A phosphotyrosine-imprinted polymer receptor for the recognition of tyrosine phosphorylated peptides

Article abstract of DOI:10.1002/chem.200801046

Hyperphosphorylation at tyrosine is commonly observed in tumor proteomes and, hence, specific phosphoproteins or phosphopeptides could serve as markers useful for cancer diagnostics and therapeutics. The analysis of such targets is, however, a challenging

Full text of DOI:10.1002/chem.200801046

DOI: 10.1002/chem.200801046
A Phosphotyrosine-Imprinted Polymer Receptor for the Recognition of
Tyrosine Phosphorylated Peptides
Marco Emgenbroich,[a] Cristiana Borrelli,[a] Sudhirkumar Shinde,[a] Issam Lazraq,[a]
Filipe Vilela,[a] Andrew J. Hall,[a, e] Joakim Oxelbark,[b] Ersilia De Lorenzi,[b]
Julien Courtois,[c] Anna Simanova,[c] Jeroen Verhage,[c] Knut Irgum,[c] Kal Karim,[d] and
Bçrje Sellergren*[a]
9516
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 9516 –9529

FULL PAPER
Abstract: Hyperphosphorylation at ty-
rosine is commonly observed in tumor
proteomes and, hence, specific phos-
phoproteins or phosphopeptides could
serve as markers useful for cancer diag-
nostics and therapeutics. The analysis
of such targets is, however, a challeng-
ing task, because of their commonly
low abundance and the lack of robust
and effective preconcentration tech-
niques. As a robust alternative to the
commonly used immunoaffinity tech-
niques that rely on phosphotyrosine-
C
ing affinities for the pTyr template, in
the range of that observed for corre-
sponding antibodies, and a clear prefer-
ence for pTyr over phosphoserine
(pSer). In further analogy to the anti-
bodies, the imprinted polymer was ca-
pable of capturing short tyrosine phos-
phorylated peptides in the presence of
an excess of their non-phosphorylated
counterparts or peptides phosphorylat-
ed at serine.
Introduction
dant proteins.[2] An urgent need for alternative affinity tech-
niques has therefore emerged.
Protein phosphorylation and dephosphorylation is a key reg-
ulating mechanism of biological processes and, therefore, is
a post-translational modification of profound biological im-
portance.[1] Defects in the function or level of the enzymes
carrying out the modification, that is, the kinase/phosphatase
switch, have been implicated as the major cause of several
diseases, including cancer. Comprehensive mapping of the
phosphoproteome is, therefore, an important objective in
todayꢁs proteomics research.[2–5] In this context, it has
proven particularly difficult to obtain a comprehensive view
of proteins phosphorylated at tyrosine (Scheme 1).[4] This is
because the prefractionation techniques used for this pur-
pose commonly lack the selectivity, sensitivity, and robust-
ness required for reproducibly extracting these low-abun-
In this area, molecularly imprinted polymers could play
an important role, thus complementing currently used im-
munological and chemical methods. Molecular imprinting
has resulted in a range of robust polymer-based receptors,
predominantly for small lipophilic target molecules.[6,7] The
technique entails copolymerization of mono- and bifunction-
al monomers in the presence of a template, which is there-
after removed to leave sites that can be reoccupied by the
template or closely related compounds. Regarding biological
receptors, they are distinguished by their robustness and
ease of synthesis, which has led to their assessment in a
range of molecular-recognition-based applications that
target small molecules.
Although molecularly imprinted polymers (MIPs) have
proven their value for the enrichment of low-molecular-
weight analytes, their use in the enrichment of peptide or
protein target molecules has so far met with limited suc-
cess.[8] Epitope imprinting, which uses shorter peptide se-
[a] Dr. M. Emgenbroich, C. Borrelli, S. Shinde, I. Lazraq, Dr. F. Vilela,
Dr. A. J. Hall, Priv. Doz. Dr. B. Sellergren
INFU, Technische Universität Dortmund
Otto Hahn Strasse 6
ꢀ
ꢀ
quences that correspond to the N or C termini of the
target, has, in this context, emerged as a promising strat-
egy.[9–11]
44221 Dortmund (Germany)
Following this approach to the synthesis of phospho-pep-
tide or protein receptors, we reasoned would require the for-
mation of a tight binding site for the phosphorylated side
chain. Binding of this residue should be strong enough to
overcome unfavorable secondary interactions and the site
should be accessible for protein targets. In a first attempt to
meet these requirements, we decided to make use of our re-
cently reported urea-based host monomers[12–14] in combina-
tion with an N,O-protected pTyr template (Scheme 2) and
to probe the recognition properties on a peptide level. We
anticipated that these neutral hydrogen-bond receptors
would provide sufficient binding energy while not suffering
from the charge-dependent sequence bias commonly ob-
served for positively charged chelating receptors.[15]
E-mail: borje@infu.uni-dortmund.de
[b] Dr. J. Oxelbark, Prof. Dr. E. De Lorenzi
Department of Pharmaceutical Chemistry
University of Pavia, Via Taramelli 12
27100 Pavia (Italy)
[c] Dr. J. Courtois, Dr. A. Simanova, Dr. J. Verhage, Prof. K. Irgum
Department of Chemistry
Umeå University
90187 Umeå (Sweden)
[d] Dr. K. Karim
Cranfield Health
Cranfield University at Silsoe
Silsoe MK454DT (UK)
[e] Dr. A. J. Hall
Sunderland Pharmacy School
University of Sunderland
Sunderland SR13SD (UK)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200801046.
Chem. Eur. J. 2008, 14, 9516 –9529
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9517

B. Sellergren et al.
EDMA was purified by the following
procedure prior to use: The received
material was washed consecutively
with 10% aqueous NaOH, water,
brine, and water. After drying over
MgSO4, pure dry EDMA was obtained
by distillation under reduced pressure.
All the other reagents were used as re-
ceived. The anhydrous solvents were
stored over the appropriate molecular
sieves. The other solvents were of re-
agent grade or higher. The functional
monomer N-3,5-bis(trifluoromethyl)-
phenyl-N’-4-vinylphenylurea (1) was
synthesized as reported previously.[14]
The substrates Fmoc-TyrOMe, Fmoc-
GluOMe, and Fmoc-LysOMe (Tyr=
tyrosine, Glu=glutamic acid, Lys=
lysine) were synthesized following
standard procedures from the corre-
sponding amino acid methyl esters and
N-fluoren-9-yl methoxycarbonyloxy-
succinimide.
Scheme 1. The main motifs formed by post-translational phosphorylation of proteins.
The materials used for the MALDI-
TOF mass-spectrometric analysis were
as follows: Aqueous solutions were
prepared using Milli-Q water filtered
through
pore).
(DHB),
a
0.2-mm membrane (Milli-
2,5-Dihydroxybenzoic acid
a-cyano-4-hydroxycinnamic
acid (CHCA), and phosphoric acid
were purchased from Sigma-Aldrich
(Milwaukee, USA).
Scheme 2. Prepolymerization complexes formed between monourea monomer 1 and receptor monomer 2 with
Fmoc-pTyrOMe and the procedure for the preparation of the corresponding imprinted polymers. The poro-
gens used were THF for P1 and DMF for P2.
Apparatus and methods: NMR spectra
were recorded using
a Bruker Ad-
vance DRX300 spectrometer unless
otherwise stated. Elemental analysis
was performed at the Department of
Organic Chemistry, Johannes Guten-
Experimental Section
berg Universität (Mainz) on a Heraeus CHN rapid analyzer (Hanau,
Germany). FT-IR spectroscopy was performed on a NEXUS FT-IR spec-
trometer (Thermo Electron Corporation, Dreieich, Germany). Nitrogen
sorption measurements were performed on a Quantachrome Autosorb
6B automatic adsorption instrument (Quantachrome Corporation, Boyn-
ton Beach, FL). Prior to the measurements, samples (100–150 mg) were
heated at 40–608C under high vacuum (10ꢀ5 Pa) for at least 12 h. The
specific surface areas S were evaluated by using the BET method, the
specific pore volumes Vp by following the Gurvitch method, and the aver-
age pore diameter Dp by using the Barrett-Joiner-Halenda (BJH) theory
applied to the desorption branch of the isotherm.
Materials:
Phosphoroxytrichloride
(POCl3),
N-methylmorpholine
(NMM), 1-naphthyl dihydrogenphosphate (NP), tetraethylammonium
tetrafluoroborate (TEATFB), trimethylolpropane trimethacrylate
(TRIM), ethylene glycol dimethacrylate (EDMA), triethylamine (TEA),
2,2-diphenyl-1-picrylhydrazyl (DPPH), 1,3-bis(isocyanatemethyl)benzene,
and tetrabutylammonium hydroxide (TBAOH) were Aldrich products
obtained from Sigma-Aldrich (Milwaukee, USA); toluene from Fischer
(Zürich, Switzerland); 1,2,2,6,6-pentamethylpiperidine, benzoin methyl
ether (BME), and 3-[(methacryloyl)oxypropyl]trimethoxysilane (g-
MAPS) from Fluka (Buchs, Switzerland); N,N-dimethylformamide
(DMF) from Riedel-deHan (Seelze, Germany); 2,2’-azobisisobutyroni-
trile (AIBN) from SERVA (Heidelberg, Germany); acetonitrile (MeCN)
and methanol from J.T. Baker (Phillipsburg, NJ, USA); N,N’-azo-bis(2,4-
dimethyl)valeronitrile (ABDV) from Wako Chemicals GmbH (Neuss,
Germany); 2,2,4-trimethylpentane from Merck (Darmstadt, Germany);
4-aminostyrene from Lancaster (UK); THF and dichloromethane from
Acros (Geel, Belgium); and deuterated dimethyl sulfoxide ([D6]DMSO)
from Deuterio GmbH (Kastellaun, Germany).
Measurement of swelling: NMR tubes were filled during intermittent vi-
brations up to 1 cm with dry polymer particles and weighed. Solvent
(1 mL) was added and the particles allowed to soak in the solvent for
24 h. The particles were then allowed to settle and the bed height of the
swollen particles was measured. The swelling factor was calculated as the
ratio of the bed height of the swollen particles to the bed height of the
dry particles.
1,1’-[1,3-Phenylenebis(methylene)]bis[3-(4-vinylphenyl)urea] (2): 1,3-Bi-
s(isocyanatemethyl)benzene (0.78 mL, 5 mmol) was added to a stirred so-
lution of 4-aminostyrene (1.19 g, 10 mmol) in dry THF (40 mL) at room
temperature and under a flow of nitrogen. The reaction proceeded over-
night after which a white precipitate formed. The precipitate was filtered
and dried under vacuum to afford the desired product as a white amor-
The analytes Fmoc-pSerOH and Fmoc-pTyrOH (Fmoc=(9-fluorenylme-
thyl)carbamate, pSer=phosphoserine) were purchased from Bachem
GmbH (Weil am Rhein, Germany); the peptide angiotensin and phos-
phoangiotensin from Calbiochem–Merck (Darmstadt, Germany); the
zeta-chain-associated protein kinase 70 kDa (ZAP-70) peptides and ref-
erence peptides Ser-436, pSer-436, Ser-357, pSer-357, and pThr-295 (Ser=
serine, pThr=phosphothreonine) were generous gifts from Prof. Rainer
Bischoff (University of Groningen, Netherlands; ZAP-70) and Priv. Doz.
Dr. Rainer Lehmann (University Hospital Tübingen, Germany), respec-
tively.
1
phous powder (yield: 1.49 g, 70%). H NMR (300 MHz, [D6]DMSO): d=
4.25 (d, 4H), 5.05 (d, 2H), 5.61 (d, 2H), 6.59 (dd, 4H), 7.1–7.4 (m, 12H),
8.58 ppm (s, 2H); elemental analysis calcd (%) for 2: C 73.2, H 6.14, N
13.14; found: C 72.3, H 6.0, N 13.0.
9518
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 9516 –9529

Phosphopeptide Recognition by an Imprinted Polymer Receptor
FULL PAPER
1-(4-Vinylphenyl)-3-[3,5-bis(trifluromethyl)phenyl]thiourea (3): 3,5-Bis-
(trifluoromethyl)phenylisothiocyanate (3.5 mmol) was added to a stirred
solution of 4-aminostyrene (3.5 mmol) in dry THF (20 mL) under nitro-
gen. The solution was heated to reflux overnight and then the solvent
was evaporated under reduced pressure. The resulting solid residue was
purified by column chromatography (silica gel, CH2Cl2) to give the de-
set processors, 2-GB memory, and a 20-GB fixed drive. This system was
used to execute the software packages SYBYL 7.0 (Tripos Inc., St. Louis,
Missouri, USA). The molecular model of monomer/template (dianion)
complexes were minimized by using Tripos Force Fields, Gasteiger—
Hückel charges, and the Powell minimization method and refined with
the molecular mechanics method by applying an energy minimization
with the MAXIMIN2 command. Energy minimization was performed on
the monomer/template complexes to a value of 0.001 kcalmolꢀ1, and
these complexes were then used for calculating the binding energy of
complexation of the template to the monomers by using a computational
docking program FlexiDock. The energy calculations were made based
on a site-point matching score and the resulting energies cannot be equa-
ted with interaction enthalpies.
1
sired product in 60% yield. H NMR (300 MHz, [D6]DMSO): d=5.22 (d,
1H), 5.28 (d, 1H), 6.76 (dd, 1H), 7.51 (s, 4H), 7.77 (m, 1H), 8.34 (s, 2H),
9.47 (s, 1H), 9.56 ppm (s, 1H); 13C NMR (300 MHz, [D6]DMSO): d=
114.21, 117.18, 122.22, 123.7, 124.13, 124.24, 124.94, 126.87, 129.85, 130.17,
130.50, 130.82, 134.37, 136.33, 138.44, 142.04, 179.81 ppm; FAB MS: m/z:
390.0 [M+], 391.0 [M+H]+; elemental analysis calcd for C17H12F6N2S: C
52.31, H 3.10, N 7.18, S 8.21; found: C 52.19, H 3.14, N 7.22, S 8.25.
N-(9-Fluorenylmethoxycarbonyl)-O’-phosphonotyrosine methyl ester
(Fmoc-pTyrOMe): The synthesis of Fmoc-pTyrOMe followed a previous-
ly reported procedure starting from Fmoc-TyrOMe.[16] POCl3 (1.34 mL,
14.4 mmol) and NMM (0.95 mL, 8.6 mmol) were added to Fmoc-
TyrOMe (3.00 g, 7.2 mmol) in dry dichloromethane (75 mL). The solution
was stirred for 3 h, whereby the conversion was monitored by TLC
(chloroform/acetone, 19:1). An additional portion of POCl3 (0.5 mL) and
NMM (0.4 mL) were added, and the solution stirred for another 4 h. The
organic phase was washed with 1n HCl (2”) and water (1”) and there-
after evaporated. The residue was taken up in acetone (20 mL), stirred
for 5 min, and evaporated. This procedure was repeated three times. The
product was purified twice by column chromatography (silica gel, chloro-
form/methanol, 9:1, 1% acetic acid) to yield 2.13 g (61.3%). The purity
was estimated by reversed-phase HPLC to be approximately 95% based
on peak areas (column: C-18 Luna, mobile phase: MeCN/water (50:50,
v/v, 1% TEA), UV: l=254 nm). 1H NMR (300 MHz, [D6]DMSO): d=
2.75–3.10 (m, 2H), 3.6 (s, 4H), 4.1–4.3 (m, 4H), 7.0–8.0 ppm (m, 12H);
31P NMR (300 MHz, [D6]DMSO): d=ꢀ1.523 ppm (s, PO4ꢀ); FAB MS:
m/z: 496.16 [MꢀH]ꢀ; elemental analysis calcd (%) for the dihydrated
complex: C 56.3, H 5.29, N 2.63; found: C 56.7, H 5.15, N 2.75.
Polymer preparation
Crushed monoliths: Imprinted polymers P1 and P2 were prepared in the
following manner: The bis(tetrabutylammonium) salt of Fmoc-pTyrOMe
(template; 0.5 mmol), urea monomer (P1: 1 mmol 1; P2: 0.5 mmol 2),
methacrylamide (4 mmol), and EDMA (20 mmol) were dissolved in THF
for P1 or DMF for P2 (5.6 mL). The initiator ABDV (1% w/w of total
monomer) was added to the solution. The solution was transferred to a
glass ampoule, cooled to 08C, and purged with a flow of dry nitrogen for
10 min. The tubes were then flame-sealed while still cooling, and the
polymerization initiated by placing the tubes in a thermostatted water
bath preset at 508C. The tubes were broken after 24 h and the polymers
lightly crushed. They were washed thereafter with MeOH/0.1n HCl (1:1,
3”), and extracted in a Soxhlet apparatus with methanol for 24 h. This
process was followed by further crushing and sieving, whereby the frac-
tion of 25–36 mm was used for packing the HPLC columns to evaluate
the binding properties. Nonimprinted polymers (PN1 or PN2) were pre-
pared in the same manner described above, but with the omission of the
template molecule from the prepolymerization solution.
Capillary monolithic supports: Polyimide-coated capillaries of 250 mm i.d.
and 360 mm o.d. were obtained from Polymicro Technologies (Phoenix,
Arizona, USA). These capillaries were pretreated according a procedure
based on our previous study.[17] The capillaries were washed with acetone
and deionized water, flushed for approximately 5 min with 1m aqueous
NaOH, sealed in the filled state, etched by heating in a circulating air
oven at 1208C for 2 h, and cooled to room temperature. A washing pro-
cedure of deionized water and acetone (15 min each in sequence) was
employed and the final drying took place in a vacuum oven at 608C for
at least 1 h. A silanization reaction to introduce methacrylic anchoring
groups onto the surface was carried out by filling the capillary with a
mixture of g-MAPS in DMF (1:1, v/v) containing 0.01% DPPH. The ca-
pillaries were sealed as above and heated at 1208C for 6 h. Finally, the
capillaries were washed with acetone and dried in a vacuum oven at
608C for at least 3 h.
Bis(tetrabutylammonium) 1-naphtyl phosphate (TBA2NP): A solution of
tetrabutylammonium hydroxide in methanol (1m, 4.46 mL, 2 equiv) was
added to 1-naphtyl phosphate (0.50 g, 2.2 mmol) in dry methanol
(10 mL). The reaction mixture was stirred at room temperature for 2 h.
The solvent was removed under vacuum and the oily residue dried over
P2O5. The bis(tetrabutylammonium) salt of the template Fmoc-pTyrOMe
was synthesized in a similar manner.
Tetrabutylammonium hydrogen-1-naphtyl phosphate (TBAHNP): A so-
lution of tetrabutylammonium hydroxide (1m, 2.2 mL, 1 equiv) in metha-
nol was added to 1-naphthyl phosphate (0.50 g, 2.2 mmol) in dry metha-
nol (10 mL) and the resulting solution was stirred at room temperature
for 2 h. The solvent was removed and the residue dried over P2O5 to give
a light-brown solid.
1H NMR spectroscopic titrations and estimation of the complex stoichio-
metries and association constants: The complex stoichiometry was first
assessed using the Job method of continuous variation. Stock solutions of
the host monomer and guest (2 mm in [D6]DMSO, respectively) were
combined in NMR tubes, thereby resulting in the following molar ratios:
0:10, 2:8, 3:7, 4:6, 5:5, 6:4, 7:3, 8:2, 10:0m. NMR spectra were thereafter
recorded and the proton signals, which could be monitored for all the
mixing ratios, were used for the evaluation of the complex stoichiometry.
All 1H NMR spectroscopic titrations were performed in dry deuterated
solvents. The association constants K for the interaction between the
hosts and guests were determined by titrating an increasing amount of
guest (e.g., TBAHNP) into a constant amount of functional monomer
(i.e., 1 or 2). The concentration of the functional monomer was 1 mm and
the amount of added guest was 0, 0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 4.0, 6.0, and
10.0 equivalents. The complexation induced shifts (CISs) of the host urea
or vinyl protons were followed and titration curves were constructed of
CIS versus guest concentration. The raw titration data were fitted to a
1:1 binding isotherm by nonlinear regression using Microcal Origin 5.0,
from which the association constants were calculated.
A polymeric capillary core monolith was prepared in situ in a 2-m pre-
treated capillary with a solution of TRIM (40%, w/w) in a mixture of
2,2,4-trimethylpentane/toluene (70:30, w/w) as the porogen[18] and AIBN
(1% with respect to the weight of TRIM) as the initiator for thermal
polymerization, which proceeded for 24 h at 608C. This capillary core
monolith was scored and snapped into 70-mm long pieces, which were in-
dividually subjected to intensive cleaning (>30 column volumes) with
methanol on
a Shimadzu model LC-10ADVP HPLC pump (Kyoto,
Japan) with the microstepping option running at 10 mLminꢀ1. The capilla-
ries were ranked by their back pressure as measured by the HPLC
system during the washing step, and only columns that showed a constant
back pressure of 3.7–3.9 MPa were selected for the grafting step. On
each of these capillary core monolith columns, a UV transparent window
was then made by removing 40 mm of the polyimide coating in the cen-
tral part of the column with a scalpel, thus leaving 15 mm on each side to
allow fitting of the column with fittings and sleeves, obtained from Up-
church Scientific (Oak Harbor, WA, USA).
Grafting of the imprinted polymer to capillary monolithic supports: The
prepolymerizaton mixture was prepared as follows: The template Fmoc-
pTyrOMe (ꢁ2 mg) was mixed with the strong, non-nucleophilic base
1,2,2,6,6-pentamethylpiperidine[19] (2 equiv) to form the ionized template.
The grafting solution was then prepared by mixing the template (T), urea
Modeling: The workstation used to simulate monomer/template interac-
tions was a Silicon Graphics Octane with the IRIX 6.5 operating system.
The workstation was configured with two 195-MHz reduced instruction
Chem. Eur. J. 2008, 14, 9516 –9529
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9519

B. Sellergren et al.
monomer 1, and EDMA in acetonitrile (T/1/EDMA=1:4:40) to make up
a 12% (w/w) monomer solution, and AIBN (1% w/w with respect to the
combined monomer weight) was finally added as the initiator. This solu-
tion was introduced in an empty fused-silica capillary (1 m”250 mm i.d.)
and methanol was propelled by an HPLC pump to force the solution into
each 70-mm column. After 10 column volumes had been pumped through
the column, the column was sealed at both ends by GC septa and photo-
grafting was carried out in a UV Spectrolinker XL-1500 (Spectronics
Corporation, Westbury, NY, USA) at l=365 nm to establish the imprint-
ed polymer layer. The temperature inside the photografting chamber was
kept at (ꢀ10ꢂ3)8C, and the light intensity in the polymerization zone
was 3.3 mWcmꢀ2, as determined by an International Light Model IL1400
radiometer with a model XRL140B probe (Newburyport, MA, USA).
After the polymerization was completed, a 1-mm piece was trimmed
from each capillary end, and thereafter the capillaries were individually
flushed by at least 30 column volumes of methanol at 10 mLminꢀ1. The
back pressure was measured again during the washing step, and columns
that did not show an increase in the back pressure of (0.4ꢂ0.1)MPa
were discarded. The nonimprinted reference polymers were prepared in
the same way, but the template complex was omitted from the grafting
mixture.
where C is the analyte concentration in the mobile phase, Fv is the solu-
tion flow rate, teq is the breakthrough time as determined with the area
method, t0 is the measured void time of the column, tea and tep are the
extra column times from the autosampler and pump, respectively, deter-
mined by replacing the column with a zero dead volume connector (tea
was determined by injecting from the autosampler and tep by running a
step gradient with subsequent determination of the breakthrough times),
and Vc is the geometrical volume of the column tube.
Igor Pro v.3.14 (WaveMetrics inc., Lake Oswego, OR, USA) was used
for nonlinear fitting of theoretical isotherms to experimental data, and
best fits were evaluated with the Fisher test.[21] The adsorption isotherm
models evaluated were Langmuir [Eq.(2)], bi-Langmuir [Eq. (3)], and
Freundlich [Eq. (4)] where q* is the concentration in the stationary phase
at equilibrium with concentration C and C is the concentration in the
mobile phase.
*
q
q
q
¼ qsbC=ð1 þ bCÞ
ð2Þ
ð3Þ
ð4Þ
*
*
¼ qs1b1C=ð1 þ b1CÞ þ qs2b2C=ð1 þ b2CÞ
m
¼ aC
HPLC evaluation: The 25—36-mm particle-size fraction was sedimented
repeatedly (methanol/water, 80:20) to remove fine particles and then
slurry-packed into HPLC columns (30”4.6 mm i.d. or 50”4.6 mm i.d.)
using the same solvent mixture as the pushing solvent. Subsequent analy-
ses of the polymers were performed using an Agilent HP1050 or HP1100
system equipped with a UV diode-array detector (DAD) and a worksta-
tion. Analyte detection was performed at l=260 and 220 nm, depending
on the analyte, and a flow rate of 0.5 mLminꢀ1. The retention factor k
was calculated as k=(tꢀt0/t0), where t is the retention time of the analyte
and t0 is the retention time of the void marker (acetone or sodium ni-
trate).
The Langmuir models [Eqs. (2) and (3)] assume that one [Eq. (2)] or two
[Eq. (3)] distinguishable classes of sites are present on the surface, each
with saturation capacity qs and binding constant b. The Freundlich iso-
therm [Eq. (4)], on the other hand, assumes sites to have a Gaussian dis-
tribution of binding strengths. Herein, the width of the Gaussian distribu-
tion describes the degree of heterogeneity through the index m.
Solid-phase extraction: Solid-phase extraction (SPE) experiments were
performed off-line by using HPLC columns (30”4.6 mm i.d.) packed
with P1 and PN1 and manual fraction collection at the detector outlet.
The hardware consisted of an Agilent HP 1050 system equipped with a
binary pump, a diode-array UV detector, and workstation. Analyte de-
tection was performed at l=260 and 220 nm, depending on the analyte,
and a flow rate of 0.5 mLminꢀ1. The SPE experiments comprised a 2-h
conditioning step using the loading solvent A, a loading step also using
the loading solvent A, and an elution step using a stronger eluent B. The
Micro-LC evaluation: A micro-HPLC system (Shimadzu, Kyoto, Japan)
consisting of an LC-10 ADVP pump and a SPD-10 AVP UV-detector
fitted with a 35-nL cell from LC Packings (Amsterdam, The Netherlands)
was used in the chromatographic evaluation with pure acetonitrile as the
eluent. All the analytes were dissolved in acetonitrile and injected at
room temperature using a microinjector (Upchurch, Oak Harbor, WA,
USA) with a 35-nL external capillary loop and an electric actuator. The
concentration of the injected analytes was set at 20 mgmLꢀ1 (chosen for
the best peak shape, retention properties, and detection) and special care
was taken to keep this value accurate from one sample to another. Data
acquisition was performed using Clarity software (DataApex, Prague,
Czech Republic). All the spectra were recorded for 40 min at a flow rate
of 4 mLminꢀ1. An equilibration period of 20 min was allowed between
runs to decrease the amount of analyte retained on the column. As a
result of peak asymmetry, the retention parameters were calculated
based on the center of gravity of the peaks. Peak-shape considerations
were used for selecting the concentration of the injected solutions.
loading consisted of injecting single peptides or
a peptide mixture
(10 mL) and passing the load solvent through the column for a given
time. Either one or two fractions were collected in the load and elution
steps. After each run, the columns were regenerated by continuous wash-
ing with MeOH (single peptide runs) or MeOH with 0.1% trifluoroacetic
acid (TFA; peptide mixture runs) for at least 2 h. The single-peptide
standards were prepared by diluting a peptide stock solution (100 mL,
1 mgmLꢀ1 in Millipore water) to 1 mL with mobile phase A. The model
peptide mixtures comprised either all of the nine peptides shown in
Table 6 dissolved in water or a limited set of them. The non-pTyr contain-
ing peptides were present at a concentration of 11 mgmLꢀ1, whereas the
pTyr peptides (pAng and pZAP-70) were present at a concentration of
11 or 0.11 mgmLꢀ1
.
Frontal analysis: Frontal analysis was performed using columns (50”
4.6 mm) packed with P1 and PN1 to determine single-component adsorp-
tion isotherms. The concentration steps were assessed by using the stair-
case method with stock solutions of Fmoc-pTyrOMe (0.001 gLꢀ1 and
0.01 gLꢀ1). Ten steps in each series (10% steps) gave 20 experimental
points over a 100-fold concentration range. The step times in each series
were chosen to allow complete equilibration of the mobile phase with
the stationary phase. Before starting the next, higher concentration
series, washing times were typically set to more than three times the
equilibration time because as this procedure was shown experimentally
to afford correct and reproducible breakthrough times. Staircase chroma-
tograms were recorded at l=210 and 265 nm, depending on the analyte
concentration, to allow a sufficient signal-to-noise ratio and ensure a
linear detector response. All frontal chromatograms were evaluated
using the area method[20] as implemented with a Microsoft Excel work-
sheet. The adsorbed amount q* is given in Equation (1), which is adapted
from the step series to staircase:
MALDI-TOF mass spectrometry: Mass-spectrometric analysis of the
fractions collected during the SPE experiments was performed using a
MALDI reflector time-of-flight mass spectrometer (Autoflex II mass
spectrometer, Brucker-Daltonics GmbH, Bremen, Germany) equipped
with a Scout-384 source unless otherwise stated. Ions were generated by
irradiation of analyte/matrix deposits by nitrogen laser at l=337 nm and
analyzed with an accelerating voltage of 25 kV in the reflector mode and
in the positive-ion mode. Data collection, in terms of the scanning condi-
tions and the number of scans, was performed identically for all samples
unless otherwise noted. The spectra were collected by accumulating
1000 laser shots, the scanning was performed by using the RP Pepmixt
Par method, and the mass spectra were analyzed with flex Control soft-
ware (Brucker Daltonic FLEXControl).
The sample preparation prior to the MALDI experiments shown in
Table 7 was as follows: Aliquots (0.5 mL) of the collected load and elute
fractions were evaporated to dryness in a vacuum oven at room tempera-
ture. The residue was dissolved in phosphoric acid in MeOH (1.3%,
100 mL). The matrix solution was prepared by dissolving DHB (40 mg) in
MeCN/water (1:1, v/v; 1 mL). The matrix solution (1 mL) and the concen-
*
*
q
¼ q n þ ðCnþ1ꢀCnÞFv½teqꢀðt0ꢀteaÞꢀtepꢃ=½VcꢀFvðt0ꢀteaÞꢃ
ð1Þ
nþ1
9520
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 9516 –9529

Phosphopeptide Recognition by an Imprinted Polymer Receptor
FULL PAPER
trated SPE aliquots (1 mL) were then deposited together on the target
plate by using the RP Pepmixt Par method.
Slight modifications to this procedure were introduced for the samples in
given in Figures S22 and S23 in the Supporting Information. In Figure
S22, the matrix solution was prepared by dissolving DHB (20 mg) in
water (1 mL), whereas in Figure S23, the matrix solution was prepared
by dissolving DHB (25 mg) in MeCN/water (1:1, v/v; 1 mL) containing
1% phosphoric acid. In Figure S23, 1-mL aliquots were collected and the
residue was directly dissolved in the DHB matrix solution (20 mL) and a
1-mL aliquot of this matrix solution was then deposited onto the target
plate.
In Figure 10, the collected fractions were directly used for MALDI analy-
sis. Saturated matrix solutions of DHB and CHCA in MeCN/water (1:1,
v/v) were prepared and used for phosphorylated and non-phosphorylated
peptides, respectively. The matrix solution (1 mL) and aliquots (1 mL) of
the collected fractions were subsequently deposited together onto the
target plate. The MALDI analysis was in this case performed on a Voy-
ager-DE (Applied Biosystem) instrument and by using the 1—2-k reflec-
tor method.
Table 1. Association constants, stoichiometries and CIC for complexes
formed between urea host monomers and naphthyl phosphate (NP)
guests in [D6]DMSO.
Results and Discussion
Host
Guest
Proton
K
Complex CIS[a]
(H/G) [ppm]
[a]
”10ꢀ3 [mꢀ1
]
A
ACHTREUNG
Monomer/template complex formation: 1,3-Disubstituted
ureas have long been exploited as neutral hosts for complex-
ing oxyanion guests.[22,23] They establish cyclic hydrogen
bonds that act as a twofold donor for the acceptor (carbox-
ylate, phosphate, or sulfonate ions). The affinity for the
guest increases with the acidity of the urea protons (donor
ability) and the basicity of the oxyanion (acceptor ability),
but is also related to the ability of the host to self-associate
(poor acceptor ability) and, hence, its solubility.[24] Common-
ly, thioureas are used in the host design because they are
more acidic and more soluble than ureas and, thus, form
stronger hydrogen bonds with a given acceptor.[23] We previ-
ously found that the polymerizable 1,3-diaryl urea 1 dis-
played a binding constant of approximately K=8800mꢀ1 to-
wards tetrabutylammonium benzoate in DMSO,[14] which is
in agreement with other reported diarylurea receptors.[23]
The monomers could be used to imprint carboxylates, thus
resulting in polymers that recognize the guest with high af-
finity and selectivity in water-rich media.[13]
monomer
1
1
2
TBAHNP NH (7,10)
2.7ꢂ0.3
1:1
2:1
1:1
3.31
TBA2NP
TBA2NP
CH (13)
>10[b]
ꢀ0.14
ꢀ0.20
3.43
1.53
ꢀ0.23
CH (30,32) >10[b]
NH (13,16) n.d.
TBAHNP NH (7,10)
TBA2NP CH (13)
3
3
1.1ꢂ0.1
1:1
2:1
>10[b]
[a] Average binding constants K and complexation induced shifts (CIS)
based on the shift values of the resonance signals indicated. [b] A low es-
timate that represents the inverse of the lowest concentration of free
ligand, resuling in host saturation; the two binding sites of the divalent
NP was assumed to interact identically and independently with the urea
host monomer 1.
all the remaining protons (see Figures S3–S7 in the Support-
ing Information). The signals, which could be monitored
throughout the titration, were used to calculate the free and
bound concentrations and if possible the association con-
stants from the resulting binding curve obtained through
nonlinear regression.
As a first step in our evaluation of hosts to complex the
phosphate species, we decided to compare this host mono-
mer with the supposedly more potent thiourea analogue 3.
Monomer 3 was synthesized analogously to the urea ana-
logue in one step by adding aminostyrene to 3,5-bis(trifluor-
ophenyl)thiocyanate. Tetrabutylammonium hydrogen-1-
naphthyl phosphate (TBAHNP) and bis(tetrabutylammoni-
um) 1-naphtyl phosphate (TBA2NP) were chosen as mono-
and dianionic guests, respectively, thus mimicking the phe-
nylphosphate substituent of the template Fmoc-pTyrOMe.
The receptor monomer solutions (1 mm in [D6]DMSO) were
titrated with a standard solution of the anion guest in up to
tenfold molar excess. Table 1 shows CISs of key protons, the
corresponding binding constants (K), and the complex stoi-
chiometries determined by the Job method of continuous
variation[25] or obtained from other sources.
Considering first the relative complex stabilities involving
monoureas 1 and 3, the monotetrabutylammonium salt (i.e.,
TBAHNP) was used as a monoanionic guest to uniquely
promote the formation of 1:1 complexes. After confirmation
of the 1:1 stoichiometry from the Job plots, the 1:1 binding
model was used to determine the respective association con-
stants. Surprisingly, the oxourea monomer formed the most
stable complexes (K=2675mꢀ1), which were more than two-
fold stronger than the corresponding thiourea complexes
(K=1089mꢀ1). This behavior obviously contrasts with most
previous findings on thiourea/oxyanion complexes but
agrees with a recent report by Roussel et al.[26] These au-
thors attributed the effect to a preference of the diaryl-
thiourea species for an E,Z conformation in contrast to the
preference of the oxyurea for a Z,Z conformation.
We then went on to assess the urea receptor monomer 2,
an analogue of receptors that were used previously as phos-
phate receptors.[27] As 2 was designed to complex the dia-
The titration was accompanied by pronounced downfield
shifts of the urea protons together with significant shifts for
Chem. Eur. J. 2008, 14, 9516 –9529
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9521

B. Sellergren et al.
nion in a 1:1 stoichiometry, we tested its ability to complex
TBA2NP. Because of difficulties in removing the residual
methanol from TBA2NP, the urea protons could not be
clearly distinguished throughout the titration, although the
maximum shifts (i.e., CIS) were in agreement with reported
values for similar hosts.[23] The vinyl protons, however, lev-
eled off at a 1:1 host/guest stoichiometry (Figure 1) and the
very low free concentrations allowed only a minimum value
of K to be estimated as 10000mꢀ1.
ꢀ
Figure 1. CIS of (H2C=) (solid symbols), and (=CH ) (open symbols) of
1 (diamonds) and 2 (squares) as a function of the total concentration C
of the guest TBA2NP in [D6]DMSO. The dashed lines indicate the guest
concentrations that correspond to 2:1 (left line) and 1:1 (right line) host/
guest stoichiometries.
Figure 2. Ball-and-stick models of the minimum-energy complexes
formed between urea monomers 1 (A) and 2 (B) with Fmoc-pTyrOMe.
A) The interaction energy for the complex was ꢀ544 kcalmolꢀ1; the hy-
drogen-bond lengths for the phosphate–urea interactions were 1.915 and
2.250 Š (top monomer) and 1.916 and 1.925 Š (bottom monomer); the
shortest carbon–carbon distance between the pTyr phenyl ring and the
phenyl ring of the bottom monomer was 3.440 Š. B) The interaction
energy for the complex was ꢀ252 kcalmolꢀ1; only one single intermolecu-
lar hydrogen bond was indicated with a length of 2.651 Š.
Probing the interactions between the monoureas and the
phosphate dianion in contrast showed the vinyl protons to
level off at a 2:1 host/guest stoichiometry. Although in this
case as well, only a minimum value of K could be estimated,
the steepness of the CIS curves indicate that the complexes
are stronger in this case with the complex involving oxourea
analogue 1 appearing somewhat more stable than that of
the thiourea derivative 3, the latter displaying a somewhat
shallower curve.[28]
We were still puzzled about the strong complexation ten-
dency of the monoureas, especially with respect to the bis-
urea receptor monomer. In spite of the presumed ability of
the bis-urea to donate four converging hydrogen bonds to
the phosphate dianion guest, it displayed weaker complexes.
Further insight into the origin of these differences was ob-
tained by molecular modeling.
Molecular modeling of the host–guest complexes was per-
formed using the genetic algorithm-based FlexiDock pro-
gram for docking ligands into receptor active sites. This pro-
gram works on a receptor–ligand pair in which the receptor
backbone atoms are fixed in space, but the ligand is mobile
(rotation/translation can be applied). The modeling gave
minimum-energy complex geometries and their relative in-
teraction energies with the latter lacking physical meaning
being used for ranking purposes only. The lowest-energy
complex for Fmoc-pTyrOMe and 1 (Figure 2A), with an in-
teraction energy of ꢀ544 kcalmolꢀ1, features the bis(trifluor-
omethyl)phenyl substituents of both urea ligands pointing in
the same direction, thus allowing four strong hydrogen
bonds to develop, and the styryl substituent of one of the li-
gands placed at p-stacking distance from the tyrosine phenyl
group. This arrangement should result in a tight cavity com-
plementary to the phenylphosphate group of pTyr. Con-
versely, the corresponding complex with 2 (Figure 2B) is
poorly defined, thus resulting in an interaction energy of
only ꢀ252 kcalmolꢀ1. The orientation prevents the engage-
ment of all the urea protons in hydrogen bonding with the
phosphate group. Although the computational approach has
arguable limitations, the relative interaction energies of the
two complexes are interestingly in agreement with both the
titration data (see above) and the affinities exhibited by the
corresponding imprinted polymers (see below).
Polymer preparation: Having established the potency of the
urea monomers 1 and 2 to complex phosphates, we turned
our attention to the polymer preparation. Polymers P1 and
P2 were prepared using monomers 1 and 2 in 2:1 and 1:1
stoichiometric ratios, respectively, to the template Fmoc-
pTyr-OMe (Scheme 2). Nonimprinted polymers PN1 and
PN2 were prepared identically to the imprinted polymers
but with the template omitted. Methacrylamide was added
as a supplementary monomer to provide additional hydro-
9522
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 9516 –9529

Phosphopeptide Recognition by an Imprinted Polymer Receptor
FULL PAPER
gen-bond stabilization[13] and EDMA as a crosslinking mo-
nomer with THF or DMF as the solvents for P1 and P2, re-
spectively. The choice of solvent was guided by the solubility
of the urea/template complex in the monomer mixture. Con-
ventional azo-initiated thermal polymerization at 508C sub-
sequently afforded the imprinted and nonimprinted poly-
mers. The polymers were crushed and sieved to a 25—36-
mm particle-size fraction and subjected to template removal
by washing with acidic methanol, followed by extraction
with methanol in a Soxhlet apparatus. On the basis of the el-
emental analysis of the remaining phosphorous atoms in the
polymers, more than 95% of the template was removed by
this treatment.
order observed herein. However, if the template acts as a
solvating agent for the growing chains, the phase separation
would be delayed, which in turn would lead to a polymer
with a more gel-like morphology.[32] This latter explanation
would be in better agreement with the observed differences
between the polymers (Table 2).
Chromatographic characterization: Imprinting effects were
assessed by chromatography by using the crushed polymer
monoliths as the stationary phase. Our first goal was to in-
vestigate how well the polymers discriminated the template
from other amino acid derivatives that contained side chains
with an expected affinity for the urea motif.
To investigate whether the imprinted and control poly-
mers were otherwise comparable in terms of morphology
and composition, the polymers were characterized by using
elemental analysis, X-ray photoelectron spectroscopy (XPS),
IR spectroscopy, cross-polarization/magic-angle-spinning
(CP-MAS) 13C NMR spectroscopy, and nitrogen sorption
analysis (see the Supporting Information). Only the nitrogen
sorption technique, which shows the porous properties of
the materials, and associated swelling tests gave evidence
toward the differences between the imprinted and nonim-
printed polymers (Table 2). Thus, all the polymers except P1
exhibited mesoporous morphology with surface areas larger
than 200 m2 gꢀ1 and average pore diameters of roughly 4 nm.
This finding contrasted with P1, which showed a lower sur-
face area and pore volume, but on the other hand exhibited
a higher swelling factor than the other materials.
Thus, Fmoc amino acid methyl esters were injected onto
the columns in an acetonitrile-rich mobile phase buffered
with triethylamine (Figure 3). Basic conditions were used to
Table 2. Physical properties of Fmoc-pTyrOMe imprinted and nonim-
printed polymers.[a]
Figure 3. Chromatographic characterization of the imprinted polymers
using amino acid derivatives as test solutes. Retention factors for the
amino acid analytes on columns packed with the imprinted and nonim-
printed control polymers. The mobile phase for P1 and PN1 was: MeCN/
water 90:10 (v/v) 1% triethylamine), whereas for P2 and PN2 it was:
MeCN/water 93:7 (v/v) 1% triethylamine.
Polymer
S
Vp
[mLgꢀ1
Dp
[nm]
Swelling
[mLmLꢀ1
]
A
U
]
N
P1
PN1
P2
0.076
0.24
0.22
0.66
5.3
4.3
3.8
3.8
1.9
1.2
1.8
1.9
247
208
342
PN2
[a] The Brunauer–Emmett–Teller (BET) specific surface area S, specific
pore volume Vp, and average pore diameter Dp were calculated from the
nitrogen adsorption isotherms, whereas the swelling in mLmLꢀ1 was de-
termined by soaking 1 mL of a packed bed of polymer particles in
MeCN/water 90:10 (v/v) +1% TEA) as described in the Experimental
Section.
promote deprotonation of the template and other analytes,
thus allowing more stable quadruple hydrogen bonds to de-
velop.[14] Figure 3 shows that P1 and P2 exhibited a strong
affinity for the template Fmoc-pTyrOMe, but that the other
amino acids were only weakly retained. Polymers PN1 and
PN2, on the other hand, exhibited no affinity for any of the
analytes under these conditions. Polymer P1 exhibited stron-
ger template retentitivity than P2, and this difference was
magnified when assessing the polymers in a competitive
phosphate-buffered mobile phase. Here, only P1 retained
the template, whereas breakthrough was seen on P2 (see
Figure S16 in the Supporting Information). The fact that P1
still retains the template to a significant extent reflects the
tight complex formed between 1 and Fmoc-pTyrOMe (see
above). This result also corroborates the relative stabilities
of the complexes obtained from the NMR spectroscopic ti-
trations (Table 1) and the modeling results (Figure 2), with
reservation for the somewhat more competitive solvent used
when preparing P2.
Morphological differences induced by the template mole-
cules are frequently reported for studies on imprinting but
these differences may go in different directions, that is, re-
sulting in polymers with either higher or lower surface
areas.[29–31] This behavior is particularly true in noncovalent
imprinting, in which the template may exert a complex
effect that influences the solvation of the growing chains
and reactivity ratios. Moreover, crosslinking levels may be
influenced by divalent templates capable of complexing two
monomersin different opolymer chains. Given the 2:1 stoi-
chiometry of the urea/template complex, the latter effect is
a plausible cause of morphological differences but not in the
Chem. Eur. J. 2008, 14, 9516 –9529
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9523

B. Sellergren et al.
Micro-liquid chromatography: Micro-liquid chromatography
(micro-LC) using capillary-column formats is an attractive
technique for separation problems that require low sample
loads and mass-spectrometric interfacing.[5] These needs
apply in the proteomics area in which sample volumes are
strongly limited but the demand for detection sensitivity and
selectivity is high. To shrink the imprinted polymers to
match the micro-LC format, we used our recently reported
technique based on grafting imprinted polymer layers onto
Mobile-phase dependence: To investigate whether the reten-
tion behavior prevailed in water-rich mobile phases, the re-
tention of the phosphorylated analytes was measured in
mobile phases with different water contents (Figure 5). Plot-
flow-through poly(TRIM) monoliths with unreacted surface
A
double bonds as anchor points.[33] Thus, imprinted and non-
imprinted capillaries were prepared and subsequently as-
sessed in the micro-LC mode. Figure 4 shows the elution
profiles obtained after separate injections of the amino acid
analytes with an acetonitrile-rich mobile phase modified
with a cationic ion-pair reagent (IPR).
Figure 5. Retention factors in micro-LC for the amino acid analytes
Fmoc-pTyrOMe (solid circles), Fmoc-pTyrOH (open squares), and
Fmoc-pSerOH (solid triangles) on an imprinted polymer capillary as a
function of the water content in acetonitrile/water mobile-phase mixtures.
A retention factor of >50 implies that no peaks were observed within
the 30-min runtime. Conditions: flow-rate=5 mLminꢀ1, column dimen-
sion=4 cm”320 mm i.d.; injections=100 mgLꢀ1 in a 180-nL loop. The
column was regenerated with MeOH between each run.
ting the retention factor versus the water content resulted in
the typical bell-shaped curves observed for several imprint-
ed polymer systems. This result is explained by a change in
the retention mode from an electrostatically driven mode in
water-poor systems to a desolvation (hydrophobic) retention
mode at higher water contents. It can be seen that both ef-
fects contain a strong selective contribution. Thus, both
pTyr-containing analytes are strongly retained at both low
and high water contents, whereas pSer is only retained in
the water-poor system. However, the addition of an IPR to
the water-poor mobile phase reveals a selectivity for pTyr
over pSer in this system too (Table 3).
Considering that the IPR had essentially no effect on the
retention on the nonimprinted control polymer, this interest-
ing behavior suggests that the IPR facilitates the access to
or cobinds with the template in the imprinted sites. The re-
sults shown in Figure 6 support this hypothesis. Thus, where-
as the addition of the template identical tetrabutylammoni-
um IPR causes a strong increase in retention of Fmoc-pTyr-
OMe, a more modest effect was observed upon the addition
of the less complementary tetraethylammonium IPR
(Figure 6, inset). Thus, the polymers exhibit a memory effect
not only for the hydrogen-bonded ligand but also for the
bulky countercation (Scheme 3).
Figure 4. Micro-LC of amino acid analytes using poly(TRIM) capillaries
A
coated with Fmoc-pTyrOMe imprinted and nonimprinted (inset) poly-
mers. Mobile-phase A: MeCN/water 90:10 (v/v) 120 ppm tetraethylam-
monium tetrafluoroborate for 10 min; mobile-phase B: MeOH for
15 min; mobile-phase A for 10 min. Arrows indicate elution of pTyr ana-
lytes.
The high affinity and selectivity displayed by the imprint-
ed polymer was also confirmed by using this format, and
strong eluents were required to elute the pTyr analytes. This
procedure resulted in clearly eluting peaks only for runs in-
volving pTyr analytes on P1, with the notable absence of an
elution peak for pSer. Thus, whereas the polymer fully re-
tains the pTyr derivative, most of the pSer analogue breaks
through within 10 minutes. In contrast to the imprinted ca-
pillary, the control capillary was incapable of retaining any
of the analytes. Another encouraging point was the ob-
served robustness of the phases and their preparation, which
is reflected in the almost identical retention factors mea-
sured for multiple runs on several independently prepared
imprinted capillaries (see Figure S17 in the Supporting In-
formation).
Other mobile-phase compositions also resulted in strong
imprinting effects (see the Supporting Information). For in-
stance, a water-rich mobile phase (80% water) gave stron-
9524
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 9516 –9529

Phosphopeptide Recognition by an Imprinted Polymer Receptor
FULL PAPER
Table 3. Retention factors for phosphorylated amino acid analytes on a
Fmoc-pTyrOMe imprinted capillary as a function of the ion pair reagent
concentration in MeCN/water: 80/20 (v/v).[a]
the presence of 0.1% TFA as a modifier may be less expect-
ed. However, the pH value of 0.1% TFA (aqueous) is 1.9,
which is slightly higher than the pKa of a phosphate mono-
ester (pKa =1.1).[3] With a charge of almost ꢀ1, the phos-
phate analytes are still expected to interact strongly with the
imprinted receptor. This expectation was supported by Flexi-
Dock modeling of the phosphate monoanion bound to the
receptor optimized for binding the dianion. The binding en-
ergies (ca. 250 kcalmolꢀ1) were similar to those obtained for
A
k
k
k
A
]
(Fmoc-pTyrOMe)
(Fmoc-pTyrOH)
(Fmoc-pSerOH)
0
5.3
18
0.11
0.19
0.50
0.99
3.05
0.07
0.09
–
0.11
0.11
0.25
50
100
215[c]
no elution[b]
no elution[b]
no elution[b]
[a] Conditions: flow-rate=5 mLminꢀ1, column=4 cm imprinted polymer
capillary column (320 mm i.d.), injection: 100 mgLꢀ1 of indicated solutes
in a 180-nL loop; conditioning was performed with MeOH between each
run. [b] No peak observed within 30 min. [c] Equal to 1 mm.
the dianion binding to the bis(urea) receptor.
A
Frontal analysis: The binding-energy distribution of the
polymers was obtained from single-component adsorption
isotherms determined by staircase frontal analysis.[34] Thus,
the isotherms of Fmoc-pTyrOMe on P1 and PN1 (Figure 7)
were obtained by using the three different mobile phases
that produced the zonal elution profiles shown in Figure 6
and Figure S18 in the Supporting Information.
Figure 6. Elution profiles of Fmoc-pTyrOMe on P1 as a function of the
IPR concentration. Tetrabutylammonium hydroxide (TBAOH) was
added to the mobile phase consisting of MeCN and sodium carbonate
buffer (10 mm, pH 9.8) in a ratio of 50:50 (v/v) to reach the final concen-
trations indicated. Inset: the same experiment but with tetraethylammo-
nium tetrafluoroborate (TEATFB) as the IPR. Conditions: column=50”
4.6 mm, DAD l=260, 210 nm, flow-rate=0.5 mLminꢀ1, injection=5 mL
of 0.2 mm stock solutions in acetonitrile.
Scheme 3. A possible binding site resulting from ternary complex im-
printing.
Figure 7. Adsorption isotherms obtained by frontal analysis chromatogra-
phy of Fmoc-pTyrOMe on P1 (filled symbols) and PN1 (open symbols)
using mobile phase A: MeCN/(sodium carbonate buffer (0.01m),
TBAOH (0.01m), pH 9.8) 50:50 (v/v) (squares); B: MeCN/(sodium car-
bonate buffer (0.01m), pH 9.8) 20:80 (v/v) (circles); C: MeCN/water
50:50 (v/v) 0.1% TFA (triangles). A and B show all isotherms within two
different concentration intervals. C=concentration in the mobile phase,
q=concentration in the stationary phase.
ger imprinting effects when buffered at high pH, thus again
providing stronger quadruple hydrogen-bond formation.
The retention effects observed under acidic conditions in
Chem. Eur. J. 2008, 14, 9516 –9529
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9525

B. Sellergren et al.
Table 5. Fisher values and Freundlich heterogeneity indices of the frontal
analysis binding isotherms.[a]
The isotherms of P1 exhibited clear curvature in the low-
concentration range, whereas an almost linear increase in
the adsorbed amount q with solute concentration C was ob-
served at higher loads. The isotherms of PN1 appeared
linear throughout the concentration range investigated with
slopes that appeared highest when using the basic mobile
phase B, followed by the acidic mobile phase C and the
IPR-modified mobile phase A. This finding reflects a de-
crease in the nonspecific contribution to binding in the same
order. Interestingly, the slopes of the same portions of the
P1 isotherms showed a different order.
Here, mobile phase A gave the highest slope, followed by
B, and finally C, the latter difference becoming particularly
apparent in the low-concentration interval. By using the
IPR-modified mobile phase A, the end result is, thus, a
strong imprint-related capacity increase. This finding again
suggests that both template components used in the imprint-
ing step are required for efficient access to the imprinted
sites (Scheme 3).
The isotherms were subsequently fitted to mono-Lang-
muir, bi-Langmuir, and Freundlich isotherm models[35] and
resulted in the isotherm parameters given in Table 4 and
Table 5. All the isotherms obtained from PN1 were fitted
with the mono-Langmuir isotherm model, but the weak cur-
vature strongly decreases the accuracy of the derived param-
eters. This behavior is not the case for P1, which showed
good fits to both the Freundlich and bi-Langmuir models
depending on the mobile phase used. The parameters de-
rived from the bi-Langmuir fittings are given in Table 4
from which we discern one class of high-energy binding sites
with binding constants of 105–106 mꢀ1, which are 10–
100 times less abundant than lower affinity sites in the K=
104 mꢀ1 range. It can further be noted that the choice of the
mobile phase influences the relative abundance of these
sites more than their associated binding constants, which
agrees with the above observations.
Polymer
Mobile
phase[b]
Fisher value
Heterogeneity
index
Freundlich
Langmuir
P1
PN1
P1
PN1
P1
PN1
A
A
B
B
C
C
2100
16000
7000
5000
15000
4900
64000
14000
22000
23000
10800
8800
0.79
1.00
0.64
0.93
0.84
0.94
[a] The isotherms were fitted with the bi-Langmuir (P1) or mono-Lang-
muir (PN1) adsorption models. [b] A: MeCN/(sodium carbonate
(10 mm)+TBAOH (10 mm), pH 9.8) 50:50 (v/v); B: MeCN/(sodium car-
bonate (10 mm), pH 9.8) 20:80 (v/v); C: MeCN/water 50:50 (v/v)+0.1%
TFA.
contrasts with the isotherm obtained by using the IPR
mobile phase A, in which a much higher Fisher value (i.e.,
64000) was obtained from the bi-Langmuir fitting. This find-
ing may indicate that the IPR-modified phase decreases the
binding-site heterogeneity of the MIPs, at least in the low-
sample-load regime that is probed here.
In view of the high binding constant (K=7.6”105 mꢀ1) ob-
tained using a water-rich mobile phase (80% water), a com-
parison of the affinities with those observed for antibodies
elicited to react with pTyr seems justified.[36] The latter ex-
hibited binding affinities under optimal conditions that
amounted to K=106–107 mꢀ1, thus leading to the conclusion
that our synthetic receptors bind pTyr almost as strongly as
their biological counterparts, albeit under different condi-
tions.
Phosphopeptide recognition: Crucial to the utility of the re-
ported imprinting strategy would be the extent to which
these pTyr-selective sites would cross-react with peptides
containing this epitope. This behavior is far from evident
given the size of the template and the lack of pore-system
control in conjunction with the formation of the imprinted
sites. An answer to this question was provided by the use of
a small set of test peptides available in the mono- and non-
phosphorylated forms and including Tyr, Ser, and Thr as
phosphorylation sites (Table 6).
The Fisher values in Table 5 reflect which of the models
provides the best fit to a particular isotherm, a higher
number, thus indicating a better fit. For instance, the iso-
therm obtained by using the acidic mobile phase C shows a
stronger adherence to a Freundlich model. This behavior
We first investigated the retention of the stable neuropep-
tide angiotensin II (Ang), containing an internal tyrosine
residue present in both the non-phosphorylated (Ang) and
monophosphorylated (pAng) forms. To better understand
the retention mechanism of these peptides, they were inject-
ed onto P1 and PN1 using a series of TFA-modified binary
acetonitrile/water mobile-phase mixtures. The acidic modifi-
er TFA was anticipated to act as an anionic IPR, thus pro-
viding effective solubilization of the peptides[37] without dis-
rupting the phosphate binding-site interactions (see above).
The sometimes strong retention of the peptides (see
Figure 9 and Figures S19–S20 in the Supporting Informa-
tion) and their weak UV chromophores precluded accurate
measurement of the retention times; therefore, we decided
instead to record the portion of total peptide injected that
eluted with minor retention within the first 10 min after in-
Table 4. Parameters derived from fitting of the frontal analysis binding
isotherms with Langmuir models.[a]
Polymer
Mobile
phase[b]
qs1
[mm]
K1
[mꢀ1
qs2
[mm]
K2
[mꢀ1
G
]
N
]
P1
PN1
P1
PN1
P1
PN1
A
A
B
B
C
C
1.4
[c]
0.5
0.8
0.8
0.6
8.0”103
[c]
0.05
0.03
0.007
2.5”105
–
15”103
8.0”103
9.0”103
5.5”103
7.6”105
–
8.7”105
–
[a] The isotherms were fitted with the bi-Langmuir (P1) or mono-Lang-
muir (PN1) adsorption models. [b] A: MeCN/(sodium carbonate
(10 mm)+TBAOH (10 mm), pH 9.8) 50:50 (v/v); B: MeCN/(sodium car-
bonate (10 mm), pH 9.8) 20:80 (v/v); C: MeCN/water 50:50 (v/v)+0.1%
TFA. [c] No fitting possible as a result of very low curvature of the iso-
therm.
9526
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 9516 –9529

Phosphopeptide Recognition by an Imprinted Polymer Receptor
FULL PAPER
Table 6. Tyr and Ser-containing model peptides used to probe the phosphoselectivity of the polymers.[a]
and PN1 systems using a mobile
phase containing 80% water.
Whereas pAng appears almost
Peptide
Non-phosphorylated
m/z
Phosphorylated
m/z
ZAP-70
Ang
Ser-436
Ser-357
pThr-295
ALGADDSYYTAR
DRVYIHPF
CDFRSFRSVT
AHRHRGSARLHPPLNHS
–SQVGLpTRRSRTE
1303
1047
1305
1944
ALGADDSpYYTAR
DRVpYIHPF
CDFRpSFRSVT
1383
1127 quantitatively retained on P1, it
1385
2025
breaks through on PN1. Mean-
while, Ang breaks through
completely on all columns.
The results were supported
by MALDI-TOF mass-spectro-
AHRHRGpSARLHPPLNHS
1471
[a] The phosphorylated peptides are indicated by the letter p; for example, pAng=phosphorylated angioten-
sin.
jection (Figure 8A). To avoid carrying over effects from the
strong retention, the columns were regenerated after each
run by methanol.[38] A strong phosphopeptide preference
was evident, as seen in the large difference in the percent-
age of eluted peptide on comparing the two peptide forms.
Ang was poorly retained on both P1 and PN1, except in
water-rich mobile phases (>80% water), in which hydro-
phobic nonspecific binding becomes prevalent. pAng also
breaks through completely, but only on PN1 and with the ex-
ception for water-poor mobile phases, in which some reten-
tion is observed. Polymer P1 retains the phosphorylated
peptide selectively and mainly in water-poor and water-rich
mobile phases. An example is shown in Figure 9 of these P1
Figure 9. Elution profiles of Ang and pAng injected on P1 and PN1 with
MeCN/water 20:80 (v/v) 0.1% TFA as mobile phase A. Reconditioning
was performed using MeOH (+0.1% TFA) as mobile phase B. Method:
0–30 min 100% A; 30–40 min 100% B. Injection: 10 mL of peptide
(0.1 mgmLꢀ1) in mobile phase A; flow-rate=0.5 mLminꢀ1, l=260 nm.
metric analysis of fractions collected prior to and after the
switch from mobile phase A to the eluting mobile phase B
(Figure 10). Here, we chose to focus on the water-poor
mobile phase A (MeCN/water, 95:5), thus showing low non-
specific hydrophobic binding and high pAng selectivity. The
Ang peptides exhibited minimal fragmentation in the
MALDI experiment and appeared as peaks, thereby agree-
ing with the molecular mass of the parent peptide. This pro-
cedure allows rough estimates of the peptide contents in
each fraction from the relative peak intensities to be made.
As can be seen in Figure 10, the MALDI-TOF mass-spec-
trometric results support the UV detection results shown in
Figure 8A. Thus, Ang is recovered from mobile phase A on
both columns, whereas pAng is selectively retained on P1.
This fraction can be recovered by using a somewhat stronger
eluent (MeOH+0.1% TFA), but the lower intensity of the
corresponding elution peak would suggest the recovery to
be incomplete. However, because MALDI can, at the most,
give estimates of the relative peptide abundances, conclu-
sions concerning the mass balance would be premature.
Having established the selectivity of P1 for a phosphory-
lated versus a nonphosphorylated peptide, we turned to in-
Figure 8. Results from SPE experiments of peptide analytes on P1 and
PN1. Effect of the MeCN/water ratio (+0.1% TFA) on the breakthrough
portion (based on peak area) of A) Ang (squares) and p-Ang (circles)
and B) Ser-436 (squares) and pSer-436 (circles) on P1 (open symbols)
and PN1 (solid symbols) within the first 10 minutes of elution. The pep-
tides (10 mL) were injected as 0.1 mgmLꢀ1 solutions in the mobile phase.
Chem. Eur. J. 2008, 14, 9516 –9529
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9527

B. Sellergren et al.
or absence of the peptides and to provide a rough measure
of their abundance. Unfortunately, distinct peaks were dis-
played only for Ang and pAng, whereas the resolution and
sensitivity was insufficient to clearly identify the remaining
peptides. For instance, the masses of ZAP-70 and Ser-436
and their phosphorylated counterparts were within the in-
strument resolution, thus precluding their separate identifi-
cation. On the other hand, Ser-357, pSer-357, and pThr-295
appeared with very low signal intensities and could not be
identified in a reliable way.[39] To partly circumvent these
problems, additional SPE experiments were performed on
model peptide mixtures of decreased complexity (see Fig-
ures S22 and S23 in the Supporting Information).
Collectively, the SPE results revealed a preferential reten-
tion of pTyr-containing peptides at both spiking levels. Thus,
whereas the non-phosphorylated peptides or those phos-
phorylated at Ser are recovered mainly in the load fractions,
the pTyr peptides pAng and pZAP-70 are selectively re-
tained by P1.
Figure 10. MALDI-TOF MS analysis of fractions collected with a 10-min
interval after separate injection of Ang or pAng on P1 or PN1 using the
loading mobile phase A (Load) and after switching to an eluting mobile
phase B (Elute). Load (A): MeCN/water 95:5 (v/v) 0.1% TFA; elute
(B): MeOH+0.1% TFA. The relative peak intensities were calculated
with reference to the total peak intensity of the mass spectra. SPE
method: 0–10 min 100% A; 10–20 min 100% B. Injection: 10 mL of pep-
tide (0.1 mgmLꢀ1) in the load mobile phase A. Flow rate: 0.5 mLminꢀ1
.
vestigate whether the receptors could discriminate between
pTyr- and pSer-containing peptides. For this purpose, we
performed measurements on the breakthrough fractions of
the Ser-containing reference peptides Ser-436 and pSer-436
in analogy to the experiment performed on Ang and pAng.
The results shown in Figure 8B contrast with the results ob-
tained for the pTyr peptide Ang because no evidence for a
phosphate-related selectivity was seen. Thus, Ser-436 and
pSer-436 behave in a seemingly identical manner on both P1
and PN1, thus generally resulting in somewhat lower recov-
eries at both high and low aqueous contents relative to the
Ang results.
Although encouraging, the results discussed so far are
based on separate injections of the peptide analytes and
may not reflect the ability of the polymer to enrich phos-
phorylated peptides from peptide mixtures. We, therefore,
performed SPE experiments that used two model peptide
mixtures comprising pTyr and non-pTyr peptides present in
weight ratios of 1:1 and 1:100, respectively (Table 7). Direct
MALDI-TOF mass-spectrometric analysis of the load and
elute fractions was performed to demonstrate the presence
Conclusion
The results show that combinations of binding motifs from
host–guest chemistry with conventional imprinting may be
very rewarding. Thus, the stable complexes formed between
the diarylurea host monomers and quaternary ammonium
phosphate salts result in exceptionally tight binding sites
when imprinted. With binding constants for the amino acid
template exceeding K=107 mꢀ1 in a aqueous-rich solvent
(80% water), the pTyr-imprinted polymers compare favora-
bly to pTyr antibodies, which display upper affinities in the
K=106–107 mꢀ1 range.[36] The nature of the binding site could
be deduced from NMR spectroscopic titrations and molecu-
lar-modeling experiments. These data all suggested the for-
mation of tight complexes between two diarylurea mono-
mers and one phosphate group through quadruple hydrogen
bonding possibly aided by a p–p interaction between one of
the monomer ligands and the tyrosine phenyl group.
The sites exhibit sufficient binding energy to bind shorter
peptides containing phosphorylated tyrosine, whereby the
templating induces clear pTyr selectivity. The ability of these
sites to discriminate between
Table 7. Peak intensities of peptides identified by MALDI-TOF mass spectrometric analysis of fractions col-
pSer-
and
pTyr-containing
lected during SPE experiments performed using P1 or PN1.[a]
shorter peptides, along with the
apparently small charge-depen-
dent sequence bias, seems
promising for future applica-
tions of the polymers as robust
and generic pTyr-selective SPE
phases. The approach also ap-
pears suited for the design of
sequence specific phases, for in-
stance, targeting disease bio-
markers or for more advanced
peptide fractionation.
P1
PN1
pY/non-pY Peptide[b]
Ang
Load 1 Load 2 Elute 1 Elute 2 Load 1 Load 2 Elute 1 Elute 2
2325
893
0
121
0
0
31
0
0
36
0
0
1018
0
0
25
0
1751
5525
1009
919
55
512
0
317
0
0
63
0
0
81
0
0
0
0
0
0
0
1:1
p Ang
Ser-436/ZAP-70[c] 1118
Ang
1632
66
0
0
0
0
1:100
p Ang
38
0
Ser-436/ZAP-70[c] 1065
[a] The fractions were collected at 5-min intervals after injection (10 mL) of a model peptide mixture on P1 or
PN1 using a loading mobile phase A (Load) and afterward switching to an eluting mobile phase B (Elute).
Mobile phase: 1–10 min: A=MeCN/water 95:5 (v/v) 0.1% TFA; 10–20 min: B=MeOH (0.1% TFA). [b] The
peptide mixture consisted of nine peptides each at a concentration of 11 mgmLꢀ1 (pY/non-pY=1:1) or all at
11 mgmLꢀ1 except for pAng and pZAP-70, which were present at a concentration of 0.11 mgmLꢀ1 in water
(pY/non-pY=1:100). [c] Nonresolved peak assigned to Ser-436 and ZAP-70.
9528
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 9516 –9529

Phosphopeptide Recognition by an Imprinted Polymer Receptor
FULL PAPER
[20] G. Guiochon, S. Golshan-Shirazi, A. Katti, Fundamentals of Prepa-
rative and Nonlinear Chromatography, Academic Press, New York,
1994.
[21] Y. Chen, M. Kele, I. Quinones, B. Sellergren, G. Guiochon, J. Chro-
matogr. 2001, 927, 1–17.
[22] E. Fan, S. A. Van Arman, S. Kincaid, A. D. Hamilton, J. Am. Chem.
Soc. 1993, 115, 369–370.
[23] D. Esteban Gomez, L. Fabbrizzi, M. Licchelli, E. Monzani, Org.
Biomol. Chem. 2005, 3, 1495–1500.
Acknowledgements
This project was financially supported by the European Community
within the Human Potential Programme under the contract HPRN-CT-
2002–00189 (AquaMIP). The authors are grateful to Prof. Rainer Bis-
choff (University of Groningen, The Netherlands) and Priv. Doz. Dr.
Rainer Lehmann (University Hospital Tübingen, Germany) for valuable
discussion and their generous gifts of peptides. We also wish to thank Dr.
Panagiotis Manesiotis and Prof. Lihua Zhang for experimental advice
and assistance.
[24] M. C. Etter, T. W. Panunto, J. Am. Chem. Soc. 1988, 110, 5896–
5897.
[25] K. A. Connors, Binding Constants. The Measurement of Molecular
Complex Stability, Wiley, New York, 1987.
[26] C. Roussel, M. Roman, F. Andreoli, A. Del Rio, R. Faure, Chirality
2006, 18, 762–771.
[27] P. Bühlmann, S. Nishizawa, K. P. Xiao, Y. Umezawa, Tetrahedron
1997, 53, 1647–1654.
[28] The differences between the tendency of the monomer to complex
the phosphate probe became more obvious when deprotonation was
attempted in situ with triethylamine as the base; on using this
weaker base, ion pairing competes with urea complexation and
leads to significantly weaker complexes with poorly defined stoi-
chiometries. The relative stabilities of the monomer/phosphate com-
plexes agreed with the slopes of the isotherms and the maximum
CIS values obtained (see Figures S8–S11 in the Supporting Informa-
tion).
[1] S. A. Johnson, T. Hunter, Nat. Methods 2005, 2, 17–24.
[2] B. Bodenmiller, L. N. Mueller, M. Mueller, B. Domon, R. Aeber-
sold, Nat. Methods 2006, 4, 231–237.
[3] M. R. Larsen, T. E. Thingholm, O. N. Jensen, P. Roepsdorff, T. J. D.
Jçrgensen, Molecules Molecular and Cellular Proteomics 2005, 4.7,
873–886.
[4] J. Rush, A. Moritz, K. A. Lee, A. Guo, V. L. Goss, E. J. Spek, H. X.
Zhang, X.-M. Zha, R. D. Polakiewicz, M. J. Comb, Nat. Biotechnol.
2005, 23, 94–101.
[5] T. Miliotis, P.-O. Ericsson, G. Marko-Varga, R. Svensson, J. Nilsson,
T. Laurell, R. Bischoff, J. Chromatogr. B 2001, 752, 323–334.
[6] C. Alexander, H. S. Andersson, L. I. Andersson, R. J. Ansell, N.
Kirsch, I. A. Nicholls, J. OꢁMahony, M. J. Whitcombe, J. Mol. Recog-
nit. 2006, 19, 106–180.
[7] Molecularly Imprinted Polymers, Man-made Mimics of Antibodies
and Their Applications in Analytical Chemistry (Ed.: B. Sellergren),
Vol. 23, Elsevier Science B.V., Amsterdam, 2001.
[8] N. W. Turner, C. W. Jeans, K. R. Brain, C. J. Allender, V. Hlady,
D. W. Britt, Biotechnol. Prog. 2006, 22, 1474–1489.
[9] A. Rachkov, N. Minoura, Biochim. Biophys. Acta Protein Struct.
Mol. Enzymol. 2001, 1544, 255–266.
[10] M. M. Titirici, A. J. Hall, B. Sellergren, Chem. Mater. 2003, 15, 822–
824.
[11] H. Nishino, C.-S. Huang, K. J. Shea, Angew. Chem. 2006, 118, 2452–
2456; Angew. Chem. Int. Ed. 2006, 45, 2392–2396.
[29] G. Wulff, Angew. Chem. 1995, 107, 1958–1979; Angew. Chem. Int.
Ed. Engl. 1995, 34, 1812–1832.
[30] B. Sellergren, J. Chromatogr. A 1994, 673, 133–141.
[31] B. Sellergren, K. J. Shea, J. Chromatogr. 1993, 635, 31.
[32] This behavior agrees with the results obtained for polymers pre-
pared in the presence of good solvents (see ref. [31]).
[33] J. Courtois, G. Fischer, B. Sellergren, K. Irgum, J. Chromatogr. A
2006, 1109, 92–99.
[34] P. Sajonz, M. Kele, G. Zhong, B. Sellergren, G. Guiochon, J. Chro-
matogr. 1998, 810, 1–17.
[35] K. D. Shimizu in Molecularly Imprinted Materials: Science and Tech-
nology (Eds.: M. Yan, O. Ramstrçm) 2005, pp. 419–434.
[36] S. Ruff-Jamison, R. Campos-Gonzalez, J. R. J. Glenney, J. Biol.
Chem. 1991, 266, 6607–6613.
[12] P. Manesiotis, A. J. Hall, M. Emgenbroich, M. Quaglia, E. de Loren-
zi, B. Sellergren, Chem. Commun. 2004, 2278–2279.
[13] J. L. Urraca, A. J. Hall, M. C. Moreno-Bondi, B. Sellergren, Angew.
Chem. 2006, 118, 5282–5285; Angew. Chem. Int. Ed. 2006, 45, 5158–
5161.
[37] D. Guo, C. T. Mant, R. S. Hodges, J. Chromatogr. 1987, 386, 205–
222.
[38] The conditioning solvent and the time of column conditioning
strongly influenced the retentivity of the pTyr peptides; thus, insuffi-
cient conditioning led to breakthrough of the peptides and low re-
covery in the elution step, whereas extensive conditioning caused a
strong retentitivity (for example, see Figure 9).
[39] The serine-phosphorylated peptides exhibited limited stability in the
TFA-modified eluents and partial decomposition of the samples
may have occured prior to analysis.
[14] A. J. Hall, P. Manesiotis, M. Emgenbroich, M. Quaglia, E. De Lor-
enzi, B. Sellergren, J. Org. Chem. 2005, 70, 1732–1736.
[15] A. Ojida, Y. Mito-oka, K. Sada, I. Hamachi, J. Am. Chem. Soc.
2004, 126, 2454–2463.
[16] C. Krog-Jensen, Lett. Peptide Sci. 1999, 6, 193–197.
[17] J. Courtois, M. Szumski, E. Bystrçm, A. Iwasiewicz, A. Shchukarev,
K. Irgum, J. Sep. Sci. 2006, 29, 14–24.
[18] C. Viklund, K. Irgum, Macromolecules 2000, 33, 2539–2544.
[19] R. Z. Sommer, H. I. Lipp, L. L. Jackson, J. Org. Chem. 1971, 36,
824–828.
Received: May 30, 2008
Published online: October 10, 2008
Chem. Eur. J. 2008, 14, 9516 –9529
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9529