 0 a
[
L ] )
work, we apply an experimental protocol where the
ligand/receptor is first incubated off-line and washed
and then the ligands are released into the mass spec-
trometer for identification.
b
1
+ [L ]Ka
0
where Ka is the association constant of the ligand for
the macromolecule, when there is a single binding site
for the ligand.
Resu lts a n d Discu ssion
Extension of this analysis suggests that when the
receptor is challenged with saturating amounts of
ligands, the concentration of ligands in the release phase
will be proportional to their affinities (for each class of
compound competing for a single binding site). Besides
identifying the ligands upon their release from the PUF
chamber, electrospray mass spectrometry may be used
to roughly estimate the relative concentrations and
therefore affinities of each ligand by comparing the
relative increase in ion abundances in the release phase
to those in the original library mixture. Within certain
limitations, one might use the PUF approach to quickly
assess which compounds in a mixture have the highest
affinity for a receptor, even in a crude reaction mixture.
Quantitative conclusions about ligand macromolecule
interactions assume a completely homogeneous mixture
of ligand and macromolecule during the PUF analysis.
Consequently in the present case, where the binding cell
is unstirred, this is only likely to be true in a relative
sense.
The general protocol for this screening procedure is
schematically outlined in Figure 1. In the present work,
we investigated whether this system might be applied
to the screening of solution-phase libraries produced by
conventional synthetic means. The general idea was to
apply a synthetic scheme which would likely produce a
family of homologues in a one-pot reaction and then,
without further characterization, screen the crude
mixture using the PUF/MS system. Potentially inter-
esting compounds identified by their mass would then
be further characterized, thus focusing the characteriza-
tion effort only on those compounds of interest.
A full understanding of the PUF approach to screen-
ing requires an intimate and quantitative understand-
ing of many interrelated variables, e.g., ligand affinity
constant, the efflux of unbound compound from the cell,
cell mixing and flow rate, receptor concentration, indi-
vidual library compound concentrations, protein polar-
ization on the membrane, and ionization during elec-
trospray. However, much can be gleaned by assuming
the system is first brought to a steady-state condition,
where the amount of bound ligand is defined by the law
of mass action, and then allowing the system to be
perturbed releasing bound ligands. Thus, if the binding
cell in Figure 1 is filled with macromolecule of con-
centration [M0] and then eluted with a library ligand
of concentration [L0] until steady state is reached,
the effluent from the cell and the free ligand concen-
tration in the cell will have a ligand concentration
of [L0]. Under these conditions, the concentration of
As a model system we chose adenosine deaminase
(ADA) and derivatives of the inhibitor EHNA [(+)-
-
9
erythro-9-(2-hydroxy-3-nonyl)adenine, Kd ) 7 × 10
9
M]. ADA is an enzyme which metabolizes adenosine
and is of some significance in the inactivation of
adenosine analogues used in cancer and AIDS treat-
ment. A number of analogues have been developed for
the inhibition of this enzyme, including EHNA.
Synthesis of potential EHNA derivatives was carried
out in three steps as shown in Figure 2. An equimolar
mixture of the eight methyl ketones was brominated

4
008 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 25
Zhao et al.
F igu r e 4. Computer-reconstructed mass chromatograms
+
(
from LC-MS analysis in Figure 3) showing [M + H] ions of
EHNA type analogues expected from the reaction given in
Figure 2. Each doublet represents four stereoisomers.
Ta ble 1. Electrospray Tandem Mass Spectrometry of
Synthetic EHNA Librarya
F igu r e 2. Synthesis of EHNA analogues. For the same “pot”
reaction, all eight side chains were introduced into the
syntheses at the same time and were themselves prepared in
one “pot”. Isolations of individual compounds were not per-
formed. The only purification was simple liquid-liquid extrac-
tion. Derivatives 6b,d ,h were also synthesized individually (as
their diastereomers) in the conventional manner and tested
for ADA inhibition (Table 2).
library
compd
ion fragment (m/z)
[MH - 46]+ [MH - H2O]+ [MH - NH3]+
(mass)
a
6
6
6
6
6
b (278) 136 (100) 232 (15.3)
h (284) 136 (100) 238 (6.2)
c (292) 136 (100) 246 (2.0)
d (306) 136 (100) 260 (1.8)
e (334) 136 (100) 288 (4.0)
260 (11.4)
266 (29.8)
274 (5.4)
288 (5.7)
316 (6.6)
261 (8.2)
267 (24.8)
275 (1.3)
289 (1.3)
317 (2.0)
a
Library was analyzed as a mixture without chromatographic
separation. Masses corresponding to all expected library com-
pounds were observed in the positive ion electrospray mass
spectrum of the mixture except the compound at m/z 250 (see LC-
MS analysis in Figure 5).
able chemical heterogeneity. Nevertheless, computer-
reconstructed mass chromatograms (Figure 4) indicated
that protonated molecules consistent with seven of the
potential products were present at approximately equal
abundance (each of the doublets probably represents a
set of diastereomers, data not shown for -nC3H7
product). A mass corresponding to the eighth product,
R ) (CH3)2CdCH-CH2-, did not appear in the crude
reaction mixture.
In addition to molecular weight confirmation of seven
EHNA analogues in the reaction mixture using LC-
MS, electrospray tandem mass spectra were obtained
in another set of analyses which provided additional
structural information of components of the library. A
summary of fragment ions common to all the EHNA
analogues is shown in Table 1. Fragment ions were
formed following protonation at either the primary
amino group or the hydroxyl group as indicated in the
structures in Table 1. The most abundant fragment ion
in each tandem mass spectrum corresponded to proto-
nated adenine, m/z 136, formed by elimination of the
substituent group from the adenine nitrogen (see struc-
ture in Table 1). Other significant fragment ions
F igu r e 3. Reverse-phase LC-MS profile of crude EHNA
library as prepared in Figure 2.
with trimethylbromosilane-dimethyl sulfoxide in dry
acetonitrile at room temperature. Adenine was then
treated with 2 mol equiv of the crude R-bromo ketones
(assumed) in N,N-dimethylacetamide in the presence
of potassium carbonate, and the presumed intermediate
ketones were reduced using sodium borohydride.
All three reactions were carried out without any
attempt to isolate intermediates or final products, other
than to remove salts and solvent. In addition to forming
multiple products by virtue of multiple competing
reagents in the same reaction vessel, the chemistry
outlined could produce many byproducts due to side
reactions. Thus, for each R group, four isomers are
expected about the two chiral centers in the molecule.
Further, other products might be expected by virtue of
side reactions from alkylation of other nitrogens in the
molecule and by other unidentified side reactions. As
shown in Figure 3, a positive ion electrospray LC-MS
analysis of this crude reaction mixture showed consider-

Screening Combinatorial Libraries via PUF/ ESMS
J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 25 4009
F igu r e 5. Protonated molecules released from ADA enzyme
and detected using the PUF/ES MS protocol outlined in Figure
1
(scan mode). Ions at m/z 359 and 397 are not present in
significant concentration in the library and are considered
artifacts.
included [MH - H2O]+ and [MH - 46] , which con-
firmed the presence of a hydroxyl group on the acyclic
side chain of adenine and loss of ammonia from the
protonated molecule, consistent with the presence of a
primary amine.
An advantage to this approach for screening is that
full structure elucidation of analogues is not carried out
until potential targets are identified. PUF/MS takes
advantage of the specificity of mass spectrometry and
the selectivity of PUF to quickly identify potential lead
compounds in crude reaction mixtures. Once a prelimi-
nary correlation of structure-activity has been estab-
lished, full characterization need only be carried out on
the smaller number of compounds of potential biological
interest.
+
F igu r e 6. Selected ion chromatograms (m/z 250, 278, 306,
and 334) of protonated library molecules washed from the
binding cell with methanol using the PUF/ES MS protocol
outlined in Figure 1: (a) in the absence of ADA, (b) in the
presence of ADA, and (c) in the presence of ADA with a 6-fold
excess of EHNA. The methanol wash (destabilizing condition)
starts at time ) 0.
Ta ble 2. Comparison of Observed PUF Binding Enhancement
of EHNA Analogues by ADA with the Biological Activity of
Selected Diastereomeric Pairs Synthesized Individually
diastereomeric ADA inhibtn inhibtn of ADA
relative ion
compd
M + 1) ion
index
([I]/[S]0.5)
relative to
E/THNA ) 1
enhancement
by PUF
a
(
6
6
b (278)
d (306)
0.00042
0.00016
0.069
1.0
2.6
0.006
1.0
5.3
<0.01
When ADA (approximately 1 nmol) was challenged
with this crude reaction mixture using the general PUF
screening protocol outlined in Figure 1, only certain
compounds were released by aqueous acetonitrile after
the initial wash phase (Figure 5), as identified by mass
6h (284)
a
Relative ion enhancement caused by the PUF process was
determined by measuring the increase in signal abundance relative
to that corresponding to E/THNA in Figure 5 (representing PUF
enhancement of bound derivatives) versus the relative abundance
of the same signal to E/THNA derivatives in Figure 4 (relative
concentration of derivatives in reaction mixture before PUF).
+
spectrometry. Thus, [M + H] ions corresponding to
protonated EHNA (m/z 278) and three of its putative
homologues (nC7H15, m/z 292; nC8H17, m/z 306; and
nC10H21, m/z 334; see Figure 2 for structures) were
detected eluting from the cell in the recovery phase.
However, protonated ions corresponding to the n-propyl
to the above were obtained for ions in the presence of
ADA versus control, where only (()-EHNA and its
linear homologues were observed to bind. Further, in
the presence of the excess EHNA, signals corresponding
to the protonated ions of the linear homologues (R )
nC H - and R ) C H -) were completely suppressed.
(m/z 236), isobutyl (m/z 250), and benzylic (m/z 284)
derivatives were not detected in these scans when
compared against appropriate controls, even though
their presence was observed in the reaction mixture
before addition of protein. There is little or no signal
at m/z 320 corresponding to the nC9H19 homologue
8
17
10 21
This suggests that binding of these compounds was
competitive with EHNA for the enzyme. The initial
burst of compounds at about 24 min is thought to be
due to nonspecific binding of the EHNA compounds to
the membrane and cell parts.
(Figure 5). This particular side chain was not intro-
duced as a reactant and thus acts as a negative control
for the experiment. The ions at m/z 359 and 397 are
considered to be artifacts, since they are not present in
significant concentration in the library control experi-
ments and were known to be significant background
ions in this instrument.
A similar series of experiments was carried out using
positive ion electrospray mass spectrometry, with se-
lected ion monitoring. In these experiments, protonated
molecules corresponding to four of the seven EHNA-type
analogues in the reaction mixture were monitored in
the presence (Figure 6b) and absence (Figure 6a) of ADA
and finally with a 6-fold excess of authentic (()-EHNA
added (relative to protein, Figure 6c). Results similar
Taken at face value, these PUF experiments can
provide a first-approximation structure-activity rela-
tionship (SAR) analysis of synthetic analogues produced
in the one-pot reaction. In general, the greater the
enhancement of a particular derivative in the eluate
from the binding cell in the release phase relative to
its starting concentration in the library, the greater its
expected interaction with the receptor protein in the cell
(see previous discussion on stirring and competitive
interaction at a single site).
Table 2 gives the relative enhancement of the ion
currents corresponding to the higher alkyl homologue
(6d ) and benzylic derivative (6h ) versus the E/THNA
diastereomers (6b) with the latter enhancement being

4
010 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 25
Zhao et al.
assigned a value of 1. As may be seen, the m/z 306 ion
alkyl homologue 6d ) is enhanced by a factor of about
, whereas the m/z 284 ion (benzylic derivative 6h )
shows less than 1/100 the enhancement of the E/THNA
derivatives. Given the fact that these compounds
appear to be acting at the active site on the enzyme (see
previous discussion of competition with authentic EHNA),
this would strongly suggest that the benzylic derivative
MHz). Chemical shifts are denoted in ppm (δ) relative to TMS
as the internal reference. Unless otherwise noted the NMR
(
5
spectra were obtained using CDCl
3
as the solvent.
Synthetic mass spectra were obtained on either a Finnigan
MAT 90 double-focusing mass spectrometer or the Hewlett-
Packard mass spectrometer described below. Tandem MS
data were acquired using a Bruker (Billerica Mall, MA)
ESQUIRE ion trap mass spectrometer equipped with electro-
spray ionization.
The pumping and injection apparatus for the mass spec-
trometer interface consisted of a Hewlett-Packard (Palo Alto,
CA) 1090L gradient HPLC system (50 µL/min) equipped with
a Rheodyne (Cotati, CA) model 8125 injector and photodiode
array UV/vis absorbance detector. UV data were collected by
monitoring absorbance at 260 nm. Mass spectrometry was
carried out with a Hewlett-Packard mass spectrometer equipped
with a ChemStation data system and nebulizer-assisted elec-
trospray LC-MS interface. During analysis the quadrupole
analyzer was maintained at 120 °C. Nitrogen (80 psi) was
used for nebulization of the eluate, and the nitrogen bath gas,
for evaporation of solvent from the electrospray, was held at
300 °C and a flow rate of 10 L/min.
(
6h ) is inactive relative to the EHNA family and as a
first approximation that the higher homologues may be
more active.
To verify these conclusions, we prepared the diaster-
eomers of EHNA (i.e., EHNA and THNA, 6b), the linear
homologues 6d , and the benzylic derivative 6h and
bioassayed them for their ADA inhibition index ([I]/
[
S]0.5) using the method of Schaeffer.10 The synthetic
approach was the same as for the one-pot reaction
except that the reactions were carried out individually
and the intermediates were purified by chromatography.
The final derivatives (6b,d ,h ) were greater than 99%
pure by HPLC (UV detection) and showed the expected
ESMS signals at m/z 278, 306, and 284, respectively,
as the only ion currents. As may be seen in Table 2,
relative to an E/THNA activity of 1, the derivative 6d
is more than 2 times as active and the benzylic deriva-
tive 6h has about 6/1000 the activity of the EHNA
family. These results are in good agreement with that
predicted by the observed relative ion enhancement in
the PUF experiments, also shown in Table 2.
The synthesis and PUF screening of the EHNA
derivatives presented in this paper can be carried out
by one person in approximately 2-3 days with about
equal time spent on the chemistry and mass spectrom-
etry. To the extent that such a system is generally
applicable, standard solution-phase chemistry might
effectively be used in combination with PUF/electro-
spray mass spectrometry to quickly provide a first-line
screen of related derivatives and direction to the chemist
in search of biologically active compounds.
Bin d in g Cell. The binding cell was fashioned from an
HPLC preparative in-line solvent filter (Upchurch Scientific,
Oak Harbor, WA; cat. #A-333) in which the filter element was
replaced by a semipermeable membrane (YM10, MWCO
1
0 000; Amicon, Beverly, MA), which was cut to exactly the
same size as the filter element. When the membrane was
inserted between the two internal halves of the cell body and
the cell assembled with the outer compression elements, the
system formed two internal chambers of different sizes, again
separated by the semipermeable membrane. The larger of the
two chambers was used as the inlet side (where the enzyme
was held) and the smaller chamber the outlet side of the
binding cell. Both the inlet and outlet were connected to 0.005-
in. i.d. tubing (PEEK, Upchurch) via finger tight fittings
(Upchurch). To enhance flow from the outlet side of the cell,
small (ca. 0.001 in.) grooves were cut in the outlet face of the
cell body. These grooves were radial in nature, terminating
at the center hole for the outlet port. The total volume of the
larger chamber in the cell in this configuration was ap-
proximately 80 µL.
On e-P ot Syn th esis of EHNA An a logu es. The R-halo
ketones were prepared as a mixture using the trimethylbro-
mosilane-dimethyl sulfoxide method of Pagnoni.1 Briefly, to
a stirred solution of a mixture of eight methyl ketones (1 mmol
each) in dry acetonitrile (20 mL) were added trimethylbro-
mosilane (1.16 mL, 8.8 mmol) and then dimethyl sulfoxide
(0.63 mL, 8.8 mmol) dropwise. The eight ketones consisted of
2-hexanone, 2-nonanone, 2-decanone, 2-undecanone, 2-tride-
canone, 5-methyl-2-hexanone, 6-methyl-5-hepten-2-one, and
benzylacetone. The reaction mixture was kept at room tem-
perature for 2 h, poured into water (80 mL), and subsequently
extracted with ethyl ether (3 × 35 mL). The ether layer was
dried over anhydrous sodium sulfate and then evaporated in
vacuo to a brown liquid which was used without further
purification.
1
Exp er im en ta l Section
(()-EHNA was purchased from Sigma Chemical Co. (St.
Louis, MO). Adenosine deaminase enzyme (ADA) was pur-
chased from Boehringer Mannheim (Indianapolis, IN) as a
suspension in 3.2 M (NH ) SO . The ADA was pelleted by
4 2 4
centrifugation, the supernatant decanted, and the pellet
dissolved in phosphate buffer before being used. Enzyme
concentrations were measured by the BCA method using a
commercially available kit from Pierce (Rockford, IL). Molar-
ity of the enzyme was calculated using a MW of 41 250.5,
which was determined by using matrix-assisted laser desorp-
A mixture of the R-bromo ketones (assumed 8 mmol),
adenine (540 mg, 4 mmol), anhydrous potassium carbonate
(552 mg, 4 mmol), and N,N-dimethylacetamide (20 mL) were
added together and then stirred at 110 °C for 1 h. After
cooling, the reaction mixture was filtered, and the filtrate was
evaporated in vacuo to dryness giving a brown solid which was
used without further purification in the following reduction
step.
7
tion time-of-flight mass spectrometry.
Reagents used in the synthesis were purchased from Aldrich
Chemical Co. (Milwaukee, WI). Column chromatography was
monitored by TLC (silica gel 60 F254; Rieldel-de Ha e¨ n AG,
Germany). Silica gel for column chromatography was pur-
chased from Sigma Chemical Co. (St. Louis, MO; 10-40 µm
for flash chromatography and 200-400 mesh for open column).
TLC plates were visualized with UV light or by treatment with
phosphomolybdic acid (PMA). All single intermediates and
final products were judged homogeneous by TLC and charac-
terized by NMR and MS. Individually synthesized final
diastereomers were also subjected to HPLC, using a Hypersil
BDS-C18 2- × 250-mm column with 5-µm particle (Hewlett-
Packard) at a flow rate of 0.22 mL/min. For the library, a
Hypersil C18 4- × 100-mm column (Hewlett-Packard) was used
and LC carried out at 0.60 mL/min. HPLC grade solvents
were obtained from Fisher Scientific Inc. (Itasca, IL). NMR
spectra were recorded on a Varian XL 300 spectrometer (300
A mixture of NaBH
4
(610 mg, 16 mmol) and the brown solid
obtained from the previous alkylation step (assumed 8 mmol)
in ethanol (25 mL) was allowed to stir at room temperature
for 1 h. The reaction was then quenched with acetic acid and
3
neutralized with saturated NaHCO and the majority of the
ethanol removed in vacuo. After dilution with water, the
mixture was extracted with ethyl ether (3 × 10 mL). The ether
layer was dried and evaporated in vacuo to produce a yellow
liquid which was used directly in the binding studies.
This yellow liquid was analyzed by reverse-phase electro-
spray LC-MS. Thus, 3 µL of the oil diluted 20× with

Screening Combinatorial Libraries via PUF/ ESMS
J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 25 4011
undecanyl)adenine (5d ; 90 mg, 56%) as a white solid after
methanol was injected onto a C18 column (Hewlett-Packard
Hypersil, 10 cm × 4 mm) attached to the aforementioned
electrospray mass spectrometer system. The column was
eluted with a water(A)/methanol(B) linear gradient (12-45%
B, 0-20 min; 45-85% B, 20-70 min; 85-95% B, 70-75 min;
and 95% B, isocratic) at a flow rate of 60 µL/min. Mass spectra
were scanned from 100 to 500 amu at a rate of 1 scan/s. The
results of these studies are shown in Figure 3. The MS/MS
analyses were carried out by direct infusion of the yellow liquid
in a solution of acetonitrile/water (1:1, v/v) containing 1% acetic
acid at 10 µL/min. The results of this study are summarized
in Table 1.
purification by flash chromatography: TLC (CHCl
) 0.52; NMR δ 0.85 (3H, t, CH ), 1.21 (12H, m, 6CH
2.05, 2.26 (2H, m, CH ), 2.28 (3H, s, CH ), 5.38 (1H, q, CH),
6.56 (2H, s, NH ), 7.97 (1H, s, Ade-2 or -8H), 8.35 (1H, s, Ade-2
or -8H); δ 13.99, 22.50, 25.80, 27.59, 28.79, 28.97, 29.06, 30.94,
31.61 (CH (CH , CH ), 62.27 (-CH(Ade)CO-), 118.97, 139.38,
150.01, 152.92, 155.77 (Ade 5 C’s), 203.34 (-COCH ).
Reduction of 5d (89 mg, 0.29 mmol) in EtOH (3 mL) with
NaBH (4 mg, 0.11 mmol) gave the diastereomeric mixture of
9-(2-hydroxy-3-undecanyl)adenines as a white solid (6d ; 89 mg,
100%): TLC one spot, R ) 0.41 (ethyl acetate/ethanol, 9:1);
HPLC >99.0% (75% MeOH/H O, 20.4 and 22.9 min, corre-
sponding to the diastereomers); ESMS gave a single ion
current at m/z 306; NMR δ 0.85 (3H, t, CH ), 1.08/1.29, 1.19
(15H, 2d, 1m, CH CH(OH), 6CH ), 1.98, 2.08 (2H, m, CH CH-
(Ade)), 4.26 (1H, m, CH CH(OH) or C 17 CH(Ade)), 4.38 (1H,
m, CH CH(OH) or C 17CH(Ade)), 6.70 (2H, s, NH ), 7.86/7.91
(1H, 2s, Ade-2 or -8H), 8.28 (1H, s, Ade-2 or -8H); δ 13.99,
20.10, 20.77, 22.51, 26.12, 27.29, 29.04/29.08, 29.20/29.22,
31.66 (CH (CH , CH ), 62.31/62.71, 68.37/69.26 (CH(OH)CH-
(Ade)C 17), 119.22/119.48, 140.19/140.69, 149.58/149.84, 152.30/
152.35, 155.79/155.82 (five Ade ring carbons); HRMS (EI) for
O calcd 305.2216, found 305.2208.
9-(2-Hyd r oxy-3-ben zyl)a d en in e (6h ). The diastereomeric
3
/MeOH, 9:1)
R
f
H
3
2
),
2
3
2
C
3
2
)
7
3
3
4
f
Syn th esis of Selected Dia ster eom er ic EHNA An a -
logu es. 9-(2-Hyd r oxy-3-n on yl)a d en in e (6b). The required
R-halo reagent 3b was prepared by the trimethylbromosilane-
2
H
3
1
2
dimethyl sulfoxide method of Pagnoni. Briefly, trimethyl-
bromosilane (1.0 mL, 7.58 mmol) and then DMSO (0.54, 7.58
mmol) were added slowly at room temperature to a stirred
solution of 2-nonanone (2b; 0.98 g, 6.89 mmol) in dry aceto-
nitrile (20 mL). The mixture was stirred for 2 h, poured into
3
2
2
3
8
H
3
8
H
2
C
ice-cold water (50 mL), and extracted with CHCl
The organic layer was dried (Na SO ) and evaporated to give
a residue which was subjected to column chromatography
hexane/CHCl , 3:2) to yield 3-bromo-2-nonanone (3b; 0.76 g,
0%): TLC R
t, CH ), 1.30 (12H, m, 4 CH
CH ), 4.23 (1H, t, CH).
A stirred mixture of the 3-bromo-2-nonanone (3b; 448 mg,
mmol), anhydrous K CO (248 mg, 1.8 mmol), and adenine
4; 243 mg, 1.8 mmol) in N,N-dimethylacetamide (5 mL) was
3
(3 × 20 mL).
3
2
)
7
3
2
4
8
H
(
3
16 27 5
C H N
5
f
) 0.44 (CHCl
3
/hexane, 1:1); NMR δ
H
0.89 (3H,
3
2
), 1.96 (2H, m, CH ), 2.37 (3H, s,
2
mixture of aromatic EHNA analogues 6h and its intermediates
were prepared in a manner similar to the undecanyl analogues
6d , described above. Bromination of the ketone (1.58 g, 10.7
mmol) gave, after column chromatography (hexane/EtOAc, 10:
1), 3-bromo-benzylacetone (3h ; 1.38 g, 6.1 mmol, 67%) as a
3
2
2
3
(
kept at 70 °C for 30 min. The insoluble solids were then
filtered and washed with hot ethanol. Ethanol from the
filtrate and N,N-dimethylacetamide were removed under
vacuum, and the brown residue so obtained was purified by
flash chromatography (EtOAc/MeOH, 93:7) to give 9-(2-keto-
colorless oil: TLC single spot, R
1); NMR δ 2.32 (3H, s, CH CO), 3.31 (2H, dd, PhCH
(1H, t, CHBr), 7.19-7.35 (5H, m, C ).
Alkylation of adenine (718 mg, 5.32 mmol) with the bro-
moketone 3h (158 mg, 0.70 mmol) yielded 9-(2-keto-3-benzyl)-
adenine (5h ; 793 mg, 53%) after column chromatography
f
) 0.49 (hexanes/EtOAc, 10:
H
3
2
), 4.47
6
H
5
3
-nonyl)adenine (5b; 274 mg, 55%): TLC R
MeOH, 93:7); NMR δ 0.84 (3H, t, CH ), 1.22 (8H, m, 4CH
), 2.28 (3H, s, CH ), 5.39 (1H, q, CH),
), 7.98 (1H, s, Ade-2 or -8H), 8.35 (1H, s, Ade-2
(CDCl ) 13.93, 22.39, 25.83, 27.64, 28.52, 30.99,
(CH , CH ), 62.34 (C 13CH(Ade)CO-), 119.05,
39.40, 150.06, 153.01, 155.98 (five Ade ring carbons), 203.46
CO).
To a solution of keto derivative 5b (186 mg, 0.68 mmol) in
EtOH (3 mL) was added aqueous NaBH (3 mL, 9 mg, 0.24
f
) 0.55 (EtOAc/
H
3
2
),
(EtOAc/EtOH, 9:1): TLC one spot, R
9:1); NMR δ 2.21 (3H, s, CH CO), 3.47 (2H, dd, PhCH
(1H, q, CH(Ade)CO), 6.10 (2H, s, NH ), 7.01-7.22 (5H, m,
), 7.76 (1H, s, Ade-2 or -8H), 8.32 (1H, s, Ade-2 or -8H);
28.25 (CH ), 36.91 (CH ), 63.74/63.90 (-CHCOCH ), 127.38,
128.71, 128.84, 135.25 (benzene 6 C’s), 119.20, 139.67, 150.00,
153.01, 155.57 (Ade 5 C’s), 202.93 (-COCH ).
Reduction of 5h (320 mg, 1.14 mmol) with NaBH
0.41 mmol) gave 9-(2-hydroxy-3-benzyl)adenine (6h ; 290 mg,
90%): TLC R ) 0.27 (EtOAc/EtOH, 9:1); HPLC purity >99.0%
(30% MeOH/H O, 12.6 and 13.8 min, two diastereomers);
1.07/1.43
), 4.25-4.52 (2H, m, BzCH(Ade)-
), 6.30/6.36 (2H, 2s, NH ), 6.80-6.84, 6.94-6.98,
f
) 0.36 (EtOAc/EtOH,
2
6
.05, 2.27 (2H, m, CH
.76 (2H, s, NH
2
3
H
3
2
), 5.51
2
2
or -8H); δ
3
1
C
3
6 5
C H
1.33 (CH
3
2
)
5
3
6
H
δ
C
3
2
3
(CH
3
3
4
(15.6 mg,
4
mmol). The mixture was stirred at room temperature for 1.5
h and then the reaction quenched by adding acetic acid. The
f
2
solution was neutralized with saturated aqueous NaHCO
then extracted with ethyl acetate (3 × 15 mL). The organic
phase was dried (MgSO ) and then evaporated to give a
diasteromeric mixture of 9-(2-hydroxy-3-nonyl)adenine (6b;
3
and
ESMS gave a single ion current at m/z 284; NMR δ
(3H, 2d, CH ), 3.30 (2H, t, CH
3 2
H
4
CH(OH)CH
3
2
7.14-7.19 (5H, m, benzene H), 7.50 (1H, s, Ade-2 or -8H), 8.31/
8.32 (1H, 2s, Ade-2 or -8H); δ 20.59/20.92 (CH ), 33.84/38.10
(CH ), 65.16/66.54/67.80/68.84 (-CH(Ade)CH(OH)CH ), 126.82/
1
9
87 mg, 100%): TLC one spot, R
:1); HPLC >99.0% (42% MeOH/H
f
) 0.41 (ethyl acetate/ethanol,
c
3
2
O, 12.1 and 13.1 min,
2
3
corresponding to the diastereomers); ESMS gave a single ion
current at m/z 278; NMR δ 0.78 (3H, t, CH ), 1.08/1.28, 1.18
CH(OH), 4CH ), 1.96, 2.06 (2H, m, CH CH-
Ade)), 4.25 (1H, m, CH CH(OH) or C 13CH(Ade)), 4.40 (1H,
CH(OH) or C 13CH(Ade)), 5.3 (1H, broad, OH), 6.92
2H, s, NH ), 7.90/7.95 (1H, 2s, Ade-2 or -8H), 8.28 (1H, s,
Ade-2 or -8H); δ 13.83, 19.96/20.65, 22.32, 25.98, 27.37,
8.67, 31.37/31.58 (CH (CH , CH ), 61.93/62.26, 68.23/69.09
CH(OH)CH(Ade)C 13), 118.93/119.19, 140.00/140.52, 149.53/
49.83, 152.20, 155.75/155.79 (five Ade ring carbons); HRMS
EI) for C14 O calcd 277.1903, found 277.1884.
-(2-Hyd r oxy-3-u n d eca n yl)a d en in e (6d ). The diaster-
126.89, 128.66, 128.74, 137.24/137.45 (phenyl C), 119.56/
119.85, 140.98/141.39, 149.00/149.23, 151.95/151.97, 155.63/
155.69 (Ade C’s); ESMS gave a single ion current at m/z 284
(M + 1); HRMS (EI) for C15H N O calcd 283.1433, found
17 5
283.1421.
H
3
(
(
11H, 2d, 1m, CH
3
2
2
3
6
H
m, CH
(
3
6
H
2
Scr een in g P r otocol. Immediately before each experiment,
the membrane was removed from its protective envelope and
rinsed by floating glossy side down in a beaker of distilled
water for 1 h while changing the water three times. After
mounting in the ultrafiltration cell (glossy side toward the
larger of the two chambers which will contain the enzyme),
the membrane was flushed with 90% aqueous methanol or 90%
aqueous acetonitrile for 15-20 min at 0.1 mL/min to remove
organics present on its surface. The system was then equili-
brated with binding buffer until UV detection showed a stable
baseline.
The crude library oil as prepared above (100 µL) was
dissolved in methanol (1 mL) and then diluted approximately
300-fold with potassium phosphate buffer (50 mM, pH 7.5).
The control experiment was carried out first and was exactly
the same as for the binding experiment except that buffer
without enzyme was used. For the binding experiment, the
aqueous buffer solution of the library (100 µL) was mixed with
C
2
(
1
(
3
2
)
5
3
6
H
23 5
H N
9
eomeric mixture of EHNA-type analogues 6d and its inter-
mediates were prepared in a manner similar to the nonane
derivatives described above. Bromination of 2-undecanone
(
3
2d ; 1.03 mL, 5 mmol) in dry acetonitrile (10 mL) gave
-bromo-2-undecanone (3d ; 0.80 g, 65%) as a colorless oil after
chromatography (cyclohexane/CHCl , 3:2): TLC (cyclohexane/
CHCl , 3:2) R ) 0.50; NMR δ 0.88 (3H, t, CH ), 1.27 (12H,
m, 6CH ), 1.96 (2H, m, CH ), 2.36 (3H, s, CH ), 4.22 (1H, t,
3
3
f
H
3
2
2
3
1
2
CH).
Alkylation of adenine (4; 71 mg, 0.53 mmol) with 3-bromo-
-undecanone (3d ; 174 mg, 0.70 mmol) gave 9-(2-keto-3-
2

4
012 J ournal of Medicinal Chemistry, 1997, Vol. 40, No. 25
Zhao et al.
phosphate buffer (150 µL) containing ADA enzyme (8.5 µM,
total 1.02 nmol) and the mixture allowed to incubate at room
temperature for 10 min before injection (200 µL) into the
binding cell. The binding cell was then flushed with water at
are the concentration of inhibitor and substrate, respectively.
For each plot, the reaction was carried out in triplicate at five
different inhibitor concentrations. The average initial velocity
was then used in the aforementioned plots. Results normal-
ized to the EHNA diastereomeric inhibitors are presented in
Table 2.
1
00 µL/min to reduce salts and unbound and weakly bound
ligands present in the library. To prevent buffer from entering
the mass spectrometer during this wash phase, the flow was
directed to waste. After 6 min of washing, the flow was
directed to the mass spectrometer and the cell flushed for
another 10 min with water. Mass spectra were collected by
scanning the range m/z 100-400 over ca. 6 s at unit mass
resolution. At 16-min postinjection, the eluant was changed
to 90% acetonitrile at 50 µL/min (methanol produced similar
results) to disrupt the ligand/enzyme complex and release
ligands into the mass spectrometer as a bolus. To enhance
electrospray ionization, water/acetonitrile/acetic acid (47.5:
Ack n ow led gm en t. We thank Hewlett-Packard for
generously providing the mass spectrometers used in
this study.
Refer en ces
(
1) (a) DeWitt, S. H.; Czarnik, A. W. Combinatorial organic synthesis
using Parke-Davis’s diversomer method. Acc. Chem. Res. 1996,
29, 114-122. (b) Gordon, E. M.; Gallop, M. A.; Patel, D. V.
Strategy and tactics in combinatorial organic synthesis. Ap-
plications to drug discovery. Acc. Chem. Res. 1996, 29, 144-
4
7.5:5, v/v/v) was added after the cell at 10 µL/min.
Compounds were analyzed using selected ion chromato-
1
54. (c) Still, W. C. Discovery of sequence-selective peptide
binding by synthetic receptors using encoded combinatorial
libraries. Acc. Chem. Res. 1996, 29, 155-163. (d) Lowe, G.
Combinatorial chemistry. Chem. Soc. Rev. 1995, 309-317. (e)
Ellman, J . A. Design, synthesis, and evaluation of small-molecule
libraries. Acc. Chem. Res. 1996, 29, 132-143.
grams to define their release after the changeover to the
organic acetonitrile phase. Typically, there was a pronounced
release of compounds when the acetonitrile first reached the
binding cell. This initial signal then returned to baseline
before the bound ligands began to elute at about 10 min after
acetonitrile reached the binding chamber. This latter signal
lasted about 25 min under the conditions used. Spectra were
summed over this last area, and control spectra (summed over
exactly the same time period) were subtracted. The resulting
mass spectrum was reproducible and is shown in Figure 5.
Com p etitive Bin d in g P r otocol. In the competitive bind-
ing studies, three experiments were performed in which the
library was subjected to the protocol described above: (1)
without ADA enzyme, (2) with ADA enzyme, and (3) with ADA
enzyme and an excess of externally added, authentic (()-
EHNA. Controls and data analyses were performed es-
sentially the same as described above, except that (1) selected
ion monitoring (2000-ms dwell time) was used for ion detection
and (2) the library incubation was carried out in the presence
of (()-EHNA (4.15 nmol) equal to 6 times the enzyme
concentration. The amount of enzyme used in these experi-
ments was 0.69 nmol, and 90% methanol was used for the
release phase. The results of these three experiments are
presented in Figure 6.
(
2) (a) Gallop, M. A.; Barrett, R. W.; Dower, W. J .; Fodor, S. P.;
Gordon, E. M. Applications of combinatorial technologies to drug
discovery. 1. Background and peptide combinatorial libraries.
J . Med. Chem. 1994, 37, 1233-1251. (b) Gordon, E. M.; Barrett,
R. W.; Dower, W. J .; Fodor, S. P.; Gallop, M. A. Applications of
combinatorial technologies to drug discovery. 2. Combinatorial
organic synthesis, library screening strategies, and future direc-
tions. J . Med. Chem. 1994, 37, 1385-1401. (c) Desai, M. C.;
Zuckermann, R. N.; Moos, W. H. Recent advances in the
generation of chemical diversity libraries. Drug Dev. Res. 1994,
33, 174-188.
3) Kelly, M. A.; Liang, H.; Sytwu, I.-I.; Vlattas, I.; Lyons, N. L.;
Bowen, B. R.; Wennogle, L. P. Characterization of SH2-ligand
interactions via library affinity selection with mass spectrometric
detection. Biochemistry 1996, 35, 11747-11755.
4) Scott, J . K.; Smith, G. P. Searching for peptide ligands with an
epitope library. Science 1990, 249, 386-390.
(5) Houghten, R. A.; Pinilla, C.; Blondelle, S. E. Generation and use
of synthetic peptide combinatorial libraries for basic research
and drug discovery. Nature (London) 1991, 354, 84-86.
6) Houghten, R. A.; Dooley, C. T. The use of synthetic peptide
combinatorial libraries for the determination of peptide ligands
in radioreceptor assays: opiate peptides. Bioorg. Med. Chem.
Lett. 1993, 3, 405-412.
7) van Breemen, R. B.; Huang, C.-R.; Nikolic, D.; Woodbury, C. P.;
Zhao, Y.-Z.; Venton, D. L. Pulsed ultrafiltration-mass spec-
trometry: A new method for screening combinatorial libraries.
Anal. Chem. 1997, 69, 2159-2164.
(
(
(
(
Bioassay of Selected Diaster eom er ic Mixtu r es of EHNA
An a logu es. The inhibitory activity of three selected EHNA
analogues, i.e., 6b,d ,h , was measured by the method of
1
0
Schaeffer.
The inhibition index [I]/[S]0.5, i.e., the ratio of
(8) (a) Chen, C.; Venton, D. L. Pulsed ultrafiltration analysis of
ligand and macromolecule interactions. 206th National Meeting
of the American Chemical Society, Chicago, IL, August 1993;
Abstract. (b) Chen, C. J .; Chen, S.; Woodbury, C. P.; Venton, D.
L. Pulsed ultrafiltration characterization of binding I: A new
method. Unpublished results.
concentration of inhibitor to that of substrate for 50% inhibi-
tion, was used to compare the inhibitory potency of the
compounds. All compounds were assayed as a mixture of two
diastereomers. The ratio of the diastereomers, isolated by
chromatography, was close to 1:1 based on HPLC analyses.
All enzyme reactions were carried out at 25 °C in phosphate
buffer (50 mM, pH 7.6) and monitored at 264 nm by the
decrease of the absorbance of the substrate. The amount of
enzyme was chosen so that the decrease in absorbance was
constant in the first 10 min. Briefly, 100 µL of enzyme was
incubated with 800 µL of inhibitor solution at various concen-
trations for 10 min. The reaction was started by adding 100
(
9) North, T. W.; Cohen, S. S. Aranucleosides and aranucleotides
in viral chemotherapy. Pharmacol. Ther. 1979, 4, 81-108.
(10) Schaeffer, H. J .; Schwender, C. F. Enzyme inhibitors. 26.
Bridging hydrophobic and hydrophilic regions on adenosine
deaminase with some 9-(2-hydroxy-3-alkyl)adenines. J . Med.
Chem. 1974, 17, 6-8.
(
11) Bellesia, F.; Ghelfi, F.; Grandi, R.; Pagnoni, U. M. Regioselective
R-bromination of carbonyl compounds with trimethylbromosi-
lane-dimethyl sulphoxide. J . Chem. Res. 1986, 428-429.
12) Bergeron, R. J .; Hoffman, P. Application of N-Phenyltrifluo-
romethanesulfonamides to the synthesis of pyrazines. J . Org.
Chem. 1980, 45, 161-163.
(
µL of an 0.66 mM solution of adenosine. To determine [I]/[S]0.5
,
v
0
/v was plotted versus [I]/[S], where v is the initial velocity
0
of the uninhibited reaction, v is the initial velocity of inhibited
reaction at various inhibitor concentrations, and [I] and [S]
J M960729B